Fire protection control apparatus for selectively controlling firefighting facilities in road tunnel according to fire detection location and control method thereof

Abstract

One embodiment of the present invention relates to a fire control system and a control method thereof that selectively controls road tunnel fire fighting equipment according to a fire detection location, intelligently selecting and operating related fire fighting equipment by reflecting the fire detection location and the environmental characteristics within the tunnel.

Description

Technology Field

[0001] The following embodiments relate to a fire control system and a control method thereof that selectively controls road tunnel fire fighting equipment according to a fire detection location, intelligently selecting and operating related fire fighting equipment by reflecting the fire detection location and the environmental characteristics within the tunnel. Background Technology

[0002] Due to the enclosed linear structure of road tunnels, heat and smoke spread rapidly in the event of a fire, posing a very high risk of large-scale casualties. To prevent this, various firefighting and disaster prevention facilities, such as jet fans, fire hydrants, and evacuation guidance lights, are installed throughout the entire tunnel.

[0003] Conventional tunnel fire control systems have adopted a simple method of activating all facilities collectively within the entire tunnel or a specific zone upon fire detection. However, in actual fire situations, the diffusion range of toxic gases and the zone of heat influence become highly uneven depending on the airflow direction, longitudinal gradient, and fire intensity inside the tunnel.

[0004] If facilities are controlled collectively without considering these physical characteristics, unnecessary equipment located outside the fire's impact zone is activated, leading not only to the waste of electricity and firefighting water but also to the problem of insufficient resources being allocated to critical areas requiring intensive control. Furthermore, operating jet fans in inappropriate locations can have the counterproductive effect of pushing smoke toward evacuees. Therefore, there is an urgent need for precise control technology that reflects the dynamic spread characteristics of a fire in real time and selectively activates only the optimal group of facilities. Prior art literature

[0005] (Patent Document 0001) KR 10-2926502 B (Patent Document 0002) KR 10-2671001 B (Patent Document 0003) KR 10-2900602 B The problem to be solved

[0006] The problem that an embodiment of the present invention aims to solve is to provide a fire control system and a control method thereof that selectively controls road tunnel fire fighting equipment according to fire detection locations, by modeling an asymmetric dynamic spatial influence field reflecting real-time airflow components and tunnel inclination and calculating spatial correlation for each facility to overcome the limitations of resource waste and inefficient response caused by the conventional batch control method described above. means of solving the problem

[0007] According to one embodiment, the method comprises: a step of collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; a step of generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; a step of mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; a step of generating a fire control package including a control command for each of the group of equipment to be controlled based on the group of equipment to be controlled and target operating parameters for each equipment; and a step of transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of determining the group of equipment to be controlled and target operating parameters for each equipment includes: a step of loading a previously stored equipment layout table corresponding to the fire detection location and the section identifier; Based on the above-mentioned equipment layout table and fire detection location, the method may include: a step of filtering candidate equipment among firefighting equipment according to the equipment separation distance between the fire detection location and the installation location information of the firefighting equipment, whether they are in the same section, and whether they are in adjacent sections, and determining the candidate equipment as a group of equipment to be controlled; and for each firefighting equipment included in the group of equipment to be controlled, a step of querying target operating parameters corresponding to a section identifier and a fire intensity index from the equipment control profile, and performing unit conversion and range restriction so that the target operating parameters conform to the parameter schema for each equipment type to determine the target operating parameters for each equipment.

[0008] Additionally, the detection sensors include: a smoke sensor, a gas sensor, a temperature sensor, a flame detection sensor, a camera, and an airflow measurement sensor, and the step of determining the candidate equipment as a group of control target equipment comprises: a step of calculating a tunnel longitudinal airflow velocity component by projecting the airflow velocity vector included in the sensor data onto the longitudinal unit vector of the road tunnel, and defining upstream and downstream directions according to the sign of the airflow velocity component; a step of generating a set of deflection parameters reflecting the deflected diffusion of fire heat and combustion gases based on the longitudinal gradient angle of the tunnel longitudinal profile table corresponding to the section identifier mapped to the fire detection location and the tunnel longitudinal airflow velocity component; a step of setting the fire detection location as the center point of a three-dimensional spatial coordinate system and modeling a dynamic spatial influence field in the form of an asymmetric ellipsoid in which the major axis expands in the downstream direction and the major axis contracts in the upstream direction based on the set of deflection parameters; a step of classifying the candidate equipment filtered from the equipment placement table into fire extinguishing equipment, smoke exhaust equipment, and evacuation guidance equipment according to the equipment type, and mapping the three-dimensional installation coordinates of each classified candidate equipment to the dynamic spatial influence field. For each of the above candidate facilities, a step of calculating a spatial association index by applying a distance weight inversely proportional to the distance from the fire detection location and a type weight corresponding to the operational purpose for each facility type; and a step of determining only the candidate facilities whose spatial association index exceeds a preset activation threshold value as the group of facilities to be controlled.

[0009] And, the step of calculating the spatial correlation index comprises: for each of the candidate facilities, a step of calculating the longitudinal distance component and the lateral distance component between the 3D coordinates of the fire detection location and the 3D installation coordinates of the candidate facility; a step of determining whether the candidate facility is positioned upstream or downstream relative to the fire detection location based on the upstream and downstream directions, and selecting an upstream attenuation coefficient or a downstream amplification coefficient according to the determination result; a step of determining an upward / downward slope correction coefficient according to the longitudinal gradient based on the longitudinal gradient angle, and correcting the slope correction coefficient based on the difference in elevation between the installation height of the candidate facility and the fire detection location; and a step of calculating the effective separation distance for each candidate facility based on the longitudinal distance component, lateral distance component, and slope correction coefficient, and the upstream attenuation coefficient or downstream amplification coefficient. The method may include the step of calculating a distance weight inversely proportional to the effective separation distance, wherein if the effective separation distance is less than a preset minimum distance, the minimum distance is applied as a lower limit value, and if the effective separation distance exceeds a preset maximum distance, the distance weight is set to 0; the step of querying a type weight corresponding to each of the fire extinguishing equipment, smoke exhaust equipment, and evacuation guidance equipment from a type weight table preset to correspond to the operating purpose of each equipment type; and the step of calculating a spatial association index for each candidate equipment normalized to a value between 0 and 1 by multiplying the distance weight and the type weight.

[0010] In addition, the method comprises the steps of: collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each piece of equipment; generating a fire control package including control commands for each of the group of equipment to be controlled and target operating parameters for each piece of equipment based on the group of equipment to be controlled and target operating parameters for each piece of equipment; and transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of generating the fire control package includes: loading an evacuation path graph that defines evacuation nodes including evacuable exits, evacuation connecting passages, and emergency exits within the road tunnel, and the connection relationships between the evacuation nodes based on the fire detection location, fire intensity index, and section identifier; Based on the above evacuation path graph and fire detection location, the method may include the step of updating the evacuation path graph to exclude risk nodes and risk links included in the fire impact zone and the smoke spread expected zone, and calculating an optimal evacuation path for each starting node in the updated evacuation path graph; and based on the optimal evacuation path, the method may include the step of generating a visual guidance control command including an evacuation direction arrow, evacuation distance, exit identification information, and a warning message on a display unit included in the evacuation guidance device.

[0011] Additionally, the detection sensor comprises: a camera, a smoke sensor, a gas sensor, and an airflow measurement sensor, and the step of calculating the optimal evacuation path for each starting node comprises: a step of calculating the Available Safe Egress Time (ASET) for each node and link, defined as the time until the toxic gas and smoke layer spreading from the fire detection location reaches each evacuation node and connecting link on the evacuation path graph to create a survival limit environment, based on time-series smoke concentration data included in the sensor data and longitudinal airflow velocity components within the road tunnel; a step of decelerating and correcting the average walking speed of evacuees based on section-by-section evacuee crowd density data extracted from images captured by the camera and a preset section-by-section longitudinal gradient angle of the tunnel floor, and calculating the Required Safe Egress Time (RSET) for each node and link, which is the cumulative required evacuation time required to pass through each connecting link from the starting node based on the decelerating and corrected average walking speed. A step of updating the evacuation path graph such that, for each individual connection link of the evacuation path graph, if it is determined that the required evacuation time (RSET) for a specific starting node exceeds the limit arrival time (ASET), the corresponding connection link is determined to be a dangerous link, and the traffic weight of the dangerous link is set to a preset value of infinity to be excluded from graph search; and a step of, for valid connection links where the required evacuation time (RSET) is within the limit arrival time (ASET), calculating a bottleneck penalty index based on the crowd density data and performing a dynamic weight update to reflect the bottleneck penalty index in the traffic weight of the valid connection links.and, for the updated evacuation path graph reflecting the above bottleneck penalty index, calculate the expected inflow traffic for each evacuation path node based on the physical maximum capacity pre-set for each evacuation path node, and if there is an evacuation path node where the inflow traffic exceeds the physical maximum capacity, apply a load balancing-based time-series routing algorithm to detour the optimal evacuation path of the evacuee corresponding to the lower-priority departure node to the next-priority evacuation path node or the tunnel exit in the upstream direction of the airflow to finally determine the optimal evacuation path; may be included.;

[0012] And, the step of calculating the above Limit Attainment Time (ASET) by node and link comprises: a step of calculating the longitudinal movement speed of a ceiling jet spreading along the tunnel ceiling from a fire detection location based on the longitudinal airflow velocity component within the road tunnel included in the sensor data and the fire intensity index; a step of calculating the longitudinal distance component between the fire detection location and the upper ceiling coordinates of the evacuation node or connecting link for each evacuation node and connecting link included in the evacuation path graph, and predicting the ceiling arrival time, when smoke first reaches the upper ceiling of each evacuation node and connecting link, by evacuation node and connecting link based on the longitudinal distance component and the longitudinal movement speed; a step of calculating the volume expansion rate index and mass generation rate index of soot particles based on the slope and integral value in the concentration increase section for the time-series smoke concentration data; and a step of querying the ceiling height, reference elevation, effective cross-sectional area, and longitudinal gradient angle of the corresponding tunnel section from a previously stored tunnel cross-sectional profile table based on the section identifier mapped to the fire detection location. A step of calculating, by node and link, the smoke layer descent time required for the smoke layer reaching the ceiling to descend vertically to a preset reference altitude based on the above-mentioned volumetric expansion rate index, mass generation rate index, effective cross-sectional area, and longitudinal gradient angle; a step of calculating, by node and link, the toxic gas saturation time at which the evacuee's ability to function is determined to be lost by applying a Fractional Effective Dose model based on the component analysis results of the toxic gas concentration data included in the sensor data; and a step of calculating, by node and link, the limit visibility loss time, defined as the point at which visibility decreases below a preset limit visibility threshold, by applying a preset light attenuation coefficient to the soot concentration calculated from the time-series smoke concentration data.and may include the step of deriving the physical smoke settlement time by summing the ceiling arrival time and the smoke layer descent time, comparing the physical smoke settlement time, the toxic gas saturation time, and the limit visibility loss time to determine the shortest time as the Limit Attainment Time (ASET) as the point in time for establishing the survival limit environment of the corresponding evacuation node and connecting link, and mapping the determined Limit Attainment Time (ASET) to the nodes and links of the evacuation path graph. Effects of the invention

[0013] According to one embodiment, candidate equipment is filtered primarily based on fire detection location and equipment layout information, and target operating parameters for each piece of equipment are converted and finalized according to a schema, thereby enabling precise control specialized for the fire location.

[0014] Furthermore, by modeling an 'asymmetric ellipsoidal dynamic spatial influence field' that fuses longitudinal airflow velocity and longitudinal gradient within the tunnel to reflect the deflected diffusion of fire heat and combustion gases, it is possible to select a group of facilities optimized for the actual risk range. This involves broadly including downstream facilities where smoke is pushed by the airflow while efficiently limiting upstream facilities. By calculating a spatial correlation index based on the purpose of each facility type—such as firefighting, smoke exhaust, and evacuation guidance—it provides the effect of rationally determining control priorities.

[0015] In addition, longitudinal and lateral distance components are derived based on the 3D coordinates of candidate facilities, and the effective separation distance is precisely calculated by applying upward / downward correction coefficients based on tunnel inclination and amplification / attenuation coefficients based on airflow direction. Through this, facilities located at points with high fire impact can be reliably activated even if the actual physical distance is far, and the success rate of fire suppression and evacuation guidance can be increased by strategically concentrating firefighting resources through a normalized spatial correlation index.

[0016] In addition, by modeling the evacuation route within the tunnel as a graph structure and excluding risk factors in fire impact and smoke spread zones in real time, it is possible to calculate an optimal evacuation route isolated from flames and smoke and intuitively guide evacuees through visual devices.

[0017] In addition, by utilizing camera and various sensor data to precisely calculate the Attempted Time to Set (ASET) and Required Time to Set (RSET) for each node and link, it is possible to select only valid paths through which evacuees can safely pass before reaching the danger zone. In particular, by reflecting bottleneck penalties based on crowd density in the path weights and applying a load balancing algorithm that considers the maximum capacity of evacuation facilities, it prevents congestion toward specific exits and provides the effect of maximizing the evacuation efficiency of the entire tunnel.

[0018] Furthermore, by analyzing the velocity of the ceiling jet stream to predict the time it takes for smoke to reach the ceiling, and by deriving the smoke layer descent time by considering the volumetric expansion rate of soot particles and the tunnel cross-sectional area, the physical point of smoke settlement can be predicted with extreme accuracy. In addition, by comparing the toxic gas saturation time (FED model) with the time of loss of critical visibility to determine the shortest time as the survival limit, there is a safety assurance effect that allows evacuees to be guided to a safe path at the optimal timing before their operational capabilities are lost. Brief explanation of the drawing

[0019] FIGS. 1 and 2 are schematic diagrams illustrating a fire control system that selectively controls road tunnel fire fighting equipment according to a fire detection location according to an embodiment of the present invention. FIG. 3 is a flowchart illustrating a fire control method for selectively controlling road tunnel firefighting equipment according to a fire detection location according to an embodiment of the present invention. FIG. 4 is a flowchart showing the steps of determining the group of equipment to be controlled and the target operating parameters for each piece of equipment in a fire control method for selectively controlling road tunnel fire fighting equipment according to a fire detection location according to an embodiment of the present invention. Specific details for implementing the invention

[0020] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0021] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0022] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0023] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or coupled with that other component, or that there may be other components in between.

[0024] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0025] In particular, where a 'step' in this specification is described as 'comprising' one or more detailed steps or sub-steps, said 'step' may be interpreted as including its own basic processing step while simultaneously performing the described detailed steps as well.

[0026] For example, if it is stated that 'a step of doing B to A' includes 'a step of doing D to C; a step of doing F to E; and a step of doing H to G,' the 'step of doing B to A' may be interpreted not merely as the basic operation of doing B to A, but as a configuration that performs detailed procedures together, such as a step of doing D to C, a step of doing F to E, and a step of doing H to G.

[0027] Accordingly, the above configuration does not exclude various sub-procedures included within the scope of execution of the corresponding step, and may be included within the scope of the present invention even if other procedures or means performing substantially the same or equivalent functions are substituted.

[0028] Expressions such as 'end part', 'both ends', 'one end', 'other end', and 'side end' of a component can be interpreted as referring to at least / any one of the end parts of that component.

[0029] In the description of the present invention, 'a method in which a device comprises a processor, a memory, a communication module, and a non-transient storage medium, and a program stored in the non-transient storage medium is executed by the processor,' the term 'method' may be interpreted as referring to the program stored in the non-transient storage medium itself or a part of the program.

[0030] The term 'Return' as used in the description of the present invention may refer to a result value being output, returned, or returned from a method, procedure, function, etc. used in a given program language / structure.

[0031] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application. For example, the term 'artificial intelligence model' may be selected from one or more of known general artificial intelligence models.

[0032] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0033] According to one embodiment, the method comprises: collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; generating a fire control package including a control command for each of the group of equipment to be controlled and target operating parameters for each equipment based on the group of equipment to be controlled and target operating parameters for each equipment; and transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of generating the fire control package includes: identifying whether the spraying of a fire extinguishing fluid by the fire extinguishing equipment has ended based on the operating status data of the fire extinguishing equipment among the firefighting equipment, and generating a drainage start condition that initiates automatic drainage control according to the result of identifying whether the spraying has ended. The present invention provides a method for controlling an automatic drainage device for discharging residual firefighting fluid after use of a road tunnel firefighting facility, comprising the step of generating a drainage control command including control parameters for each of a drainage valve, a drainage pump, and a drainage line switching valve included in the drainage facility of the road tunnel when the above drainage initiation condition is satisfied.

[0034] The step of collecting time-synchronized sensor data and equipment status data is to collect raw data streams including sensor-specific measurements and equipment-specific operating statuses from multiple detection sensors and multiple firefighting equipment installed in a road tunnel, and to generate a data set aligned with a common time standard to eliminate time axis mismatches caused by different sampling periods and communication delays.

[0035] Here, detection sensors are devices that detect physical and chemical changes within a tunnel and may include smoke sensors, temperature sensors, flame detection sensors, cameras, and airflow measurement sensors, and sensors included in / equipped with firefighting equipment (flow sensors, pressure sensors, etc.) may also be referred to / classified / interpreted / defined as being included in 'detection sensors'.

[0036] Firefighting equipment is equipment installed for fire suppression and disaster prevention, and may include at least one of fire extinguishing equipment, drainage equipment, ventilation equipment, and evacuation guidance equipment. Additionally, time-synchronized sensor data and equipment status data refer to data in which the collection time is normalized based on a timestamp per sensor or a status update time reported by an equipment controller, and corrected to be aligned with the server's reference time or a common time axis.

[0037] The step of collecting time-synchronized sensor data and equipment status data can be performed to suppress misjudgments caused by time errors between sensors in the subsequent fire event information generation step by normalizing the collection time based on the timestamps for each sensor included in the sensor data, identifying sensor channels where timestamps are missing or drift exceeds a threshold, marking such channels as missing channels, or adopting only interpolable intervals as valid intervals. In addition, at least one of the valve open / closed status, pump operating status, flow rate and pressure, and spray pattern setting value of the fire extinguishing equipment, the valve open / closed status, pump operating status of the drainage equipment, the position of the drainage line switching valve, the power cut-off status, and the fault flag included in the equipment status data can be aligned so that they are loaded on the same time axis.

[0038] The step of generating fire event information is to determine whether a fire event has occurred based on the time-synchronized sensor data and equipment status data, and to generate structured fire event information by calculating the fire detection location and fire intensity index.

[0039] Here, the occurrence of a fire event may be a determination result calculated by combining at least one of exceeding a sensor threshold, spatial distribution, temporal duration, and camera analysis results, and may be configured to exclude events corresponding to false alarms. Additionally, the fire detection location may be expressed by longitudinal coordinates of the tunnel or a predefined tunnel zone number, and the fire intensity index may be a single scalar index or grade value calculated by combining at least one of the temperature rise rate, smoke concentration rise rate, gas concentration change amount, flame detection duration, flame occupancy area, or required level of fire extinguishing fluid injection flow rate.

[0040] The step of generating fire event information may involve performing unit normalization, noise removal, missing value correction, and outlier clipping for each sensor type, calculating the probability of fire occurrence based on the simultaneous rise pattern and spatial distribution of multiple sensors, and determining that a fire event has occurred if the probability of fire occurrence is greater than or equal to a preset judgment threshold. Additionally, the fire detection location may be determined by referring to a sensor placement map containing sensor installation coordinates and tunnel changeage information to set the installation location of the maximum response sensor as a primary candidate location, and then applying a correction value based on the response intensity distribution of adjacent sensors.

[0041] The step of mapping fire detection locations to section identifiers of road tunnels and querying previously stored equipment control profiles corresponding to section identifiers and fire intensity indicators to determine the group of equipment to be controlled and the target operating parameters for each piece of equipment is a step of mapping the fire detection locations to section identifiers at the tunnel section level and querying equipment control profiles using section identifiers and fire intensity indicators as keys to determine the set of equipment requiring automatic control and the target parameters for each piece of equipment.

[0042] Here, the section identifier may be a unique number assigned to manage the entire length of the tunnel by dividing it into units of a certain distance, and the equipment control profile may be a control logic table in which operating equipment and operating levels are predefined according to fire location and intensity.

[0043] The step of mapping fire detection locations to section identifiers of road tunnels and determining target equipment groups and target operating parameters for each piece of equipment may involve defining a section relationship for the section identifier that includes the fire detection section and adjacent sections, limiting priority candidates to only those installed in the fire detection section and adjacent sections using an equipment placement table, and then determining the target equipment group by applying priority rules for each equipment type indicated by the equipment control profile. Additionally, when determining target operating parameters for each piece of equipment, unit conversion and range restriction may be performed by referring to the parameter schema for each equipment type; for example, pump rotation speed may be limited to within the inverter allowable range, valve opening rate may be clipped to a range from 0 to 100, and drainage line switching valves may be mapped to one of the discrete states of isolation, bypass, or reservoir transfer.

[0044] The step of generating a fire control package involves generating a package structure including control commands and command application sequences for each facility based on the control target facility group and target operating parameters for each facility, and can be converted into a data set in the form of a communication protocol that each facility controller can interpret. In addition, metadata including precedence dependencies between commands, interlock conditions, number of retries, timeouts, and whether an acknowledgment is required can be configured to prevent facility conflicts caused by indiscriminate simultaneous operation. In one embodiment of the present invention, a drainage control command for discharging residual fire extinguishing fluid after use of the fire extinguishing facility can be configured to be included in the fire control package.

[0045] The step of automatically controlling the group of equipment to be controlled by transmitting a fire control package to equipment controllers involves transmitting the generated fire control package to equipment controllers at the tunnel site via a communication network and performing closed-loop control to execute control commands included in the package according to execution acknowledgments and feedback status received from the equipment controllers. At this time, the package identifier, creation time, application section identifier, fire intensity indicator, and command sequence number are included to prevent duplicate execution, and for equipment that does not receive an acknowledgment within a certain period, fault-tolerant operations such as retransmission or reduced package execution may be performed.

[0046] The step of identifying whether the spraying of a fire extinguishing fluid of a fire extinguishing facility has ended, and generating a drainage start condition for initiating automatic drainage control according to the identification result of whether the spray has ended, is a step of determining a state in which the spray can be considered to have ended by referring to the operating state of the fire extinguishing facility included in the facility status data, and generating a drainage start condition to initiate drainage control only when such determination is established. For example, the spray end can be determined by combining at least one of whether the spray valve of the fire extinguishing facility has switched to a closed state, whether the spray flow rate is maintained for a continuous preset time or longer below a preset end flow rate threshold, or whether the pump drive signal is in a stopped state, and hysteresis may be applied by setting the end threshold and the resumption threshold differently from each other. In addition, the drainage start condition may be configured by further combining safety conditions including at least one of the availability state of the drainage facility, the initial position of the drainage line switching valve, whether the sump water level is above the minimum operating water level, and whether there is no overcurrent or overheating alarm of the drainage pump.

[0047] The step of generating a drainage control command when the drainage initiation condition is satisfied is to calculate control parameters for each of the drainage valve, drainage pump, and drainage line switching valve included in the drainage facility of the road tunnel, and to generate a drainage control command including the calculated control parameters.

[0048] Here, the drain line switching valve can be controlled to select one of the following paths based on the contamination level of the discharged fluid: general sewer network outflow, bypass outflow, or emergency storage tank isolation transfer. The drain control command may include a target rotational speed or target discharge pressure of the drain pump, an operating duration, a target value for the drain valve opening rate and opening / closing sequence, and a switching valve position control signal. Additionally, to suppress bottlenecks during the discharge process, pulsation control parameters may be generated so that the opening actuator of the drain valve repeatedly opens and closes at a preset cycle. The pulsation control parameters may include at least one of a repetition cycle, an open holding time, a closed holding time, the number of repetitions, or a water level-based termination condition.

[0049] Additionally, the detection sensor comprises: a multi-channel water quality sensor that collects hydrogen ion concentration (pH), electrical conductivity data, and fluid temperature data for residual fire extinguishing fluid flowing into a primary sump located below the road surface of the road tunnel; an ultrasonic reflectance sensor that collects ultrasonic reflectance data including reflection intensity, reflection time delay, and spectral characteristics of the reflected signal based on the reflection signal of the ultrasonic transmitted into the main drainage line connected to the primary sump and received by reflecting from the inner wall of the main drainage line, the fluid-air boundary, and suspended solids within the fluid; and a differential pressure sensor that collects differential pressure data including the pressure difference (ΔP) between the upstream pressure and the downstream pressure by measuring the upstream pressure and the downstream pressure of a predetermined section of the main drainage line, respectively; and the step of generating the drainage control command comprises: a step of calculating a contamination profile to evaluate whether hazardous chemicals are contained in the residual fire extinguishing fluid and the possibility of thermal runaway fire originating from the electric vehicle battery. A step of evaluating the risk of pipe clogging of the main drainage line by molten vehicle parts, tire debris, or fire debris based on the measurements of the ultrasonic reflectance sensor and differential pressure sensor; and a step of, if the analysis result of the contamination profile determines that the residual extinguishing fluid exceeds a preset toxicity threshold or the risk of pipe clogging is above a reference value, generating isolation control signals and bypass control signals for the drainage line switching valve to block outflow to external natural water systems and general sewer networks, generating an operation control signal to operate a special drainage pump combined with a macerator for physically crushing solid debris, generating a transfer path control signal to forcibly transfer the contaminated fluid to a sealed emergency storage tank connected to the switching valve, and generating pulsation control parameters to repeatedly open and close the opening actuator of the drainage valve at a preset cycle to prevent bottlenecks during the discharge process;The drainage control command may include: an isolation control signal, a bypass control signal, an operation control signal, a transfer path control signal, and pulsation control parameters linked to the contamination profile and the pipeline blockage risk.;

[0050] The above-mentioned sensing sensors include a multi-channel water quality sensor, an ultrasonic reflectance sensor, and a differential pressure sensor, and sensor data collected from them can be used as input to generate a drainage control command.

[0051] Here, the multi-channel water quality sensor can be installed to come into contact with the residual firefighting fluid flowing into the primary sump located beneath the road surface of the road tunnel, and can be configured to simultaneously measure the hydrogen ion concentration (pH), electrical conductivity (EC), and fluid temperature of the residual firefighting fluid in multiple channels. For example, the multi-channel water quality sensor can be configured to integrate the pH electrode, conductivity electrode, and temperature sensor into a single housing, or to place them separately at multiple depths to measure even the vertical distribution.

[0052] pH and electrical conductivity data collected by multi-channel water quality sensors can be utilized as basic indicators to estimate the type of fire extinguishing agent used for fire suppression or changes in the concentration of ionic substances leached by the fire. In particular, since patterns of rapid pH fluctuations or abnormal increases in electrical conductivity may appear when electrolytes or toxic components are mixed into the fluid during thermal runaway in electric vehicle batteries, the aforementioned pH and electrical conductivity data can be used as indicators to assess the likelihood of a fire originating from thermal runaway. Furthermore, since fluid temperature data reflects thermal effects caused by residual fire heat or the incorporation of high-temperature debris, it can be used for control purposes that consider the heat resistance limits of drainage facilities or pump protection logic; when combined with pH and electrical conductivity data from the same time period, it can enhance the reliability of contamination assessment.

[0053] The ultrasonic reflectance sensor can collect ultrasonic reflectance data by transmitting an ultrasonic signal into the main drainage line connected to the primary sump and analyzing the reflected signal received from the inner wall of the main drainage line, the fluid-air boundary, or floating solids in the fluid.

[0054] Here, ultrasonic reflectance data can be configured to include reflection intensity, reflection time delay (round-trip propagation time), and spectral features of the reflected signal. For example, reflection intensity may reflect the density of suspended solids or the possibility of sediment presence, reflection time delay may indicate changes in the position of the reflection boundary due to changes in the height of sediment within the pipe or the effective cross-section, and spectral features may be used to classify the particle size or material properties of foreign substances based on scattering characteristics.

[0055] A differential pressure sensor can generate differential pressure data including the pressure difference (ΔP) between the upstream and downstream pressures by measuring the upstream and downstream pressures, respectively, in a designated section of the main drainage line. This differential pressure data represents the degree of pressure loss of the fluid passing through the drainage pipe and can be used as an indicator to suggest the progression of physical blockage within the pipe when the pressure difference increases despite the flow rate remaining constant or similar. Therefore, by cross-referencing ultrasonic reflectance data with differential pressure data, the risk of pipe clogging can be assessed more reliably by simultaneously reflecting increases in suspended matter and sediment, as well as increases in flow resistance.

[0056] The step of generating the drainage control command described above involves determining the outflow path of the drainage line based on the contamination level of the residual digestion fluid and the risk of pipeline blockage, and generating a structured drainage control command by calculating valve, pump, and actuator control parameters linked thereto. This step can be subdivided into a step of calculating a contamination profile, a step of evaluating the risk of pipeline blockage, and a step of generating control signals including isolation, bypass, crushing, transfer, and pulsation control according to the contamination level or blockage risk.

[0057] The step of calculating the contamination profile involves analyzing the temporal variation and combination patterns of pH, electrical conductivity, and fluid temperature data collected from multi-channel water quality sensors to generate a result that quantitatively evaluates the presence of hazardous chemicals in the residual extinguishing fluid and the potential for electric vehicle battery thermal runaway fires. For example, the amount of pH change, the amount of electrical conductivity change, and the rate of temperature rise over a specific time interval can be configured as features. These features are then mapped to a pre-established contamination reference table or classification rule to calculate scores for the likelihood of hazardous chemical inclusion and thermal runaway origin. Subsequently, a contamination profile can be generated as a structured data set including toxicity grades, hazardous substance indices, and environmental impacts. Additionally, the contamination profile may be configured to include the sensor identifier used for analysis, the applied reference table version, and the time of creation, enabling traceability within subsequent control logic.

[0058] The step of evaluating the risk of pipe blockage in the main drainage line involves cross-referencing measurements from ultrasonic reflectance sensors and differential pressure sensors to quantify whether molten vehicle parts, tire debris, or fire debris are blocking the drain pipe and the severity thereof. The pipe blockage risk can be defined as a probabilistic or score-based indicator representing the extent to which the drainage system's water flow capacity has deteriorated compared to normal levels, and the system can be configured to activate special control logic for drainage pump protection and discharge stabilization if a risk level exceeding a threshold is calculated.

[0059] If the analysis of the contamination profile determines that the residual digestion fluid exceeds a preset toxicity threshold or that the risk of pipeline blockage is above a standard threshold, isolation control signals and bypass control signals for the drainage line switching valve can be generated to block outflow to external natural water systems and general sewer networks. The isolation control signal may be a signal that drives the switching valve to an isolation position to close the flow path normally connected to the natural drainage route, and the bypass control signal may be a signal that drives the switching valve to a bypass position to change the flow path to a closed emergency storage tank capable of treating contaminants or a separate treatment line. Accordingly, the environmental outflow of toxic fluids can be physically blocked.

[0060] The step of generating an operation control signal to operate a special drainage pump equipped with a macerator for physically crushing solid debris can be performed to prevent large chunks of debris from damaging the pump impeller or piping or causing sudden blockage. The macerator is a crushing device that cuts and grinds solids mixed in the fluid to finely pulverize them, and can be configured to operate in conjunction with the special drainage pump to finely pulverize substances causing blockage in the pipeline, thereby increasing discharge efficiency.

[0061] The step of generating a transfer path control signal to forcibly transfer the contaminated fluid to a closed emergency reservoir connected to a switching valve is a process designed to enable post-purification treatment by temporarily storing the fluid containing toxic substances in a space isolated from the outside. The transfer path control signal may include a command to establish a flow path so that the fluid flows into the intended reservoir in conjunction with the position setting of the switching valve, and may be expressed as a combination of multiple valve states, such as “reservoir connection line open, external discharge line closed, bypass line blocked,” depending on the facility configuration.

[0062] The step of generating pulsation control parameters that repeatedly open and close the opening actuator of a drain valve at a preset cycle to prevent bottlenecks during the discharge process is a control technique intended to mitigate, through hydrodynamic fluctuations, the phenomenon in which drainage is interrupted when solid matter gets stuck at the valve inlet or a constricted section of the pipeline. The pulsation control parameters may include a period (frequency) for periodically changing the valve opening rate, an open holding time, a closed holding time, a number of repetitions, and a termination condition, and may be configured to induce hydraulic fluctuations to induce the re-suspension of stagnant foreign matter and the resolution of local blockages.

[0063] The drainage control command is generated in conjunction with the calculated contamination profile and pipeline blockage risk, and can be configured to transmit isolation control signals, bypass control signals, operation control signals, transfer path control signals, and pulsation control parameters to the drainage facility system by structuring them into a single control package. In this case, the drainage control command can be configured to include a command identifier, application time, applicable facility identifier, timeout, and whether an acknowledgment is required, so that the facility controller can prevent duplicate execution and receive feedback on the execution status.

[0064] And, the step of calculating the contamination profile comprises: generating aligned water quality data by performing time synchronization based on sensor-specific timestamps and noise reduction based on moving averages to align the hydrogen ion concentration (pH), electrical conductivity data, and fluid temperature data according to the same time standard; calculating a water quality change feature vector including a change in pH (ΔpH), a change in electrical conductivity (ΔEC), and a change in temperature (ΔT) based on the aligned water quality data; mapping the change in pH (ΔpH), the change in electrical conductivity (ΔEC), and the change in temperature (ΔT) included in the water quality change feature vector to hazardous chemical indicator sections of a pre-set contamination standard table by fire type, querying reference scores and weights corresponding to each of the change in pH (ΔpH), the change in electrical conductivity (ΔEC), and the change in temperature (ΔT), and calculating a hazardous chemical inclusion probability score normalized between 0 and 1 by weighted summing the queried reference scores and weights; A step of mapping the pH change amount (ΔpH), electrical conductivity change amount (ΔEC), and temperature change amount (ΔT) included in the above water quality change feature vector to the thermal runaway origin indicator section of the above fire type-specific contamination standard table, querying reference scores and weights corresponding to each of the pH change amount (ΔpH), electrical conductivity change amount (ΔEC), and temperature change amount (ΔT), and calculating a thermal runaway fire origin probability score normalized between 0 and 1 based on the queried reference scores and weights; a step of calculating a combination score by applying a non-linear combination equation including linear combination terms and interaction terms to the hazardous chemical inclusion probability score and the thermal runaway origin probability score, and calculating an integrated contamination score normalized between 0 and 1 by applying a sigmoid function to the combination score; a step of determining a contamination grade corresponding to a preset toxicity threshold section based on the integrated contamination score, and generating a contamination profile including the contamination grade, the integrated contamination score, and the water quality change feature vector.and may include the step of generating a structured pollution profile record that can be referenced in the generation step of a drainage control command by adding the profile creation time, version information of the pollution standard table by fire type applied, and the sensor identifier of the multi-channel water quality sensor to the pollution profile above.

[0065] The step of generating aligned water quality data is to generate aligned water quality data by performing time synchronization based on sensor-specific timestamps and noise reduction based on moving averages so that hydrogen ion concentration data, electrical conductivity data, and fluid temperature data collected from multi-channel water quality sensors can be compared at the same time reference even if they have different sampling periods or communication delays.

[0066] Here, time synchronization refers to ensuring the consistency of calculations by aligning the timestamps attached by each water quality sensor during data transmission to a common time axis corresponding to the server's reference time or a preset master clock. This can be performed by resampling using at least one of the following methods: selecting the nearest value, linear interpolation, or segment averaging, in alignment with the grid time of the common time axis. Additionally, the system may be configured to identify channels where sensor-specific timestamps are missing or drift exceeds a preset allowable range and mark them as missing channels, or to exclude segments where drift occurs as invalid segments and include only the remaining valid segments in the alignment target.

[0067] Moving average-based noise reduction refers to a filtering technique that smoothly extracts the trend of water quality changes by applying a moving average or weighted moving average with a preset window length to each channel on a common time axis to reduce transient spikes or electrical interference included in the sensor's measurements. Accordingly, it can be configured to mitigate high-frequency noise and suppress misjudgment of contamination caused by instantaneous spikes.

[0068] The step of calculating a water quality change feature vector is to calculate a water quality change feature vector including a change in hydrogen ion concentration, a change in electrical conductivity, and a change in temperature based on the sorted water quality data.

[0069] Here, a water quality change feature vector may refer to one-dimensional array data having components of the difference between a reference value of a pre-set reference interval—such as a normal state before a fire or just before the drainage start condition is met—and a real-time measurement value at the current time of analysis. For example, the change in hydrogen ion concentration can be defined as the difference between the representative hydrogen ion concentration value of the reference interval and the current hydrogen ion concentration value, and the change in electrical conductivity and the change in temperature can also be defined in the same way.

[0070] The reference interval may be set to at least one of the following: immediately after fire suppression, immediately before the drainage initiation condition is met, or a pre-set past time window; and the reference value may be determined to be at least one of the following: the average value, the median value, or the average value after removing outliers. Accordingly, rather than the absolute water quality value in a static state, the “change relative to the reference” may be configured as a feature quantity to mitigate the influence of sensor offset or absolute value deviation.

[0071] The step of calculating the hazardous chemical inclusion probability score is to map the change in hydrogen ion concentration, the change in electrical conductivity, and the change in temperature included in the water quality change feature vector to the hazardous chemical indicator section of a pre-set pollution level standard table for each fire type, respectively, to look up the standard score and weight corresponding to each change amount, and then calculate the hazardous chemical inclusion probability score normalized to between 0 and 1 by weighted summing the looked-up standard score and weight.

[0072] Here, the hazardous chemical indicator range may refer to an area defined by segmenting the range of water quality changes that suggest the possibility of contamination with toxic chemical components exceeding the level of general fire debris.

[0073] The contamination standard table by fire type may be configured to define intervals for hydrogen ion concentration change, electrical conductivity change, and temperature change for each fire type, and to store a standard score and weight for each interval. The weighted summation may be performed by summing the values ​​obtained by multiplying the standard score by the weight for each change, and by applying lower and upper limits to the calculated sum to normalize it to between 0 and 1, it may be configured to be expressed as a standardized value ranging from 0, corresponding to no contamination, to 1, corresponding to a state with a very high probability of containing highly toxic substances.

[0074] The step of calculating the probability score of origin of a thermal runaway fire involves mapping the change in hydrogen ion concentration, the change in electrical conductivity, and the change in temperature included in the water quality change feature vector to the thermal runaway origin indicator section of the pollution level standard table for each fire type, respectively, querying the standard score and weight corresponding to each change, and calculating the probability score of origin of a thermal runaway fire normalized to between 0 and 1 based on the queried standard score and weight. Here, the thermal runaway origin indicator section may refer to a data section predefined to reflect a specific pattern of change in hydrogen ion concentration, a pattern of rapid increase in electrical conductivity, and a pattern of sustained high temperature that may appear due to battery electrolyte leakage or combustion.

[0075] The thermal runaway fire origin probability score is set to have the same normalization range as the hazardous chemical inclusion probability score, so that it can be configured to facilitate mutual comparison and combination in the subsequent combination stage.

[0076] The step of calculating the integrated contamination score is to calculate a combined score by applying a non-linear combination formula including a linear combination term and an interaction term to the hazardous chemical inclusion probability score and the thermal runaway fire origin probability score, and to calculate an integrated contamination score normalized between 0 and 1 by applying a sigmoid transformation to the combined score.

[0077] Here, the non-linear combination equation may be configured to include a linear combination term that applies a combination weight to each of the two scores, and an interaction term proportional to the product of the two scores or the product thereof, so that the risk level increases further when both scores are calculated to be high simultaneously. The sigmoid transformation may be configured so that the output is limited to between 0 and 1 even if the input range of the combination score varies, thereby allowing the integrated contamination score to be directly input into the threshold-based control logic.

[0078] The step of determining the pollution level grade and generating a pollution level profile involves determining a pollution level grade corresponding to a preset toxicity threshold range based on the integrated pollution level score, and generating a pollution level profile including the pollution level grade, the integrated pollution level score, and a water quality change feature vector. Here, the pollution level grade may refer to status information classified into multiple stages such as “Safe,” “Caution,” “Warning,” and “Severe” according to the integrated pollution level score, and the pollution level profile may be structured as judgment basis data for determining whether to isolate, bypass, or transfer to a retention tank in the subsequent drainage control command generation step.

[0079] The step of generating a contamination profile record is to generate a structured contamination profile record that can be referenced in the generation step of a drainage control command by adding the profile generation time, version information of the contamination standard table by fire type applied, and the sensor identifier of the multi-channel water quality sensor to the contamination profile.

[0080] Here, the structured contamination profile record can be configured in a standard data format that combines not only the analysis results but also the version of the reference table used in the analysis and sensor source information, and can be configured to ensure traceability and integrity as a basis data for determining the drainage path.

[0081] According to one embodiment, the method comprises: collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; generating a fire control package including a control command for each of the group of equipment to be controlled and target operating parameters for each equipment based on the group of equipment to be controlled and target operating parameters for each equipment; and transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of generating fire event information includes: generating a fused input feature vector by performing unit normalization by sensor type, noise removal, and missing value correction on the sensor data collected from the detection sensors; and calculating a fused fire score with weights applied by sensor type based on the fused input feature vector. The present invention provides a control method for a fire identification device that identifies a fire in a road tunnel by fusing a plurality of sensor data, comprising the step of determining that a fire event has occurred when the fused fire score is greater than or equal to a preset fire judgment threshold, and calculating a fire detection location and a fire intensity index according to the judgment result.

[0082] Based on the sensor data and equipment status data above, the step of generating fire event information including whether a fire event has occurred, the fire detection location, and a fire intensity index is to determine whether it is an actual fire or a false alarm by fusion analyzing the observation patterns of multiple sensors included in a time-synchronized data set, and if the determination result is confirmed to be a fire, to generate structured fire event information by calculating the fire detection location and the fire intensity index together.

[0083] Here, the fire detection location can be expressed as longitudinal coordinates within the tunnel or a predefined zone number, and the fire intensity index can be calculated as a scalar value or a grade value based on at least one of the temperature rise rate, the smoke and gas concentration rise rate, the flame detection duration, and the change in the area of ​​a camera-based heat source or flame candidate. Additionally, the fire event information may be configured to include an event identifier, a judgment time, a fire detection location, a fire intensity index, and a summary value of the sensor features used as the basis for judgment.

[0084] The step of generating a fused input feature vector by performing unit normalization by sensor type, noise removal, and missing value correction involves preprocessing the input so that sensor data with different physical quantities and scales can be compared and combined within the same computation frame, thereby configuring it into a fused input feature vector in the form of a multidimensional array. For example, the absolute temperature and temperature rise rate of a temperature sensor, the concentration and concentration rise rate of a smoke sensor, the amount of concentration change of a gas sensor, the detection duration of a flame detection sensor, the area ratio of a camera-based flame or smoke candidate region, and the airflow direction and velocity components of an airflow measurement sensor can each be standardized according to pre-set unit conversion rules and normalization ranges. Additionally, noise reduction filters such as moving average or median filters can be applied, and segments missing due to communication errors can be corrected using at least one of linear interpolation or preserving the previous value. Furthermore, by configuring the fused input feature vector so that it does not consist solely of a single point-in-time value but includes temporal summary features such as average values, slopes, peak values, and rising edge features within a pre-set time window, misjudgments sensitive to instantaneous spikes can be suppressed.

[0085] The step of calculating a fusion fire score with weights applied according to sensor type involves selecting key feature quantities designated for each sensor type among the components of the fusion input feature vector, and applying basic weights and reliability-based correction weights according to sensor type to calculate a fusion fire score in the form of a single scalar. For example, weights reflecting direct signs of fire may be assigned to feature quantities based on smoke, gas, temperature, and flame detection sensors; weights reflecting visual evidence may be assigned to camera-based vision feature quantities; and weights contributing to the verification of the consistency of combustion product diffusion may be assigned to airflow feature quantities. Additionally, so that the calculated fusion fire score can be directly used for threshold determination, it may be configured to be calculated as a value normalized between 0 and 1 by applying lower and upper limits to the weighted sum result or by applying a normalization transformation.

[0086] The step of determining that a fire event has occurred when the fusion fire score is above a preset fire judgment threshold, and calculating the fire detection location and fire intensity index according to the judgment result, is to generate a fire confirmation alarm when the calculated score exceeds a threshold value, but to suppress false alarms, to further verify whether the state where the fusion fire score is above the threshold is maintained for a continuous preset time or whether a simultaneous rise pattern in multiple sensors is satisfied, and then finally confirm that a fire has occurred. The fire detection location can be determined as the final location by setting the installation location of the maximum response sensor as a primary candidate by referring to the sensor placement map and reflecting a correction value based on the response intensity distribution of adjacent sensors, and the fire intensity index can be calculated as the combined result of the multiple feature quantities and subsequently input for equipment control profile lookup and control parameter determination.

[0087] The step of mapping the fire detection location to a section identifier of a road tunnel and querying a pre-stored equipment control profile corresponding to the section identifier and fire intensity indicator to determine the group of equipment to be controlled and target operating parameters for each piece of equipment is a step of mapping the physical location where the fire occurred to a section identifier, which is a system management unit, and confirming a response scenario optimized for the corresponding fire event by querying a pre-defined equipment control profile using the section identifier and fire intensity indicator as keys.

[0088] Here, the equipment control profile may be a predefined logic table determining which equipment to operate at what level depending on the location and intensity of the fire, and the group of equipment to be controlled may be determined as a set of fire extinguishing, ventilation, and evacuation guidance equipment that must be operated in response to the fire. Additionally, the target operating parameters may include specific control values ​​that the equipment must perform, such as the rotational speed of the pump, the target opening rate of the valve, and the rotational direction and speed of the jet fan; they may be determined as values ​​that are limited within the allowable range for each equipment type or have unit conversions applied, and may be configured to be replaced with alternative equipment depending on the equipment's failure flag or availability status.

[0089] The step of generating a fire control package is a step of generating structured package data including equipment-specific control commands and command application sequences based on the group of equipment to be controlled and equipment-specific target operating parameters, and converting the determined control strategy into an integrated data set in the form of a communication protocol that can be interpreted by the field equipment controller.

[0090] Here, the fire control package bundles individual commands for multiple heterogeneous facilities into a single execution unit. It can be configured to suppress facility conflicts caused by indiscriminate simultaneous operation by including not only target values ​​for each facility identifier but also precedence dependencies between commands, interlock conditions, timeouts, number of retries, and whether an acknowledgment is required.

[0091] The step of automatically controlling a group of target equipment by transmitting a fire control package to equipment controllers involves transmitting the generated fire control package to field equipment controllers via a wired or wireless communication network and performing closed-loop control that tracks command execution based on execution acknowledgments and feedback status received from the equipment controllers. For example, fault-tolerant operations can be performed by retransmitting if an acknowledgment is not received, or by creating and executing a reduced package excluding the equipment if a failure in some equipment is detected. Additionally, the system can be configured to continuously monitor equipment status data during control execution, so that if abnormal situations such as overcurrent, overheating, inability to operate, or sudden pressure changes are detected, an emergency stop or safety mode switching command is applied first.

[0092] Additionally, the detection sensors include: a smoke sensor, a gas sensor, a temperature sensor, a flame detection sensor, a camera, and an airflow measurement sensor, and the step of calculating the fused fire score comprises: a step of deriving a fire estimation point based on features based on the smoke sensor, gas sensor, temperature sensor, and flame detection sensor included in the fused input feature vector; a step of extracting the movement speed vector and driving trajectory of vehicles driving in the tunnel by applying an object tracking algorithm to the camera-based vision features included in the fused input feature vector; a step of generating an airflow-based spatiotemporal delay profile that hydrodynamically models the delay time for combustion products to diffuse and reach the smoke sensor and gas sensor spaced upstream and downstream from the fire estimation point based on the airflow direction and airflow speed features included in the fused input feature vector; and a step of identifying the time of peak occurrence based on the time-series detection features of the smoke sensor and gas sensor included in the fused input feature vector. A step of calculating environmental synchronization reliability for each of the smoke sensor and gas sensor by verifying whether the time of occurrence of the peak value is included within the valid arrival time window indicated by the airflow-based spatiotemporal delay profile; a step of calculating a spatial movement velocity vector in which detection events are sequentially activated along the longitudinal direction of the tunnel based on the detection value rising edge feature quantity of the temperature sensor, gas sensor, and smoke sensor included in the fused input feature vector; and a step of calculating velocity cosine similarity between the spatial movement velocity vector and the movement velocity vector.If the velocity cosine similarity is greater than or equal to a preset synchronization threshold, the sensor detection event corresponding to the fire estimation point is determined as a False Alarm, and the fire estimation point is mapped to a section identifier of a road tunnel to determine the section containing the fire estimation point, and a motion noise penalty index is applied to the section containing the fire estimation point; a step of generating an initial dynamic weight redistribution matrix based on a preset basic weight for each detection sensor type; a step of generating a first adjusted dynamic weight redistribution matrix by up-adjusting the weight components corresponding to the smoke sensor and gas sensor in the initial dynamic weight redistribution matrix based on the environmental synchronization reliability; and a step of generating a second adjusted dynamic weight redistribution matrix for the section to which the motion noise penalty index is applied by attenuating the weight components corresponding to the smoke sensor, gas sensor, and temperature sensor in the first adjusted dynamic weight redistribution matrix and increasing the weight components corresponding to the flame detection sensor and the camera-based vision feature quantity. and may include the step of calculating the fusion fire score by performing a matrix multiplication operation between the second adjustment dynamic weight redistribution matrix and the fusion input feature vector.

[0093] The detection sensor is a group of sensors for capturing signs of fire in a road tunnel from various angles, and is configured to include a smoke sensor, a gas sensor, a temperature sensor, a flame detection sensor, a camera, and an airflow measurement sensor. Here, the smoke sensor and the gas sensor provide the concentration of combustion products and the trend of concentration change, the temperature sensor and the flame detection sensor provide thermal characteristics and flame detection persistence (e.g., flicker or radiation components in specific wavelength bands), the camera provides flame / smoke candidate regions and vehicle driving status as image-based features, and the airflow measurement sensor provides the longitudinal airflow direction and airflow velocity components within the tunnel, which can be used as inputs for interpreting the time delay of combustion product diffusion.

[0094] The step of deriving a fire estimation point based on features of a smoke sensor, gas sensor, temperature sensor, and flame detection sensor is to calculate primary candidate coordinates that are highly likely to have a fire by analyzing the spatial response intensity distribution of physical sensing data included in the fused input feature vector. For example, after evaluating whether there exists a group of sensors in a specific section where the temperature rise edge intensity of the temperature sensor increases, the concentration rise rate of the nearby smoke sensor and the concentration change amount of the gas sensor increase together, and the detection duration of the flame detection sensor increases simultaneously, the fire estimation point can be set by inversely calculating the center point based on the installation coordinates of the sensor group. At this time, the fire estimation point may be determined as the location of the maximum value of a single sensor or as a corrected location obtained by weighted averaging the response intensity distributions of adjacent sensors, and may be expressed as at least one of tunnel length direction coordinates or a section identifier coordinate system.

[0095] The step of extracting the movement velocity vectors and driving trajectories of vehicles traveling within a tunnel by applying an object tracking algorithm to camera-based vision features is a step of generating mobility clues to determine whether the sensor response near the estimated fire point is caused by a fixed fire source or by the moving heat source, exhaust gas, dust, etc. of the vehicles. For example, the object tracking algorithm can track the inter-frame displacement of the bounding box center point or vehicle key point of the same vehicle object in consecutive frames, calculate a movement velocity vector for each vehicle based on the temporal change of displacement, and generate a driving trajectory by connecting position histories across multiple frames. Additionally, the representative movement velocity vector in the section containing the estimated fire point can be calculated as the average or median of the movement velocity vectors for each vehicle and used as a standard for a subsequent comparison step.

[0096] The step of generating an airflow-based spatiotemporal delay profile based on airflow direction and velocity characteristics involves hydrodynamically modeling the delay time for combustion products generated at the fire estimation point to diffuse upstream and downstream by the airflow and reach the smoke sensor and gas sensor, thereby generating a predicted arrival time value to be used for verification. For example, the longitudinal distance between the fire estimation point and the installation location of each smoke sensor and gas sensor can be calculated, and the effective movement speed of the combustion products can be set based on the airflow direction and velocity components provided by the airflow measurement sensor. Subsequently, the estimated arrival time for each sensor can be calculated as the ratio of the effective movement speed to the distance. Additionally, by adding an allowable error time to the estimated arrival time, considering turbulence, changes in tunnel cross-section, and airflow disturbances caused by vehicle traffic, the system can be configured to define an effective arrival time window for each sensor.

[0097] The step of identifying the point in time of peak occurrence based on time-series detection features of smoke and gas sensors involves deriving the time when the valid detection signal reaches its maximum value for each sensor channel, or the representative time of the interval where it rises above a threshold and converges to the maximum value, thereby defining the point in time when the actual observed combustion product is reached. At this time, to exclude one-off peaks caused by sensor noise, the system may be configured to adopt only peaks that persist for a certain period of time or longer after the rising edge as valid peaks.

[0098] The step of calculating environmental synchronization reliability by verifying whether the time of peak occurrence falls within the valid arrival time window indicated by the airflow-based spatiotemporal delay profile is a step of examining physical probability by quantifying the degree of agreement between the arrival time predicted by the airflow model and the time when the sensor signal actually forms a peak. For example, environmental synchronization reliability normalized between 0 and 1 can be calculated by assigning high reliability if the time of peak occurrence is within the valid arrival time window and attenuating reliability as it deviates from the window. Additionally, the system can be configured to distinguish between the smoke sensor and the gas sensor under evaluation based on the direction of diffusion dominance, calculate reliability by direction, and then map it to the reliability for each sensor.

[0099] The step of calculating a spatial movement velocity vector based on the rising edge features of detection values ​​from temperature sensors, gas sensors, and smoke sensors is a step of calculating the spatial progression speed of sensor responses using the propagation pattern in which detection events are sequentially activated along the tunnel length direction. For example, the time at which a rising edge is first observed from multiple sensors is defined as the activation time for each sensor, and the propagation speed for each segment is calculated based on the difference in activation time relative to the length distance for each pair of adjacent sensors, and then integrated into a spatial movement velocity vector having a length component.

[0100] The step of calculating the velocity cosine similarity between the spatial movement velocity vector and the movement velocity vector is a comparison step for determining whether the cause of the sensor response is fire spread or noise originating from the moving vehicle by quantifying the degree of alignment between the propagation direction and velocity of the sensor response and the movement direction and velocity of the moving vehicle. For example, the degree of directional alignment of the two vectors can be calculated using an inner product-based normalization method, wherein directional component-centered normalization is applied to prevent excessive influence from differences in vector magnitudes, and the calculation can be configured to focus on the tunnel length component.

[0101] The step of determining a non-fire alarm and applying a motion noise penalty index when the velocity cosine similarity is above a preset synchronization threshold is a step of identifying cases where there is a high probability that a sensor response will be triggered by the mobility factors of a moving vehicle, and generating a damping index to reduce the fire detection sensitivity of the corresponding section. For example, if the similarity is above the threshold, a sensor detection event corresponding to a fire-estimation point may be determined as a non-fire alarm candidate, and after determining the section by mapping the fire-estimation point to a section identifier of a road tunnel, a motion noise penalty index may be applied to that section. Here, the motion noise penalty index may be configured to increase in proportion to at least one of the degree of exceeding the threshold, the number of tracked vehicles, the variance of vehicle speeds, or the sequential strength of sensor activation.

[0102] The step of generating an initial dynamic weight redistribution matrix based on basic weights for each detection sensor type is a step of preparing a basic weight matrix that reflects the importance or basic reliability for each sensor type. For example, since smoke sensors and gas sensors reflect direct indications of combustion products, their basic weights may be set relatively high, and since camera-based vision features and flame detection sensors provide visual grounds, separate basic weights may be set, and the matrix may be configured to correspond one-to-one with the component bundles of the fused input feature vectors.

[0103] The step of generating a first adjustment dynamic weight redistribution matrix based on environmental synchronization reliability is to upwardly adjust the weight components corresponding to smoke sensors and gas sensors in the initial dynamic weight redistribution matrix to strengthen the contribution of sensors that have shown a response that is temporally aligned with the airflow model. Conversely, if the environmental synchronization reliability is calculated to be low, the corresponding weight components may be configured to be attenuated, and the upward and attenuation adjustments may be performed using at least one of a linear proportional method or a segment-based rule method.

[0104] The step of generating a second adjusted dynamic weight redistribution matrix for sections to which a motion noise penalty index is applied is a step of modifying the weighting strategy to enable robust fire judgment even in noisy situations by attenuating the weighting components corresponding to smoke sensors, gas sensors, and temperature sensors, and increasing the weighting components corresponding to flame detection sensors and camera-based vision features, in sections with a high probability of motion noise. For example, in cases where there is a high risk that smoke, gas, and temperature data will be distorted by vehicle movement, the system may be configured to suppress false alarms by placing higher weight on flame detection persistence or the maintenance of vision-based flame candidate regions.

[0105] The step of calculating a fused fire score by performing a matrix multiplication operation between the second adjusted dynamic weight redistribution matrix and the fused input feature vector is a step of calculating a single fire score by integrating the fused input feature vector with the contribution by sensor type and feature quantity adjusted. For example, the fused fire score can be calculated by applying each weight component of the matrix to the corresponding component of the fused input feature vector and then performing a weighted summation, and the calculated score can be configured to be normalized to a range between 0 and 1 so that it can be directly compared with a subsequent fire judgment threshold. In addition, by configuring the process so that a weight snapshot of how the weights by sensor type were adjusted is recorded during the calculation of the fused fire score, traceability of the basis for the fusion judgment can be ensured in false alarm analysis and post-hoc verification.

[0106] And, the step of calculating the fire detection location and fire intensity index comprises: a step of estimating a suspected ignition vehicle based on a set of candidate heat source pixels per vehicle included in the camera-based vision features; a step of recognizing the vehicle type and external dimensions of the suspected ignition vehicle from the camera-based vision features, and loading a basic fire profile including a base heat release rate and a fire growth model corresponding to the vehicle type from a pre-established vehicle fire database; a step of identifying a high-temperature area recording a temperature above a preset danger temperature in the ceiling of the road tunnel based on temperature sensor-based features to which the second adjusted dynamic weight redistribution matrix is ​​applied, and calculating the longitudinal expansion rate of the high-temperature area; a step of determining the fire intensity index at the current time by matching the expansion rate of the high-temperature area with the fire growth model included in the basic fire profile; a step of dividing the internal space of the road tunnel into a three-dimensional grid and calling a trajectory mapping table in which the deflection trajectory of the fire heat column according to a plurality of fire intensity and airflow velocity conditions is mapped by prior computational fluid dynamics simulation; The method may include: a step of setting the installation coordinates of a ceiling temperature sensor in which the highest temperature is detected among a plurality of ceiling temperature sensors included in the fused input feature vector as a ceiling peak node; a step of querying specific deflection trajectory data corresponding to the fire intensity index and airflow velocity feature quantity from the trajectory mapping table; a step of deriving the coordinates of an ignition origin intersecting the tunnel road surface by performing a 3D trajectory search that traces downward in the reverse direction of the airflow along the deflection angle indicated by the specific deflection trajectory data from the ceiling peak node; and a step of confirming the ignition origin coordinates as the fire detection location and including them in the fire event information.

[0107] The step of estimating a vehicle suspected of ignition involves identifying a vehicle suspected of being the cause of the fire within the tunnel by utilizing a set of vehicle-specific heat source candidate pixels included in camera-based vision features. For example, a bounding box or segment mask for each vehicle object may be determined first in the camera image, and a vehicle-specific heat source score may be calculated by combining at least one of the area ratio occupied by the set of heat source candidate pixels within each vehicle object region, the average intensity of the heat source candidate pixels, and the temporal persistence of the heat source candidate pixels. In this case, the set of heat source candidate pixels may be defined as a set of pixels in which the temperature equivalent value of an infrared camera is greater than or equal to a preset heat source threshold, or a set of pixels in a visible light camera in which both the flame candidate color distribution and blinking characteristics are simultaneously satisfied.

[0108] In the process of constructing a set of candidate heat source pixels, to exclude light source components unrelated to fire, such as saturated pixels caused by headlights, brake lights, and reflectors, additional verification may be performed on whether the inter-frame movement trajectory of candidate heat source pixels accompanies the movement of the vehicle object, whether the same pixel is maintained at high intensity for a certain period of time or longer, and whether smoke candidate textures or boundary diffusion patterns around the candidate heat source pixels are present. Therefore, the vehicle suspected of ignition can be configured to be estimated as the vehicle with the maximum "time cumulative value of heat source scores per vehicle object," rather than the maximum value of a single frame.

[0109] The step of recognizing the vehicle type and exterior dimensions of the suspected ignition vehicle and loading a basic fire profile is a step of calling a basic fire profile that includes a baseline heat release rate and a fire growth model typically expected for the vehicle type, based on the exterior information of the suspected ignition vehicle. For example, vehicle type recognition can be performed by using a classification model that takes vision features such as vehicle front shape, body proportions, estimated distance between axles, and the presence or absence of a cargo bed as input, and can identify the vehicle as at least one vehicle class among passenger cars, SUVs, buses, trucks, motorcycles, and electric vehicles. The exterior dimensions can be determined by estimating the actual spatial length, width, height, or representative cross-sectional area of ​​the vehicle through distance conversion using camera correction parameters and known reference dimensions such as tunnel lane width and wall reference lines.

[0110] A vehicle fire database storing a basic fire profile can store, for each vehicle class, at least one of a base heat release rate indicating the initial heat release level, a fire growth model parameter indicating the growth pattern over time, and a fuel load estimate affecting the formation characteristics of a heat column in the roof in the form of a record.

[0111] Here, the fire growth model may be composed of a multi-segment model in which the growth rate varies by segment, such as a “growth segment immediately after ignition,” a “flashover or rapid growth segment,” and a “normal combustion segment,” and the transition condition for each segment may be defined as at least one of the elapsed time, the rate of temperature rise, or the duration of flame detection.

[0112] The step of identifying high-temperature regions and calculating longitudinal expansion rates involves spatially grouping regions recording temperatures above a critical threshold in the tunnel ceiling using temperature sensor-based features to which a second adjusted dynamic weight redistribution matrix is ​​applied, and calculating the longitudinal expansion rate of the corresponding regions. For example, among multiple ceiling temperature sensors, sensors exceeding a critical temperature threshold may be labeled as high-temperature sensors, and a continuous section of high-temperature sensors with adjacent longitudinal coordinates may be defined as a high-temperature region. The longitudinal expansion rate can be calculated by tracking the upstream and downstream boundary coordinates of the high-temperature region for each time frame and dividing the change in boundary coordinates by the observation time interval; furthermore, moving average-based smoothing may be applied to suppress noise in the expansion rate.

[0113] The step of determining the fire intensity index involves calculating the fire intensity index at the current point in time by comparing and matching the longitudinal expansion rate of the high-temperature region with the fire growth model of the underlying fire profile. For example, if the fire growth model loaded from the vehicle fire database includes an expected range regarding “at what speed the high-temperature region of the ceiling expands over time,” the fire intensity index can be determined as a grade of “weak, medium, or strong” depending on which segment of the expected range the observed expansion rate falls into. As another example, the current state parameters of the model may be back-estimated to minimize the error between the observed expansion rate and the expansion rate predicted by the model, and the back-estimated state parameters may be defined as the scalar value of the fire intensity index.

[0114] The step of calling the 3D grid network and trajectory mapping table is to discretize the internal space of the tunnel into 3D grid units and load the trajectory mapping table, in which the deflection trajectory of the fire heat column according to fire intensity and airflow velocity conditions is mapped in advance by computational fluid dynamics simulation, into a queryable state.

[0115] Here, the three-dimensional grid can be defined as a set of cells divided into pre-set grid intervals for the tunnel length, width, and height directions, and each cell can be used as an index to represent the average direction, average velocity, and temperature gradient of the airflow. The trajectory mapping table can be configured to store at least one of the deflection angle of the hot column centerline, the spatial path of the deflected centerline, and the uncertainty width of the path as a value, using a combination of the fire intensity indicator section and the airflow velocity section as a key.

[0116] The step of setting the ceiling peak node involves selecting the sensor with the highest detected temperature among the ceiling temperature sensors included in the fused input feature vector, and defining the installation coordinates of that sensor as the ceiling peak node. In this case, if there are multiple sensors with the highest temperature, a single ceiling peak node can be determined by applying the temperature rise rate per sensor, the detection persistence of the flame detection sensor, or the density of the camera-based heat source candidate pixel set as a tiebreaker.

[0117] The step of querying bias trajectory data is to select an item corresponding to the determined fire intensity index and airflow velocity feature from the trajectory mapping table to obtain candidate bias trajectory data connecting the ceiling peak node and the ignition origin.

[0118] Here, the airflow velocity feature may include at least one of the longitudinal velocity component of the airflow measuring sensor, the effective wind speed correction value according to the jet fan operating state, and the wind speed fluctuation component according to the vehicle piston effect, and when querying the trajectory mapping table, the trajectory corresponding to the section to which the current wind speed belongs may be selected, or an intermediate trajectory may be synthesized by linear interpolation between trajectories of adjacent sections.

[0119] The step of deriving the ignition origin coordinates by performing a 3D trajectory search is a step that starts from the ceiling peak node, performs downward backtracking in the reverse direction of the airflow along the deflection angle indicated by the retrieved deflection trajectory data, and determines the point where the backtracking path intersects the tunnel road surface as the ignition origin coordinates. For example, backtracking can be implemented by moving stepwise from the ceiling peak node to adjacent cells within the grid, and determining the next direction of movement by a weighted composite direction of the “reverse airflow direction” and the “deflection trajectory centerline direction” at each movement step. The longitudinal and transverse coordinates at the point where the backtracking path reaches below the road surface height can be derived as the ignition origin coordinates, and if multiple road surface intersection points are calculated, a candidate that minimizes the distance difference between the camera-based suspected ignition vehicle location, the fire estimation point, and the center coordinates of the high-temperature area can be selected as the final intersection point.

[0120] The step of determining the fire detection location and including it in the fire event information is to determine the derived ignition origin coordinates as the fire detection location and record them as a field of the fire event information along with the fire intensity index. For example, the fire detection location may be expressed as at least one of the values ​​obtained by converting the longitudinal component of the ignition origin coordinates into tunnel changeage coordinates, or the segment identifier containing the said coordinates. Additionally, the fire event information may be configured to include a suspected ignition vehicle identifier, a vehicle type recognition result, a version of the trajectory mapping table used, and a flag indicating whether backtracking was successful, so as to ensure traceability of the basis for judgment in subsequent firefighting equipment control and post-analysis.

[0121] According to one embodiment, the method comprises: collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; generating a fire control package including a control command for each of the group of equipment to be controlled and target operating parameters for each equipment based on the group of equipment to be controlled and target operating parameters for each equipment; and transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of generating the fire control package includes: identifying a ventilation equipment among the group of equipment to be controlled based on the fire detection location and fire intensity index, and generating operating mode information for the ventilation equipment by determining upstream / downstream directions and target smoke exhaust directions based on the fire detection location. The present invention provides a control method for a firefighting control device that automatically controls a ventilation facility when a fire occurs in a road tunnel, comprising the step of: setting ventilation control parameters including rotation direction, rotation speed, opening rate, and operation sequence for a jet fan, exhaust fan, supply fan, damper, and ventilation port included in the ventilation facility based on the above-mentioned driving mode information and target driving parameters, and generating a ventilation control command according to the above-mentioned ventilation control parameters.

[0122] The step of collecting time-synchronized sensor data and equipment status data involves collecting sensor-specific measurements and equipment-specific operating statuses collected from detection sensors and firefighting equipment distributed at multiple points within a road tunnel as raw data, and generating a data set aligned with a common time axis corresponding to a server reference time in order to reduce time axis mismatches in which the same physical event is recorded at different times due to differences in sensor-specific sampling cycles and communication delays.

[0123] Time-synchronized sensor data and equipment status data refer to data from heterogeneous devices with different communication cycles that are aligned to the same point in time based on the server's reference time. This allows for consistent matching of airflow conditions and fan operation status at a specific time. For example, by normalizing the timestamps of each sensor channel to the server's reference time and resampling them using at least one of nearest value selection, linear interpolation, or interval averaging to align with a common time grid, aligned data can be generated that enables comparison between sensors at the same analysis time. Furthermore, by identifying channels where timestamps are missing, drift, or abnormal delays exceed the allowable range and marking them as missing channels, or by excluding drift-occurring sections as invalid and adopting only valid sections, misjudgments caused by time errors between sensors can be suppressed in subsequent fire detection.

[0124] The step of generating operation mode information by identifying ventilation equipment and determining the target smoke exhaust direction is to identify equipment corresponding to ventilation equipment among the group of equipment to be controlled based on the fire detection location and fire intensity index, define upstream and downstream directions based on the fire detection location, and then generate operation mode information of the ventilation equipment by determining the target smoke exhaust direction to induce smoke and toxic gases.

[0125] Here, upstream and downstream directions can be defined based on the tunnel's direction of travel or the direction of airflow, and the target smoke exhaust direction may refer to the target direction of the airflow set to suppress the inflow of smoke toward the direction where evacuees are concentrated. Additionally, the operating mode information may consist of information specifying a control algorithm type suitable for the current situation among a reverse airflow prevention mode, a concentrated smoke exhaust mode, a critical wind speed maintenance mode, or a zone separation mode, depending on the progression of the fire. By configuring it to include an application section identifier, an application start time, a priority control target fan group, a basic opening policy for dampers and vents, and interlock conditions required for mode switching, consistent control can be performed without conflicts between facilities during the subsequent ventilation control parameter setting stage.

[0126] The step of setting ventilation control parameters and generating ventilation control commands is to determine ventilation control parameters, including rotation direction, rotation speed, opening rate, and operation sequence, for each of the jet fan, exhaust fan, supply fan, damper, and ventilation port included in the ventilation equipment, based on the operation mode information and the target operation parameters, and to generate ventilation control commands according to the determined ventilation control parameters. For example, the jet fan may be set to a rotation direction corresponding to the target smoke exhaust direction and may be set so that the target rotation speed increases stepwise according to the fire intensity index, and the exhaust fan and supply fan may be set to have operation parameters corresponding to the target flow rate or target pressure. In addition, the damper and ventilation port may be set to have an opening rate to form a target smoke exhaust path, and the operation sequence may include a starting sequence that prevents sudden pressure fluctuations by starting the jet fan sequentially.

[0127] Ventilation control commands may consist of individual device-specific drive signals containing finally calculated rotational speed and opening rate information, and may be configured as a command set in which equipment-specific control parameters are packaged into protocol data interpretable by the field equipment controller, including an equipment identifier, target value, application time, application sequence number, timeout, number of retries, and whether an acknowledgment is required. Additionally, when ventilation control commands are transmitted as part of a fire control package, they may be configured to include a package identifier, event identifier, application section identifier, and version information of the operating mode information to prevent duplicate execution by the equipment controller and ensure the integrity of the execution history.

[0128] Additionally, the detection sensor comprises: a smoke sensor, a temperature sensor, and a camera, and the step of generating operating mode information of the ventilation facility comprises: detecting congested vehicles existing in an upstream section relative to the fire detection location based on images captured by the camera included in the detection sensor; calculating the upstream congested vehicle density based on the inter-vehicle distance, lane occupancy rate, and number of vehicles per unit length of the congested vehicles; tracking driving vehicles existing in a downstream section relative to the fire detection location based on the object tracking results of images captured by the camera; aggregating the longitudinal components of the driving speed vectors of the driving vehicles to calculate the average moving speed in the downstream direction; and applying a piston effect correction model including an effective cross-sectional reduction factor based on the upstream congested vehicle density and an effective flow coefficient based on the average moving speed in the downstream direction to calculate a longitudinal airflow correction value induced by the piston effect in the tunnel. A step of obtaining the longitudinal gradient of the fire occurrence section from a previously stored tunnel longitudinal profile table based on a section identifier mapped to the fire detection location; a step of calculating a reference critical wind speed for preventing back-layering based on the fire intensity index and the longitudinal gradient of the fire occurrence section; a step of calculating a dynamic critical velocity by correcting the reference critical wind speed with the longitudinal airflow correction value; and a step of calculating the real-time descent depth of the smoke layer formed on the ceiling based on the smoke concentration feature of the smoke sensor included in the detection sensor and the smoke layer boundary estimation result of the image captured by the camera.The method may include: a step of calculating a limit flow rate for the occurrence of a plug-holing phenomenon based on the real-time descent depth and the effective opening area of ​​the ventilation opening; and a step of generating operating mode information of the ventilation equipment to include the dynamic critical wind speed and the limit flow rate.

[0129] The step of detecting congested vehicles is to identify vehicles that remain in a virtually stationary state within an upstream section relative to the fire detection location, based on images captured by a camera included in the detection sensor. For example, the upstream section may be defined based on a section identifier mapped to the fire detection location and a preset upstream length range. After detecting candidate vehicle objects by applying an object detection model to the camera images, vehicles whose speed is below a preset congestion threshold speed based on the inter-frame object tracking results may be determined as congested vehicles if the state is maintained for a continuous preset time or longer. At this time, the upstream direction may be defined as the airflow inflow side or the side where a vehicle enters the direction of the fire source relative to the fire detection location. The method may be configured to exclude vehicles whose tracking reliability has dropped sharply due to lane changes, obstruction, or light reflection, or to suppress overestimation of the set of congested vehicles by merging duplicate detections between multiple cameras.

[0130] The step of calculating the upstream congested vehicle density involves calculating the inter-vehicle distance, lane occupancy rate, and number of vehicles per unit length for the detected congested vehicles, and quantifying the upstream congested vehicle density by combining the calculated values. For example, the inter-vehicle distance can be calculated as the longitudinal position of adjacent congested vehicles classified as the same lane, the lane occupancy rate can be calculated as the ratio of the vehicle object area to the lane area within the section of interest, and the number of vehicles per unit length can be calculated by dividing the number of congested vehicles by the length of the section of interest. Additionally, a coordinate transformation model reflecting the camera installation height, tilt angle, and focal length can be applied to convert the vehicle position in image coordinates into the tunnel longitudinal distance, and the congested vehicle density can be configured to be calculated as a density index normalized between 0 and 1 by weighted summing the inverse component of the inter-vehicle distance, the lane occupancy rate component, and the number of vehicles per unit length component. Accordingly, it is interpreted that the higher the congested vehicle density, the smaller the effective cross-section through which air can flow within the tunnel, and this can be input for subsequent piston effect correction.

[0131] The step of tracking moving vehicles involves connecting moving vehicles located in a downstream section relative to the fire detection location in consecutive frames based on the object tracking results of the camera images, and generating a trajectory for each vehicle. For example, the downstream direction may be defined as the direction in which fire smoke and toxic gases are induced, and object tracking may be performed based on the center point of the bounding box of a vehicle object or the inter-frame displacement of a vehicle keypoint. Additionally, in sections where branching or merging occurs during tracking, external features and motion continuity constraints may be applied to stably maintain the same vehicle identifier.

[0132] The step of calculating the average travel speed in the downstream direction is to calculate the average travel speed in the downstream direction by aggregating the tunnel longitudinal component among the travel speed vectors of the vehicles. For example, the travel speed vector may be calculated by dividing the change in longitudinal position of each vehicle's trajectory by time, and the average travel speed in the downstream direction is calculated as the average or median value of the longitudinal speeds of each vehicle within a pre-set aggregation time window. If outliers such as rapid acceleration or rapid deceleration are included, the system may be configured to apply the average value after removing the outliers. The average travel speed in the downstream direction may be used as a basic variable to quantify the effect of moving vehicles dragging surrounding air.

[0133] The step of calculating the longitudinal airflow correction value is to calculate the correction component of the longitudinal airflow induced by the piston effect in the tunnel by applying a piston effect correction model that includes an effective cross-sectional reduction factor based on the upstream congested vehicle density and an effective flow coefficient based on the downstream average travel speed.

[0134] Here, the piston effect can be defined as a phenomenon in which a vehicle moves within a tunnel and forms a longitudinal airflow by pushing or pulling air, and the effective cross-section reduction factor can be defined as a damping factor between 0 and 1 to reflect that the tunnel's effective cross-section decreases as the density of the congested vehicles increases. Additionally, the effective flow rate factor can be defined as an amplification factor with a value greater than or equal to 0 to reflect that the longitudinal flow rate induced by the vehicle cluster increases as the average downstream moving speed increases. Accordingly, the longitudinal airflow correction value is calculated as the result of combining the “value obtained by applying the effective flow rate factor to the average downstream moving speed” and the “value dampened by the effective cross-section reduction factor due to congestion,” and the sign and direction of application of the correction value can be configured to be consistently assigned according to the definition of upstream / downstream directions based on the fire detection location.

[0135] The step of obtaining the longitudinal gradient is to query the longitudinal gradient of the fire occurrence section from a pre-stored tunnel longitudinal profile table based on a section identifier mapped to the fire detection location. For example, the tunnel longitudinal profile table may be configured to store the longitudinal gradient, section length, elevation reference point, and gradient direction for each section identifier, and the longitudinal gradient may also be used to determine whether the upstream or downstream direction is an ascending gradient or a descending gradient. Additionally, the longitudinal gradient may be used as a correction term for calculating the wind speed that prevents backflow, as a factor reflecting the tendency of high-temperature smoke to move upward along the slope.

[0136] The step of calculating the reference critical wind speed is to calculate the reference critical wind speed for preventing backflow based on the fire intensity index and the longitudinal gradient of the fire occurrence section. For example, the reference critical wind speed may be set to increase as the fire intensity index increases, and may be configured to add to the reference critical wind speed by reflecting that the tendency for smoke backflow increases due to high-temperature buoyancy and slope when the longitudinal gradient is upward. Additionally, the reference critical wind speed may be implemented by calculating it by querying a pre-established critical wind speed lookup table or by applying a correction coefficient corresponding to the fire intensity index section and the longitudinal gradient section.

[0137] The step of calculating the dynamic critical wind speed is to calculate the dynamic critical wind speed by correcting the reference critical wind speed with the longitudinal airflow correction value. For example, if the longitudinal airflow correction value acts in the same direction as the target smoke exhaust direction and it is evaluated that a certain level of longitudinal airflow has already been secured, the reference critical wind speed can be subtracted to lower the dynamic critical wind speed; conversely, if the longitudinal airflow correction value acts as a reverse component, the reference critical wind speed can be added to increase the dynamic critical wind speed. Additionally, the system can be configured to ensure stability as a control input by applying a lower limit so that the dynamic critical wind speed is not calculated to be less than zero, and by applying an upper limit so that it does not exceed the equipment operating limit.

[0138] The step of calculating the real-time descent depth is a step of calculating the real-time descent depth of the smoke layer formed in the ceiling portion based on the smoke concentration feature of the smoke sensor and the smoke layer boundary estimation result of the camera image.

[0139] The step of calculating the occurrence limit flow rate is a step of calculating the occurrence limit flow rate that must be limited to prevent plughole phenomena from occurring, based on the real-time descent depth and the effective opening area of ​​the ventilation opening.

[0140] Here, the plug-holing phenomenon can be defined as a phenomenon in which only relatively clean air at the bottom of the smoke layer is drawn in when the suction force of the exhaust system is excessive or the smoke layer is thin, and the limit flow rate can be defined as the maximum allowable exhaust capacity to preferentially discharge only smoke without such a phenomenon occurring. For example, the effective opening area can be calculated by reflecting the opening state of the vent, the damper opening rate, the grille loss coefficient, and the attenuation of the effective area due to contamination or blockage, and the limit flow rate can be set to decrease when the real-time descent depth is shallow and a thin smoke layer is formed, and to increase when the real-time descent depth is deep and a sufficient smoke layer is secured. Additionally, the limit flow rate can be implemented by calculating it by querying a limit flow rate table corresponding to the combination interval of the “effective opening area” and the “real-time descent depth,” or by applying a pre-set hydrodynamic correction coefficient.

[0141] The step of generating operating mode information is a step of structuring and generating operating mode information of the ventilation equipment to include the dynamic critical wind speed and the generated limit flow rate. For example, the operating mode information may be configured to include a target smoke exhaust direction, an applicable section identifier, a mode start time, a dynamic critical wind speed, a generated limit flow rate, and summary values ​​of upstream congested vehicle density and downstream average travel speed that serve as the basis for calculating the values, and the operating target values ​​of the jet fan, exhaust fan, supply fan, damper, and ventilation port may be limited in the subsequent ventilation control parameter setting step so that the constraints for blocking backflow and the constraints for avoiding plugholes are simultaneously satisfied.

[0142] And, the step of calculating the real-time descent depth comprises: a step of highlighting smoke candidate pixels by performing a difference operation with a preset background image on a time-series frame of video captured by the camera; a step of performing a spatial derivative operation of vertical pixel brightness on the time-series frame and extracting pixel inflection points where the absolute value of the brightness change rate exceeds a preset boundary threshold; a step of deriving a two-dimensional visual smoke boundary line by connecting the pixel inflection points according to horizontal connectivity constraints and inter-frame continuity constraints; a step of calculating the visual smoke bottom elevation by applying a camera back-projection model reflecting the camera's installation elevation, tilt angle, and focal length, and converting the two-dimensional visual smoke boundary line into a three-dimensional tunnel spatial coordinate system; and a step of generating a vertical distribution profile expressed in the form of a continuous function by interpolating smoke concentration values ​​at each installation elevation of smoke sensors obtained from a group of smoke sensors arranged in multiple stages at different elevations on the tunnel sidewall. Based on the above vertical distribution profile, a step of identifying as a physical sensing altitude the altitude at which the light attenuation rate reaches a reference threshold corresponding to the limit visibility distance for human evacuation; a step of calculating a soot concentration index from the smoke concentration characteristic of the smoke sensor group, and updating the first reliability weight and the second reliability weight so as to exponentially decrease the first reliability weight assigned to the visual smoke bottom altitude and relatively increase the second reliability weight assigned to the physical sensing altitude as the soot concentration index increases; a step of deriving a corrected smoke altitude by cross-weighting the sum of the visual smoke bottom altitude and the physical sensing altitude based on the first reliability weight and the second reliability weight; and a step of calculating the base smoke layer descent thickness by subtracting the corrected smoke altitude from the highest ceiling height of the road tunnel.and may include the step of calculating a thermal expansion fluctuation correction value proportional to the time-series temperature rise rate of the tunnel ceiling, and adding the thermal expansion fluctuation correction value to the base smoke layer descent thickness to determine the real-time descent depth.

[0143] The step of calculating the real-time descent depth involves estimating the lower boundary of the smoke layer formed on the tunnel ceiling based on camera images, but to compensate for situations where the image boundary becomes unstable due to dense smoke, the step calculates the real-time descent depth that can be directly input for evacuation safety and smoke exhaust control by cross-fusing the physical sensing results of groups of smoke sensors placed at different altitudes on the side walls. At this time, the real-time descent depth can be defined as the amount of descent of the lower altitude of the smoke layer relative to the maximum ceiling height of the tunnel, and the method can be configured to first derive a corrected smoke altitude by combining the image-based lower altitude and the sensor-based lower altitude with a reliability weight, and then determine the descent depth from the corrected smoke altitude.

[0144] The step of highlighting smoke candidate pixels is a step of highlighting newly emerging smoke areas as candidates relative to fixed structures within the tunnel by performing a difference operation with a pre-set background image for each time-series frame captured by the camera.

[0145] Here, the background image is a video of the tunnel interior during normal conditions without fire, and can be set as a reference frame from the normal operating section prior to the fire or as a background model based on the median of frames accumulated over a certain period. Additionally, the difference operation can be implemented as a preprocessing step that removes fixed terrain features and emphasizes fluid components by utilizing the difference in pixel values ​​between the current frame and the background image. Furthermore, to reduce false positives caused by lighting flicker or headlight reflections, it can be configured to select a channel less sensitive to changes in illumination, or to remove sporadic noise components by applying post-processing such as morphological open / close to the difference result. Moreover, in cases where camera vibration or minute viewpoint changes are present, it can be configured to perform image stabilization correction for frame alignment before performing the difference operation.

[0146] The step of extracting pixel inflection points utilizes the fact that the bottom of the smoke layer is generally formed in a horizontal direction to perform a spatial derivative operation of the vertical pixel brightness for the time series frame, and extracts the point where the absolute value of the brightness change rate exceeds a preset boundary threshold as a boundary candidate.

[0147] Here, the spatial derivative of vertical pixel brightness refers to an operation that identifies points along the vertical axis of the image where the brightness change between upper and lower pixels becomes abrupt, and the boundary threshold can be set as the minimum reference value for brightness changes caused by the contrast difference between the smoke layer and the clear air layer. Additionally, if the boundary of the smoke layer appears blurry, the system can be configured to perform differentiation after mitigating high-frequency noise by applying Gaussian smoothing or box smoothing; furthermore, considering cases where overall contrast is degraded due to dense smoke, the threshold may be set to an adaptive value based on the frame-by-frame average contrast or histogram distribution.

[0148] The step of deriving a two-dimensional visual smoke boundary involves connecting extracted pixel inflection points according to horizontal connectivity constraints and inter-frame continuity constraints to form candidate curves at the bottom of the smoke layer for each frame. Here, the horizontal connectivity constraint is a geometric rule requiring that adjacent inflection points be smoothly connected in the horizontal direction of the tunnel, and can be set to connect only when the horizontal distance and height difference between adjacent inflection points are within an allowable range. Additionally, the inter-frame continuity constraint is a rule reflecting physical consistency that the smoke layer boundary does not abruptly jump or disappear over time, and can be implemented by limiting the amount of change in the boundary position of the next frame based on the boundary position of the previous frame. Furthermore, considering cases where local boundary contamination occurs due to vehicle headlights, sign reflections, dust, etc., at least one of the minimum connection length condition, discontinuous section interpolation condition, and curvature-based smoothing condition can be applied to generate a highly reliable two-dimensional visual smoke boundary.

[0149] The step of calculating the elevation of the bottom of the visual smoke is to calculate the elevation component of the bottom of the smoke layer by converting the derived two-dimensional visual smoke boundary line into a three-dimensional tunnel space coordinate system through a camera back-projection model.

[0150] Here, the camera back-projection model refers to a model that inversely calculates pixel coordinates of a 2D image into 3D spatial coordinates by reflecting parameters such as the camera's installation elevation, tilt angle, and focal length. It can be implemented by converting pixel coordinates into rays in the tunnel coordinate system and then calculating the point where the ray intersects the tunnel sidewall plane or a reference cross-section. Additionally, if the same boundary line is sampled multiple times in the horizontal direction within the frame, the average or median value of the sampled elevation values ​​is adopted as the visual smoke bottom elevation, and the model can be configured to perform back-projection after applying distortion correction parameters to reduce the effects of lens distortion or perspective distortion.

[0151] The step of generating a vertical distribution profile involves interpolating smoke concentration values ​​at different installation altitudes obtained from a group of smoke sensors arranged in multiple stages on the tunnel sidewalls according to altitude, thereby generating a vertical distribution profile that expresses the change in smoke concentration according to altitude in the form of a continuous function. Here, the vertical distribution profile can be defined as data representing the change in smoke concentration according to the height from the tunnel floor to the ceiling as a single curve function, and can be generated using at least one of linear interpolation, segmental spline interpolation, or interpolation methods with added monotonicity constraints. Additionally, considering sensor-specific offsets or communication delays, the system may be configured to accept only concentration values ​​aligned by time based on the same analysis time as inputs, and to exclude sensor channels where outliers are detected or to apply mitigation weights.

[0152] The step of identifying the physical sensing altitude is to determine the altitude at which the light attenuation rate reaches a reference threshold corresponding to the limit visibility distance for human evacuation, based on the vertical distribution profile, as the physical sensing altitude.

[0153] Here, the light attenuation rate is an indicator representing the degree to which visibility is reduced by smoke concentration, and can be derived by mapping the concentration value or the rate of change of concentration of the smoke sensor to a pre-set conversion table, and the reference threshold can be pre-set to correspond to the minimum visibility distance standard for ensuring evacuation safety. Accordingly, the altitude that first satisfies the reference threshold on the vertical distribution profile or the lower boundary altitude of the section exceeding the reference threshold can be configured to be identified as the physical sensing altitude, and the physical sensing altitude can be used as a direct measurement-based altitude value contrasted with the camera-based altitude.

[0154] The step of updating the reliability weights is to calculate a smoke concentration index from the smoke concentration feature of the smoke sensor group, and to update the first reliability weight and the second reliability weight so as to exponentially decrease the first reliability weight assigned to the visual smoke bottom height and relatively increase the second reliability weight assigned to the physical sensing height as the smoke concentration index increases.

[0155] Here, the smoke concentration index can be calculated by combining at least one of the average or maximum value of smoke concentration by altitude, the rate of increase of smoke concentration, and the spatial variance of smoke concentration. If the smoke concentration index exceeds a certain standard, it is determined that the visibility of the camera image is reduced and the possibility of boundary misjudgment increases, so the first reliability weight can be configured to rapidly attenuate. Additionally, the two weights can be normalized so that their sum becomes 1 so that they can be directly used in subsequent combinations, or minimum and maximum limits can be applied to prevent excessive biased combinations from occurring.

[0156] The step of deriving the corrected smoke height is to calculate the corrected smoke height as the optimal smoke bottom height by cross-weighting the visual smoke bottom height and the physical sensing height based on the updated first reliability weight and second reliability weight. For example, the first reliability weight may be dominant to prioritize image-based boundaries under normal conditions, but the second reliability weight may become dominant to prioritize sensor-based height when the smoke concentration index rises due to dense smoke. Additionally, the method may be configured to ensure the stability of the control input by additionally applying a smoothing filter to the corrected smoke height of the previous time point so that the calculated corrected smoke height does not change abruptly between frames.

[0157] The step of calculating the base smoke layer descent thickness involves subtracting the corrected smoke elevation from the highest ceiling height of the road tunnel to determine how much the bottom of the smoke layer has descended from the ceiling, thereby calculating the base smoke layer descent thickness. Here, the highest ceiling height can be retrieved from tunnel cross-section design values ​​or a ceiling height table by section identifier, and in sections where slope or cross-section variation exists, the local ceiling height of the section mapped to the fire detection location can be used. Additionally, if the corrected smoke elevation is expressed based on road surface elevation, the system can be configured to perform the subtraction calculation after converting to the same coordinate system to maintain alignment with the road surface elevation reference point.

[0158] The step of determining the real-time descent depth involves calculating a thermal expansion fluctuation correction value proportional to the time-series temperature rise rate of the tunnel ceiling and adding this thermal expansion fluctuation correction value to the base smoke layer descent thickness to determine the final real-time descent depth. For example, the thermal expansion fluctuation correction value can be set to reflect the fact that buoyancy and changes in flow structure caused by high-temperature flow increase as the temperature rise rate measured by the ceiling temperature sensor increases; furthermore, to prevent the real-time descent depth from increasing unrealistically due to excessive correction, the system may be configured to apply an upper limit to the correction value or apply different correction coefficients for each fire intensity indicator section.

[0159] According to one embodiment, the method comprises: a step of collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; a step of generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; a step of mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; a step of generating a fire control package including a control command for each of the group of equipment to be controlled based on the group of equipment to be controlled and target operating parameters for each equipment; and a step of transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of determining the group of equipment to be controlled and target operating parameters for each equipment includes: a step of loading a previously stored equipment layout table corresponding to the fire detection location and the section identifier; Based on the above-mentioned equipment layout table and fire detection location, the method may include: a step of filtering candidate equipment among firefighting equipment according to the equipment separation distance between the fire detection location and the installation location information of the firefighting equipment, whether they are in the same section, and whether they are in adjacent sections, and determining the candidate equipment as a group of equipment to be controlled; and for each firefighting equipment included in the group of equipment to be controlled, a step of querying target operating parameters corresponding to a section identifier and a fire intensity index from the equipment control profile, and performing unit conversion and range restriction so that the target operating parameters conform to the parameter schema for each equipment type to determine the target operating parameters for each equipment.

[0160] The step of loading the above-mentioned equipment layout table is to load a previously stored equipment layout table into memory corresponding to the fire detection location and the section identifier, thereby preparing a set of equipment metadata that can be used for filtering candidate equipment and calculating separation distances thereafter.

[0161] Here, the equipment layout table may be configured to include an equipment identifier for each fire fighting equipment, an equipment type code, an installation section identifier, installation location information, and an equipment controller identifier, and the installation location information may be expressed as at least one of the tunnel longitudinal coordinates (chain ridge), a transverse offset, and an installation elevation. Additionally, the equipment layout table may be configured to further include operational status metadata such as whether each equipment is normally available, a fault flag, whether it is in inspection mode, and the recent response delay, so that it can be used as a basis for automatically excluding unavailable equipment or selecting replacement equipment in a subsequent step.

[0162] The step of determining the above candidate equipment as a group of equipment to be controlled involves filtering candidate equipment among firefighting equipment based on the equipment placement table and fire detection location, the equipment separation distance between the fire detection location and the installation location information of the firefighting equipment, whether they are in the same section, and whether they are in an adjacent section, and confirming the filtered candidate equipment as a group of equipment to be controlled. For example, the equipment separation distance can be calculated as the difference between the longitudinal coordinates of the fire detection location and the longitudinal coordinates of the candidate equipment as a basic component, and if necessary, corrected to an effective separation distance by additionally reflecting a lateral offset and the difference in installation elevation. Additionally, whether they are in the same section can be determined by whether the installation section identifier of the candidate equipment matches the section identifier mapped to the fire detection location, and whether they are in an adjacent section can be determined by whether the section identifier and the candidate equipment section are directly connected in a section adjacency relationship table or a section graph. Accordingly, equipment in the same section is adopted as a priority candidate, equipment in the adjacent section is adopted as an additional candidate only if the fire intensity index is above a certain level or if the number of available equipment in the same section is insufficient, and equipment whose effective separation distance exceeds a preset maximum allowable distance is excluded from the candidates. Furthermore, by configuring the candidate equipment filtering results to record both the equipment-specific selection reason code and the exclusion reason code, traceability of control decisions can be ensured.

[0163] The step of determining the target operating parameters for each facility is to query the target operating parameters corresponding to the section identifier and fire intensity index from the facility control profile for each firefighting facility included in the control target facility group, and to determine the target operating parameters for each facility by performing unit conversion and range restriction so that the queried target operating parameters conform to the parameter schema for each facility type.

[0164] Here, the equipment control profile may be implemented in the form of a lookup table that provides a set of target operating parameters using at least one of a section identifier, a fire intensity indicator section, and an equipment type code as a key, and the target operating parameters may be configured to include at least one of rotation direction, rotation speed, target flow rate, target pressure, opening rate, operation holding time, and start delay time depending on the equipment type. In addition, the parameter schema for each equipment type defines the parameter items, units, types of values, and minimum and maximum value ranges allowed for each equipment, for example, the rotation speed of a jet fan may be limited to a value in units of revolutions per minute, the opening rate of a damper may be limited to a value in units of percentage, and the rotation direction may be limited to one of the pre-set enumerated values.

[0165] Therefore, if parameters retrieved from the equipment control profile are stored in different units, unit conversion is performed by referring to the equipment-specific conversion rule table; if the converted value exceeds the allowable range of the schema, it can be configured to be finalized by truncating it to a lower or upper limit. For example, if the opening rate is defined in percentage units from 0 to 100 but the input specification of the field controller is a digital value from 0 to 4095, the system can be configured to convert the percentage value to a digital value and then truncate it to within the allowable range of the schema to finalize the applicable value. Furthermore, by adding and storing the application time, application step number, and profile version information to the finalized equipment-specific target operating parameters, it is possible to clearly trace which profile was applied with which value during replay and subsequent audits of the same event.

[0166] Additionally, the detection sensors include: a smoke sensor, a gas sensor, a temperature sensor, a flame detection sensor, a camera, and an airflow measurement sensor, and the step of determining the candidate equipment as a group of control target equipment comprises: a step of calculating a tunnel longitudinal airflow velocity component by projecting the airflow velocity vector included in the sensor data onto the longitudinal unit vector of the road tunnel, and defining upstream and downstream directions according to the sign of the airflow velocity component; a step of generating a set of deflection parameters reflecting the deflected diffusion of fire heat and combustion gases based on the longitudinal gradient angle of the tunnel longitudinal profile table corresponding to the section identifier mapped to the fire detection location and the tunnel longitudinal airflow velocity component; a step of setting the fire detection location as the center point of a three-dimensional spatial coordinate system and modeling a dynamic spatial influence field in the form of an asymmetric ellipsoid in which the major axis expands in the downstream direction and the major axis contracts in the upstream direction based on the set of deflection parameters; a step of classifying the candidate equipment filtered from the equipment placement table into fire extinguishing equipment, smoke exhaust equipment, and evacuation guidance equipment according to the equipment type, and mapping the three-dimensional installation coordinates of each classified candidate equipment to the dynamic spatial influence field. For each of the above candidate facilities, a step of calculating a spatial association index by applying a distance weight inversely proportional to the distance from the fire detection location and a type weight corresponding to the operational purpose for each facility type; and a step of determining only the candidate facilities whose spatial association index exceeds a preset activation threshold value as the group of facilities to be controlled.

[0167] The step of calculating the longitudinal airflow velocity component of the tunnel and defining the upstream and downstream directions is to derive the longitudinal airflow velocity component by projecting the airflow velocity vector provided by the airflow measurement sensor onto the longitudinal unit vector of the road tunnel, and to consistently define the upstream and downstream directions according to the sign of the longitudinal airflow velocity component.

[0168] Here, the longitudinal unit vector is a reference direction vector representing the tunnel extension direction, which can be normalized according to the tunnel's direction of progress or a reference direction designated by the manager, and projection refers to an operation that extracts only the longitudinal component by decomposing the airflow velocity vector in the direction of the said longitudinal unit vector. For example, if the longitudinal airflow velocity component is positive, the corresponding direction is defined as the downstream direction, and if it is negative, the corresponding direction is defined as the upstream direction; however, the definition result can be configured to be fixedly stored based on an event identifier so that the upstream / downstream definition does not change within the same event even if the reference direction changes during tunnel operation.

[0169] The step of generating a set of deflection parameters is to obtain a longitudinal gradient angle from a tunnel longitudinal profile table corresponding to a section identifier mapped to a fire detection location, and to generate a set of deflection parameters reflecting the deflection diffusion of fire heat and combustion gases based on the obtained longitudinal gradient angle and the tunnel longitudinal airflow velocity component.

[0170] Here, the tunnel longitudinal profile table may include at least one of the longitudinal gradient angle, section length, elevation reference point, and gradient direction for each section identifier, and the longitudinal gradient angle may be used to determine whether the gradient is uphill or downhill in the downstream direction. Additionally, the deflection parameter set may be configured to include at least one of a downstream amplification factor that amplifies diffusion in the downstream direction, an upstream damping factor that dampens diffusion in the upstream direction, a buoyancy enhancement factor according to the longitudinal gradient, and a transport control factor according to the magnitude of the airflow velocity component, and said coefficients may be determined by querying a pre-set rule table or by selecting a coefficient combination corresponding to the segmentation result of the airflow velocity component and the longitudinal gradient angle. Accordingly, the deflection diffusion characteristic in which the fire heat column tilts in a specific direction when the tunnel slope and airflow conditions are combined can be quantified in the form of parameters.

[0171] The step of modeling the dynamic spatial influence field is to set the fire detection location as the center point of a three-dimensional spatial coordinate system and to model the dynamic spatial influence field in the form of an asymmetric ellipsoid in which the major axis is extended downstream and the major axis is contracted upstream based on the set of bias parameters.

[0172] Here, the dynamic spatial influence field can be defined as a spatial range around a fire detection location where the need for firefighting equipment control is relatively high, expressed in the form of a weighted field or a probability field, and the asymmetric ellipsoid can refer to an ellipsoid in which the longitudinal axis length is set differently for the upstream and downstream directions. For example, the longitudinal radius of the dynamic spatial influence field is set to increase in proportion to the airflow velocity component and the downstream amplification factor in the downstream direction, and to decrease in the upstream direction by the upstream damping factor; the transverse radius and the elevation radius can be configured to be corrected according to at least one of the tunnel cross-sectional width, ceiling height, turbulence intensity due to vehicle traffic, or fire intensity indicator section. Additionally, since the influence field modeling results can be updated over time for the same event, it can be configured to ensure the reproducibility of subsequent control decisions by storing snapshots of the influence field parameters at each update time.

[0173] The step of classifying candidate facilities and mapping them to a dynamic spatial influence field is to classify candidate facilities filtered from a facility layout table into fire extinguishing facilities, smoke exhaust facilities, and evacuation guidance facilities according to facility type, and to map the 3D installation coordinates of each classified candidate facility to the dynamic spatial influence field coordinate system.

[0174] Here, fire extinguishing equipment may include at least one of an indoor fire hydrant, a sprinkler, a pressure pump, and a water discharge valve; smoke exhaust equipment may include at least one of a jet fan, an exhaust fan, a damper, and a ventilation opening; and evacuation guidance equipment may include at least one of an emergency light, an evacuation guidance sign, and an emergency broadcasting device. Additionally, the 3D installation coordinates may be expressed as at least one of a change-based longitudinal coordinate, a transverse offset relative to the tunnel centerline, and an installation elevation. If the coordinates are stored in different reference coordinate systems, they may be configured to be unified into the influence field coordinate system by referring to a coordinate transformation mapping table and then mapped. Accordingly, whether each candidate facility is included within the risk influence field or its relative proximity to the center point of the influence field may be provided as an input for the subsequent calculation of the correlation.

[0175] The step of calculating the spatial correlation index is to calculate the spatial correlation index by applying a distance weight inversely proportional to the distance from the fire detection location to each candidate facility and a type weight corresponding to the operational purpose of each facility type.

[0176] Here, the facility separation distance can be calculated by combining at least some of the longitudinal distance component, lateral distance component, and elevation difference component between the fire detection location and the candidate facility installation coordinates. To reflect the asymmetry of the dynamic spatial influence field, the system may be configured to use an effective separation distance with an upstream damping coefficient applied to the upstream candidate facility and an effective separation distance with a downstream amplification coefficient applied to the downstream candidate facility. Additionally, the distance weight is set to increase as the effective separation distance decreases; however, if the effective separation distance is less than a preset minimum distance, the minimum distance is applied as a lower limit to prevent an excessive surge in weights, and if the effective separation distance exceeds a preset maximum distance, the distance weight is set to 0 to naturally exclude facilities outside the influence zone. The type weight can be assigned differently depending on the operational purpose of each facility type; for example, a type weight table may be configured such that a relatively high weight for fire extinguishing purposes is assigned to fire extinguishing facilities near the fire detection location, a relatively high weight for smoke induction purposes is assigned to smoke exhaust facilities in the downstream direction, and a relatively high weight for guidance purposes is assigned to evacuation guidance facilities on the evacuation path. The spatial association index is calculated as a result of combining the distance weights and type weights mentioned above, and can be configured to be normalized to a range between 0 and 1 so that subsequent threshold comparisons are possible.

[0177] The step of determining the group of facilities to be controlled is to determine only the candidate facilities whose spatial association index exceeds a preset activation threshold as the group of facilities to be controlled.

[0178] Here, the activation threshold value may be set differently depending on at least one of the fire intensity indicator range, tunnel operation policy, minimum control range by equipment type, or accident response manual. Furthermore, to prevent excessive simultaneous control within the same equipment type, it may be configured to select only the top K items per equipment type or to additionally apply minimum separation constraints between equipment. Additionally, by configuring the confirmed control target equipment group result to record candidate equipment identifiers, spatial association indices, applied distance and type weights, activation threshold values, and impact field parameter snapshots together, traceability can be ensured for reference during the subsequent fire control package generation stage and for post-audits.

[0179] And, the step of calculating the spatial correlation index comprises: for each of the candidate facilities, a step of calculating the longitudinal distance component and the lateral distance component between the 3D coordinates of the fire detection location and the 3D installation coordinates of the candidate facility; a step of determining whether the candidate facility is positioned upstream or downstream relative to the fire detection location based on the upstream and downstream directions, and selecting an upstream attenuation coefficient or a downstream amplification coefficient according to the determination result; a step of determining an upward / downward slope correction coefficient according to the longitudinal gradient based on the longitudinal gradient angle, and correcting the slope correction coefficient based on the difference in elevation between the installation height of the candidate facility and the fire detection location; and a step of calculating the effective separation distance for each candidate facility based on the longitudinal distance component, lateral distance component, and slope correction coefficient, and the upstream attenuation coefficient or downstream amplification coefficient. The method may include the step of calculating a distance weight inversely proportional to the effective separation distance, wherein if the effective separation distance is less than a preset minimum distance, the minimum distance is applied as a lower limit value, and if the effective separation distance exceeds a preset maximum distance, the distance weight is set to 0; the step of querying a type weight corresponding to each of the fire extinguishing equipment, smoke exhaust equipment, and evacuation guidance equipment from a type weight table preset to correspond to the operating purpose of each equipment type; and the step of calculating a spatial association index for each candidate equipment normalized to a value between 0 and 1 by multiplying the distance weight and the type weight.

[0180] The step of calculating the longitudinal distance component and the transverse distance component is, for each of the candidate facilities, aligning the 3D coordinates of the fire detection location and the 3D installation coordinates of the candidate facilities into the same coordinate system, and then decomposing the distance difference between the two coordinates into a longitudinal component and a transverse component suitable for the tunnel structure.

[0181] Here, 3D coordinates can be defined as coordinates including tunnel longitudinal coordinates (e.g., changeridge), a transverse offset relative to the tunnel centerline, and installation elevation. If the installation coordinates of the candidate equipment are stored in a local coordinate system for each section, they can be configured to be converted to the fire detection location coordinate system by referring to a section identifier-based coordinate transformation table and then calculated. Additionally, since the longitudinal distance component represents the separation in the tunnel extension direction between the fire detection location and the candidate equipment, and the transverse distance component represents the separation in the tunnel width direction, they can be used as inputs to determine geometric relationships, such as whether the equipment is adjacent to the same lane as the fire source or adjacent to the opposite side wall.

[0182] The above longitudinal distance component can be calculated as the absolute value of the difference between the longitudinal coordinates of the fire detection location and the longitudinal coordinates of the candidate facility, or as a sign value including directionality, and the above transverse distance component can be calculated as the absolute value of the offset difference in the tunnel width direction. In addition, if the tunnel cross-section includes a curve or widening section, the transverse distance component can be converted into a normalization ratio with respect to the cross-section width of the section rather than a simple offset difference to ensure the possibility of comparison between different sections, and if necessary, the installation elevation difference component can be processed separately to be reflected in the longitudinal gradient correction step.

[0183] The step of determining whether the candidate facility is positioned upstream or downstream and selecting an upstream attenuation coefficient or a downstream amplification coefficient involves determining, based on the previously defined upstream and downstream directions, whether the longitudinal coordinates of the candidate facility are upstream or downstream of the fire detection location, and selecting a directional coefficient that reflects the asymmetry of risk diffusion caused by airflow according to the determination result. For example, if the candidate facility is located downstream, the downstream amplification coefficient may be selected by reflecting the dominant direction of combustion gas diffusion, and if the candidate facility is located upstream, the upstream attenuation coefficient may be selected by reflecting the relatively low probability of diffusion against the wind.

[0184] The above direction coefficient can be managed as a rule table selected stepwise according to at least one of the magnitude of the tunnel longitudinal airflow velocity component, the fire intensity indicator section, and the longitudinal gradient angle section, and can be fixedly stored linked to an event identifier so that the upstream / downstream judgment criteria and the coefficient selection results do not change within the same event. In addition, if the upstream / downstream judgment is unstable because the candidate equipment is located near the same longitudinal coordinates as the fire detection location, a neutral coefficient is applied by placing a pre-set neutral section, or if the absolute value of the longitudinal distance component is less than the minimum judgment distance, it can be treated as equipment in the same section and configured to apply a separate priority rule.

[0185] The step of determining upward / downward slope correction coefficients based on the longitudinal gradient and correcting them based on the elevation difference involves determining the slope correction coefficient to reflect the phenomenon where the buoyancy behavior of smoke and heat is reinforced along the slope—that is, the chimney effect—based on the longitudinal gradient angle, and further correcting the slope correction coefficient by reflecting the elevation difference between the installation elevation of the candidate equipment and the fire detection location. For example, if the upstream direction has an upward gradient, the slope correction coefficient applied to the upstream candidate equipment may be increased because the risk on the upstream side may increase due to backflow and upward buoyancy influences; conversely, if the downstream direction has an upward gradient, the downstream correction coefficient may be increased because it is determined that downstream diffusion is more favorable.

[0186] The above altitude difference-based correction can be performed by increasing the correction coefficient to reflect the increased possibility of heat exposure and smoke retention above the smoke layer when the candidate equipment is installed at an altitude higher than the fire detection location, and decreasing the correction coefficient when the candidate equipment is installed at a lower altitude. In this case, the altitude difference can be configured to be converted into an altitude normalized with respect to the tunnel design road surface altitude reference point or the altitude reference point for each section, and then compared; furthermore, minimum and maximum values ​​can be assigned to the slope correction coefficient to prevent subsequent distance weights from becoming unstable due to excessive increases or decreases in the slope correction coefficient.

[0187] The step of calculating the effective separation distance for each candidate facility is a step of calculating the effective separation distance for each candidate facility based on the longitudinal distance component, transverse distance component, and slope correction coefficient, and the upstream attenuation coefficient or downstream amplification coefficient, thereby calculating a perceived distance that reflects the physical influence of the fire rather than a simple straight-line distance. For example, the effective separation distance can be adjusted by synthesizing the longitudinal distance component and the transverse distance component according to a pre-set combination rule, then weighting and correcting with a slope correction coefficient, so that the effective separation distance for downstream facilities is relatively reduced by the amplification coefficient, and the effective separation distance for upstream facilities is relatively increased by the attenuation coefficient.

[0188] The above coupling rule may be defined as a weighted coupling method in which the longitudinal distance component is the main component, but the influence of the transverse component is adjusted according to tunnel cross-sectional characteristics or equipment type. For example, it may be configured to reflect a large weight for the longitudinal component in the case of smoke exhaust equipment, and a relatively large weight for the transverse component and altitude-related correction in the case of evacuation guidance equipment. In addition, it may be configured to apply a lower limit to the transverse component or the slope correction coefficient to prevent the result of calculating the effective separation distance from becoming abnormally small.

[0189] The step of calculating distance weights is to calculate distance weights for each candidate facility inversely proportional to the effective separation distance, wherein if the effective separation distance is less than a preset minimum distance, the minimum distance is applied as a lower limit value, and if the effective separation distance exceeds a preset maximum distance, the distance weight is set to 0. Here, the reason for applying the minimum distance as a lower limit value is to prevent the weight from diverging excessively when the effective separation distance approaches 0, and the reason for setting the weight to 0 when the maximum distance is exceeded is to suppress unnecessary control loads by effectively excluding facilities outside the fire influence zone from control candidates.

[0190] The above minimum and maximum distances may be set differently depending on the type of equipment; for example, since fire extinguishing equipment requires control centered on the proximity section, the maximum distance may be configured to be relatively small, while since smoke exhaust equipment has a downstream long-distance influence zone, the maximum distance may be configured to be relatively large. Additionally, to suppress sudden changes in distance weights when the effective separation distance fluctuates over time of the same event, a smoothing filter or a section-by-section step value snapping rule may be applied.

[0191] The step of querying type weights is to query type weights corresponding to fire extinguishing equipment, smoke exhaust equipment, and evacuation guidance equipment, respectively, from a pre-configured type weight table set to correspond to the operational purpose of each equipment type. Here, the type weight can be defined as a strategic importance score assigned differentially to each equipment group depending on whether fire suppression, smoke exhaust, or human evacuation guidance is prioritized, and the type weight table can be configured to be provided with at least one of the fire intensity indicator section, upstream / downstream direction, and priority of the protected section as a key.

[0192] The step of calculating the spatial association index for each candidate facility is to derive a raw association value for each candidate facility by multiplying the distance weight and the type weight, and then normalize the raw association value so that it can be compared within a range between 0 and 1 to calculate the spatial association index for each candidate facility. For example, normalization can be performed by scaling by a relative ratio based on the maximum raw association value within the set of candidate facilities, or by mapping to a normalization function having a pre-set saturation interval, and can be configured so that the closer the spatial association index is to 1, the more the facility is a core facility that has a high need for operation in the current fire situation.

[0193] By configuring the spatial association index calculation results to record the longitudinal distance component, transverse distance component, selected direction coefficient type, slope correction coefficient, effective separation distance, distance weight, type weight, and normalization reference value used in the calculation for each candidate facility identifier, it is possible to ensure reference for the basis in the subsequent control target facility group determination stage and traceability in post-audits.

[0194] According to one embodiment, the method comprises: a step of collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; a step of generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; a step of mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; a step of generating a fire control package including a control command for each of the group of equipment to be controlled and target operating parameters for each equipment based on the group of equipment to be controlled and the target operating parameters for each equipment; and a step of transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of generating the fire control package includes: a step of loading an evacuation path graph that defines evacuation nodes including evacuable exits, evacuation connecting passages, and emergency exits within the road tunnel, and the connection relationships between the evacuation nodes based on the fire detection location, fire intensity index, and section identifier. Based on the above evacuation path graph and fire detection location, the method may include the step of updating the evacuation path graph to exclude risk nodes and risk links included in the fire impact zone and the smoke spread expected zone, and calculating an optimal evacuation path for each starting node in the updated evacuation path graph; and based on the optimal evacuation path, the method may include the step of generating a visual guidance control command including an evacuation direction arrow, evacuation distance, exit identification information, and a warning message on a display unit included in the evacuation guidance device.

[0195] The step of loading the evacuation path graph is to load the evacuation path graph, which includes evacuation nodes serving as units for evacuation judgment within the road tunnel and connection relationships between said evacuation nodes, from a storage unit or database and load it into memory based on the fire detection location, the fire intensity indicator, and the section identifier.

[0196] Here, an evacuation node may be defined as a node including at least one of an evacuation-possible exit, an evacuation connecting passage, and an emergency exit within a road tunnel, and each evacuation node may be configured to include at least one attribute among a node identifier, an installation section identifier, tunnel longitudinal coordinates, a transverse offset, an elevation, a node type, a physical maximum capacity per node, and an identifier of an evacuation guidance device associated with the node. Additionally, the connection relationship is modeled as a link representing a passage section movable between evacuation nodes, and may be configured to include at least one of a link identifier, a starting node, an ending node, a link length, a link width, a longitudinal gradient angle of the link, a traversable direction within the link, and a basic traversal weight per link. In this case, the evacuation path graph is data representing the connection status of passages that evacuees can actually move through in the form of a mathematical network, and can be used as virtual map data to exclude dangerous sections and search for a safe escape route in a subsequent step.

[0197] The step of updating the evacuation path graph and calculating the optimal evacuation path for each starting node is to derive the fire impact zone and the smoke spread prediction zone based on the fire detection location, then update the evacuation path graph by determining evacuation nodes and connecting links included in the fire impact zone or the smoke spread prediction zone as dangerous nodes and dangerous links and excluding them from graph search, and to calculate the optimal evacuation path for each starting node on the updated evacuation path graph. For example, the fire impact zone may be determined according to a zone identifier mapped to the fire detection location and a pre-set influence radius rule, and the smoke spread prediction zone may be determined by reflecting the dominant spread direction and spread range indicated by the tunnel longitudinal airflow direction, longitudinal gradient, and fire intensity indicator. In addition, the exclusion of risk nodes and risk links may be performed by at least one of the following methods: removing the corresponding node or link from the graph, setting the traffic weight of the corresponding node or link to a preset infinite value, or applying a penalty to the traffic weight in proportion to the risk level; and the update result may be configured to record the reason for the risk assessment, the time of application, and the version information of the update rule used. The optimal evacuation path may be calculated by searching such that the cumulative traffic weight from the starting node to the arrival candidate node classified as a safe node is minimized; and in cases where multiple paths with the same cost exist, the path may be determined by applying at least one of the node capacity, link width, or detour distance as an additional criterion, thereby preventing the generation of a path that leads evacuees into a risk area.

[0198] The step of generating a visual guidance control command is to generate visual guidance information in the form of a control command, including at least one of an evacuation direction arrow, evacuation distance, exit identification information, and a warning message, on a display unit included in an evacuation guidance device based on the optimal evacuation path for each starting node. For example, the evacuation direction arrow may be determined to indicate the direction of the next moving node of the optimal evacuation path from the current node or link on the graph where the location of the evacuation guidance device is installed is mapped, the evacuation distance may be calculated by accumulating the link length from the current location node to the arrival node along the optimal evacuation path, and the exit identification information may consist of at least one of an emergency exit number, an evacuation connecting passage number, or an exit zone name based on the type and identifier of the arrival node. Additionally, the warning message may be generated in a form such as “No Passage,” “Detour Evacuation,” or “Caution: Smoke Spread Section,” depending on at least one of the result of excluding a dangerous node or dangerous link, whether a detour is necessary, and the risk level of the current section. The above visual guidance control command may be packaged to include at least one of a device identifier, an application time, a priority, a command validity period, and a display update cycle, and may be configured to include a fire event identifier and command version information to prevent conflicts with existing guidance commands when a path is updated for the same fire event, and accordingly, the evacuation guidance display board or wall indicator may be controlled to output the calculated path information as visual information that evacuees can intuitively recognize.

[0199] Additionally, the detection sensor comprises: a camera, a smoke sensor, a gas sensor, and an airflow measurement sensor, and the step of calculating the optimal evacuation path for each starting node comprises: a step of calculating the Available Safe Egress Time (ASET) for each node and link, defined as the time until the toxic gas and smoke layer spreading from the fire detection location reaches each evacuation node and connecting link on the evacuation path graph to create a survival limit environment, based on time-series smoke concentration data included in the sensor data and longitudinal airflow velocity components within the road tunnel; a step of decelerating and correcting the average walking speed of evacuees based on section-by-section evacuee crowd density data extracted from images captured by the camera and a preset section-by-section longitudinal gradient angle of the tunnel floor, and calculating the Required Safe Egress Time (RSET) for each node and link, which is the cumulative required evacuation time required to pass through each connecting link from the starting node based on the decelerating and corrected average walking speed. A step of updating the evacuation path graph such that, for each individual connection link of the evacuation path graph, if it is determined that the required evacuation time (RSET) for a specific starting node exceeds the limit arrival time (ASET), the corresponding connection link is determined to be a dangerous link, and the traffic weight of the dangerous link is set to a preset value of infinity to be excluded from graph search; and a step of, for valid connection links where the required evacuation time (RSET) is within the limit arrival time (ASET), calculating a bottleneck penalty index based on the crowd density data and performing a dynamic weight update to reflect the bottleneck penalty index in the traffic weight of the valid connection links.and, for the updated evacuation path graph reflecting the above bottleneck penalty index, calculate the expected inflow traffic for each evacuation path node based on the physical maximum capacity pre-set for each evacuation path node, and if there is an evacuation path node where the inflow traffic exceeds the physical maximum capacity, apply a load balancing-based time-series routing algorithm to detour the optimal evacuation path of the evacuee corresponding to the lower-priority departure node to the next-priority evacuation path node or the tunnel exit in the upstream direction of the airflow to finally determine the optimal evacuation path; may be included.;

[0200] The above detection sensor is a group of sensors configured to include a camera, a smoke sensor, a gas sensor, and an airflow measurement sensor, for multi-faceted verification of the safety of the evacuation route and real-time estimation of the physical spread of fire and the evacuation environment. Here, the camera provides visual conditions such as evacuees and vehicle objects within the tunnel, crowd distribution, and smoke diffusion patterns; the smoke sensor provides smoke concentration and concentration rise rates by section or altitude; and the gas sensor may provide at least one of carbon monoxide, carbon dioxide, oxygen concentration, or toxic gas indicators. Additionally, the airflow measurement sensor provides an airflow velocity vector including wind direction and wind speed, and the airflow velocity vector is converted into a longitudinal airflow velocity component through projection onto a longitudinal unit vector of the tunnel, which can be used as an input to predict the dominant diffusion direction and arrival time of smoke and toxic gases.

[0201] The step of calculating the optimal evacuation path for each starting node involves calculating the time to reach the limit and the required evacuation time for each node and link on the evacuation path graph, respectively; excluding dangerous links or updating traffic weights based on the comparison of the two times; and then determining the optimal path for each starting node by performing load balancing so as not to exceed the capacity of the evacuation connecting passages. Here, the starting node corresponds to a node at the starting point where evacuees may be present within the tunnel, the link corresponds to the actual passage section between the starting node and the arrival node, and the optimal evacuation path can be defined as the path searched to minimize the accumulated traffic weight.

[0202] The step of calculating the time to reach the limit by node and link is to predict, by node and link, the time until the toxic gas and smoke layer spreading from the fire detection location reach each node and link and create a survival limit environment, based on the time-series smoke concentration data included in the sensor data and the longitudinal airflow velocity component within the road tunnel. Here, the time to reach the limit may be defined as the remaining time until an environment is first formed at a specific node or link where normal walking and breathing of an evacuee become difficult, and the survival limit environment may be defined as a state in which at least one of reduced visibility, accumulation of toxic gas, or loss of breathing height due to the descent of the smoke layer exceeds a critical level.

[0203] Time-series smoke concentration data used to calculate the time to reach the limit can be aligned with concentration values ​​collected from a smoke sensor along a time axis, and then smoothed using at least one of a moving average, a median filter, or a pre-set time window-based interval average to suppress sensor noise and instantaneous spikes. Additionally, features such as the start time of the concentration rise interval, the rise rate, and the time to reach the peak can be extracted, and said features can be used as a basis for determining whether smoke diffusion is actually in progress and whether the diffusion speed is accelerating.

[0204] In calculating the time to reach the limit, the longitudinal airflow velocity component serves as a value to provide the dominant direction and velocity of smoke and toxic gases; it can be combined with the longitudinal distance component from the fire detection location to each node or link to predict the time to reach. For example, if the longitudinal airflow velocity component is above a certain magnitude in the downstream direction, the estimated time to reach for downstream nodes and links is reduced; conversely, if the upstream component is dominant or a reverse airflow is formed, the estimated time to reach for upstream nodes and links is reduced. In this case, during sections where the airflow velocity component changes rapidly, the system can be configured to ensure the stability of the prediction by applying the average value over a recently set period or a conservative lower limit.

[0205] The step of calculating the required evacuation time by node and link comprises decelerating the average walking speed of evacuees based on the evacuee crowd density data by section extracted from the video captured by the camera and the pre-set longitudinal gradient angle of the tunnel floor by section, and calculating the cumulative required evacuation time by node and link required to pass through each connecting link from the starting node based on the decelerating-corrected average walking speed. Here, the required evacuation time can be defined as the cumulative travel time required to reach a specific link or a specific node from the starting node, and can be calculated as the cumulative sum of the passage times per link.

[0206] Crowd density data can be calculated by detecting evacuee objects in camera images and combining the number of evacuees within the interval of interest corresponding to the link with the effective width or effective area of ​​the link. For example, evacuee object detection can be performed by calculating a bounding box or segment mask using a person detection model, and inter-frame tracking-based deduplication, occlusion correction, or confidence threshold-based filtering can be applied to suppress cases where the number of evacuees is under- or over-counted. Additionally, crowd density can be calculated as an average value within a specific time window so that instantaneous crowd shaking is not excessively reflected in the required evacuation time.

[0207] The step of decelerating the average walking speed of evacuees is a step of adjusting the average walking speed by section by reflecting the crowd walking correlation, in which walking speed decreases as crowd density increases, and reflecting the effect of increased walking burden when the longitudinal gradient angle is an uphill or downhill gradient. For example, the base walking speed based on crowd density can be determined by querying a pre-set speed-density lookup table, and the deceleration correction based on the longitudinal gradient angle can be configured so that a larger deceleration factor is applied in an uphill gradient. Accordingly, the passage time per link is calculated based on the value obtained by dividing the link length by the deceleration-corrected average walking speed, and an additional delay term may be added in bottleneck sections where the link width narrows.

[0208] The step of updating the evacuation path graph is to determine, for individual connection links in the evacuation path graph, that if it is determined that the required evacuation time for a specific starting node exceeds the limit arrival time, the corresponding connection link is identified as a dangerous link, and the passage weight of the dangerous link is set to a preset value of infinity so that it is excluded from graph search.

[0209] Here, the override determination may be performed by comparing the value obtained by subtracting a preset safety margin time from the time to reach the limit with the required evacuation time, and the safety margin time may be dynamically adjusted based on fire intensity indicators or sensor reliability. Additionally, the time to reach the limit, required evacuation time, comparison time, and risk reason code used as the basis for determining a risk link may be configured to be recorded as link metadata.

[0210] The step of performing dynamic weight updates is to calculate a bottleneck penalty index based on crowd density data for valid connection links where the required evacuation time is within the limit arrival time, and to update the link weights so as to reflect the bottleneck penalty index in the traffic weights of the valid connection links.

[0211] Here, the bottleneck penalty index can be calculated to be proportional to at least one of the link width, crowd density within the link, the rate of change of density over time, or the inflow rate at the link entry point, and as the penalty index increases, the traffic weight of the corresponding link increases, thereby inducing the link to be preferentially avoided during graph navigation. Additionally, weight updates can be performed repeatedly at regular intervals to reflect the movement and alleviation of bottlenecks due to crowd movement in real time.

[0212] The step of finally determining the optimal evacuation path by applying a load-balancing-based time-series routing algorithm involves calculating the expected inflow traffic per node based on the physical maximum capacity of each evacuation connecting passage node for an updated evacuation path graph reflecting a bottleneck penalty index, and determining the optimal path by diverting the path of evacuees corresponding to lower-priority departure nodes to the next-priority evacuation connecting passage node or the tunnel exit in the upstream direction of the airflow if there exists an evacuation connecting passage node where the inflow traffic exceeds the physical maximum capacity.

[0213] Here, an evacuation connecting passage node can be defined as an evacuation node connecting to an opposite tunnel or evacuation zone, and the maximum capacity can be defined as the number of people who can pass through per unit time or the number of people who can stay at the same time. Additionally, the expected inflow traffic can be calculated by aggregating the expected number of evacuees per starting node and the distribution of arrival nodes assigned as a result of graph search, and lower-priority starting nodes can be determined based on at least one of the risk level of the starting node, the number of evacuees at the starting node, or the urgency of the required evacuation time at the starting node. Furthermore, time-series routing can be configured to simultaneously secure evacuation efficiency and safety margin while suppressing evacuation connecting passage overcrowding and secondary bottlenecks by updating sensor data and crowd data at preset update cycles and recalculating the path by re-evaluating risk links and bottleneck penalties.

[0214] And, the step of calculating the above Limit Attainment Time (ASET) by node and link comprises: a step of calculating the longitudinal movement speed of a ceiling jet spreading along the tunnel ceiling from a fire detection location based on the longitudinal airflow velocity component within the road tunnel included in the sensor data and the fire intensity index; a step of calculating the longitudinal distance component between the fire detection location and the upper ceiling coordinates of the evacuation node or connecting link for each evacuation node and connecting link included in the evacuation path graph, and predicting the ceiling arrival time, when smoke first reaches the upper ceiling of each evacuation node and connecting link, by evacuation node and connecting link based on the longitudinal distance component and the longitudinal movement speed; a step of calculating the volume expansion rate index and mass generation rate index of soot particles based on the slope and integral value in the concentration increase section for the time-series smoke concentration data; and a step of querying the ceiling height, reference elevation, effective cross-sectional area, and longitudinal gradient angle of the corresponding tunnel section from a previously stored tunnel cross-sectional profile table based on the section identifier mapped to the fire detection location. A step of calculating, by node and link, the smoke layer descent time required for the smoke layer reaching the ceiling to descend vertically to a preset reference altitude based on the above-mentioned volumetric expansion rate index, mass generation rate index, effective cross-sectional area, and longitudinal gradient angle; a step of calculating, by node and link, the toxic gas saturation time at which the evacuee's ability to function is determined to be lost by applying a Fractional Effective Dose model based on the component analysis results of the toxic gas concentration data included in the sensor data; and a step of calculating, by node and link, the limit visibility loss time, defined as the point at which visibility decreases below a preset limit visibility threshold, by applying a preset light attenuation coefficient to the soot concentration calculated from the time-series smoke concentration data.and may include the step of deriving the physical smoke settlement time by summing the ceiling arrival time and the smoke layer descent time, comparing the physical smoke settlement time, the toxic gas saturation time, and the limit visibility loss time to determine the shortest time as the Limit Attainment Time (ASET) as the point in time for establishing the survival limit environment of the corresponding evacuation node and connecting link, and mapping the determined Limit Attainment Time (ASET) to the nodes and links of the evacuation path graph.

[0215] The step of calculating the longitudinal movement speed of the ceiling jet airflow is a step of estimating the longitudinal movement speed of heat and smoke generated at the fire detection location as they rapidly spread horizontally along the tunnel ceiling, using the tunnel longitudinal airflow speed component included in the sensor data and the fire intensity index as inputs.

[0216] Here, the ceiling jet airflow refers to an airflow that spreads longitudinally along a thin layer beneath the ceiling after an updraft formed by fire heat collides with the ceiling, and can be defined as a parameter that determines the upper limit of the speed at which smoke leads propagate within the tunnel. For example, the device may be configured to divide the fire intensity index into multiple intensity zones, query a pre-set basic ceiling jet movement speed for each intensity zone in a lookup table, and then correct the query value upward if the longitudinal airflow velocity component is significantly formed in the downstream direction, or correct the query value downward or update the dominant movement direction if the longitudinal airflow velocity component is formed in the upstream direction (reverse airflow). Additionally, since the prediction of the time to reach the ceiling may become unstable if the longitudinal movement speed is excessively small and approaches zero, the device may be configured to ensure the stability of the velocity estimation by applying a pre-set minimum speed lower limit or using the average airflow velocity component of the most recent pre-set time window.

[0217] The step of calculating the longitudinal distance component and predicting the ceiling arrival time is, for each evacuation node and connecting link included in the evacuation path graph, calculating the longitudinal distance component between the fire detection location and the coordinates of the upper ceiling of the node or link, and predicting the ceiling arrival time for the smoke lead to first reach the upper ceiling for each node and link based on the longitudinal distance component and the longitudinal movement speed.

[0218] Here, the ceiling reach time can be defined as the time required for the smoke influence zone to extend from the fire origin to the ceiling of a specific evacuation point, and can be utilized as a primary risk indicator representing when each point on the evacuation path is incorporated into the smoke influence zone. The upper ceiling coordinates can be defined by combining the ceiling height of the section where the evacuation node is installed with the node's longitudinal coordinates. In the case of a connecting link, it can be configured to adopt the point with the minimum ceiling height among multiple sample points constituting the link, or a conservative value between the ceiling heights of the nodes at both ends of the link. For example, if a node exists 300 meters downstream from the fire detection location and the longitudinal movement speed is calculated to be 2 meters per second, the ceiling reach time for that node can be predicted to be approximately 150 seconds. If a reverse airflow is present and the dominant direction of movement is determined to be upstream, the reach time for the downstream node can be configured to increase or be marked as reach uncertain.

[0219] The step of calculating the volume expansion rate index and the mass generation rate index is to identify the concentration increase section for the time-series smoke concentration data and to quantify the volume expansion trend and mass generation trend of soot particles based on the concentration gradient and concentration integral value in the increase section.

[0220] Here, the concentration rise section can be defined as a section where the smoke concentration continuously increases relative to a preset baseline, the concentration gradient can be defined as the increase in smoke concentration per unit time, and the concentration integral can be defined as the total amount of smoke concentration accumulated over a certain period. Additionally, to prevent distortion of the gradient and integral values ​​by instantaneous spikes or sensor noise, the system may be configured to use smoothed concentration values ​​with a moving average, median filter, or interval average applied prior to them. Furthermore, the volumetric expansion rate index can be calculated by weighted combination centered on the gradient component, and the mass generation rate index by weighted combination centered on the integral component. Accordingly, not only the simple concentration level but also the developmental pattern of changes in the smoke volume filling the tunnel space can be reflected.

[0221] The step of querying the ceiling height, reference elevation, effective cross-sectional area, and longitudinal gradient angle from the tunnel cross-section profile table is a step of loading the geometric and structural parameters of the corresponding section by calling the previously stored tunnel cross-section profile table corresponding to the section identifier mapped to the fire detection location.

[0222] Here, ceiling height refers to the height from the road surface reference to the ceiling section, and reference altitude is a reference height used for determining evacuation safety or smoke layer descent, which may be defined as at least one of the breathable altitude or the design reference altitude. Effective cross-sectional area refers to the cross-sectional area through which smoke and airflow can substantially flow, and may be set as a value that reflects attenuation caused by obstacles, facilities, traffic congestion, etc. Longitudinal gradient angle refers to the slope of the tunnel floor surface and may be used as an input to correct for smoke acceleration or stagnation, upward diffusion, and the tendency for backflow formation caused by the slope.

[0223] The step of calculating the smoke layer descent time by node and link is a step of estimating, by node and link, the time required for the smoke layer that has reached the ceiling to descend vertically to a preset reference altitude based on the volume expansion rate index, mass generation rate index, effective cross-sectional area, and longitudinal gradient angle.

[0224] Here, the smoke layer descent time can be defined as the time it takes for the smoke layer remaining at the ceiling to thicken and descend to a reference altitude corresponding to the breathing limit of evacuees. For example, the device can conservatively estimate the descent time to the reference altitude to decrease after determining correction coefficients that reflect the fact that the smoke layer thickness increases more rapidly as the mass generation rate index increases, that smoke of the same mass accumulates more quickly within the cross-section as the effective cross-sectional area decreases, and that upward movement and backflow tendencies are strengthened by buoyancy when the longitudinal gradient angle is upward. Additionally, the reference altitude can be managed as multiple values ​​according to the operational policy, and the device can be configured to adopt the descent time corresponding to the more conservative standard among the multiple reference altitudes.

[0225] The step of calculating the toxic gas saturation time by node and link is a step of estimating the time until the point at which the evacuee's ability to function is determined to be lost by applying an effective toxic dose model based on the component analysis results of the toxic gas concentration data included in the sensor data, by node and link.

[0226] Here, the component analysis results may include concentration estimates of at least one of carbon monoxide, carbon dioxide, oxygen concentration reduction, and other hazardous components, and the effective toxic dose model may be defined as a model that determines the point at which the cumulative effect of toxic gas exposure over time exceeds a threshold as the point of loss of activity capacity. For example, the device may be configured to conservatively predict the time series of toxic gas concentration increase in conjunction with the predicted results of ceiling arrival time and smoke layer descent time for a specific node or link, and then calculate the time at which the threshold is first satisfied by accumulating the predicted concentration over time as the toxic gas saturation time.

[0227] The step of calculating the limit visibility loss time by node and link is to calculate, by node and link, the point in time when visibility drops below a preset limit visibility threshold by applying a preset light attenuation coefficient to the smoke concentration calculated from time-series smoke concentration data.

[0228] Here, the light attenuation coefficient can be defined as a coefficient reflecting the correlation that the light attenuation rate increases as the smoke concentration increases, and the limit visibility threshold can be set to correspond to a visibility standard where it is difficult for an evacuee to continue moving on their own. In addition, to prevent misjudgment caused by sensor noise or temporary spikes in concentration, the limit visibility loss time can be determined after additionally checking whether a state below the threshold is maintained for a continuous preset time or longer.

[0229] The step of determining the limit arrival time and mapping it to nodes and links of the evacuation path graph involves deriving the physical smoke settlement time by summing the ceiling arrival time and the smoke layer descent time, comparing the physical smoke settlement time, the toxic gas saturation time, and the limit visibility loss time to determine the shortest time as the limit arrival time as the point in time when the survival limit environment of the corresponding node or link is established, and then recording the determined limit arrival time as the node and link metadata of the evacuation path graph. For example, if the physical smoke settlement time is calculated to be 180 seconds, the toxic gas saturation time to be 240 seconds, and the limit visibility loss time to be 150 seconds for a certain link, the limit arrival time can be determined as 150 seconds. Additionally, considering prediction errors and reduced sensor reliability, an effective limit arrival time is separately stored by subtracting a pre-set safety margin time from the limit arrival time, and by configuring risk link determination and graph search exclusion to be performed based on the effective limit arrival time, it is possible to ensure conservative evacuation guidance.

[0230] According to one embodiment, the method comprises: collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; generating a fire control package including a control command for each of the group of equipment to be controlled based on the group of equipment to be controlled and the target operating parameters for each equipment; and transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of generating the fire control package includes: determining a target stopping point and a movement path of a mobile trolley-type fire spray support device installed above an emergency lane within the tunnel based on the fire detection location and section identifier, and generating a trolley movement control command to the target stopping point. The present invention provides a method for controlling a mobile trolley-type fire extinguishing spray support device installed on an emergency roadway in a tunnel, comprising the step of: when it is determined that the trolley has reached a target stopping point, setting spray operation parameters including a spray direction, a spray flow rate, and a spray pattern for a spraying part of the fire extinguishing spray support device, and generating a spray control command to perform spraying of a fire extinguishing fluid according to the spray operation parameters.

[0231] Target operating parameters for each piece of equipment can be determined as values ​​to which unit conversion and range limits are applied to conform to the parameter schema for each piece of equipment type. For example, they may include the target stop coordinates of the trolley, the rotation angle of the spray nozzle, the rotation speed of the fan, the opening rate of the damper, the target discharge pressure of the pump, the target spray flow rate, etc., and each value can be configured to be truncated within the allowable range for each piece of equipment. In addition, by configuring to exclude unavailable equipment and replace it with alternative equipment based on the equipment's failure flag or availability status, fault-tolerant control in fire situations can be enabled.

[0232] The step of generating a fire control package is to structure individual control commands, converted into a communication protocol recognizable by each facility controller, into an executable package by grouping them by event, based on the group of facilities to be controlled and target operating parameters for each facility. Here, the fire control package may be configured to include a package identifier, a fire event identifier, an application section identifier, an application time, a command list, a command execution order, a retry rule, a timeout, and whether an acknowledgment is required; it may also be configured to include interlock conditions to prevent conflicts between facilities and version information or hash values ​​to prevent duplicate execution of the same package.

[0233] The step of determining a target stop point and a movement path and generating a trolley movement control command is to determine the longitudinal position of the target stop point and the movement path so that a mobile trolley-type fire extinguishing spray support device installed above an emergency lane in a tunnel simultaneously satisfies an effective spray range and a safe separation from the fire source, and to generate a trolley movement control command to the determined target stop point.

[0234] Here, the mobile trolley-type fire extinguishing spray support device may refer to a robotic device capable of automatically moving along rails on the tunnel ceiling or sidewall and spraying fire extinguishing fluid at the fire location, and the target stopping point may be defined as a stopping coordinate on the rail that is close to the fire detection location while ensuring device safety and spraying efficiency. Additionally, the movement path may be determined by path search with minimum travel time or minimum risk exposure as the objective function after modeling the rail topology as a graph, and by configuring to select a detour path by marking fire impact zones, zones where structural deformation due to heat is possible, or communication blind spots as risk zones and adding the cost of passing through the risk zones, secondary damage to the trolley and control failure can be suppressed.

[0235] The step of setting injection operation parameters and generating an injection control command when it is determined that the trolley has reached a target stopping point is to determine injection operation parameters including the injection direction, injection flow rate, and injection pattern for the injection part of the fire extinguishing injection support device using the result of determining the trolley's arrival as a trigger, and to generate an injection control command to perform injection of the fire extinguishing fluid according to the determined parameters.

[0236] Here, the spray unit may be a module including a nozzle and an actuator that rotates it at various angles, the spray direction may be expressed as a horizontal rotation angle and a vertical tilt angle to strike the center point of the fire source, and the spray flow rate may be defined as a target flow rate or target pressure adjusted in proportion to the fire intensity. Additionally, the spray pattern may be configured to be selected as either a direct spray or a radial spray form depending on the airflow intensity and the type of fire, and the spray control command may be implemented as a driving signal that packages the final determined angle, flow rate, pressure, and pattern information into a digital signal and transmits it to the nozzle control module.

[0237] The step of automatically controlling a group of target equipment by transmitting a fire control package to equipment controllers involves transmitting the generated fire control package to field equipment controllers via a wired or wireless communication network and tracking the execution status of each control command based on execution acknowledgments and feedback received from the equipment controllers. For example, fault-tolerant control may be performed by retransmitting if an acknowledgment is not received for a trolley movement control command or a spray control command, or by generating a reduced package excluding the equipment and switching to an alternative operation mode if a fault in a specific piece of equipment is detected. Additionally, secondary equipment accidents in a fire situation can be suppressed by including a priority rule so that an emergency stop or safety mode switching command is applied first when an abnormality such as trolley drive unit overcurrent, rail section abnormality, pump overheating, or valve position misalignment is detected.

[0238] Additionally, the detection sensor comprises: an airflow measurement sensor, a thermal imaging camera and a distance measurement sensor mounted on the fire extinguishing spray support device, and the step of generating the spray control command comprises: establishing a local three-dimensional coordinate system with the vertical bottom of the emergency lane as the origin based on the thermal image and distance measurement value included in the sensor data; specifying the lateral separation distance from the fire extinguishing spray support device to the actual lane where the fire occurred and the depth coordinates of the fire source center point on the local three-dimensional coordinate system, and generating a target fire source shape model based on the lateral separation distance and depth coordinates; hydrodynamically inversely calculating the horizontal scattering distance deflected by the longitudinal airflow until the fire extinguishing fluid leaves the spray nozzle and reaches the fire source center point based on the tunnel longitudinal airflow velocity component inside the road tunnel included in the sensor data and the airflow direction defined by the sign of the tunnel longitudinal airflow velocity component; and calculating a longitudinal aiming correction angle to advance the aiming point in the opposite direction of the airflow direction based on the horizontal scattering distance. A step of calculating a basic geometric targeting angle based on the above-mentioned lateral separation distance and depth coordinates, and adding a longitudinal aiming correction angle to the basic geometric targeting angle to determine a spray direction parameter including the horizontal rotation (Pan) angle and vertical tilt (Tilt) angle of a multi-joint actuator equipped in the spray unit; a step of determining a target spray flow rate proportional to the fire intensity index, wherein a range limit is applied so that the target spray flow rate is located within a preset minimum flow rate and maximum flow rate range; and a step of, if it is determined that the tunnel longitudinal airflow velocity component exceeds a preset scattering risk wind speed, adjusting the target spray pressure upward and setting a spray pattern in the form of a straight stream so that the droplet particle size is greater than or equal to a preset reference particle size.If the above tunnel longitudinal airflow velocity component is determined to be less than or equal to the above scattering risk wind speed, the method may include the step of setting a spray pattern in the form of a cone spray so that the droplet particle size is less than a preset reference particle size; and the step of determining a spray control command including the spray direction parameter, target spray flow rate, target spray pressure, and spray pattern.

[0239] The step of establishing a local three-dimensional coordinate system is to define a local three-dimensional coordinate system with the vertical bottom of the emergency lane as the origin, based on the thermal image and distance measurement values ​​included in the sensor data.

[0240] Here, the local 3D coordinate system is not an absolute coordinate system of the entire tunnel, but a relative coordinate system established based on the installation location of the fire extinguishing spray support device and the structure of the emergency lane, and can refer to a spatial system for precisely analyzing the relative geometric structure between the device and the fire source. For example, the vertical bottom of the emergency lane can be set as a reference corner point where the floor surface of the emergency lane meets the side wall or lane boundary, and the axes of the local 3D coordinate system can be defined as the first axis for the longitudinal direction of the tunnel, the second axis for the transverse direction from the emergency lane to the actual lane direction, and the third axis for the vertical direction from the floor to the ceiling direction. In addition, the origin and axis definitions may be set as fixed standards based on the equipment installation drawing, or they may be dynamically corrected by estimating the emergency lane boundary line and the road surface plane from the tunnel cross-section point cloud provided by the distance measuring sensor, so that all position parameters used for the aiming calculation of the spray unit can be consistently handled in a single coordinate system.

[0241] The step of specifying the lateral separation distance and depth coordinates and generating a target fire source shape model is to specify the lateral separation distance from the fire extinguishing spray support device to the actual road where the fire occurred and the depth coordinates of the fire source center point in the local 3D coordinate system, and then generate a target fire source shape model based on the lateral separation distance and depth coordinates.

[0242] Here, the lateral separation distance can be defined as the second axial distance between the injection nozzle reference point of the injection support device and the fire source center point in the actual lane, and the depth coordinate can be defined as the first axial distance component or line-of-sight distance component from the injection support device reference point to the fire source center point. For example, after dividing the area above a critical temperature in the thermal image into high-temperature area candidates, the center point of the candidate area or the center point around the maximum temperature point can be calculated as a fire source center point candidate, and the fire source center point candidate can be inversely calculated as a 3D point in the local coordinate system by combining the distance value of the distance measuring sensor and camera calibration parameters. In addition, if there are multiple fire source center point candidates, the system may be configured to select the final fire source center point based on at least one of area, temperature peak, temporal persistence, or inter-frame movement stability.

[0243] The target fire source shape model is a model representing a target coverage area based on the center point of the fire source, and can be implemented in at least one of a circular, elliptical, polygonal, or grid-based heatmap shape. For example, the outline of a high-temperature area extracted from a thermal image can be projected onto a local coordinate system to estimate the width and length of the fire source area, and an elliptical coverage model can be defined according to the estimated width and length, and the size of the model can be adjusted to expand as the fire intensity index increases. In addition, the target fire source shape model can be combined with coverage characteristics according to the spray pattern, so that a narrow target model prioritizing center point impact is selected in the direct spray type, and an expanded target model prioritizing area coverage is selected in the radial spray type.

[0244] The step of hydrodynamically inversely calculating the horizontal scattering distance is a step of inversely calculating the horizontal scattering distance deflected by the longitudinal airflow until the fire extinguishing fluid exits the spray nozzle and reaches the center point of the fire source, based on the tunnel longitudinal airflow velocity component inside the road tunnel included in the sensor data and the airflow direction defined by the sign of the tunnel longitudinal airflow velocity component.

[0245] Here, the tunnel longitudinal airflow velocity component may be a value calculated by projecting the airflow velocity vector provided by the airflow measurement sensor onto the tunnel longitudinal unit vector, and the airflow direction may be defined as a downstream or upstream direction depending on the sign of the longitudinal airflow velocity component. Additionally, the horizontal scatter distance may be defined as the distance component by which the sprayed fire extinguishing fluid is dragged longitudinally by the airflow while flying through the air and deviates from the aiming point, and the scatter distance may be calculated by reflecting the droplet particle size, spray pressure, spray pattern, flight time, and airflow velocity component. For example, the device may be configured to calculate an initial spray velocity based on the spray pressure and nozzle specifications, calculate a first flight time estimate assuming no airflow based on the target reach distance to the fire source center point and the initial spray velocity, calculate a first scatter distance estimate by reflecting the longitudinal airflow velocity component and the airflow-following characteristics of the droplet, and determine the final horizontal scatter distance by repeatedly updating the flight time and scatter distance until the estimate value fluctuates below a convergence threshold. At this time, assuming that the airflow-following characteristics of the droplets are low in the direct spray form and high in the radial spray form, it can be configured so that a larger scattering distance is calculated in the radial spray form even at the same airflow velocity.

[0246] The step of calculating the longitudinal aiming correction angle is a step of calculating the longitudinal aiming correction angle to advance the aiming point in the opposite direction of the airflow direction based on the horizontal scattering distance.

[0247] Here, advancing the aiming point may mean setting a virtual aiming point that is moved a predetermined distance in the opposite direction of the airflow direction relative to the longitudinal coordinate where the fire source center point is located, and the predetermined distance may be set as the horizontal scattering distance. Additionally, the longitudinal aiming correction angle may be calculated by combining the longitudinal distance component and the depth coordinate between the virtual aiming point and the nozzle reference point, and the sign of the aiming correction angle may be assigned to match the definition of the airflow direction. For example, if the longitudinal airflow is dominant in the downstream direction and the horizontal scattering distance is estimated to be 3 meters, and the depth coordinate of the fire source center point relative to the nozzle reference point is 15 meters, the aiming point is advanced by 3 meters in the upstream direction, and a corresponding longitudinal aiming correction angle may be calculated. Furthermore, if the airflow velocity component fluctuates, the system may be configured to use the average airflow velocity component of the most recently set time window, or to calculate the aiming correction angle using the scattering distance corresponding to the maximum value within the time window in order to conservatively reflect the risk of scattering.

[0248] The step of determining the injection direction parameter is to calculate a geometric basic targeting angle based on the lateral separation distance and depth coordinates, and to add a longitudinal aiming correction angle to the geometric basic targeting angle to determine the injection direction parameter including the horizontal rotation angle and vertical tilt angle of the multi-joint actuator equipped in the injection unit.

[0249] Here, the geometric basic targeting angle may refer to the basic aiming angle calculated using the lateral and vertical components between the nozzle reference point and the fire source center point in the local 3D coordinate system, and the longitudinal aiming correction angle may function as a correction term added to the basic aiming angle. Additionally, the horizontal rotation angle refers to the amount of left and right aiming, and the vertical tilt angle refers to the degree of up and down tilting, and can be determined as final aiming data reflecting both the geometric distance and the airflow correction value. Furthermore, if the horizontal rotation angle and the vertical tilt angle exceed the allowable range due to the mechanical limit of the actuator of the injection unit, the angle values ​​may be cut off at an upper or lower limit, and the cutting status and the reason for cutting may be recorded as metadata in the injection control command to be utilized for post-audit and fault diagnosis.

[0250] The step of determining the target injection flow rate and applying range limits is to determine the target injection flow rate in proportion to the fire intensity index, and to apply range limits so that the target injection flow rate is located within the preset minimum and maximum flow rate ranges.

[0251] Here, the target injection flow rate is a value representing the amount of extinguishing water injected according to the scale of the fire, and can be determined by a linear or stepwise lookup table so as to increase as the fire intensity index increases, and the minimum and maximum flow rates can be set by considering the pump capacity, pipe loss, nozzle allowable flow rate, and material strength limits. Accordingly, the target flow rate can be determined within a safe range so as to inject a larger amount of fluid as the flames become stronger, without exceeding the equipment limits.

[0252] The step of setting the target injection pressure and injection pattern based on whether the wind speed at risk of scattering is exceeded is to increase the target injection pressure and set the injection pattern in the form of a direct spray if the tunnel longitudinal airflow velocity component is determined to exceed the preset wind speed at risk of scattering, and to set the injection pattern in the form of a radial spray if the tunnel longitudinal airflow velocity component is determined to be at or below the wind speed at risk of scattering.

[0253] Here, the wind speed at risk of drift can be defined as the critical wind speed at which fine droplets are excessively deflected by the airflow and fail to reach the fire source, and the critical wind speed can be set based on the tunnel cross-section, nozzle installation height, target reach distance, and past empirical data. Furthermore, when the wind speed at risk of drift is exceeded, the system can be configured to switch to a direct spray form by making the droplet size relatively larger to be less affected by wind and increasing the spray pressure to enhance straightness, and when the wind speed at risk of drift is below, to enhance cooling and oxygen blocking effects by making the droplet size relatively smaller and covering a wide area in a radial spray form. Moreover, a range limit can be applied so that the upward adjustment of the target spray pressure is performed within the pump's allowable pressure range, and when the spray pattern is changed, the system can be configured so that operation sequence parameters, including the valve switching sequence, nozzle mode switching time, and stabilization waiting time, are determined together.

[0254] The step of determining the injection control command is to structure and determine the injection control command to include the injection direction parameter, target injection flow rate, target injection pressure, and injection pattern.

[0255] Here, the injection control command may be packaged to include a device identifier, a fire event identifier, an application time, an injection step number, a horizontal rotation angle, a vertical tilt angle, a target injection flow rate, a target injection pressure, an injection pattern, and an injection duration, and may include a timeout, a number of retries, and whether an acknowledgment response is required. Additionally, the injection control command may be configured to include an interlock condition prior to the start of injection, and the interlock condition may include at least one of the following: whether a trolley arrival determination is established, the pump availability status, whether the valve position is normal, whether the nozzle angle initialization is completed, and whether the emergency stop flag is released. Accordingly, by having the field controller verify the feasibility of the injection control command and then perform the injection, the aiming correction and injection pattern selection reflecting airflow obstruction are consistently applied to the actual injection execution, and the traceability of the injection execution can be ensured.

[0256] And, the step of hydrodynamically inversely calculating the horizontal scattering distance comprises: a step of calculating the initial injection velocity at the nozzle exit based on the initial injection pressure, nozzle diameter, and nozzle discharge coefficient included in the stored equipment specification information; a step of calculating the target reach distance vector from the nozzle reference point to the fire source center point based on the depth coordinate and lateral separation distance in the local 3D coordinate system; a step of calculating an initial estimate of the Time of Flight (TOF) to the fire source center point after nozzle exit under ideal conditions where no airflow exists, based on the initial injection velocity and the target reach distance vector; a step of determining the representative droplet particle size by selecting a pre-set first representative droplet particle size in the case of a direct water spray type and a pre-set second representative droplet particle size in the case of a radial spray type to correspond to the injection pattern; a step of calculating the relaxation time of the droplet based on the representative droplet particle size, the density and viscosity of the firefighting fluid, and calculating an airflow following coefficient indicating the velocity following degree of the droplet with respect to the longitudinal airflow using the relaxation time; A step of calculating a longitudinal deflection velocity component defined as an effective airflow velocity damped by an airflow following coefficient, based on the above-mentioned tunnel longitudinal airflow velocity component, airflow direction, and initial flight time estimate; a step of calculating a first horizontal scatter distance initial value using the above-mentioned longitudinal deflection velocity component and the initial flight time estimate; a step of updating an effective target reach distance vector by setting an effective target point corrected for the longitudinal component of the fire source center point based on the first horizontal scatter distance initial value, and calculating a second flight time estimate by re-estimating the flight time based on the updated effective target reach distance vector;The method may include the step of recalculating the horizontal scatter distance based on the above-mentioned second flight time estimate and longitudinal deflection velocity component, and repeatedly performing flight time estimation and horizontal scatter distance recalculation until the amount of change in the recalculated horizontal scatter distance becomes less than or equal to a preset convergence threshold to determine the final horizontal scatter distance.

[0257] The step of calculating the initial injection velocity is a step of calculating the initial injection velocity at which the extinguishing fluid is discharged from the nozzle outlet based on the initial injection pressure, nozzle diameter, and nozzle discharge coefficient included in the previously stored equipment specification information.

[0258] Here, the equipment specification information refers to data recording the design specifications of the injection nozzle and pump module, and may include the initial injection pressure, nozzle diameter, nozzle discharge coefficient, allowable pressure range, and allowable flow rate range. Additionally, the initial injection velocity can be defined as the speed determined as the internal pressure of the nozzle is converted into the kinetic energy of the fluid, and the ideal injection velocity can be calculated as a value corrected to match the actual discharge efficiency based on Bernoulli's theorem and the nozzle discharge coefficient. For example, the initial injection velocity can be calculated to have a tendency to increase as the initial injection pressure increases and decrease as the density of the extinguishing fluid increases; furthermore, the system can be configured to determine the actual initial injection velocity by calculating the ideal injection velocity based on the relationship between pressure and density and then multiplying it by the nozzle discharge coefficient.

[0259] The step of calculating the target reach distance vector is to calculate the target reach distance vector from the nozzle reference point to the center point of the flower source based on the depth coordinates and lateral separation distance specified in the local 3D coordinate system.

[0260] Here, the nozzle reference point can be defined as a reference point including the nozzle exit center of the injection unit or the nozzle rotation axis and the reference point of the nozzle exit, and the target reach vector can be defined as a physical quantity having components of direction and distance in three-dimensional space to which the firefighting fluid must travel. Additionally, the target reach vector can be configured to include components of the first axis (longitudinal), second axis (lateral), and third axis (vertical) of the local three-dimensional coordinate system, and can be used as a common input for subsequent flight time estimation and horizontal scatter distance calculation.

[0261] The step of calculating the initial flight time estimate is a step of calculating the initial flight time estimate from nozzle exit to the center point of the flower source under ideal conditions where no airflow exists, based on the initial injection velocity and the target reach distance vector.

[0262] Here, the initial flight time estimate may refer to the pure time taken for the firefighting fluid to reach the fire point assuming there is no influence from wind, and can be used as a basic time variable to calculate the horizontal dispersion distance, which is the distance pushed by the airflow. For example, it can be configured to calculate the target distance corresponding to the magnitude of the target distance vector, and then divide the target distance by the initial injection velocity to calculate the initial flight time estimate, and the target distance can be calculated as a spatial distance combining longitudinal, lateral, and vertical components.

[0263] The step of determining the representative droplet particle size is to determine the representative droplet particle size by selecting a preset first representative droplet particle size in the case of a direct spray type and selecting a preset second representative droplet particle size in the case of a radial spray type, so as to correspond to the spray pattern.

[0264] Here, the representative droplet size is a key factor determining the influence of air resistance and airflow, and can be defined as a value representing the physical size of the sprayed droplets. Additionally, the system can be configured to set a relatively large particle size as the representative value for direct spray patterns and a relatively small particle size for radial spray patterns; the representative droplet size can be pre-set by experimental values ​​for each spray pattern, distribution models for each nozzle shape, or specification tables.

[0265] The step of calculating the relaxation time and the airflow following coefficient is to calculate the relaxation time of the droplet based on the representative droplet particle size, the density and viscosity of the digestion fluid, and to calculate the airflow following coefficient, which represents the degree of velocity following of the droplet for the longitudinal airflow, using the relaxation time.

[0266] Here, the relaxation time can be defined as a time constant representing the physical responsiveness required for a droplet to reach the velocity of the surrounding airflow, and can be configured to calculate a smaller relaxation time by reflecting the tendency for droplets to be swept more easily by wind as the representative droplet particle size becomes smaller. Additionally, the airflow following coefficient can be defined as a coefficient indicating the ratio at which a droplet follows the wind speed inside the tunnel based on the relaxation time, and can be configured to approach 1 as the relaxation time relative to flight time is smaller, and approach 0 as the relaxation time relative to flight time is larger.

[0267] The step of calculating the longitudinal deflection velocity component is to calculate the longitudinal deflection velocity component defined as the effective airflow velocity damped by the airflow following coefficient, based on the tunnel longitudinal airflow velocity component, the airflow direction, and the initial flight time estimate.

[0268] Here, the effective airflow velocity can be defined as the effective velocity that pushes the droplet in the longitudinal direction by reflecting the airflow following coefficient to the actual longitudinal airflow velocity component, and the sign of the longitudinal deflection velocity component can be assigned to match the definition of the airflow direction. Accordingly, the longitudinal deflection velocity component can be used as the “physical velocity force causing deflection” in the calculation of horizontal deflection distance.

[0269] The step of calculating the initial value of the first horizontal scatter distance is a step of calculating the initial value of the first horizontal scatter distance expected to be generated by the wind based on the longitudinal deflection velocity component and the initial estimate of the flight time.

[0270] Here, the initial value of the first horizontal scatter distance can be defined as the first error distance that the droplet is dragged longitudinally by the longitudinal airflow while flying, and can be calculated to increase as the longitudinal deflection velocity component increases and the flight time increases.

[0271] The step of updating the effective target reach distance vector and re-estimating the second flight time is to update the effective target reach distance vector by setting an effective target point that corrects the longitudinal component of the fire source center point based on the first horizontal scatter distance initial value, and to calculate the second flight time estimate value based on the updated effective target reach distance vector.

[0272] Here, the effective target point can be defined as a virtual aiming point set by advancing the aiming point in the opposite direction of the airflow, and since the length and direction of the flight path change when scatter distance occurs, it can be configured to reflect the change in flight time accordingly. In addition, the updated effective target reach distance vector can be configured to be used as an input for the recalculation of the second horizontal scatter distance.

[0273] The step of recalculating the horizontal scatter distance and determining the final value through iterative convergence is to recalculate the horizontal scatter distance based on the second flight time estimate and the longitudinal deflection velocity component, and to determine the final horizontal scatter distance by repeatedly performing flight time estimation and horizontal scatter distance recalculation until the amount of change of the recalculated horizontal scatter distance becomes less than or equal to a preset convergence threshold.

[0274] Here, the convergence threshold can be defined as a reference value where it is determined that there is almost no difference in the result value even after repeated iterative calculations, and the horizontal dispersion distance at the point where it falls below the convergence threshold can be determined as the final physical deviation caused by airflow and configured to be utilized for subsequent aiming correction. Additionally, to ensure the stability of the iteration process, it can be configured to suppress abnormal divergence by applying at least one of an upper limit on the number of iterations, a lower limit on flight time, and an upper limit on horizontal dispersion distance.

[0275] According to one embodiment, the method comprises: a step of collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; a step of generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; a step of mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; a step of generating a firefighting control package including a control command for each of the group of equipment to be controlled and target operating parameters for each equipment based on the group of equipment to be controlled and the target operating parameters for each equipment; and a step of transmitting the firefighting control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of generating the firefighting control package includes: a step of loading a shielding device placement table including the installation location of a fire-resistant shielding device installed in each of the plurality of sections of the road tunnel based on the fire detection location, the fire intensity index, and the section identifier; Based on the shielding device placement table and fire detection location, a step of selecting at least one of the upstream shielding device and the downstream shielding device as a shielding device to be controlled based on the fire detection section, and determining a shielding deployment sequence including a stepwise deployment order and a target deployment amount for each shielding device to be controlled; and based on the shielding deployment sequence, a step of setting shielding deployment control parameters including an operation time for each deployment stage, a deployment speed, and a final deployment position for the actuator, unlocking unit, and deployment unit of the shielding device to be controlled, and generating a shielding deployment control command according to the shielding deployment control parameters.

[0276] The step of automatically controlling a group of target equipment by transmitting a fire control package to equipment controllers involves transmitting the generated fire control package to an equipment controller or individual control panel at the tunnel site via a wired or wireless communication network to physically operate the equipment, and tracking the execution status of each control command based on execution acknowledgments and feedback status received from the equipment controllers. For example, it may be configured to perform retransmission if an acknowledgment for a shield deployment control command is not received, or to generate a reduced package that replaces a specific shield device with an adjacent shield device capable of blocking the same passage area if the inability to operate a specific shield device is detected. Additionally, it may be configured to suppress secondary accidents in fire situations and ensure traceability at the event level by including a priority rule so that an emergency stop or safety mode switching command is applied first when an anomaly is detected during deployment, such as overcurrent, rail interference, lock position mismatch, or position feedback disconnection.

[0277] The step of loading a shielding device placement table is to load a shielding device placement table, which includes the installation locations of fire-resistant shielding devices installed in each of the multiple sections of a road tunnel based on the fire detection location, fire intensity indicator, and section identifier, from a storage unit or database and load it into memory.

[0278] Here, the fire-resistant shielding device may be implemented as at least one of a shielding screen, shielding curtain, shielding shutter, or shielding panel that is lowered and deployed to partition a passage area or evacuation area within a tunnel cross-section, and the shielding device placement table may be configured to include at least one of a shielding device identifier, an installation section identifier, installation location coordinates, an identifier of a passage area to be blocked, a number of deployment stages, a maximum deployment amount per stage, an actuator identifier, and an availability status flag.

[0279] The step of selecting a shielding device to be controlled and determining a shielding deployment sequence is to select at least one of an upstream shielding device and a downstream shielding device as a shielding device to be controlled based on the fire detection section, based on the shielding device placement table and fire detection location, and to determine a shielding deployment sequence including a stepwise deployment order and a target deployment amount for each shielding device to be controlled.

[0280] Here, the upstream and downstream sides can be consistently defined according to the expected direction of smoke diffusion defined by the sign of the tunnel longitudinal airflow velocity component calculated by the airflow measurement sensor, and the shielding device to be controlled can be configured to prioritize a device in which the longitudinal separation distance is included within a preset selected radius and the available status flag is active.

[0281] The shielding deployment sequence is a temporal and spatial operational plan that defines to what height and in what order the shielding is deployed according to a fire situation, and may be configured to include a deployment stage index, a target deployment amount for each stage, an application time for each stage, and transition conditions between stages. For example, if it is determined that evacuation vehicles or pedestrian evacuees remain, a first deployment may be performed to partially lower the shielding from the tunnel ceiling to a predetermined height, and if it is determined that the evacuation is complete with the remaining evacuees eliminated, a second deployment may be performed to fully lower the shielding to the tunnel floor. Additionally, if the fire intensity index is above a preset collapse risk threshold or the toxic gas concentration exceeds a preset survival limit, an emergency cutoff sequence may be configured to prioritize the application of a full lowering deployment regardless of whether evacuees remain.

[0282] The step of setting shielding deployment control parameters and generating a shielding deployment control command is to set shielding deployment control parameters, including a deployment stage operation time, a deployment speed, and a final deployment position, for the actuator, unlocking unit, and deployment unit of the shielding device to be controlled, based on the shielding deployment sequence, and to generate a shielding deployment control command to deploy the shielding device according to the set parameters.

[0283] Here, the actuator and the deployment unit may include a motor, roller, or winding mechanism for physically lowering and deploying the shielding membrane, and the unlocking unit may include a locking mechanism for switching the shielding membrane, which is normally in a fixed state, to a deployable state in case of fire. Additionally, the shielding deployment control parameters may be configured to include at least one of a target output torque, a target deployment speed, a deployment position feedback tolerance, an unlock holding condition, an emergency stop condition, and a stage transition trigger condition. When the deployment stage is identified as a first deployment, the deployment speed may be set to decelerate to be below a preset safety limit speed for the safety of evacuees, and when it is identified as a second deployment or an emergency cutoff sequence, the speed may be set to accelerate to converge to the maximum allowable driving speed of the actuator for cutoff efficiency.

[0284] Additionally, the detection sensor includes: a smoke sensor, a gas sensor, a temperature sensor, a camera, and an airflow measurement sensor, and the step of determining the shielding deployment sequence comprises: obtaining, from the shielding device placement table, a shielding device identifier, an installation section identifier, installation location coordinates, a blocking target passage area identifier, the number of deployment stages, a maximum deployment amount per stage, and an availability status flag for fire-resistant shielding devices installed in each of the plurality of sections; mapping the fire detection location to the fire detection section identifier, distinguishing upstream and downstream sections based on the fire detection section identifier, and classifying upstream candidate shielding device groups and downstream candidate shielding device groups; for each of the upstream candidate shielding device group and downstream candidate shielding device group, calculating the longitudinal separation distance between the fire detection location and the shielding device installation location coordinates, and primarily selecting shielding devices as control target shielding devices in which the longitudinal separation distance is included within a preset selection radius and the availability status flag is active. A step of determining a predicted smoke diffusion section based on the predicted smoke diffusion direction defined by the sign of the tunnel longitudinal airflow velocity component included in the fire intensity indicator and sensor data, and finally selecting shielding devices among the initially selected shielding devices that have a traffic area included in the predicted smoke diffusion section as a blocking target as the control target shielding devices; a step of identifying whether evacuation vehicles and pedestrian evacuees remain by determining whether evacuation vehicles and pedestrian evacuees located below the control target shielding device and in the blocking target traffic area are detected based on the image of the camera included in the detection sensor; a step of calculating the altitude of the descending boundary surface of the high-temperature smoke layer formed in the ceiling of the road tunnel based on the sensor data of the smoke sensor and temperature sensor included in the detection sensor; and a step of calculating a primary deployment target altitude within a range lower than the descending boundary surface altitude and higher than the minimum evacuation securing altitude based on the descending boundary surface altitude and a preset minimum evacuation securing altitude.A step of generating a first shielding deployment sequence including a deployment stage index, a target deployment amount for each deployment stage, and an application time for each deployment stage, such that when it is determined that at least one of the evacuation vehicle and pedestrian evacuee remains, the controlled shielding device is deployed downward from the tunnel ceiling to a first deployment target altitude; a step of continuously re-evaluating whether the evacuation vehicle and pedestrian evacuee remain in the blocked passage area based on the image of the camera while the controlled shielding device maintains the first deployment target altitude; and a step of generating a second shielding deployment sequence linked to the first shielding deployment sequence, such that when it is determined that the evacuation is completed and no more evacuation vehicles and pedestrian evacuees are detected in the blocked passage area, the controlled shielding device is fully deployed downward to the tunnel floor. The method may include: a step of generating an emergency cutoff sequence such that the controlled shielding device is forcibly lowered and deployed to the tunnel floor regardless of whether evacuation vehicles and pedestrian evacuees remain, if the above-mentioned evacuation completion state is not determined and 1) the fire intensity index is determined to be above a preset threshold for the risk of collapse of the tunnel structure, or 2) the concentration of toxic gas included in the sensor data of the gas sensor included in the detection sensor is determined to exceed a preset survival limit; and a step of setting a lock release maintenance control parameter such that, when the emergency cutoff sequence is generated, a physical locking mechanism provided at the bottom of the controlled shielding device is maintained in an unlocked state so that an isolated evacuee can push the shielding device with physical force and pass through.

[0285] The above detection sensor is a group of sensors for simultaneously determining the safety of smoke spread and shielding deployment in a fire situation within a road tunnel, and is configured to include a smoke sensor, a gas sensor, a temperature sensor, a camera, and an airflow measurement sensor.

[0286] Here, the smoke sensor may provide at least one of soot concentration, light attenuation rate, or smoke concentration rise rate; the gas sensor may provide at least one of carbon monoxide, carbon dioxide, oxygen concentration decrease, or toxic gas indicator; and the temperature sensor may provide the temperature level and temperature rise rate at at least one of the ceiling, sidewall, and road surface sections. Additionally, the camera may provide video frames for detecting evacuation vehicles and pedestrian evacuees present in the blocked passage area, and the airflow measurement sensor may provide an airflow velocity vector including wind direction and wind speed. This airflow velocity vector is projected onto a tunnel longitudinal unit vector and converted into a tunnel longitudinal airflow velocity component, which can then be used as an input for predicting the expected direction of smoke diffusion and the danger zone. Furthermore, the sensor data is configured to be time-synchronized so that values ​​aligned based on the same analysis time are used as input, thereby ensuring that the determination of the shielding deployment sequence consistently reflects the smoke layer and the remaining status of evacuees at a specific time.

[0287] The step of obtaining shielding device metadata from the shielding device placement table described above is a step of loading identifiable and controllable information for fire-resistant shielding devices installed in each of multiple sections in a table format, and then extracting the shielding device identifier, installation section identifier, installation location coordinates, blocking target passage area identifier, number of deployment stages, maximum deployment amount per stage, and availability status flag to configure as input for subsequent selection logic.

[0288] Here, the installation location coordinates may be expressed as at least one of the tunnel longitudinal coordinates (chain ridge), the transverse offset relative to the tunnel centerline, and the installation elevation, and the identifier for the traffic zone to be blocked may be defined as a code identifying a lane, emergency lane, pedestrian walkway, or evacuation passage section designed to be physically blocked when the shield is lowered and deployed. Additionally, the availability status flag serves as an indicator of whether the shield device is in a state where it can operate normally without failure, and may be configured to include flags indicating an unavailable state, such as failure, inspection mode, or inability to communicate. Furthermore, the device may be configured to prevent dangerous shield deployment caused by incorrect specifications by excluding the shield device from the candidates and recording the reason for exclusion when missing fields in the table, abnormal coordinates, or discrepancies in the number of deployment steps are detected.

[0289] The step of mapping the fire detection location to a fire detection section identifier and classifying candidate shielding device groups by distinguishing upstream and downstream sections is to determine the road tunnel section containing the fire detection location as the fire detection section identifier, distinguish upstream and downstream sections based on the fire detection section identifier, and classify upstream candidate shielding device groups and downstream candidate shielding device groups based on the installation section identifier of the shielding device placement table.

[0290] Here, the distinction between upstream and downstream can be defined according to the standard direction of travel of the tunnel or a standard longitudinal unit vector designated by the manager, and can be configured to be fixedly stored linked to an event identifier so that the upstream / downstream distinction criteria do not change within the same event. In addition, if adjacency relationships exist between sections, the control range can be progressively expanded according to fire intensity and airflow conditions by referring to a section adjacency relationship table or a section graph to hierarchize the candidate group into multiple layers by classifying the section directly adjacent to the fire detection section as the first-order adjacency section and the next-order adjacency section as the second-order adjacency section.

[0291] The step of calculating the longitudinal separation distance for each of the upstream candidate shielding device group and the downstream candidate shielding device group and selecting the primary shielding device to be controlled is to calculate the longitudinal separation distance between the fire detection location and the shielding device installation location coordinates, and to primarily select shielding devices as the shielding devices to be controlled that have a longitudinal separation distance within a preset selection radius and have an active availability status flag.

[0292] Here, the longitudinal separation distance can be calculated using the difference between the changeage coordinates of the fire detection location and the changeage coordinates of the shielding device as the base component. If necessary, it can be configured to be calculated as a corrected separation distance by referring to a linear correction coefficient or a section-by-section cumulative distance table to correct distance distortion in sections where the tunnel alignment is curved. Additionally, the selection radius can be variably set according to the fire intensity indicator section, tunnel operation policy, or shielding blocking target. By configuring shielding devices with an inactive availability status flag to be excluded from the initial selection, subsequent sequences can be generated targeting only shielding devices that are actually operational in a fire situation. Furthermore, the initial selection results can be configured to record both the selection reason and the exclusion reason, thereby ensuring traceability at the event level.

[0293] The step of determining the predicted smoke spread section and selecting the final control target shielding device is to determine the predicted smoke spread section based on the predicted smoke spread direction defined by the sign of the tunnel longitudinal airflow velocity component included in the fire intensity index and sensor data, and to finally select the shielding devices among the initially selected shielding devices that have the passage area included in the predicted smoke spread section as a target for blocking as the control target shielding devices.

[0294] Here, the predicted direction of smoke diffusion can be defined as downstream when the longitudinal airflow velocity component of the tunnel is a positive value, and upstream when it is a negative value. If the signal determination is unstable due to a low airflow velocity component, a pre-set neutral section may be established to mark the predicted diffusion direction as uncertain, or the direction may be stabilized and confirmed by applying the average value of the most recent pre-set time window. Additionally, the predicted smoke diffusion section may be set to extend a longer distance as the fire intensity index along the predicted diffusion direction increases, and the section may be adjusted to expand further if the tunnel longitudinal gradient is formed in a direction that intensifies diffusion. Accordingly, the final selection is performed not based on simple distance, but by including traffic zones with a high probability of smoke reaching via the airflow as priority blocking targets, thereby improving the suitability of the shielding deployment.

[0295] The step of identifying whether the evacuation vehicles and pedestrian evacuees remain is a step of identifying whether the evacuation vehicles and pedestrian evacuees are detected based on images from a camera included in a detection sensor, by determining whether the evacuation vehicles and pedestrian evacuees located in the lower part of the control target shielding device and the blocking target passage area are detected.

[0296] Here, the traffic zone subject to blocking can be mapped to a zone of interest defined by the traffic zone identifier in the shielding device placement table, and detection can be performed by calculating bounding boxes or segment masks of vehicle and human objects using an image-based object detection model. Additionally, to prevent the determination of whether an object remains is inconsistent due to momentary false detections, the system may be configured to apply conservative rules, such as verifying the persistence of the same object based on inter-frame tracking, adopting only detections above a reliability threshold, or determining an object as remaining only when detected for more than a set number of consecutive frames. Furthermore, since the lower area of ​​the shielding device is a high-risk area for collision when the shield is lowered and deployed, the system may be configured to record a residual location type code along with the object so that deployment speed limits or deployment stop conditions can be reflected in subsequent sequences when an object is present in the said lower area.

[0297] The step of calculating the elevation of the descending boundary surface of the high-temperature smoke layer is a step of estimating the elevation of the descending boundary surface of the high-temperature smoke layer formed in the ceiling of the road tunnel based on sensor data from the smoke sensor and the temperature sensor included in the detection sensor.

[0298] Here, the elevation of the descending boundary can be defined as the height at which the high-temperature smoke layer and the relatively low-temperature breathable zone form a boundary, and it can be calculated by combining the concentration profiles of smoke sensors installed at multiple heights and the vertical temperature gradients of temperature sensors. For example, if smoke concentration increases rapidly and temperature rises at a sensor near the ceiling, whereas the increase is relatively gradual at a sensor near the road surface, the elevation at which the rate of change of smoke concentration or temperature changes abruptly between the two sections can be estimated as the descending boundary. Additionally, if the installation height of the sensors is limited, the system can be configured to interpolate the boundary from concentration and temperature measurements by referring to a pre-set smoke layer model, and a safety correction factor can be applied to estimate the elevation of the descending boundary lower to conservatively reflect the estimation error.

[0299] The step of calculating the first deployment target altitude is a step of calculating the first deployment target altitude within a range lower than the descending boundary altitude and higher than the minimum evacuation securing altitude, based on the descending boundary altitude and a preset minimum evacuation securing altitude.

[0300] Here, the minimum evacuation securing height is the minimum effective passage height to ensure the passage of evacuation vehicles and pedestrian evacuees, and can be pre-set by the height of the evacuee's body, the height of the vehicle's roof, or by operational policy. Additionally, while it is advantageous to set the primary deployment target height as low as possible to block the spread of smoke, it must be maintained above the minimum evacuation securing height to avoid hindering the movement of remaining evacuees; therefore, it can be configured to be selected within a range that satisfies both constraints simultaneously. Accordingly, the primary deployment target height can be determined to secure a survival margin that allows people to pass while blocking smoke.

[0301] The step of generating the above first shielding deployment sequence is to generate a first shielding deployment sequence including a deployment step index, a target deployment amount for each deployment step, and an application time for each deployment step, such that when it is determined that at least one of the evacuation vehicle and pedestrian evacuee remains, the shielding device to be controlled is deployed downward from the tunnel ceiling to the first deployment target altitude.

[0302] Here, the first shield deployment sequence is identified as an evacuation securing priority sequence and can be configured to suppress collision accidents during the shield deployment process by maintaining the descent speed below a preset safety limit speed and including conditions for deceleration or temporary stop when approaching objects in the lower area. Additionally, the first shield deployment sequence can be configured to include a first deployment target altitude maintenance time, a re-evaluation cycle, and a target altitude attainment determination condition to be linked to a subsequent re-evaluation stage.

[0303] The step of continuously re-evaluating the remaining status is a step of re-evaluating, at a preset interval, whether evacuation vehicles and pedestrian evacuees remain in the blocked passage area based on the image of the camera while the control target shielding device maintains the primary deployment target altitude.

[0304] Here, re-evaluation can be reliably performed using at least one of a majority vote of detection results within a certain time window, the accumulation of continuous non-detection time, or a tracking-based object loss determination, and the re-evaluation results can be configured to be recorded as time-spanning snapshots and used as grounds for subsequent sequence transitions.

[0305] The step of generating the above second shielding deployment sequence is to generate a second shielding deployment sequence linked to the first shielding deployment sequence so that when the evacuation completion state is determined in which evacuation vehicles and pedestrian evacuees are no longer detected in the blocked passage area, the shielding device to be controlled is fully lowered and deployed to the tunnel floor.

[0306] Here, the secondary shielding deployment sequence is identified as a blocking efficiency priority sequence and can be configured to be linked with control parameters that accelerate the deployment speed to converge to the maximum allowable driving speed of the actuator or apply an upward target output torque. Additionally, the secondary shielding deployment sequence can be configured to isolate fire and toxic gas spread in the shortest possible time by dividing the remaining downward amount according to the constraint of the maximum deployment amount for each step and shortening the application time for each step.

[0307] The step of generating the above emergency cutoff sequence is to generate an emergency cutoff sequence such that, if the evacuation is not determined to be complete and the fire intensity index is above a preset threshold for the risk of collapse of the tunnel structure or the toxic gas concentration included in the sensor data of the gas sensor exceeds a preset survival limit, the controlled shielding device is forcibly lowered and deployed to the tunnel floor regardless of whether it remains.

[0308] Here, the emergency cutoff sequence serves as a top priority defense sequence designed to prevent the spread of greater damage regardless of human survival; it can be configured to immediately increase the deployment speed and target output torque within the maximum allowable range upon determination and minimize waiting times between stages. Furthermore, by configuring the system to record the reason for the emergency cutoff, the time of activation, and a summary of the sensor values ​​used, traceability at the event level can be ensured.

[0309] The step of setting the above unlock maintenance control parameter is a step of setting the unlock maintenance control parameter so that, when an emergency blocking sequence is generated, a physical locking mechanism provided at the bottom of the shielding device to be controlled remains in an unlocked state so that an isolated evacuee can push the shielding with physical force and pass through.

[0310] Here, the unlock retention control parameter may be configured to include at least one of the unlock signal retention time, a re-lock prohibition flag, a lock status feedback verification condition, and a return condition after the emergency cutoff ends. Additionally, to prevent the blocking performance from degrading due to the shield opening becoming abnormally large when the lock is fully released, it may be configured to include a weak lock mode or an elastic retention mode that allows movement only within a limited displacement range. Accordingly, it is possible to mitigate the risk of isolation caused by the emergency cutoff while suppressing the rapid loss of the shield's blocking function.

[0311] And, the step of generating the shield deployment control command comprises: a step of calculating the dynamic pressure applied by the forced ventilation airflow to the surface of the shield being deployed downward, based on the tunnel longitudinal airflow velocity component included in the sensor data and the air density within the tunnel; a step of calculating the equivalent wind load acting on the shield due to the dynamic pressure, based on the real-time airflow exposure area defined by the weight per unit volume of the material of the shield, the width of the shield, and the current deployment altitude; a step of predicting the wind pressure deflection angle at which the lower part of the shield is deflected in the downstream direction of the tunnel, based on the equivalent wind load and the effective moment arm of the lower part of the shield; and, if it is determined that the wind pressure deflection angle exceeds the vertical descent limit angle, a step of estimating the mechanical friction resistance generated at the guide rails on both sides of the shield based on a preset friction coefficient, guide rail contact force, and shield deflection angle. If it is determined that the mechanical friction resistance exceeds a reference torque corresponding to the basic output of the actuator, the method may include: a step of upwardly correcting the target output torque applied to the actuator by a correction coefficient proportional to the wind pressure deflection angle to overcome the mechanical friction resistance and maintain a vertical deployment trajectory; if the shield deployment sequence is a first shield deployment sequence, a step of setting the deployment speed to be lower than or equal to a preset safety limit speed to prevent head impact accidents of evacuees passing through the blocked passage area; if the shield deployment sequence is identified as a second shield deployment sequence or an emergency blocking sequence, a step of setting the deployment speed to be accelerated so that the deployment speed of the shield converges to the maximum allowable driving speed of the actuator based on the target output torque to block the spread of fire and toxic gases in the shortest possible time; and a step of finally determining the shield deployment control parameters to include the target output torque and deployment speed, and determining the shield deployment control command based on the shield deployment control parameters.

[0312] The step of calculating dynamic pressure is to calculate the dynamic pressure exerted by the forced ventilation airflow on the surface of the shielding membrane during downward deployment, based on the tunnel longitudinal airflow velocity component and air density within the tunnel included in the sensor data.

[0313] Here, the tunnel longitudinal airflow velocity component can be defined as a value calculated by projecting the airflow velocity vector provided by the airflow measurement sensor onto the tunnel longitudinal unit vector, and the air density inside the tunnel can be corrected in real time from temperature and pressure sensor data inside the tunnel or set to a preset standard density in the tunnel operating environment. Additionally, if the airflow velocity component fluctuates instantaneously or includes sensor noise, the system can be configured to prevent excessive fluctuation in dynamic pressure calculation by applying a moving average value or a conservative upper limit value from a recently preset time window. For example, dynamic pressure can be calculated proportionally to the value obtained by multiplying the air density by the square of the tunnel longitudinal airflow velocity component and dividing the result by 2, while the sign of the airflow velocity component is used only for predicting the deflection direction and the magnitude of the dynamic pressure is calculated based on the absolute value, thereby ensuring that the wind load magnitude assessment is performed stably even in reverse airflow situations.

[0314] The step of calculating the equivalent wind load is a step of calculating the equivalent wind load acting on the shield due to dynamic pressure based on the weight per unit volume of the material of the shield, the width of the shield, and the real-time airflow exposure area defined by the current deployment altitude.

[0315] Here, the weight per unit volume of the material can be defined as the self-weight per unit volume reflecting the density of the shielding material and gravitational acceleration, and the shielding width can be defined as the effective width of the traffic area to be blocked or the effective deployment width of the shielding. Additionally, the real-time airflow exposure area can be approximated as the product of the shielding width and the length the shielding has descended from the ceiling to the current deployment height; as the deployment height decreases (the further the shielding descends), the exposure area increases, thereby reflecting an increase in the equivalent wind load. Furthermore, if the actual exposure area is limited because the lower end of the shielding is close to the floor or an obstacle, the system can be configured to dynamically crop the exposure area based on deployment position sensors or encoder feedback. Moreover, the equivalent wind load is calculated by reflecting the exposure area and a preset drag coefficient to the dynamic pressure; however, by configuring the system to apply corrections to the effective drag coefficient based on the porosity of the shielding surface, the rib structure, or the reinforcing frame, the accuracy of wind load estimation corresponding to the actual shape of the shielding can be improved.

[0316] The step of predicting the wind pressure deflection angle is a step of predicting the wind pressure deflection angle at which the lower part of the shielding membrane is deflected in the downstream direction of the tunnel, based on the equivalent wind load and the effective moment arm of the lower part of the shielding membrane.

[0317] Here, the effective moment arm can be defined as the distance to the point of application of the equivalent wind load, assuming the upper support of the shield or the guide rail support is the rotational reference point. As the deployment length increases, the effective moment arm increases, thereby reflecting a leverage effect where the deflection angle increases even under the same dynamic pressure. Additionally, the wind pressure deflection angle can be predicted based on the relative relationship between the deflection moment generated by the equivalent wind load and the restoring moment generated by the shield's self-weight and the guide rail restraint force. In the small-angle approximation section, the deflection angle can be calculated by linearizing it to be proportional to the deflection moment. Accordingly, the deflection angle prediction can be configured to be iteratively updated at each deployment stage to continuously reflect real-time airflow changes and the deflection risk associated with the progress of deployment.

[0318] The step of estimating mechanical friction resistance is a step of estimating the mechanical friction resistance occurring at the guide rails on both sides of the shield based on the preset friction coefficient, guide rail contact force, and shield deflection angle when it is determined that the wind pressure deflection angle exceeds the vertical descent limit angle.

[0319] Here, the vertical descent limit angle can be defined as a physical critical angle at which the shield can descend stably without stopping due to friction with the guide rail, even if the shield is deflected; this limit angle can be determined by the rail clearance, the stiffness of the shield reinforcement frame, and the structure of the deployment roller. Additionally, the contact force can be calculated by reflecting geometric constraints, such as the guide rail contact force increasing as the deflection angle increases because the side of the shield adheres more strongly to the rail wall. The calculation of the contact force may include additional lateral pressure components caused by the preload of the deployment roller, shield tension, or actuator traction force. Accordingly, the mechanical friction resistance is estimated as a value proportional to the product of the friction coefficient and the contact force; however, by configuring the system to apply a conservative upper limit correction in the section where the contact force surges due to the increase in the deflection angle, the risk of rail jamming and failure to operate can be preemptively reflected.

[0320] The step of upwardly correcting the target output torque is a step of upwardly correcting the target output torque applied to the actuator by a correction coefficient proportional to the wind pressure deflection angle so as to overcome mechanical friction resistance and maintain a vertical development trajectory when it is determined that the mechanical friction resistance exceeds the reference torque corresponding to the basic output of the actuator.

[0321] Here, the reference torque can be defined as the actuator's base output torque required to overcome the shield's self-weight and basic friction within the normal deflection range, and the correction factor can be set to increase as the deflection angle approaches or exceeds the vertical descent limit angle. Additionally, an upper limit may be applied to the target output torque to ensure it does not exceed the actuator's maximum allowable torque and thermal protection limits; furthermore, since rapid application of torque correction may increase vibration or shock in the deployment section, a torque ramp-up section and a torque change rate limit may be applied together. Moreover, if torque upward correction occurs repeatedly, the shield deflection may be structurally excessive; therefore, the system can be configured to incorporate conservative recovery logic, such as reducing the deployment speed, pausing and retrying, or implementing a bypass deployment mode.

[0322] The step of setting the deployment speed to a reduced speed is a step of setting the deployment speed to a reduced speed such that the deployment speed of the shield becomes less than or equal to a preset safety limit speed in order to prevent head injury accidents of evacuees passing through the blocked passage area when the shield deployment sequence is identified as a primary shield deployment sequence.

[0323] Here, the safety limit speed can be defined as a low speed set to ensure that impact energy remains within an allowable range even if a pedestrian or vehicle comes into contact with the lower part of the descending shield, and it can be managed across multiple safety speed zones according to operational policies. Furthermore, the deceleration setting can be configured not only to simply lower the target speed but also to set the acceleration and deceleration profiles to be smooth, thereby preventing abrupt jerks during the shield's descent. Moreover, if a camera-based residual object is detected to be in close proximity to the lower region of the shield, the system can be configured to dynamically reinforce deployment speed rules by further reducing the safety limit speed or bringing the system to a temporary stop upon approaching within a certain distance.

[0324] The step of accelerating the deployment speed is a step of accelerating the deployment speed based on the target output torque so that, when the shielding deployment sequence is identified as a secondary shielding deployment sequence or an emergency cutoff sequence, the deployment speed of the shielding membrane converges to the maximum allowable driving speed of the actuator in order to cut off fire and toxic gas spread within the shortest possible time.

[0325] Here, the secondary shielding deployment sequence can be defined as a sequence that prioritizes blocking efficiency after the evacuation completion state is confirmed, and the emergency blocking sequence can be defined as a top priority sequence that is immediately triggered by emergency conditions such as a risk of collapse or exceeding the toxic gas survival limit. Additionally, the acceleration setting can be configured to actively increase the speed in sections with sufficient torque margin in conjunction with the upward correction result of the target output torque, and to limit the speed increase in sections where wind pressure deflection or frictional resistance increases rapidly. Furthermore, the maximum allowable driving speed convergence control can be configured to suppress rail interference and overcurrent caused by high-speed deployment by considering the tolerance of the deployment unit position feedback, speed limits per rail section, and emergency stop conditions together.

[0326] The step of determining the shield deployment control command is to finally determine the shield deployment control parameters to include the target output torque and deployment speed, and to determine the shield deployment control command based on the shield deployment control parameters.

[0327] Here, the shield deployment control parameters may be structured to include at least one of a deployment stage index, a target deployment height or target deployment amount, a target output torque, a target deployment speed, a stage operation time, a position feedback tolerance, a torque upper limit and a speed upper limit, a deployment completion determination condition, and an emergency stop condition. Additionally, the shield deployment control command may be packaged to include a controlled shield device identifier, a fire event identifier, an application time, a command validity period, a retry rule, a timeout, and whether an acknowledgment response is required. Furthermore, by configuring the command to attach summary values ​​of the calculation basis—such as dynamic pressure, equivalent wind load, wind pressure deflection angle, friction resistance estimate, and correction coefficient—as metadata, it is possible to ensure verification of execution suitability in the field controller and traceability in post-audits.

[0328] According to one embodiment, the method comprises: a step of collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; a step of generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity index based on the sensor data and equipment status data; a step of mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity index to determine a group of equipment to be controlled and target operating parameters for each equipment; a step of generating a fire control package including a control command for each of the group of equipment to be controlled and target operating parameters for each equipment based on the group of equipment to be controlled and the target operating parameters for each equipment; and a step of transmitting the fire control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of generating the fire control package includes: a step of loading a fire hydrant system configuration table including equipment identifiers and installation locations of fire hydrants, water supply pipes, pressurized pumps, and flow control valves included in the fire hydrant system within the road tunnel based on the fire detection location, fire intensity index, and section identifier; A smart fire hydrant control method for automatically controlling firefighting water pressure and volume in a tunnel is provided, comprising: a step of determining a target water supply section corresponding to a fire detection section based on the measurement values ​​of a pressure sensor and a flow sensor included in the fire hydrant system configuration table and detection sensor; a step of, for a fire hydrant corresponding to the target water supply section, querying a target water pressure and a target flow rate corresponding to a section identifier and a fire intensity index from the equipment control profile, and calculating fire hydrant water supply control parameters including the rotational speed of a pressure pump and the opening rate of a flow control valve to satisfy the target water pressure and target flow rate; and a step of generating a fire hydrant water supply control command according to the fire hydrant water supply control parameters.

[0329] The step of generating a fire control package is to structure control commands in the form of communication protocols interpretable by each equipment controller into an executable package by grouping them into event units, based on the group of equipment to be controlled and target operating parameters for each equipment. Here, the fire control package may be configured to include at least one of a package identifier, a fire event identifier, an application section identifier, an application time, a command list, a command execution order, a timeout, a retry rule, whether an acknowledgment is required, and an interlock condition. In the case of a fire hydrant water supply control command, it may be configured to be packaged with an interlock rule that includes a precondition, such as increasing the valve opening rate after the pump start is completed, and a condition prohibiting simultaneous control, such as invalidating the water supply command when the emergency stop flag is activated.

[0330] The step of loading a fire hydrant system configuration table is to load a fire hydrant system configuration table, which includes the equipment identifiers and installation locations of fire hydrants, water supply pipes, pressurized pumps, and flow control valves included in the fire hydrant system within the road tunnel, from a storage unit or database and load it into memory based on fire detection locations, fire intensity indicators, and section identifiers.

[0331] Here, the fire hydrant system configuration table is a table containing physical infrastructure information of a fluid supply network within a tunnel, and may be configured to include a unique identifier for each fire hydrant, the connection status of piping, and the installation coordinates of pumps and valves, and may include at least one of the following: fire hydrant installation location (chaining, lateral offset, installation elevation), nozzle diameter per fire hydrant, allowable water pressure range per fire hydrant, allowable flow rate range per fire hydrant, pipe diameter and pipe length, pump performance curve table reference key, section boundary valve identifier, and section division information. Additionally, if an anomaly such as a missing field, abnormal coordinates, nozzle diameter mismatch, or inversion of the allowable range is detected in the table, the corresponding equipment is excluded from the candidates and the reason for exclusion is recorded, thereby suppressing the risk of overpressure and overflow control caused by incorrect specifications.

[0332] The step of determining the target water supply section is a step of determining the target water supply section corresponding to the fire detection section based on the measured values ​​of the pressure sensor and flow sensor included in the fire hydrant system configuration table and the detection sensor.

[0333] Here, the target water supply section can be defined as the fire hydrants closest to the fire point where water discharge occurs and the connected water supply piping lines. It can be configured to include a combination of the section containing the fire detection location, and the upstream and downstream end sections that must be controlled together to stably supply firefighting water to that section. For example, the device can be configured to set the piping branch section connected to the group of fire hydrants closest to the fire detection location as a candidate section, and then exclude sections with excessive water loss by comparing the level and fluctuation of the upstream reference water pressure and the downstream end water pressure measured by the pressure sensor, or to confirm the section where water supply is actually possible as the target water supply section by reflecting the status of the section boundary valve. Additionally, if a pattern is observed where the actual flow rate measured by the flow sensor increases rapidly while the water pressure drops sharply, the device can be configured to mark the possibility of section saturation due to multiple discharges as a water supply limit risk and conservatively adjust the boundary or control priority of the target water supply section.

[0334] The step of calculating fire hydrant water supply control parameters is to query the target water pressure and target flow rate corresponding to the section identifier and fire intensity index from the equipment control profile for the fire hydrant corresponding to the target water supply section, and to calculate fire hydrant water supply control parameters including the rotational speed of the pressurizing pump and the opening rate of the flow control valve to satisfy the queried target water pressure and target flow rate.

[0335] Here, fire hydrant water supply control parameters can be defined as numerical values ​​for the motor rotation speed and valve opening rate required to implement the amount of water (flow rate) and spray force (water pressure) necessary to suppress a fire. The target water pressure is set to be greater than or equal to the minimum operating water pressure and within the allowable water pressure range for each fire hydrant. The target flow rate can be configured to increase proportionally to the fire intensity index but be limited by pipe losses and the maximum discharge capacity of the pump. Additionally, the device can be configured to calculate the water pressure control error between the target water pressure and the downstream end pressure using pressure sensor measurements, and the flow control error between the target flow rate and the actual flow rate of the section using flow sensor measurements. Subsequently, the device can calculate the target pump head and target rotation speed based on these control errors and a pump performance curve table.

[0336] The step of generating a fire hydrant water supply control command is a step of generating a driving signal transmitted to a pressure pump controller and a flow control valve actuator in the form of a control command so as to perform fire hydrant water supply according to the calculated fire hydrant water supply control parameters.

[0337] Here, the fire hydrant water supply control command can be packaged to include a fire event identifier, a target water supply section identifier, a pump identifier, a target rotational speed, a target opening rate per valve identifier, an application time, a command validity period, a refresh cycle, a timeout, a retry rule, and whether an acknowledgment is required. Additionally, it can be configured to include an emergency interlock condition that automatically invalidates the water supply control command or switches to a safety mode including lowering the rotational speed and gradually closing the valve when an abnormal condition, such as pump overcurrent, motor overheating, reaching a pressure at risk of pipe rupture, or valve position misalignment, is detected from the facility status data.

[0338] The step of automatically controlling a group of target facilities by transmitting a fire control package to facility controllers involves transmitting the generated fire control package to field facility controllers via a wired or wireless communication network and tracking the execution status of each control command based on execution acknowledgments and feedback received from the facility controllers. For example, if an acknowledgment for a fire hydrant water supply control command is not received, retransmission may be performed; or if the inability to operate a specific pump or valve is detected, a replacement pump or bypass piping route capable of covering the same target water supply section may be selected to regenerate a reduced package. Additionally, the system may be configured to stably maintain water pressure and flow rates in a fire situation by periodically collecting updated values ​​from pressure and flow sensors to recalculate water pressure and flow control errors, and by performing closed-loop updates to fine-tune pump rotation speeds and valve opening rates.

[0339] In addition, the step of calculating the fire hydrant water supply control parameters comprises: querying the fire hydrant identifier, fire hydrant installation location, nozzle diameter per fire hydrant, allowable water pressure range per fire hydrant, and allowable flow rate range per fire hydrant from the fire hydrant system configuration table for fire hydrants included in the target water supply section; determining the upstream reference water pressure and downstream end water pressure of the target water supply section based on the measurements of the pressure sensor and flow sensor, and calculating the section pressure drop (ΔP_drop) defined as the difference between the upstream reference water pressure and the downstream end water pressure; querying the pipe diameter, pipe length, roughness coefficient according to pipe material, and number of pipe bends of the water supply pipe included in the target water supply section from the fire hydrant system configuration table, and estimating the effective friction coefficient and local loss coefficient of the water supply pipe based on the section pressure drop and the measurements of the flow sensor. A step of determining a preliminary target flow rate for a target water supply section based on the fire intensity index above, wherein the preliminary target flow rate is determined to be limited by the allowable flow rate range for each fire hydrant included in the target water supply section and the effective friction coefficient of the water supply piping; a step of obtaining a target water pressure and a profile target flow rate corresponding to the section identifier and the fire intensity index from the equipment control profile; a step of determining a final target flow rate by limiting the profile target flow rate to within the range of the minimum and maximum values ​​of the preliminary target flow rate based on the preliminary target flow rate and the profile target flow rate; a step of calculating a water pressure control error between the target water pressure and the downstream end water pressure, and calculating a flow rate control error between the final target flow rate and the actual flow rate of the section measured by the flow sensor; and a step of calculating a target pump head and a target rotational speed for a pressure pump based on the water pressure control error, the flow rate control error, and a pump performance curve table.The method may include: a step of calculating a target distribution flow rate for each fire hydrant based on a priority for each fire hydrant, a nozzle diameter for each fire hydrant, and a current flow rate measurement for each fire hydrant, so that the final target flow rate is distributed to a plurality of fire hydrants included in the target water supply section; a step of calculating a target opening rate for each flow control valve based on the target distribution flow rate for each fire hydrant, a valve flow coefficient (Cv) table, and a section pressure drop; and a step of generating a set of fire hydrant water supply control parameters including the target rotational speed and the target opening rate.

[0340] The step of querying specification information of fire hydrants included in the above-mentioned target water supply section is to identify fire hydrants included in the target water supply section from a fire hydrant system configuration table, and to configure specification records by querying the fire hydrant identifier, fire hydrant installation location, nozzle diameter for each fire hydrant, allowable water pressure range for each fire hydrant, and allowable flow rate range for each fire hydrant for each identified fire hydrant.

[0341] Here, the fire hydrant installation location can be expressed as at least one of the tunnel longitudinal coordinates (chain ridge), transverse offset, and installation elevation, and the nozzle diameter can be defined as a representative diameter value used to calculate water discharge performance and required flow rate. Additionally, the allowable water pressure range and allowable flow rate range for each fire hydrant refer to the operational limits and the amount of water that can be supplied by design that the nozzle, hose, and connecting piping of the corresponding fire hydrant can withstand, and can be utilized as reference data to prevent equipment damage caused by excessive pressurization. Furthermore, the device may be configured to exclude fire hydrants with inverted or missing ranges from the candidates and to record the reason for exclusion.

[0342] The step of determining the upstream reference water pressure and downstream end water pressure and calculating the section pressure drop is a step of determining the upstream reference water pressure and downstream end water pressure of the target water supply section based on the measured values ​​of the pressure sensor and the flow sensor, and then calculating the section pressure drop defined as the value obtained by subtracting t...

Claims

Claim 1 The method comprises: a step of collecting time-synchronized sensor data and equipment status data from a plurality of detection sensors and a plurality of firefighting equipment within a road tunnel; a step of generating fire event information including whether a fire event has occurred, a fire detection location, and a fire intensity indicator based on the sensor data and equipment status data; a step of mapping the fire detection location to a section identifier of the road tunnel and querying a previously stored equipment control profile corresponding to the section identifier and the fire intensity indicator to determine a group of equipment to be controlled and target operating parameters for each piece of equipment; a step of generating a firefighting control package including a control command for each of the group of equipment to be controlled and target operating parameters for each piece of equipment based on the group of equipment to be controlled and target operating parameters for each piece of equipment; and a step of transmitting the firefighting control package to equipment controllers to automatically control the group of equipment to be controlled; wherein the step of determining the group of equipment to be controlled and target operating parameters for each piece of equipment includes: a step of loading a previously stored equipment layout table corresponding to the fire detection location and the section identifier; A step of filtering candidate firefighting equipment based on the equipment separation distance between the fire detection location and the installation location information of the firefighting equipment, whether they are in the same section, and whether they are in adjacent sections, based on the equipment layout table and fire detection location; and a step of determining the candidate equipment as a group of equipment to be controlled, for each firefighting equipment included in the group of equipment to be controlled, querying target operating parameters corresponding to the section identifier and fire intensity index from the equipment control profile, and performing unit conversion and range restriction so that the target operating parameters conform to the parameter schema for each equipment type;The method comprises: a detection sensor including a smoke sensor, a gas sensor, a temperature sensor, a flame detection sensor, a camera, and an airflow measurement sensor; and the step of determining the candidate equipment as a group of control target equipment comprises: projecting the airflow velocity vector included in the sensor data onto the longitudinal unit vector of the road tunnel to calculate the tunnel longitudinal airflow velocity component, and defining the upstream and downstream directions according to the sign of the airflow velocity component; generating a set of deflection parameters reflecting the deflected diffusion of fire heat and combustion gases based on the longitudinal gradient angle of the tunnel longitudinal profile table corresponding to the section identifier mapped to the fire detection location and the tunnel longitudinal airflow velocity component; setting the fire detection location as the center point of a three-dimensional spatial coordinate system and modeling a dynamic spatial influence field in the form of an asymmetric ellipsoid in which the major axis expands in the downstream direction and the major axis contracts in the upstream direction based on the set of deflection parameters; classifying the candidate equipment filtered from the equipment placement table into fire extinguishing equipment, smoke exhaust equipment, and evacuation guidance equipment according to the equipment type, and mapping the three-dimensional installation coordinates of each classified candidate equipment to the dynamic spatial influence field. A method for controlling a firefighting control system for selectively controlling a road tunnel firefighting system according to a fire detection location, comprising: a step of calculating a spatial association index for each of the above candidate facilities by applying a distance weight inversely proportional to the distance from the fire detection location and a type weight corresponding to the operational purpose of each facility type; and a step of determining only the candidate facilities for which the spatial association index exceeds a preset activation threshold value as the group of facilities to be controlled. Claim 2 In claim 1, the step of calculating the spatial correlation index comprises: for each of the candidate facilities, a step of calculating a longitudinal distance component and a lateral distance component between the 3D coordinates of the fire detection location and the 3D installation coordinates of the candidate facility; a step of determining whether the candidate facility is positioned upstream or downstream relative to the fire detection location based on the upstream and downstream directions, and selecting an upstream attenuation coefficient or a downstream amplification coefficient according to the determination result; a step of determining an upward / downward slope correction coefficient according to the longitudinal gradient based on the longitudinal gradient angle, and correcting the slope correction coefficient based on the difference in elevation between the installation height of the candidate facility and the fire detection location; and a step of calculating an effective separation distance for each candidate facility based on the longitudinal distance component, lateral distance component, and slope correction coefficient, and the upstream attenuation coefficient or downstream amplification coefficient. A control method for a firefighting control system that selectively controls a road tunnel firefighting system according to a fire detection location, comprising: a step of calculating a distance weight inversely proportional to the effective separation distance, wherein if the effective separation distance is less than a preset minimum distance, the minimum distance is applied as a lower limit value, and if the effective separation distance exceeds a preset maximum distance, the distance weight is set to 0; a step of querying a type weight corresponding to each of the fire extinguishing system, smoke exhaust system, and evacuation guidance system from a type weight table preset to correspond to the operating purpose of each equipment type; and a step of calculating a spatial association index for each candidate equipment, normalized to a value between 0 and 1, by multiplying the distance weight and the type weight. Claim 3 delete

Images