Intelligent building fire monitoring method

Abstract

The invention discloses an intelligent building fire monitoring method, and relates to the field of building fire monitoring, and the method comprises the steps: collecting three types of fire characterization parameters: temperature, smoke concentration and carbon monoxide concentration; the method comprises the following steps: acquiring fire characterization parameters of each node at an initial frequency, comparing the fire characterization parameters to trigger primary warning, then increasing the acquisition frequency for a primary warning node and adjacent nodes, and continuously acquiring multiple groups of data; fire source initial coordinates are determined through positioning anchor points and parameter gradient differences, the coordinate range is continuously corrected, and specific fire source position coordinates are provided; and dynamically generating a fire diffusion path according to the smoke diffusion path of the monitoring node in the fire source coordinate range, tracking the fire diffusion path and providing diffusion direction prediction. The accuracy and timeliness of intelligent building fire monitoring are improved, the quick response to fire dangerous cases is facilitated, and the safety of buildings and personnel is guaranteed.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of building fire monitoring, and relates to a smart building fire monitoring method. BACKGROUND

[0002] In modern urban development, smart buildings are multifunctional complexes integrating office, business, and residence. The internal electrical equipment is dense, the personnel flow is frequent, and the types of combustible materials are diverse. These factors significantly increase the risk of fire. Once a fire occurs, it will not only cause serious casualties and property losses, but also seriously affect the normal operation of the building and the public safety of the city. Therefore, monitoring the fire of the smart building and discovering fire hazards in a timely manner are crucial for protecting life and property safety and maintaining social stability.

[0003] Current technologies have proposed methods for smart building fire monitoring. For example, the invention patent with publication number CN118781734A proposes a building intelligent electrical fire monitoring system and method. Based on the multi-time point leakage current and temperature data of the target area distribution box, a space-time input tensor is constructed. Through time sequence feature coding, dynamic change patterns are extracted, and key feature enhancement is performed using a channel and space dual attention mechanism. The feature expression is optimized by fusing mean pooling and feature offset compensation. Finally, a classifier is used to realize the binary classification decision of fire alarm. This solves the problem of low early warning accuracy caused by the difficulty of capturing dynamic coupling features of current and temperature in traditional electrical fire monitoring.

[0004] Another example is the invention patent with publication number CN119992807A, which proposes a building intelligent electrical fire monitoring method and system. Based on the leakage current and temperature time series data of the floor distribution box, joint time-frequency features are constructed by time delay cross-correlation analysis. Multi-scale intrinsic mode functions and trend items are extracted by applying ensemble empirical mode decomposition. The frequency domain characteristics of the multi-scale intrinsic mode functions are analyzed and the trend features are fitted by combining Hilbert transform. The multi-dimensional features are fused to form a feature vector, which is input into a pre-trained random forest model to realize fire risk grading prediction. This solves the problem of insufficient early risk grading warning accuracy caused by the difficulty of capturing multi-scale abnormal features of current and temperature signals in traditional electrical fire monitoring.

[0005] Although the above two existing technologies have achieved certain results in smart building fire monitoring, there are still the following shortcomings: first, the diversity of smart building fire causes, and the existing technology only monitors the leakage current and temperature of the distribution box or the signal features after decomposition to infer the occurrence of fire, which is difficult to cover the entire building space, resulting in the inability to capture fire features comprehensively, especially for fires caused by flammable material burning, kitchen open fire, and other non-electrical reasons, which delays the fire response.

[0006] Secondly, the fire development has time dynamics of initial stage, spreading stage and attenuation stage. The existing technology collects data at a fixed frequency and outputs static alarm results through a pre-trained model. This fixed frequency monitoring mechanism cannot adapt to the pace of fire evolution, and continuously collects data of all distribution boxes at a high frequency, increasing system load and energy consumption.

[0007] In addition, the existing technology only analyzes the time correlation of parameters, and the fire information prediction only stays at the risk level, which cannot provide the location coordinates of the fire source, leading to incorrect evacuation or fire extinguishing direction, and unable to predict the fire spread path.

[0008] Finally, the existing technology only analyzes the parameter time sequence change of a single distribution box or the coupling window between floors, which cannot track the fire spread path and provide fire spread direction prediction, leading to personnel entering dangerous areas and firefighters unable to block key spread nodes according to the spread path. SUMMARY

[0009] In view of this, in order to solve the problems raised in the background art, the present application provides a smart building fire monitoring method.

[0010] The purpose of the present application can be achieved by the following technical solutions: a smart building fire monitoring method, comprising: distributing monitoring nodes in the building space, collecting fire characteristic parameters of each node at an initial frequency, including temperature, smoke concentration and carbon monoxide concentration.

[0011] When at least one of the fire characteristic parameters of any monitoring node exceeds the preset threshold, a primary warning is triggered.

[0012] The data acquisition frequency of the primary warning node and adjacent nodes is increased, and multiple sets of data are continuously obtained. If the fire characteristic parameters continuously increase or all exceed the threshold value for multiple times, it is determined as a fire risk event. If the fire characteristic parameter value falls back to the safe range, the warning is removed.

[0013] Taking the first node that triggers the primary warning as the reference point and based on the number and spatial distribution of the nodes that trigger the fire risk event, the three nodes with the earliest response time are selected as the positioning anchor points. The initial coordinates of the fire source are determined according to the gradient difference of the parameter values of each anchor point and the spatial position relationship.

[0014] The data of the nodes that trigger the new fire risk event is continuously received to correct the fire source coordinate range multiple times.

[0015] The smoke spread path of the monitoring nodes in the fire source coordinate range is dynamically generated to execute a hierarchical response.

[0016] Compared with the prior art, the present application has the following advantages: (1) The present application collects three types of fire characteristic parameters, temperature, smoke concentration and carbon monoxide concentration, covering the characteristics of electrical fires and fires caused by non-electrical reasons such as kitchen open fire, solving the problem of missing detection of non-electrical fires caused by single monitoring dimension.

[0017] (2) The present application collects fire characteristic parameters of each node at the initial frequency, compares the primary warning triggered by the fire characteristic parameters, and continuously obtains multiple sets of data, solving the problems of low efficiency of fixed frequency collection and missing fire source position, and the problem of increasing system load and energy consumption caused by continuous high frequency collection.

[0018] (3) The present application determines the initial coordinates of the fire source by positioning the anchor point and the parameter gradient difference, and continuously corrects the coordinate range, providing specific fire source position coordinates, solving the problem of incorrect evacuation or fire extinguishing direction caused by the inability to provide fire source position coordinates.

[0019] (4) The present application dynamically generates a fire spread path according to the smoke diffusion path of the monitoring nodes in the fire source coordinate range, traces the fire spread path and provides a spread direction prediction, solving the problems of the prior art that cannot trace the fire spread path and cannot predict the spread direction, leading to personnel entering dangerous areas and firefighters being unable to block key spread nodes, providing accurate path reference for personnel evacuation and fire rescue. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0021] Figure 1 The method embodiment of the present application is a flow chart.

[0022] Figure 2 The fire source initial coordinate positioning flow chart of the present application.

[0023] Figure 3 The fire source range correction flow chart of the present application. DETAILED DESCRIPTION

[0024] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0025] Please refer to Figure 1 As shown in the drawings, the present application provides a smart building fire monitoring method, comprising: S1. distributing monitoring nodes in the building space, collecting fire characteristic parameters of each node at an initial frequency, including temperature, smoke concentration and carbon monoxide concentration.

[0026] It should be noted that the initial frequency can be based on the general collection frequency in the field of fire monitoring as the initial frequency benchmark, balancing real-time and resource consumption. It is the basic state setting of fire monitoring, and its core purpose is to achieve the normalization coverage of the building space under the premise of balancing system load and monitoring demand.

[0027] Temperature usually uses temperature sensors to convert temperature signals into electrical signals by sensing heat radiation or heat conduction in the environment, achieving numerical collection.

[0028] Smoke concentration generally uses smoke sensors, ion type detects the influence of smoke particles on ionization current, photoelectric type detects the scattering or absorption effect of smoke on light, judges the smoke concentration and converts it into quantitative data.

[0029] Carbon monoxide concentration mainly relies on carbon monoxide sensors, which generate electrical signals through chemical reactions between carbon monoxide and internal substances of the sensor, such as electrochemical oxidation, to reflect the concentration value.

[0030] The monitoring nodes are specifically deployed as follows: according to the building plan, the building space is divided into several independent functional areas.

[0031] Specifically, the detailed building plan of the building to be monitored is obtained, including wall, door, window and floor information, the spatial structure of the building is analyzed based on the plan, different space units formed by physical structures such as walls, fire doors and stairwells are identified, and the entire building space is divided into multiple independent, logically or physically separated functional areas, such as independent offices, conference rooms, corridors, equipment rooms, warehouses, lobbies, open office areas, stairwells and elevator shafts.

[0032] In each independent functional area, key risk locations are identified based on fire dynamics characteristics. Areas with continuously running high-power electrical equipment clusters, combustible materials storage, and daily personnel flow exceeding the preset threshold are marked as high-risk areas. Areas that do not meet high-risk characteristics and have personnel activities are marked as regular risk areas.

[0033] It should be noted that the identification of key risk locations is based on the use of the area, such as computer rooms, warehouses, offices, etc., and the types, quantities, distribution density of combustible materials in the area, as well as possible ignition sources, such as electrical equipment, open flame sources, etc., are counted.

[0034] The preset thresholds are based on the limits on personnel density and combustible storage in national or local fire protection regulations, and are directly obtained by using the thresholds in the standards.

[0035] Clusters of high-power electrical equipment operating continuously under high load for extended periods of time are prone to fires caused by aging circuits, short circuits, and poor heat dissipation, and these fires can spread rapidly. For example, areas like data center computer rooms and power distribution rooms are prone to fires due to combustible materials, and high personnel turnover makes evacuation difficult and increases the risk of casualties in the event of a fire. This combination significantly increases the probability of fire and the severity of its consequences, thus marking these areas as high-risk.

[0036] Although conventional risk areas have fewer combustible materials or limited power equipment, due to the presence of human activities, fires may still endanger life safety once they occur, requiring basic coverage monitoring to balance monitoring costs and safety needs.

[0037] Grid node deployment is performed in high-risk areas, and basic coverage radius deployment is performed in general risk areas.

[0038] It's important to note that grid deployment refers to evenly distributed nodes at a high density, ensuring comprehensive coverage across the entire region. For example, in a data center room, grid nodes can capture subtle changes like localized temperature spikes and early smoke spread in real time, avoiding missed alerts due to blind spots. Fire parameters in high-risk areas change rapidly and complexly, and densely populated nodes can more accurately capture parameter gradient differences, providing sufficient data support for subsequent fire source location and fire spread path assessment, meeting the need for rapid response.

[0039] Basic coverage radius deployment is based on the monitoring range of nodes, such as the effective sensing radius of a smoke sensor. This ensures that every location in the area is within the monitoring range of at least one node, avoiding resource waste while meeting basic safety requirements. Conventional risk areas have low levels of human activity and equipment usage, resulting in fewer fire hazards. Basic coverage is sufficient to detect early-stage fires, ensuring a balanced balance between safety and cost-effectiveness.

[0040] Record the three-dimensional coordinates of all monitoring nodes and build a mapping table of adjacent node relationships.

[0041] S2. Trigger a primary warning when at least one of the fire characterization parameters of any monitoring node exceeds a preset threshold.

[0042] Specifically, the time series data sequences of the fire characterization parameters collected by all monitoring nodes at an initial frequency are received in real time.

[0043] For each monitoring node, when it is first detected that any of the fire characterization parameters of the node exceeds the threshold of the corresponding type, the node is marked to enter a primary monitoring state.

[0044] It should be noted that the temperature threshold can be set to be higher than the normal ambient temperature but not reach the critical value of open fire combustion, so as to capture the early warming trend.

[0045] The smoke concentration threshold needs to distinguish the concentration difference between normal environment and early fire smoke, so as to avoid false triggering due to non-fire factors such as cooking oil smoke.

[0046] The carbon monoxide concentration threshold needs to refer to the human safety limit value and the concentration level generated by incomplete combustion in the early stage of fire.

[0047] The time series data collected subsequently by the node is continuously monitored during the maintenance of the primary monitoring state.

[0048] If any parameter again exceeds the limit in the newly collected data, or the already exceeded parameter remains in the exceeded state for a continuous time step, it is determined that an abnormal condition is triggered.

[0049] It should be noted that the first exceeding of a single parameter may be caused by an occasional disturbance due to non-fire factors. By judging the re-occurrence of any parameter exceeding the limit, the persistence of the exceeding phenomenon can be verified. For the case where the already exceeded parameter remains in the exceeded state for a continuous time step, even if there is no new parameter exceeding the limit, the persistent exceeding state itself indicates that the risk is continuing, which filters out occasional disturbances and covers persistent risks, realizes accurate judgment of abnormal state, and provides a reliable basis for subsequent generation of primary warning events.

[0050] When the monitoring node first triggers the abnormal condition, a primary warning event is generated, and the trigger node identifier and three-dimensional coordinates associated with the primary warning event, the first exceeding parameter type and timestamp, and the historical sequence of all exceeding parameters are recorded.

[0051] At the early stage of fire occurrence, parameters such as temperature, smoke concentration, or carbon monoxide concentration will first show abnormal changes. By monitoring the exceeding of any parameter in a single node, abnormal signals can be quickly captured in the early stage of fire, which can help to gain time for subsequent risk confirmation and response, and avoid delaying the early warning opportunity due to waiting for multiple parameters to exceed the limit at the same time.

[0052] S3. The data collection frequency of the primary warning node and adjacent nodes is increased, and multiple sets of data are continuously obtained. If the fire characteristic parameter continuously increases for multiple times or all exceeds the threshold value, it is determined as a fire risk event. If the fire characteristic parameter value falls back to the safe range, the warning is removed.

[0053] The specific content of increasing the data collection frequency of the primary warning node and adjacent nodes is as follows: receiving the primary warning event and extracting the trigger node identifier and its three-dimensional coordinates in the event.

[0054] Based on the adjacent node relationship mapping table, the trigger node is taken as the core node, and all adjacent nodes having a direct topological connection relationship with the core node are retrieved through the mapping table.

[0055] Specifically, the unique identifier of the monitoring node triggering the event is extracted from the received primary warning event, and the corresponding record is retrieved in the pre-stored adjacent node relationship mapping table using the node identifier as the index. The direct adjacent node ID list field value is extracted from the retrieved record, it is confirmed that the ID exists in the current system valid node library, and it is confirmed that the connection relationship between the node and the core node is physically adjacent. All adjacent node IDs that pass the verification are output to form a direct adjacent node set. If there is no node record in the mapping table, return an empty set. If the adjacent node ID list contains invalid nodes, filter them out and output the valid nodes. The collection frequency upgrade instruction is issued to the core node and all nodes in the adjacent node set.

[0056] The fire risk event determination step is as follows: obtaining the time series data stream uploaded by the primary warning node and the adjacent node set at the upgraded frequency, and independently constructing the fire characteristic parameter change sequence for each node.

[0057] If any fire characteristic parameter of the node in the set is greater than the corresponding value of the previous period in each period of the continuous multiple collection periods, or all fire characteristic parameters exceed the corresponding type threshold value, the node is marked as a fire risk node.

[0058] It should be noted that any fire characteristic parameter in each period of the continuous multiple collection periods is greater than the corresponding value of the previous period. This is for the dynamic process of the initial fire parameter showing a continuous upward trend. This continuous increasing trend is a typical feature of fire spread. By identifying this trend, potential fire risk nodes can be locked in advance before the parameter exceeds the threshold, avoiding delays in response due to waiting for the parameter to meet the standard.

[0059] All the measured values of the fire characteristic parameters in the current collection cycle exceed the corresponding type threshold, which refers to the situation that the parameters have completely broken through the safety threshold. The simultaneous over-limit of temperature, smoke concentration, and carbon monoxide concentration means that the fire has developed to a relatively obvious stage, and at this time, the node can be directly determined to be in a clear fire risk without relying on trend judgment to confirm the authenticity of the risk.

[0060] The risk event record is generated, including the risk node identifier and three-dimensional coordinates, the trigger condition type, and the first collection cycle timestamp when the condition is met.

[0061] The specific content of the above warning release is as follows: obtaining the latest collected fire characteristic parameter data of the node in the fire risk and the primary warning.

[0062] When all the fire characteristic parameters of the fire risk node are below the corresponding type threshold in two consecutive collection cycles, the risk release condition is determined.

[0063] It should be noted that the parameters of the fire risk node may be temporarily lower than the threshold due to environmental changes, but this does not mean that the risk has been completely released. Requiring two consecutive collection cycles to be below the threshold can filter accidental parameter fluctuations, and through two consecutive verifications, it is confirmed that the parameter decline is a stable trend rather than a transient phenomenon, ensuring the reliability of the risk release judgment.

[0064] When all the fire characteristic parameters of the primary monitoring state node are below the threshold in three consecutive collection cycles, the primary warning release condition is determined.

[0065] It should be noted that the node in the primary monitoring state may be at a high collection frequency, continuously occupying system resources. By verifying the risk release for three consecutive cycles, and then terminating the primary monitoring and restoring the initial frequency, it can avoid repeated warnings due to early release or resource waste due to delayed release, and achieve efficient use of system resources.

[0066] If the node meets the risk release condition, the fire risk node marker of the node is cleared, and its collection frequency is restored to the initial frequency. If the node meets the primary warning release condition, the primary monitoring state of the node is terminated and the initial collection frequency is restored.

[0067] It should be noted that when the node meets the risk release or primary warning release condition, it means that it has escaped from the fire risk or the primary monitoring state, and there is no need to maintain high-frequency collection or special monitoring mode. At this time, the marker is cleared and the initial collection frequency is restored, which can stop the excessive monitoring of the node, reduce the resource consumption of data transmission, storage, and processing, concentrate system resources on areas that still have risks, and improve the overall operating efficiency.

[0068] When the node is out of the state, the adjacent node relationship mapping table is retrieved, and if all the adjacent nodes are not in the fire risk state or the primary monitoring state, a frequency recovery instruction is issued to the adjacent node to cancel the warning.

[0069] It should be noted that the adjacent node and the core node have a topological connection relationship, and the state change has relevance. When the core node is out of the state and the adjacent nodes are all risk-free, it indicates that the risk of the entire associated area has subsided, and by uniformly restoring the frequency of the adjacent nodes, the monitoring mode of the area can be returned to normal, avoiding local resource overload caused by continuous high-frequency monitoring of part of the nodes, and maintaining the coordination of the monitoring resource configuration in the area.

[0070] The temperature, smoke concentration and other parameters of the initial fire often change gradually, such as slow accumulation of smoke in smoldering fire, or temporary fluctuations such as temporary increase in smoke concentration caused by cooking oil smoke. It is difficult to distinguish between real fire and interference factors by collecting data at the initial frequency, and increasing the collection frequency can obtain more dense time series data to capture the subtle change trend of the parameters and provide sufficient basis for risk determination.

[0071] S4. Taking the first node triggering the primary warning as the reference point and based on the number and spatial distribution of the nodes triggering the fire risk event, the three nodes with the earliest response time are preferentially selected as the positioning anchor points, and the initial coordinates of the fire source are determined according to the gradient difference of the parameter values of each anchor point and the spatial position relationship.

[0072] The specific steps of preferentially selecting the three nodes with the earliest response time as the positioning anchor points are as follows: extracting the three-dimensional coordinates of the first node triggering the primary warning, and setting it as the reference point for positioning the fire source.

[0073] Taking the reference node coordinates as the center, excluding the risk nodes with a straight-line distance from the reference node exceeding a preset spatial correlation threshold, and forming an initial risk node set.

[0074] It should be noted that the risk nodes with a straight-line distance from the reference node exceeding the preset spatial correlation threshold may belong to other independent areas and have low relevance to the current fire source. Excluding these nodes can avoid irrelevant data interference and ensure that the data in the initial risk node set come from the core area near the fire source, improving the accuracy of subsequent positioning.

[0075] If the number of nodes is greater than or equal to three, the first collection period timestamp of each node triggering the fire risk event in the initial risk node set is extracted, the nodes are sorted from early to late according to the timestamp, and the three nodes with the earliest timestamp are selected as the positioning anchor points.

[0076] It should be noted that the temperature, smoke and other parameters generated after the fire occur will spread to the surrounding with time, and the nodes closer to the fire source will usually trigger the fire risk event earlier. Selecting the time stamp of the three earliest nodes as the positioning anchor point can maximize the guarantee that these nodes are directly affected by the fire source and provide a core reference for positioning.

[0077] If the number of nodes is less than three, the first-level adjacent nodes of each node in the initial risk node set and the reference node based on the adjacent node are retrieved.

[0078] It should be noted that the determination of the initial coordinates of the fire source needs to be based on the spatial position and parameter gradient relationship of multiple nodes, and at least three nodes can provide basic geometric reference for three-dimensional positioning. When the number of initial risk node sets is less than three, retrieving the first-level adjacent nodes can supplement the candidate nodes and avoid the positioning logic from being unable to proceed due to insufficient number of nodes.

[0079] The adjacent nodes currently in the primary monitoring state and not released from warning are screened out.

[0080] The adjacent nodes that meet the conditions are added to the initial risk node set to form an expanded risk node set.

[0081] If the number of nodes in the expanded risk node set is greater than or equal to three, the three nodes with the highest ranking are selected as the positioning anchor points.

[0082] If the number of nodes in the expanded risk node set is still less than three, all nodes in the expanded risk node set are taken as the positioning anchor points.

[0083] It should be noted that the determination of the initial coordinates of the fire source depends on the spatial position and parameter gradient relationship of multiple nodes, and three nodes are the basic threshold for accurate positioning in three-dimensional space. When the number of expanded risk node sets is greater than or equal to three, the three nodes with the highest ranking are selected as the anchor points, which can meet the minimum number requirement for positioning and ensure that these nodes are the earliest affected by the fire and the most relevant to the fire source, providing a reliable reference for positioning.

[0084] If the number of nodes after expansion is still less than three, it means that the monitoring nodes in the fire risk area respond less, such as in the early stage of fire or limited spread range, but the positioning process still needs to be based on the existing data. Taking all nodes as anchor points can use the only available node information to construct the simplest positioning basis in the case of limited data, such as determining the straight line range with two points or defining the surrounding area with a single point, avoiding the interruption of the positioning process due to insufficient number of nodes, and ensuring that the system can continuously track the fire source dynamics.

[0085] Reference Figure 2As shown in the above fire source initial coordinate determination step, the three-dimensional coordinate data of all positioning anchor points is extracted, and the fire characteristic parameter values recorded by each anchor point when the fire risk event is triggered are synchronously obtained.

[0086] The fire characteristic parameter values between the anchor points are compared, and the decreasing direction of the parameter values from high to low is identified. If the same type of parameter values of two or more anchor points present a one-way decreasing trend, it is determined that the gradient direction of the parameter is consistent with the direction of the spatial position change.

[0087] If the gradient directions of different types of parameters are inconsistent, the gradient direction of the smoke concentration parameter is prioritized as the main determination basis.

[0088] It should be noted that from the perspective of fire dynamics, smoke is one of the most widely spread and easiest to monitor characteristics in the early stage after a fire occurs. Compared to temperature and carbon monoxide concentration, the spread of smoke more directly reflects the path and direction of fire spread. The smoke produced by the fire will spread around with the air flow, and the concentration gradient is more stable in relation to the spatial position of the fire source, and is less affected by local interference factors.

[0089] The anchor point with the highest parameter value is marked as the core reference point, and the spatial orientation of the remaining anchor points relative to the core reference point is analyzed to identify the farthest anchor point in the parameter decreasing direction as the boundary reference point.

[0090] It should be noted that the anchor point with the highest parameter value is closest to the fire source location. After a fire occurs, the parameter value in the burning core area is usually the highest and gradually decreases with increasing distance. Marking it as the core reference point can determine the reference origin of the fire source positioning and ensure that the subsequent analysis is centered around the most likely fire source.

[0091] When the gradient direction is linearly related to the spatial orientation, the initial coordinates of the fire source are determined in the direction of the line connecting the core reference point to the boundary reference point, and in the spatial region close to the core reference point.

[0092] When the gradient direction is not linearly related to the spatial orientation, the initial coordinates of the fire source are determined within the geometric polygon formed by all anchor points, and in the smallest convex hull region containing the core reference point.

[0093] It should be noted that when the gradient direction is linearly related to the spatial orientation, it indicates that the fire may spread in a single direction. In this case, the initial coordinates of the fire source are determined in the direction of the line connecting the core reference point to the boundary reference point, and in the region close to the core reference point, which conforms to the rule that the highest parameter is closest to the fire source. The core reference point has the highest parameter value, which is the anchor point closest to the fire source. Locking the near-core region along the line connecting the core reference point to the boundary reference point can accurately narrow the fire source range and avoid excessive expansion of the positioning range due to the single spread direction.

[0094] When the gradient direction has no clear linear correlation with the spatial orientation, the fire spread path is more complex, and it is difficult to lock the fire source by a single direction. The initial coordinates of the fire source are determined within the geometric polygon formed by all anchor points, and the minimum convex hull region containing the core reference point can be constructed by the spatial distribution of the anchor points to cover the minimum spatial range of all affected nodes. The core reference point is included as the node with the highest parameter, which ensures that the fire source is within the region to avoid missing the positioning due to complex spread direction.

[0095] If the core reference point coordinates of the linear direction region are taken as the center point, the cubic space range is generated by extending the preset safety distance along the connecting direction, and if the convex hull region extracts the extreme points of the convex hull vertex coordinates to form the minimum circumscribed cubic range.

[0096] It should be noted that, whether it is a linear direction region or a convex hull region, by generating a cubic space range, the region where the fire source may exist can be converted into three-dimensional data containing vertex coordinates, such as a cubic with clear length, width, and height. This quantitative definition facilitates system storage, calculation, and visual display, and also provides a comparable benchmark boundary for subsequent addition of node data to correct the coordinate range.

[0097] Generate three-dimensional spatial data containing the vertex coordinates of the cubic range as the initial coordinates of the fire source.

[0098] S5. Continuously receive trigger node data of new fire risk events to correct the fire source coordinate range multiple times.

[0099] Reference Figure 3 As shown, the identification, three-dimensional coordinates, and trigger timestamp of the new fire risk event node are extracted.

[0100] Calculate the spatial distance between the new node and the current fire source coordinate range. If it is within the preset spatial correlation threshold or is a neighboring node of the nodes within the current fire source range, it is included in the correction node set.

[0101] It should be noted that by limiting the spatial range, nodes that are too far away and unrelated to the current fire source can be filtered out, reducing invalid data processing and focusing the correction work on the area that is truly affected by the fire. At the same time, including neighboring nodes in the set can capture early signals of the spread of the fire to the surrounding area, improving the timeliness of the correction of the fire source coordinate range.

[0102] If the new node is located outside the current fire source range, the fire source range is expanded in the direction of the node according to its fire characteristic parameter value gradient and spatial orientation to form a minimum convex hull region covering the new hot spot.

[0103] If the new node is located within the current fire source range, the range boundary accuracy is optimized according to its relative position of the parameter value.

[0104] When the new node is located outside the fire source range, it indicates that the fire may have spread in that direction. At this time, the new hotspot is covered by the minimum convex hull area according to its parameter gradient and spatial orientation extension range, which can not only timely include the new fire affected area, but also avoid the positioning ambiguity caused by excessive expansion of the range. When the new node is located inside, the relative height of the parameter value can reflect the fine distribution of the fire source core area. Optimizing the boundary accuracy can make the fire source range closer to the actual burning center.

[0105] After the fire occurs, the fire will continue to spread over time, and the new risk node can reflect the new trend of fire spread. By correcting the coordinate range through these node data, the fire source positioning can be gradually refined from the initial rough range, which is more consistent with the actual fire location and affected area, and can avoid the response lag caused by the initial positioning deviation.

[0106] S6. Dynamically generate fire spread path according to smoke diffusion path of monitoring nodes in fire source coordinate range and perform hierarchical response.

[0107] The specific content of the above fire spread path is as follows: extract three-dimensional coordinates and smoke concentration time series data of all monitoring nodes in the fire risk state in the fire source coordinate range.

[0108] Compare the smoke concentration gradient values of adjacent nodes to identify the decreasing direction from high to low.

[0109] If multiple adjacent node groups show continuous one-way concentration decrease, it is determined as an effective spread path segment.

[0110] It should be noted that determining the node group with continuous one-way decrease as an effective spread path segment can exclude isolated node concentration anomalies caused by local airflow disturbance and other factors, ensuring that the identified spread path conforms to the actual law of fire development, and providing accurate basic data for subsequent formation of the initial spread main path starting from the node with the highest smoke concentration in the fire source range.

[0111] Starting from the node with the highest smoke concentration in the fire source range, the initial spread main path is formed by connecting nodes in the extension direction of the effective spread path segment.

[0112] When a new node is included in the fire source range, if the smoke concentration gradient of the node is consistent with the direction of the main path, the length of the spread path is extended, and if a new decreasing direction is formed, a branch path is generated and the spread intensity level is marked.

[0113] By dynamically adjusting the path according to the gradient direction of the new node, the generated fire spread path can be more consistent with the actual burning situation, and the marked spread intensity level can distinguish the danger level of different paths, providing targeted basis for subsequent hierarchical response such as evacuation guidance and fire extinguishing resource allocation, for example, prioritizing the path with high intensity level.

[0114] The specific contents of the above-mentioned graded response are as follows: when the fire spread path only contains the main path and has no branches, and the smoke concentration in the core area does not reach the high-risk threshold, an early warning message will be sent to the fire control room, and the regional ventilation equipment will be turned on to issue a low-level alarm through the sound and light device.

[0115] When a medium-intensity branch appears in the fire spread path or the main path exceeds the safe range, video surveillance is activated and dispatch personnel are dispatched to confirm the closure of fire doors in relevant areas, and people in the affected areas are guided to evacuate through broadcasting.

[0116] When the fire covers more than a single functional area or the parameters reach a high-risk threshold, an alarm will be sent to the city fire command center and data will be uploaded to activate the automatic fire extinguishing system. The highest level alarm will be sounded in the entire building, and the elevator will be forced to land to guide all personnel to evacuate.

[0117] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0118] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0119] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0120] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0121] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A smart building fire monitoring method, characterized by: include: Distributed deployment of monitoring nodes in the building space to collect fire characteristic parameters of each node at an initial frequency, including temperature, smoke concentration, and carbon monoxide concentration; When at least one of the fire characterization parameters of any monitoring node exceeds the preset threshold, a primary warning is triggered; Increase the data collection frequency for primary warning nodes and adjacent nodes, and continuously acquire multiple sets of data. If the fire characterization parameter values ​​collected multiple times continuously increase or all exceed the threshold, it is determined to be a fire risk event. If the fire characterization parameter values ​​fall back to a safe range, the warning is lifted; Taking the first node that triggers the primary warning as the reference point and based on the number and spatial distribution of triggering nodes of the fire risk event, the three nodes with the earliest response time are preferentially selected as positioning anchor points. The initial coordinates of the fire source are determined according to the gradient difference of the parameter values ​​of each anchor point and their spatial position relationship. Continuously receive trigger node data of new fire risk events and revise the fire source coordinate range multiple times; The fire spread path is dynamically generated according to the smoke spread path of the monitoring nodes within the fire source coordinate range and a hierarchical response is performed.

2. The smart building fire monitoring method according to claim 1, characterized in that: The monitoring nodes are specifically deployed as follows: Divide the building space into several independent functional areas according to the building floor plan; Identify key risk locations within each independent functional area based on fire dynamics characteristics. Areas with continuously running high-power electrical equipment clusters, combustible materials storage, and daily personnel flow exceeding a preset threshold are marked as high-risk areas. Areas that do not meet high-risk characteristics but have personnel activities are marked as regular risk areas. Grid node deployment is implemented for high-risk areas, and basic coverage radius deployment is implemented for regular risk areas; Record the three-dimensional coordinates of all monitoring nodes and build a mapping table of adjacent node relationships.

3. The smart building fire monitoring method according to claim 1, characterized in that: The primary warning triggering steps are as follows: Receive in real time the time series data of fire characterization parameters collected by all monitoring nodes at the initial frequency; For each monitoring node, when any fire characteristic parameter of the node is detected to exceed the corresponding type threshold for the first time, the node is marked to enter the primary monitoring state; Continuously monitoring the time series data subsequently collected by the node while maintaining the primary monitoring state; If any parameter exceeds the limit again in the newly collected data, or the exceeded parameter remains in the exceeded limit state within the continuous time step, the abnormal condition is determined to be triggered; When a monitoring node triggers an abnormal condition for the first time, a primary warning event is generated, and the triggering node identifier and three-dimensional coordinates associated with the primary warning event, the first exceeded parameter type and time stamp, and the historical sequence of all exceeded parameters are recorded.

4. The smart building fire monitoring method according to claim 2, characterized in that: The specific contents of increasing the data collection frequency of primary warning nodes and adjacent nodes are as follows: Receive primary warning events and extract trigger node identifiers and their three-dimensional coordinates in the events; Based on the adjacent node relationship mapping table, the trigger node is used as the core node, and all adjacent nodes that have a direct topological connection relationship with the core node are retrieved through the mapping table; Send a collection frequency upgrade instruction to the core node and all nodes in the set of its adjacent nodes.

5. The smart building fire monitoring method according to claim 1, characterized in that: The fire risk event determination steps are as follows: Obtain the time series data stream uploaded by the primary warning node and the adjacent node set according to the upgrade frequency, and independently construct the fire characterization parameter change sequence for each node; If the parameter value of any fire characterization parameter of the node in the set is greater than the corresponding value of the previous cycle in each of multiple consecutive acquisition cycles, or the measured values ​​of all fire characterization parameters in the current acquisition cycle exceed the corresponding type threshold, the node is marked as a fire risk node; Generate risk event records including risk node identification and three-dimensional coordinates, trigger condition type, and timestamp of the first collection cycle that meets the conditions.

6. The smart building fire monitoring method according to claim 1, characterized in that: The specific contents of the warning release are as follows: Obtain the latest collected fire characterization parameter data for nodes at fire risk and primary warning; When all fire characterization parameters of a fire risk node are lower than the corresponding type threshold in two consecutive acquisition cycles, the risk elimination condition is determined; When all fire characterization parameters of the primary monitoring status node do not exceed the threshold value in three consecutive acquisition cycles, the primary warning release condition is determined; If the node meets the risk relief conditions, the fire risk node flag of the node is cleared and its collection frequency is restored to the initial frequency. If the node meets the primary warning relief conditions, its primary monitoring state is terminated and the initial collection frequency is restored. When the node is released from the state, the adjacent node relationship mapping table is retrieved. If all its adjacent nodes are not in the fire risk state or the primary monitoring state, a frequency recovery instruction is sent to the adjacent node to release the warning.

7. The smart building fire monitoring method according to claim 1, characterized in that: The specific steps of preferentially selecting the three nodes with the earliest response time as positioning anchor points are as follows: Extract the three-dimensional coordinates of the first node that triggers the primary warning and set it as the fire source positioning reference point; Taking the coordinates of the reference node as the center, exclude the risk nodes whose straight-line distance to the reference node exceeds the preset spatial association threshold to form an initial risk node set; If the number of nodes is greater than or equal to three, extract the first acquisition cycle timestamp of each node triggering the fire risk event in the initial risk node set, sort the nodes from early to late by timestamp, and select the top three nodes as positioning anchor points; If the number of nodes is less than three, retrieve the benchmark node based on the adjacent nodes and the first-level adjacent nodes of each node in the initial risk node set; Filter out the adjacent nodes that are currently in the primary monitoring state and have not been cleared of warnings; Add the selected adjacent nodes that meet the conditions to the initial risk node set to form an extended risk node set; If the number of nodes in the expanded risk node set is greater than or equal to three, the top three nodes are selected as positioning anchor points; If the number of nodes in the expanded risk node set is still less than three, all nodes in the expanded risk node set are used as positioning anchor points.

8. The smart building fire monitoring method according to claim 1, characterized in that: The steps for determining the initial coordinates of the fire source are as follows: Extract the three-dimensional coordinate data of all positioning anchor points and simultaneously obtain the fire characterization parameter values ​​recorded by each anchor point when a fire risk event is triggered; Compare the fire characterization parameter values ​​between the anchor points and identify the decreasing direction of the parameter values ​​from high to low. If there are two or more anchor points with the same type of parameters showing a unidirectional decreasing trend, it is determined that the parameter gradient direction is consistent with the spatial position change direction; If the gradient directions of different types of parameters are inconsistent, the gradient direction of the smoke concentration parameter shall be used as the main judgment basis; Mark the anchor point with the highest parameter value as the core reference point, analyze the spatial orientation of the remaining anchor points relative to the core reference point, and identify the farthest anchor point in the direction of decreasing parameter as the boundary reference point; When the gradient direction is linearly related to the spatial orientation, the initial coordinates of the fire source are defined in the direction of the line connecting the core reference point to the boundary reference point, and in the spatial area close to the core reference point; When there is no clear linear correlation between the gradient direction and the spatial orientation, the initial coordinates of the fire source are defined as the minimum convex hull area within the geometric polygon formed by all anchor points and containing the core reference point; If it is a linear area, take the core reference point coordinates as the center point, expand the preset safety distance along the connecting line to generate a cubic space range. If it is a convex hull area, extract the extreme value points of the convex hull vertex coordinates to form the minimum circumscribed cube range; Generate three-dimensional space data containing the coordinates of the vertices of the cube range as the initial coordinates of the fire source.

9. The smart building fire monitoring method according to claim 8, characterized in that: The specific contents of the corrected fire source coordinate range are as follows: Extract the identification, three-dimensional coordinates and triggering timestamp of the newly added fire risk event node; Calculate the spatial distance between the newly added node and the current fire source coordinate range. If it is within the preset spatial correlation threshold or is an adjacent node of the node within the current fire source range, then include it in the revised node set. If the newly added node is outside the current fire source range, the fire source range is expanded toward the node according to the gradient of its fire characterization parameter value and spatial orientation to form the minimum convex hull area covering the newly added hotspot; If the newly added node is located inside the current fire source range, the range boundary accuracy is optimized according to the relative position of its parameter value.

10. The smart building fire monitoring method according to claim 1, characterized in that: The specific contents of the fire spread path are as follows: Extract the three-dimensional coordinates of all monitoring nodes in the fire risk state within the fire source coordinate range and their smoke concentration time series data; Compare the smoke concentration gradient values ​​of adjacent nodes to identify the decreasing direction of concentration from high to low; If multiple adjacent node groups show a continuous unidirectional concentration decrease, they are determined to be effective diffusion path segments; Starting from the node with the highest smoke concentration within the fire source range, the nodes are connected in series along the extension direction of the effective diffusion path segment to form the initial diffusion main path; When a new node is included in the fire source range, if the smoke concentration gradient of the node is consistent with the direction of the main path, the path length is extended. If a new decreasing direction is formed, a branch path is generated and the diffusion intensity level is marked.

Images