Temporary fire linkage control system for construction site

Abstract

This application provides a temporary fire-fighting linkage control system for construction sites, belonging to the field of control system technology. The key technical points of its solution are: it includes a multi-source sensing module, an intelligent linkage module, an execution module, and an operation and maintenance module, with each module communicating and cooperating; the multi-source sensing module is deployed in the construction zone and is mobile, dynamically collecting fire data and personnel information in real time. The fire data includes smoke concentration data, temperature data, combustible gas concentration data, and flame information, and the personnel information includes personnel location data; the intelligent linkage module, based on the fire data and personnel information, performs secondary verification of the fire situation and outputs tiered linkage commands; the execution module executes the tiered linkage commands. The temporary fire-fighting linkage control system for construction sites provided by this application has the advantages of multi-dimensional accurate sensing, strong dynamic linkage capability, and flexible adaptation.

Description

Technical Field

[0001] This application relates to the field of control system technology, and more specifically, to a temporary fire-fighting linkage control system for construction sites. Background Technology

[0002] Construction sites are characterized by dense temporary buildings, numerous electrical equipment, concentrated storage of flammable materials, and frequent personnel movement, resulting in extremely high fire risks. Existing temporary fire protection systems are mostly composed of independent equipment, such as scattered fire extinguishers, manual alarm buttons, and simple sprinklers, which have three major drawbacks: First, they rely solely on smoke or heat detectors, making them susceptible to interference from construction dust and high-temperature operations, leading to false alarms or missed alarms; second, they have poor linkage, requiring manual activation of fire extinguishers and evacuation commands after an alarm, resulting in a delayed response; and third, they lack adaptability, failing to dynamically adjust the detection range and linkage logic according to the construction progress, making it difficult to meet the flexible needs of temporary work scenarios.

[0003] The above problems urgently need to be addressed. Summary of the Invention

[0004] The purpose of this application is to provide a temporary fire-fighting linkage control system for construction sites, which has the advantages of multi-dimensional accurate perception, strong dynamic linkage capability, and flexible adaptation.

[0005] This application provides a temporary fire-fighting linkage control system for construction sites, the technical solution of which is as follows: It includes a multi-source sensing module, an intelligent linkage module, an execution module, and an operation and maintenance module, with each module communicating and connecting collaboratively. The multi-source sensing module is deployed in the construction zone and is mobile, and collects fire data and personnel information in real time. The fire data includes smoke concentration data, temperature data, combustible gas concentration data and flame information, and the personnel information includes at least personnel location data. The intelligent linkage module performs secondary verification of the fire situation based on fire data and personnel information and outputs hierarchical linkage instructions. The execution module executes hierarchical linkage instructions; The operation and maintenance module generates a visual emergency command interface based on fire data, personnel information, and hierarchical linkage instructions, and pushes it to the fire and rescue vehicle navigation system.

[0006] Furthermore, in this application, the multi-source sensing module includes a smoke detector submodule, an infrared temperature detector submodule, a combustible gas sensor submodule, a video flame recognition submodule equipped with the YOLO target detection algorithm, and a UWB personnel positioning tag submodule; The smoke detector submodule collects smoke concentration data in the construction zone in real time. When the smoke concentration data exceeds the preset safety upper limit threshold, it triggers the generation of a smoke detection anomaly signal and transmits the smoke concentration data and the smoke detection anomaly signal to the intelligent linkage module. The infrared temperature sensor submodule is used to perform non-contact temperature monitoring of equipment, materials and work surfaces in the construction area, collect temperature data of the environment and target objects in real time, trigger the generation of abnormal high temperature signal when the temperature data is ≥60℃ and record the temperature change trend, and transmit the temperature data and abnormal high temperature signal to the intelligent linkage module. The combustible gas sensor submodule is used to detect combustible gas concentration data at the construction site. It presets safety thresholds for different gases (such as methane ≥ 0.5% VOL, acetylene ≥ 0.2% VOL). When the gas concentration reaches the threshold, it triggers a combustible gas exceeding the standard signal and simultaneously transmits the real-time combustible gas concentration data and combustible gas exceeding the standard signal to the intelligent linkage module. The video flame recognition submodule, equipped with the YOLO target detection algorithm, acquires video footage of the construction area in real time. It then uses the YOLO target detection algorithm to quickly identify suspected flame areas in the footage, extracts flame information, compares it with a built-in flame feature library, and generates a flame recognition result (including matching degree and suspected flame coordinates). The flame recognition result and video footage are then transmitted to the intelligent linkage module to assist in secondary verification of fire data. The flame information includes the shape, color, and dynamic characteristics of the flame. The UWB personnel positioning tag submodule is used to attach to the safety helmet of construction workers, with a positioning accuracy of ≤0.5m, and collects personnel information in real time and transmits it to the intelligent linkage module.

[0007] Furthermore, in this application, the intelligent linkage module includes a multi-source data weighted fusion submodule and a logical judgment submodule; When a single submodule in the multi-source sensing module is triggered, that submodule uses the deviation rate between the real-time acquired value S and the preset threshold S0 as the basic parameter. The multi-source data weighted fusion submodule calculates the "initial fire confidence level P1" based on the basic parameter, using the following formula: P1=∑(W i ×(S i -S 0i ) / S 0i ); Where i = 1, 2, 3, corresponding to the smoke detector submodule, infrared temperature detector submodule, and combustible gas sensor submodule, respectively; W i The weights of each sub-module in the multi-source sensing module are preset based on the interference characteristics of the construction site; S i This represents the real-time acquired value of this submodule; S 0i Indicates the security threshold of the corresponding submodule; S i-S 0i / S 0i This represents the deviation rate, with a value range of [0, 1]. The closer it is to 1, the higher the risk. The logic judgment submodule, in conjunction with the video flame recognition submodule, retrieves the real-time footage captured by the video flame recognition submodule. It then outputs a flame feature matching degree P2 (range [0, 1], where a higher matching degree indicates a closer match to standard flame features) using the YOLO target detection algorithm. Based on P1, P2 is calibrated to obtain the "final fire confidence level P," calculated using the following formula: P=α×P1+(1-α)×P2; Where α represents the calibration coefficient; The logic judgment submodule performs secondary verification of the fire situation determination: if P≥0.7, it is determined to be a suspected fire and triggers the subsequent graded linkage command output; if P<0.5, it triggers the device self-test; if 0.5≤P<0.7, it starts the delayed review mechanism: data is collected and the P value is calculated repeatedly at 3-second intervals. If the P value obtained from two consecutive calculations is ≥0.7, it is determined to be a suspected fire.

[0008] Furthermore, in this application, the intelligent linkage module also includes a fire comprehensive risk value calculation submodule and a hierarchical decision-making submodule; The fire risk calculation submodule calculates the comprehensive risk value R based on fire data and personnel information, using the following formula: R=A×(β1×A1+β2×A2)+B×(γ1×B1+γ2×B2)+C×δ; Where R represents the comprehensive risk value, used to classify fire severity levels; A represents the fire intensity coefficient, with a fixed value of A=1.2; A1 represents the flame area quantization value. A1 is generated by the video flame recognition submodule after identifying a suspected flame area using the YOLO object detection algorithm. The submodule outputs the coordinates of the suspected flame and then fits the outline of the flame area (the number of grid cells it contains) based on these coordinates. This outline, combined with the scale (the actual length represented by the grid cell side length), is converted into a flame area quantization value. The flame area quantization value = grid cell side length × grid cell side length × number of grid cells. For example, if the grid cell side length represents 0.5m, and the flame area covers 40 0.5m × 0.5m grid cells in the video frame, the flame area quantization value A1 = 40 × 0.25 = 10m. 2 A1≤5m 2 When the value is 0.2, and 5 < A1 ≤ 20m 2 Take 0.6, A1≥20m 2 Set the value to 1.0; A2 represents the core temperature quantification value. A2 is acquired by the infrared temperature sensor submodule and the range is defined as follows: 0.3 for A2≤300℃, 0.7 for 300<A2≤600℃, and 1.0 for A2>600℃. β1 represents the flame area weight, with a fixed value of β1=0.6; β2 represents the core temperature weight, with a fixed value of β2=0.4. B represents the personnel risk coefficient, with a fixed value of B=1.1; B1 represents the quantified value of the number of people in the danger zone. B1=0 is taken as 0, 1≤B1≤3 is taken as 0.5, and B1>3 is taken as 1.0. The danger zone refers to a circular / rectangular area dynamically delineated based on the comprehensive risk value R, centered on the fire center. In the case of a primary fire, the danger zone is within 5m of the fire center; in the case of a medium fire, the danger zone is within 10m of the fire center; and in the case of a high fire, the danger zone is within 15m of the fire center. Grid nodes within the danger zone are marked as high-risk nodes and are used for subsequent personnel risk calculations (such as the quantified value of the number of people in the danger zone B1) and path planning and avoidance. B2 represents the quantified distance between personnel and the fire's epicenter. The UWB personnel positioning tag submodule is attached to the construction worker's safety helmet, outputting the three-dimensional coordinates (X, Y, F, Z) of each worker in real time. p Y p Z p The positioning accuracy is ≤0.5m to ensure accurate location data; then, based on the straight-line distance formula of three-dimensional coordinates, the distance between a single person and the fire center point (X) is calculated. F Y F Z F Distance quantization value B2; And define the intervals: take 1.0 for B2≤5m, take 0.6 for 5<B2≤10m, and take 0.2 for B2≥10m; γ1 represents the weight of the number of people in the dangerous area, with a fixed value of γ1=0.5; γ2 represents the weight of the distance between people, with a fixed value of γ2=0.5; C represents the environmental risk coefficient, with a fixed value of C=0.8; δ represents the quantitative value of the combustible gas concentration exceeding the standard by a factor of 1. The combustible gas sensor submodule calculates the value based on the real-time concentration S of the combustible gas detected. i and the preset safety threshold S for the corresponding gas 0i The calculated value is: Quantitative value of combustible gas concentration exceeding the standard by multiple δ = (real-time concentration S) i -Safety threshold S 0i ) ÷ Safety threshold S 0i (Only when S) i ≥S 0i Time calculation, S i ≤S 0i The time limit exceedance multiple is 0), and the interval is defined as follows: 0 for no exceedance, 0 < δ ≤ 1 for 0.5, and δ > 1 for 1.0. The hierarchical decision-making submodule classifies the fire level according to the range of the comprehensive risk value R and outputs the corresponding hierarchical linkage instructions: When R∈[0.3, 0.5], the hierarchical decision-making submodule classifies it as a primary fire and outputs a primary command, which only controls the audible and visual alarms in the danger zone and the solenoid valves of fire extinguishers within 5m of the center of the danger zone. When R∈[0.5, 0.8], the hierarchical decision-making submodule classifies it as a medium-level fire and outputs a medium-level command on the basis of the primary command execution, controlling the start of the zone sprinkler system and the temporary power circuit controller within 10m around the danger zone; When R≥0.8, the hierarchical decision-making submodule classifies it as a high-level fire and outputs additional high-level instructions on the basis of the execution of primary and intermediate instructions. It controls and triggers the whole-site emergency broadcast and evacuation indicator lights, and pushes the coordinates of the dangerous area, personnel location data and comprehensive risk value R data to the fire command platform.

[0009] Furthermore, in this application, the video flame recognition submodule integrates an infrared thermal imaging unit and a temperature data fusion unit. The infrared thermal imaging unit collects temperature field distribution data of the construction area, and the temperature data fusion unit fuses the temperature field distribution data with the temperature data collected by the infrared temperature sensor submodule and transmits it to the intelligent linkage module to achieve accurate identification of fire data in low light and dense smoke environments. The infrared thermal imaging unit acquires temperature field distribution data of the construction area and outputs a three-dimensional temperature field data set {T} of the construction area. ir (x, y, z)}, where (x, y, z) are rasterized coordinates, T ir (x, y, z) represents the real-time temperature of this grid node; The infrared temperature sensor submodule collects temperature data T in real time. temp (i) (i=1, 2, ..., n, where n is the number of monitoring points), and simultaneously capture the coordinates (x) of abnormal high temperature signals ≥60℃. i y i , z i ); The temperature data fusion unit fuses data for each grid node (x, y, z) to obtain the final temperature value T. fuse (x, y, z), the formula is: T fuse (x, y, z) = w1 × T ir (x, y, z) + w2×T tempmap (x, y, z); Where w1 + w2 = 1; T fuse(x, y, z) represents the final temperature after fusion of grid nodes (x, y, z), used to assist in fire situation determination; w1 represents the weight of the temperature field distribution data collected by the infrared thermal imaging unit. In low light / dense smoke environments, w1=0.7 (infrared thermal imaging is more reliable in this environment); in normal lighting environments, w1=0.5 (dual-source data are equally reliable). The weight can be adaptively adjusted based on the on-site environmental conditions; T ir (x, y, z) represents the temperature of the grid node (x, y, z) acquired by the infrared thermal imaging unit, i.e., the temperature field distribution data; w2 represents the weight of the temperature data acquired by the infrared temperature sensor submodule; T tempmap (x, y, z) represents the temperature value mapped from the temperature data to the grid node. If the grid node (x, y, z) is an infrared temperature sensing monitoring point, then T is directly taken. temp (i); If it is not a monitoring point, the temperature is interpolated from the temperature of the nearest monitoring point (ensuring calibration data for the entire grid area), and the formula is: ; Where T tempmap (x, y, z) represents the interpolated temperature of the non-monitoring point grid (x, y, z) (output result, used for subsequent dual-source temperature fusion); k represents the number of neighboring infrared temperature sensing monitoring points participating in the interpolation (selecting the monitoring point closest to the current grid); i represents the index of the neighboring monitoring point (i=1, 2, ..., k); d i Represents the three-dimensional straight-line distance from the current grid (x, y, z) to the i-th nearest monitoring point; p represents the distance attenuation coefficient (controlling the attenuation rate of the weighted influence of the temperature of nearby monitoring points on the grid); T temp (i) represents the real-time temperature data of the i-th neighboring monitoring point; This represents the weight coefficient of the i-th neighboring monitoring point; The intelligent linkage module extracts the highest temperature T in the fire area based on the fused temperature data. max =max{T fuse (x, y, z)}, and define the interval: T max When ≤300℃, A2=0.3; when 300℃<T max At ≤600℃, A² = 0.7; T max When the temperature is >600℃, A2 = 1.0; A2 is included as a core parameter in the calculation of the comprehensive risk value R to support the classification of fire severity levels; The intelligent linkage module uses the fused abnormal high-temperature signal data as the real-time acquisition value Si of the infrared temperature sensor submodule, and substitutes it into the calculation formula for the initial fire confidence level P1: ; Where S 0i =60℃, deviation rate The reliability of P1 is improved based on the fused temperature value calculation, providing a basis for the calibration of the final fire confidence level P.

[0010] Furthermore, in this application, the execution module also includes an ultrasonic ranging submodule, which is linked with the intelligent linkage module to coordinate and control the pitch angle and spray pressure of the fire mobile sprinkler head to implement directional water spraying within a range of ±0.3m in the dangerous area; The intelligent linkage module determines a "suspected fire" through secondary verification, calculates the comprehensive risk value R and the fire coordinates, and then controls the linkage with the ultrasonic ranging submodule. The intelligent linkage module pushes the coordinates of the fire center point, the fire level, and the current coordinates of the fire-fighting mobile nozzle to the ultrasonic ranging submodule (providing a target reference for the ultrasonic ranging submodule, clarifying the direction and range of ranging, and avoiding blind ranging without a target). After receiving the fire coordinates, the ultrasonic ranging submodule determines the fire location based on the current coordinates (X) of the mobile fire sprinkler head. S Y S Z S ) and the coordinates of the fire center point (X F Y F Z F Automatically adjusts the initial orientation of the ranging probe (horizontally aligned with X). F -Y F Align vertically with Z F This enables directional ranging (rather than full-area scanning, thus improving efficiency); The ultrasonic ranging submodule controls the ranging probe to emit ultrasonic signals toward the fire area, and after receiving the reflected signals, calculates the straight-line distance D from the fire sprinkler head to the nearest boundary of the fire area. real Measurement accuracy ≤ ±0.1m (meets the ±0.3m water spraying accuracy requirement); The ultrasonic ranging submodule calculates the theoretical distance from the fire sprinkler head to the center of the fire using the following formula: ; The ultrasonic ranging submodule will measure the actual distance D real The theoretical distance D from the fire hose reel to the center of the fire theory Compare, if |D real -D theory If | ≤ 0.2m, the distance measurement is considered valid; if |D real -D theory If the distance exceeds 0.2m, trigger a re-measurement (to avoid single-measurement errors caused by smoke or dust). The intelligent linkage module receives the valid D from the ultrasonic ranging submodule. realThen, based on the fire data stored in the device (R value, size of the fire area), the pitch angle and spray pressure of the mobile fire sprinkler head are calculated. The intelligent linkage module is based on the installation height H of the fire-fighting mobile sprinkler head. S Measured distance D real The height of the fire's epicenter, Z F The vertical pitch angle θ and horizontal rotation angle φ are calculated using trigonometric functions to ensure that the water spraying direction is aimed at the center of the fire area; The formula for calculating the vertical pitch angle θ is: ; The horizontal distance ; The formula for calculating the horizontal rotation angle φ is: ; Example: If the installation height H of the portable fire sprinkler head S =4m, fire center point Z F =3m, measured horizontal distance D real If the horizontal distance is 5m, then θ = arctan(1 / 5) ≈ 11.3° (downward tilt), ensuring that the water spray accurately covers the center of the fire; The formula for calculating spray pressure P is: P=P0×k 校准 ; Where P is the spraying pressure, P0 is the base spraying pressure, and k 校准 This is the distance calibration coefficient; for a primary fire, P0 is 0.2 MPa, and D... real When ≤5m, k is taken as 1.0; when 5m < D real For fires ≤10m, take 1.1; for intermediate fires, take P0 as 0.3MPa, D real When ≤5m, k is taken as 1.0; when 5m < D real For fires ≤10m, take 1.2; for high-risk fires, take P0 as 0.4MPa, D real When ≤5m, k is taken as 1.0; when 5m < D real For depths ≤10m, take 1.3; The intelligent linkage module converts the calculated pitch angle θ, horizontal rotation angle φ, and spray pressure P into control commands and sends them to the ultrasonic ranging submodule. The ultrasonic ranging submodule controls the fire-fighting mobile sprinkler head to adjust to the calculated pitch angle, horizontal rotation angle, and spray pressure, and implements directional water spraying within a range of ±0.3m in the hazardous area.

[0011] Furthermore, in this application, the UWB personnel positioning tag submodule also includes an emergency alarm button unit, a voice interaction unit, and a speaker unit; the emergency alarm button unit receives a trigger operation from the construction personnel, and the construction personnel can generate an alarm signal by pressing the alarm button on the safety helmet; the voice interaction unit is used to provide feedback on the scene through real-time voice interaction with personnel, and the microphone is set on the safety helmet; the speaker unit is used to transmit personalized evacuation route broadcast information to personnel, and the speaker is set on the construction personnel's safety helmet.

[0012] The intelligent linkage module also includes a personalized evacuation route submodule, which calculates and generates personalized evacuation routes based on fire data and personnel information. The intelligent linkage module calculates the total estimated cost based on the starting point S and the current node n in the personnel information, using the following formula: f(n)=g(n)+h(n); Where f(n) represents the total estimated cost from the starting point S to node n to the ending point E (safe exit); g(n) represents the actual cumulative cost from the starting point S to node n; and h(n) represents the heuristic estimated cost from node n to the ending point E. The formula for calculating the actual cumulative cost g(n) is as follows: g(n) = g len (n)×W(n)+g fire (n) + g crowd (n); Where g len (n) represents the path length cost; W(n) represents the traffic efficiency weight; g fire (n) represents the cost of fire risk; g crowd (n) represents the cost of congestion risk; Where the path length cost g len The formula for calculating (n) is: ; Where parent(n) represents the parent node of node n (the previous node in the path); g len (parent(n)) represents the length cost from the parent node to the starting point; L represents the grid side length; The formula for calculating the traffic efficiency weight W(n) is as follows: ; Among them, the cost of fire risk g fire The formula for calculating (n) is: ; Where k fire This represents the risk weight of fire data, with a fixed value of k. fire= 10; d(n, F) represents the straight-line distance from node n to the center point F of the dangerous area, ; (X n , Y n , Z n ) are the grid coordinates of node n, and (X F , Y F , Z F ) are the grid coordinates of the fire center point; = 0.1, representing the minimum value tolerance coefficient, with a fixed value; T max represents the threshold of the core temperature of the flame, with a fixed value T max = 600 °C, which can be set to T max = 600 °C in this embodiment; the results are normalized to [0, 5], and the closer to the high-temperature area, the higher the cost; max(·) represents the maximum value function; T(n) represents the real-time temperature of grid node n; T0 represents the high-temperature risk threshold, with a fixed value T0 = 60 °C; Among them, the calculation formula for the congestion risk cost g crowd (n) is: ; Among them, g crowd (n) represents the congestion risk cost of the current grid node n (quantifying the evacuation risk caused by the gathering of people and used for the calculation of the actual cumulative cost). The larger the value, the higher the congestion risk of the node, and it needs to be avoided preferentially; k crowd represents the congestion risk weight coefficient (setting the priority of the congestion risk in path planning to balance "safe avoidance" and "smooth evacuation"), with a fixed value k crowd = 3; n is the current grid node to be evaluated, corresponding to the specific spatial position of the construction site, with coordinates (X n , Y n , Z n ); min(·, 2) represents the minimum value function (limiting the upper limit of the congestion risk cost to avoid the imbalance of algorithm decision-making caused by excessive congestion of a single node and ensuring the rationality of path planning). The logical rule is that when ≥ 2, the function result = 2, and when < 2, the function result = the calculated value within the parentheses; C(n) represents the real-time number of people gathering at grid node n (quantifying the congestion degree of the node and reflecting the risk of people扎堆 during evacuation); C0 represents the congestion critical threshold (the standard of the number of people for determining whether the node belongs to the "congestion risk area", adapting to the characteristics of the construction site passage), with a fixed value C0 = 5; Among them, the calculation formula for the heuristic estimated cost h(n) is: h(n) = P exit × α × (丨x n - x E 丨+丨yn -y E | + |z n -z E |) × L; Where h(n) represents the heuristic estimated cost from the current grid node n to the end point E (the safety exit) (quantifying the theoretical minimum cost of the future path and improving the evacuation path search efficiency); P exit represents the priority coefficient of the end point E (the safety exit) (distinguishing the main / alternate exits and preferentially selecting a better safety exit), with a fixed value. For the main exit, P exit = 1.0, and for the alternate exit, P exit = 1.2; α represents the heuristic calibration coefficient (avoiding over - estimating the path cost and ensuring the optimality of the evacuation path search), with a fixed value of α = 0.9; (x n y n z n ) represents the coordinates of the current grid node n; (x E y E z E ) represents the coordinates of the end point E (the safety exit); |x n - x E | + |y n - y E | + |z n - z E | represents the three - dimensional Manhattan distance from node n to the end point E (quantifying the spatial straight - line distance and adapting to the three - dimensional space of the construction site), obtained by summing the absolute values of the coordinate differences along the X / Y / Z axes, reflecting the spatial proximity between the node and the end point; L represents the grid side length; n is the current grid node to be evaluated; E is the end point of the path (i.e., the preset safety exit of the construction site, including the main exit and the alternate exit), and the intelligent linkage module selects the optimal exit according to the personnel information and fire data (preferentially the main exit, and switching to the alternate exit when the main exit is blocked by the fire); The personalized evacuation path sub - module generates a personalized evacuation path based on the total estimated cost f(n), the actual cumulative cost g(n), and the heuristic estimated cost h(n). The formula is: ; Where Path final represents the personalized evacuation path corresponding to the current node n; Reduce(·) represents the path reduction function, filtering collinear nodes; CollinearFilter represents the collinear filtering rule, determining whether three consecutive nodes are collinear, and if collinear, retaining the head and tail nodes; Convert(·) represents the instruction conversion function, converting the reduced grid path into a voice instruction; VoiceInstruction represents the voice instruction rule, converting the grid path into a natural language containing direction and distance; This represents the minimum cost search operator, which selects the node with the smallest total estimated cost f(n) among the nodes to be explored.

[0013] Furthermore, in this application, the operation and maintenance module also includes a cloud management platform sub-module and a local touch screen sub-module; The cloud management platform sub-module connects to the construction site BIM model. When a fire occurs, it automatically overlays fire data and personnel information onto the BIM model. The fire data includes the BIM 3D coordinates of the fire center point, the set of grid coordinates of the fire area boundary, the comprehensive risk value R, and the fire level. The personnel information includes the real-time BIM coordinates of personnel, personnel identity and tag number binding information, and the personalized evacuation route corresponding to the personnel. The local touch screen submodule generates a visual emergency command interface, which is simultaneously pushed to the fire rescue vehicle navigation system. The visual emergency command interface includes a BIM model top view and a core data panel. The BIM model top view identifies the dangerous area where the fire has occurred and the distribution of people in the surrounding area. The core data panel displays the fire level, comprehensive risk value R, number of people in the dangerous area, and operating status of fire-fighting equipment in real time.

[0014] Furthermore, in this application, the intelligent linkage module is linked with the temporary power management system at the construction site. When a fire occurs, the intelligent linkage module controls the priority to cut off the temporary power circuits within a 10m radius of the dangerous area, while preserving the power supply circuits for emergency lighting and fire-fighting equipment. The intelligent linkage module also includes a power failure linkage sub-module, which links and controls the temporary power management system at the construction site. When the intelligent linkage module determines that the fire level is ≥ medium fire, the power failure linkage submodule will trigger a fire alarm based on the coordinates of the fire center point (X). F Y F Z F The system uses BIM grid modeling data (L×L×L, where L represents the grid side length) to delineate a 10m three-dimensional spatial range around the hazardous area and automatically sends a power outage command to the temporary power management system at the construction site. The power outage command limits the scope of cutting off non-emergency temporary power circuits within a 10m three-dimensional space around the danger zone, including power circuits for construction machinery, temporary lighting, and office power; at the same time, it retains emergency lighting power circuits and dedicated power circuits for fire-fighting equipment, including power circuits for fire extinguisher solenoid valves, sprinkler systems, and ultrasonic ranging submodules. After the temporary power supply management system at the construction site executes the power outage command, it sends the execution result back to the power outage linkage submodule. If the execution fails, the power outage linkage submodule will resend the power outage command and synchronize the execution result to the local touch screen submodule for alarm display.

[0015] Furthermore, in this application, the intelligent linkage module also includes a fire prediction submodule and a dynamic adaptation submodule; When the intelligent linkage module determines that the fire level is a high-level fire, the fire prediction submodule triggers the start of fire spread trend prediction, and the dynamic adaptation submodule triggers the start of dynamic strategy adaptation mode. The fire prediction submodule loads a spatiotemporal multidimensional fire spread prediction model. The spatiotemporal multidimensional fire spread prediction model uses BIM raster modeling data (L×L×L, where L is the grid side length) as the spatial carrier and is trained and calibrated by synchronously inputting construction site environmental data collected by the fire prediction submodule, real-time fire data collected by the multi-source sensing module, and historical fire data. The construction site environmental data includes wind speed, wind direction, and humidity at the construction site. The spatiotemporal multidimensional fire spread prediction model outputs the fire spread trend prediction results for the next 3 minutes at a frequency of 1 time / second, including the spread direction, spread speed, three-dimensional boundaries of the fire scene at different time points, and the threatened sequence of high-risk areas (areas with dense combustible materials and areas with concentrated personnel). The formula for predicting the speed of fire spread is: ; Where v fire (t) represents the rate of fire spread at time t in the future; k v This represents the wind direction correction factor, with a fixed value, and the wind spreads in the same direction (k). v Take 1.2, vertical spread k v Take 0.8, back propagation k v =0.5; v represents the real-time wind speed at the construction site; T core T represents the real-time core temperature of the fire; T0 represents the critical threshold of fire risk; T max The value represents the upper limit of the flame core temperature; h represents the real-time humidity at the construction site; the effect of humidity on the decay of the spread rate is quantified by 1-0.01h. The formula for predicting the three-dimensional boundary coordinates of a fire is: ; Among them (X) Fb (t), Y Fb (t), Z Fb (t) represents the BIM coordinates of any point on the three-dimensional boundary of the fire at time t in the future (forming a boundary set, corresponding to the three-dimensional boundary of the fire at different time nodes); (X) F0 Y F0 Z F0) represents the current BIM coordinates of the fire center (provided by real-time fire data collected by the multi-source sensing module); β represents the wind direction at the construction site (with true north as 0°, provided by environmental data of the construction site collected by the fire prediction submodule); (cosβ, sinβ) correspond to the horizontal spread components; k z This represents the vertical creep correction factor (in construction scenarios, the vertical creep rate is 0.3 times that of the horizontal creep rate, k). z =0.3), adapting to the three-dimensional spatial characteristics of building floors and high-altitude work surfaces; grid association: coordinate values ​​are all based on BIM gridded modeling data (L×L×L, where L is the grid side length), ensuring accurate matching with the actual location on the construction site; The dynamic adaptation submodule dynamically updates personalized evacuation routes based on prediction results (correcting fire risk costs g). fire (n), avoid high-risk areas in advance and push the information to the UWB personnel positioning tag submodule in time, optimize the scheduling of fire-fighting equipment (adjust the fire-fighting mobile nozzles to key interception points in advance according to the predicted spread trajectory, and preset the pitch angle and spray pressure), and adjust the temporary power supply control range (cut off non-emergency power circuits within 15m of the predicted spread path in advance to prevent the fire from igniting the electrical equipment). The formula for calculating the corrected fire risk cost by the dynamic adaptation submodule is as follows: ; Where g fire (t) represents the corrected fire risk cost of the grid node at time t in the future; k fire The fire risk weighting coefficient is represented by d(n, Fb(t)); d(n, Fb(t)) represents the three-dimensional Euclidean distance from the grid node to the fire boundary at time t in the future, and the calculation formula is as follows: ; This represents the fault tolerance coefficient, with a fixed value. =0.1; k speed (t) represents the propagation speed weighting coefficient, k speed (t) = 1 + 0.5·v fire (t), the faster the spread, the higher the weight, amplifying the risk and cost of rapidly spreading areas and ensuring that the path is avoided in advance; other parameters T core T0, T max All data is publicly available as mentioned above; The preset formula for the pitch angle of a portable fire sprinkler head is: ; Where α(n, t) represents the preset pitch angle of the fire-fighting mobile sprinkler relative to the fire boundary at time t in the future; Z n The Z-axis BIM coordinates representing the installation location of the portable fire sprinkler head; d xy (n, F) b(t) represents the horizontal distance between the nozzle and the fire boundary at time t in the future, and is calculated using the following formula: ;k α The compensation angle is represented by a fixed value k. α =3°; The formula for determining the temporary power outage range is: d(n, F) path (t))≤15m; Where d(n, F) path (t) represents the shortest distance from grid node n to the fire spread path at time t in the future; 15m represents the range value for early cut-off to avoid the fire igniting electrical equipment; when node n satisfies the formula, the dynamic adaptation submodule issues an instruction to cut off the non-emergency power circuit corresponding to that node; The fire prediction submodule verifies the prediction accuracy using real-time fire data and personnel information. When the deviation between the predicted fire boundary and the actual fire boundary is greater than 0.8m, it automatically corrects the parameters of the spatiotemporal multidimensional fire spread prediction model and re-predicts. At the same time, it synchronizes the prediction results, adaptation strategies, and verification data to the cloud management platform submodule for archiving, the local touch screen submodule for visualization display, and pushes them to the fire and rescue vehicle navigation system to assist in rescue route planning. The formula for calibrating prediction accuracy is: ; Where △d represents the three-dimensional deviation between the predicted fire boundary and the actual fire boundary at time t; (X Freal (t), Y Freal (t), Z Freal (t) represents the actual fire boundary BIM coordinates (real-time fire data) collected by the multi-source sensing module at time t; the correction trigger condition is △d > 0.8m, and the model parameters (k) are automatically adjusted after triggering. v k z ), and regenerate the prediction results.

[0016] Other features and advantages of this application will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of this application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0017] Beneficial effects: ① Enhanced Sensing Precision: Reduces false alarms and missed alarms, covering risks across all scenarios. Adopting a multi-source sensing architecture, it overcomes the limitations of traditional single sensors, combining an infrared thermal imaging unit to achieve accurate identification of fire data in low-light and dense smoke environments, effectively filtering interference from construction dust, welding sparks, etc., reducing false alarm rates; quantifying sensing thresholds and feature matching degrees makes fire data and personnel information collection quantifiable and verifiable, providing accurate data support for subsequent decision-making. ② Intelligent linkage: precise hierarchical response and improved response efficiency. A secondary verification mechanism of "initial confidence P1 + flame matching degree P2" is established. Through weighted fusion and calibration algorithms (P=α×P1+(1-α)×P2), the false triggering of a single signal is avoided, and the accuracy of fire judgment is improved. Based on the hierarchical linkage logic of comprehensive risk value R, differentiated emergency actions are matched according to "primary-intermediate-advanced" (such as fire extinguishing within 5m → power cut-off within 10m → evacuation of the entire site). This avoids the waste of resources caused by "over-response" and prevents the fire from spreading due to "insufficient response". The response time is shortened to the second level. Cross-system linkage design prioritizes cutting off temporary power circuits within 10m when a fire occurs and retains emergency power supply. This blocks electrical fire sources and electric shock risks from the source and ensures the safe implementation of emergency actions. ③ Personalized Evacuation: Ensuring personnel safety and mitigating secondary risks. Personalized evacuation routes are generated based on fire data and personnel information, incorporating fire risk (g) analysis. fire ), congestion risk (g crowd The system combines path finalization (Pathfinal=Convert(...)) with traffic efficiency (W) to achieve precise "one person, one road" planning, prioritizing the avoidance of high-temperature areas and congested nodes to reduce the risk of stampedes; the combination of path instruction conversion (Pathfinal=Convert(...)) and UWB tag-based targeted broadcasting transforms complex spatial paths into concise voice commands of "direction + distance," adapting to rapid understanding in emergency situations and improving evacuation efficiency; 3D grid modeling and BIM coordinate adaptation cover three-dimensional spaces such as construction site floors and high-altitude work surfaces, solving the problem that traditional planar paths cannot adapt to complex construction environments; ④ Collaborative Rescue: Breaking down information barriers and improving rescue efficiency. The operation and maintenance module features a dual-end design. A cloud-based BIM model overlays fire data and personnel information data, while a local touchscreen generates a visual command interface, which is simultaneously pushed to the fire and rescue vehicle navigation system. Rescuers can grasp the situation on the ground and plan the optimal rescue route while en route. In high-risk fires, personnel location data and comprehensive risk value R are pushed, achieving seamless collaboration between "on-site evacuation and off-site rescue," shortening rescue response time, and increasing the probability of trapped personnel being rescued. ⑤Scenario adaptability: The system is stable and reliable, meeting the needs of temporary construction. The "zoned deployment and mobility" design of the multi-source sensing module dynamically adjusts the detection range according to the construction progress, adapting to scenarios such as changes in temporary work surfaces and relocation of material stacks; ultrasonic ranging directional water spraying achieves precise fire extinguishing within ±0.3m in hazardous areas, reducing water loss to equipment and materials in non-hazardous areas, and balancing fire protection effectiveness with the protection of construction property; ⑥ Application Value: Cost Reduction and Efficiency Improvement, Enhanced Safety Management. An integrated intelligent architecture replaces traditional independent equipment splicing, reducing manual inspection and operation costs and improving maintenance efficiency; full-process quantitative indicators (confidence level P, risk value R, positioning accuracy ≤0.5m) shift fire management from "experience-based" to "data-driven," contributing to safety and compliance at construction sites; its flexible design adaptable to temporary construction scenarios allows for rapid deployment at various construction sites, including residential, municipal, and industrial sites, demonstrating strong versatility and broad application prospects. Attached Figure Description

[0018] Figure 1 An architecture diagram of a temporary fire-fighting linkage control system for construction sites provided in this application; Figure 2 This is an architecture diagram of the multi-source sensing module in this application; Figure 3 This is an architecture diagram of the intelligent linkage module of this application; Figure 4 This is an architecture diagram of the video flame recognition submodule of this application; Figure 5 This is an architecture diagram of the operation and maintenance module of this application; Figure 6 This is an architecture diagram of the execution module of this application; Figure 7 This is an architecture diagram of the UWB personnel location tag submodule in this application. Detailed Implementation

[0019] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0020] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0021] Please refer to Figures 1 to 7As shown, this application provides a temporary fire-fighting linkage control system for construction sites, including a multi-source sensing module, an intelligent linkage module, an execution module, and an operation and maintenance module, with each module communicating and cooperating. The multi-source sensing module is deployed in the construction zone and is mobile, dynamically collecting fire data and personnel information in real time. The fire data includes smoke concentration data, temperature data, combustible gas concentration data, and flame information, and the personnel information includes at least personnel location data. The intelligent linkage module performs secondary verification of fire situation determination and outputs tiered linkage commands based on the fire data and personnel information. The execution module executes the tiered linkage commands. The operation and maintenance module generates a visual emergency command interface based on the fire data, personnel information, and tiered linkage commands and pushes it to the fire rescue vehicle navigation system.

[0022] Specifically, it includes a multi-source sensing module, an intelligent linkage module, an execution module, and an operation and maintenance module, with each module communicating and connecting collaboratively. The multi-source sensing module is deployed in the construction zone and is mobile, dynamically collecting fire data and personnel information in real time. The fire data includes smoke concentration data, temperature data, combustible gas concentration data, and flame information, while the personnel information includes at least personnel location data. The zoned deployment and mobility characteristics adapt to the temporary scenarios of dynamically adjusted work surfaces at the construction site, distinguishing it from traditional fixed-installation sensing equipment. Mobile acquisition clarifies its ability to flexibly cover the detection range as the construction progresses. By collecting multi-source fire data, it breaks through the limitations of traditional single sensing, and personnel location data provides a foundation for subsequent personalized evacuation and rescue positioning.

[0023] The intelligent linkage module, based on fire data and personnel information, performs secondary verification of fire situation determination and outputs tiered linkage instructions: the fire data and personnel information are used to clarify the data basis for decision-making and avoid rough judgments without data support; the secondary verification of fire situation determination solves the problem of false alarms from traditional single sensors; and the tiered linkage instructions reflect the accuracy of the response, which is different from traditional fire protection systems that do not have tiered and disordered responses.

[0024] The execution module executes hierarchical linkage instructions; the operation and maintenance module generates a visual emergency command interface based on fire data, personnel information, and hierarchical linkage instructions and pushes it to the fire and rescue vehicle navigation system: "Generating a visual emergency command interface" transforms abstract fire data, personnel, and instruction data into intuitive information, reducing the threshold for on-site command; "Pushing to the fire and rescue vehicle navigation system" breaks down the information barriers between on-site emergency response and external rescue, realizing the coordination of "on-site situation - off-site rescue", and solving the pain points of information isolation and unclear rescue paths in traditional fire protection systems.

[0025] In some preferred embodiments, the multi-source sensing module includes a smoke detector submodule, an infrared temperature detector submodule, a combustible gas sensor submodule, a video flame recognition submodule equipped with the YOLO target detection algorithm, and a UWB personnel positioning tag submodule. The smoke detector submodule collects smoke concentration data within the construction zone in real time. When the smoke concentration data exceeds a preset safety upper limit threshold, it triggers the generation of a smoke detection anomaly signal and transmits the smoke concentration data and the smoke detection anomaly signal to the intelligent linkage module. The infrared temperature detector submodule is used for non-contact temperature monitoring of equipment, materials, and work surfaces within the construction area. It collects environmental and target object temperature data in real time. When the temperature data is ≥60℃, it triggers the generation of an abnormal high-temperature signal and records the temperature change trend, transmitting the temperature data and the abnormal high-temperature signal to the intelligent linkage module. The combustible gas sensor submodule is used for targeted detection of combustible gases at the construction site. Gas concentration data is used to set safety thresholds for different gases (e.g., methane ≥ 0.5% VOL, acetylene ≥ 0.2% VOL). When the gas concentration reaches the threshold, a combustible gas exceeding the standard signal is triggered, and the real-time concentration data and combustible gas exceeding the standard signal are simultaneously transmitted to the intelligent linkage module. The video flame recognition submodule, equipped with the YOLO target detection algorithm, collects video footage of the construction area in real time, quickly identifies suspected flame areas in the footage using the YOLO target detection algorithm, extracts flame information and compares it with the built-in flame feature library to generate flame recognition results (including matching degree and suspected flame coordinates), and transmits the flame recognition results and video footage to the intelligent linkage module to assist in secondary verification of fire data. The flame information includes the shape, color, and dynamic characteristics of the flame. The UWB personnel positioning tag submodule is used to attach to the safety helmets of construction workers, with a positioning accuracy of ≤ 0.5m, and collects personnel information in real time and transmits it to the intelligent linkage module.

[0026] Specifically, the multi-source sensing module includes a smoke detector submodule, an infrared temperature detector submodule, a combustible gas sensor submodule, a video flame recognition submodule equipped with the YOLO target detection algorithm, and a UWB personnel positioning tag submodule; The smoke detector submodule collects smoke concentration data within the construction zone in real time. When the smoke concentration exceeds a preset safety threshold, it triggers the generation of a smoke detection anomaly signal and transmits both the smoke concentration data and the anomaly signal to the intelligent linkage module. This visualizes the fire data as specific outputs of smoke concentration data and anomaly signals, ensuring that the sensing function is quantifiable and verifiable. Unlike traditional smoke detectors that only alarm without providing data, this module achieves real-time concentration data transmission, providing a quantitative basis for the smoke dimension to the intelligent linkage module's secondary verification of the fire data, thus reducing the risk of false alarms caused by construction dust. Both the smoke concentration data and the smoke detection anomaly signal fall within the scope of fire data.

[0027] The infrared temperature sensor submodule is used for non-contact temperature monitoring of equipment, materials, and work surfaces within the construction area. It collects real-time temperature data of the environment and target objects. When the temperature data is ≥60℃, it triggers the generation of an abnormal high-temperature signal and records the temperature change trend. The temperature data and abnormal high-temperature signal are then transmitted to the intelligent linkage module. Through non-contact monitoring and full coverage of equipment / materials / work surfaces, it overcomes the shortcomings of traditional contact temperature sensors in adapting to dynamic construction scenarios (such as high-altitude work surfaces and large equipment), clarifying the scenario adaptability of temperature sensing. It focuses on capturing high-temperature points ≥60℃ and recording temperature change trends, breaking through the limitation of traditional temperature sensors that only monitor single-point temperatures. It provides the intelligent linkage module with predictive data for high-temperature spread, enabling early detection of fire data and enhancing fire data prediction capabilities. It complements the smoke detector submodule, providing preliminary data support for subsequent flame identification and gas detection verification. Temperature data and abnormal high-temperature signals both fall within the scope of fire data.

[0028] The combustible gas sensor submodule is used for targeted detection of combustible gas concentration data at the construction site. It presets safety thresholds for different gases (e.g., methane ≥ 0.5% VOL, acetylene ≥ 0.2% VOL). When the gas concentration reaches the threshold, a combustible gas exceedance signal is triggered, and the real-time combustible gas concentration data and exceedance signal are simultaneously transmitted to the intelligent linkage module. This addresses the pain point of detecting multiple types of gases at construction sites: by targeted detection and preset thresholds for different gases, it overcomes the shortcomings of traditional gas sensors that use a single universal threshold, adapting to various combustible gas scenarios generated by painting, welding, and other operations at construction sites, thus improving detection accuracy. Specific thresholds (0.5% VOL, 0.2% VOL) ​​quantify exceedance judgments, avoiding ambiguous alarms and providing a clear basis for the intelligent linkage module to determine the gas risk level. Combustible gas leaks are included in the fire data early warning scope, enabling preventative measures before combustion, supplementing the gas leak risks that smoke and heat sensors cannot cover, and strengthening the comprehensiveness of multi-source sensing. Both combustible gas concentration data and combustible gas exceedance signals fall within the scope of fire data.

[0029] The video flame recognition submodule, equipped with the YOLO target detection algorithm, acquires real-time video footage of the construction area. Using the YOLO algorithm, it quickly identifies suspected flame areas in the footage, extracts flame information, and compares it with a built-in flame feature library to generate flame recognition results (including matching degree and suspected flame coordinates). The flame recognition results and video footage are then transmitted to the intelligent linkage module for secondary verification of fire data. The flame information includes the flame's shape, color, and dynamic characteristics. The YOLO algorithm and feature comparison clarify the intelligent implementation path for flame recognition, transforming the secondary verification by the intelligent linkage module into reliable sensory data, thus addressing the pain point of false alarms from traditional single sensors. By outputting quantitative results of matching degree and coordinates, rather than simply judging the presence or absence of flames, the intelligent linkage module can make accurate decisions based on the matching degree (e.g., ≥0.7 for suspected flames). Simultaneously, the coordinate data provides location information for subsequent targeted firefighting and area power outages. Through the extraction of multiple features including shape, color, and dynamics, it differentiates itself from interference items such as welding sparks and strong light reflections in construction scenarios, improving the accuracy of flame recognition, meeting adaptability to construction scenarios, and enhancing anti-interference capabilities. The flame recognition results and video footage both fall under the category of fire data.

[0030] The UWB personnel positioning tag submodule is used to attach to the safety helmets of construction workers, with a positioning accuracy of ≤0.5m. It collects personnel information in real time and transmits it to the intelligent linkage module. The specific implementation of personnel information collection is clearly defined: the location data of personnel information is visualized to ensure the reliability of personnel positioning (utilizing the behavioral principle that construction workers must wear safety helmets); the quantitative indicator of ≤0.5m positioning accuracy distinguishes it from traditional GPS (meter-level error) positioning schemes, providing core data support for the intelligent linkage module to generate precise evacuation routes for each person; the design of attaching to the safety helmet is adapted to construction site management regulations, ensuring that the positioning tag is "carried on the body and not easily detached," avoiding personnel positioning failure. Simultaneously, real-time data transmission allows the intelligent linkage module to dynamically update personnel locations, improving the timeliness of evacuation and rescue.

[0031] In some preferred embodiments, the intelligent linkage module includes a multi-source data weighted fusion submodule and a logical judgment submodule; When a single submodule in the multi-source sensing module is triggered, that submodule uses the deviation rate between the real-time acquired value S and the preset threshold S0 as the basic parameter. The multi-source data weighted fusion submodule calculates the "initial fire confidence level P1" based on the basic parameter, using the following formula: P1=∑(W i ×(S i -S 0i ) / S 0i ); Where i = 1, 2, 3, corresponding to the smoke detector submodule, infrared temperature detector submodule, and combustible gas sensor submodule, respectively; Wi The weights of each sub-module in the multi-source sensing module are preset based on the interference characteristics of the construction site; S i This represents the real-time acquired value of this submodule; S 0i Indicates the security threshold of the corresponding submodule; S i -S 0i / S 0i This represents the deviation rate, with a value range of [0, 1]. The closer it is to 1, the higher the risk. The logic judgment submodule, in conjunction with the video flame recognition submodule, retrieves the real-time footage captured by the video flame recognition submodule. It then outputs a flame feature matching degree P2 (range [0, 1], where a higher matching degree indicates a closer match to standard flame features) using the YOLO target detection algorithm. Based on P1, P2 is calibrated to obtain the "final fire confidence level P," calculated using the following formula: P=α×P1+(1-α)×P2; Where α represents the calibration coefficient, α=0.4, which adapts the reliability of video recognition in complex environments; The logic judgment submodule performs secondary verification of the fire situation determination: if P≥0.7, it is determined to be a suspected fire and triggers subsequent hierarchical linkage; if P<0.5, it triggers equipment self-test; if 0.5≤P<0.7, it starts the delayed review mechanism: data is collected and P value is calculated repeatedly at 3-second intervals. If the P value obtained from two consecutive calculations is ≥0.7, it is determined to be a suspected fire.

[0032] Specifically, if P ≥ 0.7, it is judged as a suspected fire, triggering subsequent tiered linkage: by quantifying the high confidence threshold (P ≥ 0.7) for fire data judgment, the judgment criteria for suspected fires are quantifiable and verifiable, avoiding the defects of traditional subjective judgment or vague judgment based on a single threshold, and ensuring the consistency of judgment results; directly triggering tiered linkage instructions ensures both the timeliness of fire data response (without delaying the golden time for handling) and reserves space for subsequent review through the description of suspected fires, taking into account both safety and rigor.

[0033] If P < 0.5, a self-check is triggered: quantifying the "low confidence threshold" (P < 0.5) for fire data judgment, clarifying that the signal in this scenario is likely interference (such as construction dust, equipment mis-triggered) rather than real fire data, avoiding the activation of emergency response for invalid signals, and reducing construction stoppages and resource waste caused by false alarms; unlike the limitations of traditional systems that only alarm without investigation, the self-check identifies sensor faults (such as smoke detector filter blockage, infrared temperature sensor zero-point drift) or signal interference sources, ensuring the long-term reliability of sensing equipment; enhancing the system's fault tolerance and ease of operation and maintenance, transforming signal anomalies into equipment maintenance needs, eliminating potential faults in advance, avoiding missed alarms due to equipment failure when a real fire occurs, and improving the overall stability of the system.

[0034] If 0.5 ≤ P < 0.7, a delayed verification mechanism is activated: data is collected and the P value is calculated repeatedly at 3-second intervals. If the P value obtained from two consecutive calculations is ≥ 0.7, it is judged as a suspected fire. Differentiated processing rules are designed for the medium confidence interval (0.5 ≤ P < 0.7) to solve the shortcomings of traditional black-and-white judgments. It avoids blindly triggering linkage (avoiding false alarms) and directly ignoring (avoiding missed alarms), balancing response speed and judgment accuracy. The specific parameters of delayed verification are clearly defined (3-second interval, two consecutive compliances) to make the rules implementable and reproducible, adapting to the characteristics of instantaneous interference at construction sites (such as brief welding sparks, dust drifting by), and filtering single-frame / single-instantaneous interference signals. The judgment conditions for suspected fires are further refined. Data reliability is improved by repeated collection and calculation to ensure that real fire data is not missed in medium-risk scenarios (such as fire data with weak initial signals but continuous strengthening). At the same time, the impact of occasional interference is reduced by the continuous compliance requirement, strengthening the anti-interference capability of the secondary verification mechanism.

[0035] In some preferred embodiments, the intelligent linkage module further includes a fire comprehensive risk value calculation submodule and a hierarchical decision-making submodule; The fire risk calculation submodule calculates the comprehensive risk value R based on fire data and personnel information, using the following formula: R=A×(β1×A1+β2×A2)+B×(γ1×B1+γ2×B2)+C×δ; Where R represents the comprehensive risk value, used to classify fire severity levels; A represents the fire intensity coefficient, with a fixed value of A=1.2; A1 represents the flame area quantization value. A1 is generated by the video flame recognition submodule after identifying a suspected flame area using the YOLO object detection algorithm. The submodule outputs the coordinates of the suspected flame and then fits the outline of the flame area (the number of grid cells it contains) based on these coordinates. This outline, combined with the scale (the actual length represented by the grid cell side length), is converted into a flame area quantization value. The flame area quantization value = grid cell side length × grid cell side length × number of grid cells. For example, if the grid cell side length represents 0.5m, and the flame area covers 40 0.5m × 0.5m grid cells in the video frame, the flame area quantization value A1 = 40 × 0.25 = 10m. 2 A1≤5m 2 When the value is 0.2, and 5 < A1 ≤ 20m 2 Take 0.6, A1≥20m 2 Set the value to 1.0; A2 represents the core temperature quantification value. A2 is acquired by the infrared temperature sensor submodule and the range is defined as follows: 0.3 for A2≤300℃, 0.7 for 300<A2≤600℃, and 1.0 for A2>600℃. β1 represents the flame area weight, with a fixed value of β1=0.6; β2 represents the core temperature weight, with a fixed value of β2=0.4. B represents the personnel risk coefficient, with a fixed value of B=1.1; B1 represents the quantified value of the number of people in the danger zone. B1=0 is taken as 0, 1≤B1≤3 is taken as 0.5, and B1>3 is taken as 1.0. The danger zone refers to a circular / rectangular area dynamically delineated based on the comprehensive risk value R, centered on the fire center. In the case of a primary fire, the danger zone is within 5m of the fire center; in the case of a medium fire, the danger zone is within 10m of the fire center; and in the case of a high fire, the danger zone is within 15m of the fire center. Grid nodes within the danger zone are marked as high-risk nodes and are used for subsequent personnel risk calculations (such as the quantified value of the number of people in the danger zone B1) and path planning and avoidance. B2 represents the quantified distance between personnel and the fire's epicenter. The UWB personnel positioning tag submodule is attached to the construction worker's safety helmet, outputting the three-dimensional coordinates (X, Y, F, Z) of each worker in real time. p Y p Z p The positioning accuracy is ≤0.5, ensuring accurate location data; then, based on the straight-line distance formula of three-dimensional coordinates, the distance between a single person and the fire center point (X) is calculated. F Y F Z F Distance quantization value B2; And define the intervals: take 1.0 for B2≤5m, take 0.6 for 5<B2≤10m, and take 0.2 for B2≥10m; γ1 represents the weight of the number of people in the dangerous area, with a fixed value of γ1=0.5; γ2 represents the weight of the distance between people, with a fixed value of γ2=0.5; C represents the environmental risk coefficient, with a fixed value of C=0.8; δ represents the quantitative value of the combustible gas concentration exceeding the standard by a factor of 1. The combustible gas sensor submodule calculates the value based on the real-time concentration S of the combustible gas detected. i and the preset safety threshold S for the corresponding gas 0i The calculated value is: Quantitative value of combustible gas concentration exceeding the standard by multiple δ = (real-time concentration S) i -Safety threshold S 0i ) ÷ Safety threshold S 0i (Only when S) i ≥S 0i Time calculation, S i ≤S 0i The time limit exceedance multiple is 0), and the interval is defined as follows: 0 for no exceedance, 0 < δ ≤ 1 for 0.5, and δ > 1 for 1.0. The hierarchical decision-making submodule classifies the fire level according to the range of the comprehensive risk value R and outputs the corresponding hierarchical linkage instructions: When R∈[0.3, 0.5], the hierarchical decision-making submodule classifies it as a primary fire and outputs a primary command, which only links and controls the audible and visual alarms in the dangerous area and the solenoid valves of fire extinguishers within 5m of the center point of the dangerous area; this avoids the resource waste problem of traditional all-area alarms or fire extinguishing; it balances response speed and interference control, only activating necessary equipment, which can quickly alert on-site personnel and deal with the initial fire, without affecting construction in non-risk areas, and is suitable for the needs of emergency response and production on construction sites.

[0036] When R∈[0.5, 0.8], the hierarchical decision-making submodule classifies the fire as intermediate level and outputs intermediate level commands on the basis of the primary command execution, controlling the activation of the zoned sprinkler system and the temporary power circuit controller within 10m of the dangerous area; the risk range of intermediate level fire (R∈[0.5, 0.8]) is defined, corresponding to the fire expansion and risk escalation scenarios, and the response upgrade is achieved by superimposing commands, reflecting the progressive nature of hierarchical linkage; the spraying pressure (0.3MPa), power cut-off range (within 10m) and delay (≤1 second) of the sprinkler system are clearly defined, which both expands the fire control range through spraying and cuts off the electrical fire source, solving the defects of traditional linkage measures that are single and cannot stop the spread of fire; zonal control is emphasized to avoid equipment damage and construction stagnation caused by full-area spraying, while ensuring the power supply of fire-fighting equipment and emergency lighting through precise power cut-off, taking into account both fire control effect and safety guarantee.

[0037] When R≥0.8, the hierarchical decision-making submodule classifies the fire as a high-level fire and, in addition to the execution of primary and intermediate commands, outputs an advanced command to control and trigger the entire site's emergency broadcast and evacuation indicator lights. It also pushes the coordinates of the hazardous area, personnel location data, and the comprehensive risk value R to the fire command platform. By defining the risk threshold for a high-level fire (R≥0.8), corresponding to severe fires or situations where personnel are trapped, the highest level of emergency response is initiated to ensure personnel safety. This includes controlling and activating the entire site's emergency broadcast (with clear volume) and evacuation indicator lights (in directional mode) to achieve rapid personnel evacuation, pushing precise data to the fire command platform, and establishing a seamless link between on-site emergency response and off-site rescue, addressing the pain points of ambiguous information and unclear routes in traditional rescue efforts. This achieves a full-chain response of fire control, evacuation, and rescue coordination. Through quantified parameters and precise data push, it improves evacuation efficiency and rescue success rate, enhancing the integrity and coordination of fire rescue plans.

[0038] In some preferred embodiments, the video flame recognition submodule integrates an infrared thermal imaging unit and a temperature data fusion unit. The infrared thermal imaging unit collects temperature field distribution data of the construction area, and the temperature data fusion unit fuses the temperature field distribution data with the temperature data collected by the infrared temperature sensor submodule and transmits it to the intelligent linkage module to achieve accurate identification of fire data in low light and dense smoke environments. The infrared thermal imaging unit acquires temperature field distribution data of the construction area and outputs a three-dimensional temperature field data set {T} of the construction area. ir (x, y, z)}, where (x, y, z) are rasterized coordinates, T ir (x, y, z) represents the real-time temperature of this grid node; The infrared temperature sensor submodule collects temperature data T in real time. temp (i) (i=1, 2, ..., n, where n is the number of monitoring points), and simultaneously capture the coordinates (x) of abnormal high temperature signals ≥60℃. i y i , z i ); The temperature data fusion unit fuses data for each grid node (x, y, z) to obtain the final temperature value T. fuse (x, y, z), the formula is: T fuse (x, y, z) = w1 × T ir (x, y, z) + w2×T tempmap (x, y, z); Where w1 + w2 = 1; T fuse (x, y, z) represents the final temperature after fusion of grid nodes (x, y, z), used to assist in fire situation determination; w1 represents the weight of the temperature field distribution data collected by the infrared thermal imaging unit. In low light / dense smoke environments, w1=0.7 (infrared thermal imaging is more reliable in this environment); in normal lighting environments, w1=0.5 (dual-source data are equally reliable). The weight can be adaptively adjusted based on the on-site environmental conditions; T ir (x, y, z) represents the temperature of the grid node (x, y, z) acquired by the infrared thermal imaging unit, i.e., the temperature field distribution data; w2 represents the weight of the temperature data acquired by the infrared temperature sensor submodule; T tempmap (x, y, z) represents the temperature value mapped from the temperature data to the grid node. If the grid node (x, y, z) is an infrared temperature sensing monitoring point, then T is directly taken. temp (i); If it is not a monitoring point, the temperature is interpolated from the temperature of the nearest monitoring point (ensuring calibration data for the entire grid area), and the formula is: ; Where T tempmap (x, y, z) represents the interpolated temperature of the non-monitoring point grid (x, y, z) (output result, used for subsequent dual-source temperature fusion); k represents the number of neighboring infrared temperature sensing monitoring points participating in the interpolation (selecting the monitoring point closest to the current grid); i represents the index of the neighboring monitoring point (i=1, 2, ..., k); d iRepresents the three-dimensional straight-line distance from the current grid (x, y, z) to the i-th nearest monitoring point; p represents the distance attenuation coefficient (controlling the attenuation rate of the weighted influence of the temperature of nearby monitoring points on the grid); T temp (i) represents the real-time temperature data of the i-th neighboring monitoring point; This represents the weight coefficient of the i-th neighboring monitoring point; The intelligent linkage module extracts the highest temperature T in the fire area based on the fused temperature data. max =max{T fuse (x, y, z)}, and define the interval: T max When ≤300℃, A2=0.3; when 300℃<T max At ≤600℃, A² = 0.7; T max When the temperature is >600℃, A2 = 1.0; A2 is included as a core parameter in the calculation of the comprehensive risk value R to support the classification of fire severity levels; The intelligent linkage module uses the fused abnormal high-temperature signal data as the real-time acquisition value Si of the infrared temperature sensor submodule, and substitutes it into the calculation formula for the initial fire confidence level P1: ; Where S 0i =60℃, deviation rate The reliability of P1 is improved based on the fused temperature value calculation, providing a basis for the calibration of the final fire confidence level P.

[0039] Specifically, by integrating an infrared thermal imaging unit, the traditional single-vision recognition is upgraded to a dual-modal perception combining vision and infrared thermal imaging. This differs from traditional flame recognition solutions that rely solely on visible light, laying the hardware foundation for adaptability to complex environments. Traditional visible light flame recognition is prone to missed or false alarms in scenarios such as nighttime construction (low light) and early-stage fires (dense smoke) due to light obstruction and blurred images. Infrared thermal imaging, however, is unaffected by light or smoke and can penetrate obstructions to capture high-temperature areas. Complementing visible light vision recognition, it significantly improves the recognition accuracy in complex environments. It also supports the secondary verification of fire situation determination in the intelligent linkage module. The fused temperature field distribution data provides objective temperature evidence for flame feature matching, avoiding misjudgments based solely on visual features, enhancing the reliability of fire situation data determination, and indirectly improving the accuracy of subsequent hierarchical linkage.

[0040] In some preferred embodiments, the execution module further includes an ultrasonic ranging submodule, which is linked with the intelligent linkage module to coordinate and control the pitch angle and spray pressure of the fire-fighting mobile sprinkler head to implement directional water spraying within a range of ±0.3m in the dangerous area; The intelligent linkage module determines a "suspected fire" through secondary verification, calculates the comprehensive risk value R and the fire coordinates, and then controls the linkage with the ultrasonic ranging submodule. The intelligent linkage module pushes the coordinates of the fire center point, the fire level, and the current coordinates of the fire-fighting mobile nozzle to the ultrasonic ranging submodule (providing a target reference for the ultrasonic ranging submodule, clarifying the direction and range of ranging, and avoiding blind ranging without a target). After receiving the fire coordinates, the ultrasonic ranging submodule determines the fire location based on the current coordinates (X) of the mobile fire sprinkler head. S Y S Z S ) and the coordinates of the fire center point (X F Y F Z F Automatically adjusts the initial orientation of the ranging probe (horizontally aligned with X). F -Y F Align vertically with Z F This enables directional ranging (rather than full-area scanning, thus improving efficiency); The ultrasonic ranging submodule controls the ranging probe to emit ultrasonic signals toward the fire area, and after receiving the reflected signals, calculates the straight-line distance D from the fire sprinkler head to the nearest boundary of the fire area. real The measurement accuracy is ≤ ±0.1m, which meets the ±0.3m water spraying accuracy requirement; The ultrasonic ranging submodule calculates the theoretical distance from the fire sprinkler head to the center of the fire using the following formula: ; The ultrasonic ranging submodule will measure the actual distance D real The theoretical distance D from the fire hose reel to the center of the fire theory Compare, if |D real -D theory If | ≤ 0.2m, the distance measurement is considered valid; if |D real -D theory If the distance exceeds 0.2m, a re-measurement is triggered to avoid single-measurement errors caused by smoke or dust. The intelligent linkage module receives the valid D from the ultrasonic ranging submodule. real Then, based on the fire data stored in its own database, the pitch angle and spray pressure of the mobile fire sprinkler head are calculated. The intelligent linkage module is based on the installation height H of the fire-fighting mobile sprinkler head. S Measured distance D real The height of the fire's epicenter, Z F The vertical pitch angle θ and horizontal rotation angle φ are calculated using trigonometric functions to ensure that the water spraying direction is aimed at the center of the fire area; The formula for calculating the vertical pitch angle θ is: ; The horizontal distance ; The formula for calculating the horizontal rotation angle φ is: ; Example: If the installation height H of the portable fire sprinkler head S =4m, fire center point Z F =3m, measured horizontal distance D real If the horizontal distance is 5m, then θ = arctan(1 / 5) ≈ 11.3° (downward tilt), ensuring that the water spray accurately covers the center of the fire; The formula for calculating spray pressure P is: P=P0×k 校准 ; Where P is the spraying pressure, P0 is the base spraying pressure, and k 校准 This is the distance calibration coefficient; for a primary fire, P0 is 0.2 MPa, and D... real When ≤5m, k is taken as 1.0; when 5m < D real For fires ≤10m, take 1.1; for intermediate fires, take P0 as 0.3MPa, D real When ≤5m, k is taken as 1.0; when 5m < D real For fires ≤10m, take 1.2; for high-risk fires, take P0 as 0.4MPa, D real When ≤5m, k is taken as 1.0; when 5m < D real For depths ≤10m, take 1.3; The intelligent linkage module converts the calculated pitch angle θ, horizontal rotation angle φ, and spray pressure P into control commands and sends them to the ultrasonic ranging submodule. The ultrasonic ranging submodule controls the fire-fighting mobile sprinkler head to adjust to the calculated pitch angle, horizontal rotation angle, and spray pressure, and implements directional water spraying within a range of ±0.3m in the hazardous area.

[0041] Specifically, the ultrasonic ranging submodule provides distance feedback, and the intelligent linkage module provides a fire data location benchmark. The two work together to achieve dynamic adaptation of the parameters of the fire-fighting mobile sprinkler head, solving the shortcomings of traditional sprinklers with fixed spray ranges that cannot adapt to fire data in different locations. It enhances system synergy, ensuring that the actions of the execution modules are based on accurate fire data, rather than operating independently. Unlike the crude mode of traditional fire-fighting mobile sprinklers that manually adjust angles / pressures or spray with fixed parameters, it reduces human intervention through automatic control adjustment, improves the timeliness of fire data handling, and adapts to the needs of scenarios where the location of fire data changes dynamically at the construction site.

[0042] Implementing directional water spraying within ±0.3m of the hazardous area quantifies the precise fire control indicators (±0.3m) of the execution module, making directional water spraying verifiable and quantifiable. This avoids the problems of traditional sprinklers having too large a coverage area and poor targeting, significantly improving fire extinguishing efficiency. It avoids water waste (spraying only in hazardous areas, in line with green construction concepts) and prevents equipment and materials in non-hazardous areas from being damaged by water immersion, balancing fire protection effectiveness and construction property protection. It supports a tiered linkage logic: adapting to tiered fire response, it achieves "point-to-point" fire control through precise directional water spraying for fires of different ranges and levels, avoiding excessive interference caused by full-area spraying, and enhancing the accuracy and flexibility of the system response.

[0043] In some preferred embodiments, the UWB personnel positioning tag submodule also includes an emergency alarm button unit, a voice interaction unit, and a speaker unit. The emergency alarm button unit receives a trigger operation from the construction worker, who can generate an alarm signal by pressing the alarm button on their safety helmet. The voice interaction unit is used to provide feedback on the situation on-site through real-time voice interaction with the worker, and the microphone is located on the safety helmet. The speaker unit is used to transmit personalized evacuation route broadcast information to the worker, and the speaker is located on the worker's safety helmet.

[0044] The intelligent linkage module also includes a personalized evacuation route submodule, which calculates and generates personalized evacuation routes based on fire data and personnel information. The intelligent linkage module calculates the total estimated cost based on the starting point S and the current node n in the personnel information, using the following formula: f(n)=g(n)+h(n); Where f(n) represents the total estimated cost from the starting point S to node n to the ending point E (safe exit); g(n) represents the actual cumulative cost from the starting point S to node n; and h(n) represents the heuristic estimated cost from node n to the ending point E. The formula for calculating the actual cumulative cost g(n) is as follows: g(n) = g len (n)×W(n)+g fire (n) + g crowd (n); Where g len (n) represents the path length cost; W(n) represents the traffic efficiency weight; g fire (n) represents the cost of fire risk; g crowd (n) represents the cost of congestion risk; Wherein, the path length cost g len The formula for calculating (n) is: ; Where parent(n) represents the parent node of node n (the previous node in the path); g len (parent(n)) represents the length cost from the parent node to the starting point; L represents the grid side length; The formula for calculating the traffic efficiency weight W(n) is as follows: ; Among them, the cost of fire risk g fire The formula for calculating (n) is: ; Where k fire This represents the risk weight of fire data, with a fixed value of k. fire =10; d(n, F) represents the straight-line distance from node n to the center point F of the danger zone. ; (X) n Y n Z n Let (X) be the raster coordinates of node n. F Y F Z F () represents the grid coordinates of the fire's center point; =0.1, representing the minimum tolerance coefficient, which takes a fixed value; T max This represents the flame core temperature threshold, with a fixed value of T. max =600℃, this embodiment can be set to T max =600℃; the result is normalized to [0, 5], the closer to the high temperature zone, the higher the cost; max() represents the maximum value function; T(n) represents the real-time temperature of grid node n; T0 represents the high temperature risk threshold, with a fixed value of T0=60℃; Among them, the cost of congestion risk g crowd The formula for calculating (n) is: ; Where g crowd (n) represents the congestion risk cost of the current grid node n (quantifying the evacuation risk caused by the gathering of people at the node, used for calculating the actual cumulative cost). The larger the value, the higher the congestion risk of the node, and the more it needs to be avoided; k crowd This represents the congestion risk weighting coefficient (which sets the priority of congestion risk in route planning, balancing "safety avoidance" and "smooth evacuation"), with a fixed value of k. crowd =3; n is the current grid node to be evaluated, corresponding to the specific spatial location of the construction site, with coordinates (X, Y, Z). n Y n Z n); min(·, 2) represents the minimum value function (limiting the upper bound of the congestion risk cost, avoiding algorithm decision imbalance caused by excessive congestion at a single node, and ensuring the rationality of path planning). The logical rule is that when ≥2, the function result = 2. When <2, the function result = the calculated value within the parentheses; C(n) represents the real-time number of people gathering at grid node n (quantifying the congestion degree of the node and reflecting the risk of people clustering during evacuation); C0 represents the congestion critical threshold (the personnel quantity standard for determining whether a node belongs to the "congestion risk area", adapting to the characteristics of the construction site passage), and the fixed value is C0 = 5; Among them, the calculation formula for the heuristic estimated cost h(n) is: h(n) = P exit ×α×(|x n - x E | + |y n - y E | + |z n - z E |)×L; Among them, h(n) represents the heuristic estimated cost from the current grid node n to the end point E (safe exit) (quantifying the theoretical minimum cost of the future path and improving the evacuation path search efficiency); P exit represents the priority coefficient of the end point E (safe exit) (distinguishing the main / backup exits and preferentially selecting a better safe exit), with fixed values. For the main exit, P exit = 1.0, and for the backup exit, P exit = 1.2; α represents the heuristic calibration coefficient (avoiding overestimation of the path cost and ensuring the optimality of the evacuation path search), with a fixed value of α = 0.9; (x n , y n , z n ) represents the coordinates of the current grid node n; (x E , y E , z E ) represents the coordinates of the end point E (safe exit); |x n - x E | + |y n - y E | + |z n - z E | represents the three-dimensional Manhattan distance from node n to the end point E (quantifying the spatial straight-line distance and adapting to the three-dimensional space of the construction site), which is obtained by summing the absolute values of the coordinate differences on the X / Y / Z axes and reflects the spatial proximity between the node and the end point; L represents the grid side length; n is the current grid node to be evaluated; E is the path end point (i.e., the preset safe exit of the construction site, including the main exit and the backup exit), and the intelligent linkage module selects the optimal exit according to the personnel information and fire data (preferentially the main exit, and switches to the backup exit when the main exit is blocked by the fire); The personalized evacuation route submodule calculates and generates personalized evacuation routes based on the total estimated cost f(n), the actual cumulative cost g(n), and the heuristic estimated cost h(n), using the following formula: ; Where Path final This represents the personalized evacuation path corresponding to the current node n; Reduce() represents the path simplification function, filtering out collinear nodes; CollinearFilter represents the collinearity filtering rule, determining whether three consecutive nodes are collinear, and retaining the first and last nodes if they are collinear; Convert() represents the instruction conversion function, converting the simplified grid path into a voice instruction; VoiceInstruction represents the voice instruction rule, converting the grid path into natural language containing direction and distance; This represents the minimum cost search operator, which selects the node with the smallest total estimated cost f(n) among the nodes to be explored.

[0045] It should be noted that the significance of the personalized evacuation route submodule in generating personalized evacuation routes based on the location starting point S and current node n in the personnel information lies in achieving precise adaptation of "one person, one route": Unlike the traditional fire protection's extensive mode of "uniform broadcast instructions (such as 'all personnel evacuate to the exit')" and "fixed route planning," personalized evacuation routes are generated based on each construction worker's real-time UWB positioning data (≤0.5m accuracy), with the starting point being the individual's current location and the ending point being the optimal safe exit, thus solving the pain point of "disconnection between evacuation routes and individual locations." Dynamic integration of multi-dimensional risk constraints: By integrating fire risk g... fire Congestion risk g crowdThe system achieves high traffic efficiency (W) by avoiding dynamic fire hazard areas (T≥60℃), congested areas (≥5 people per grid), and static obstacles (scaffolding / barriers) in real time during evacuation route generation. This overcomes the shortcomings of traditional routes that "only consider distance and ignore real-time risks," enabling evacuation routes to have "dynamic adjustment" capabilities. It also connects multiple functional modules in a closed loop: as the core hub of "UWB positioning → intelligent linkage decision-making → UWB tag voice broadcasting → fire rescue coordination," personalized evacuation routes transform abstract positioning and risk data into executable evacuation instructions, directly supporting functions such as "UWB tag voice directional broadcasting" and "fire rescue vehicle navigation system push," ensuring a closed loop across the entire system's "perception, decision-making, execution, and rescue" chain. In addition, it adapts to dynamic work scenarios: Construction site work surfaces move frequently, and temporary facilities (such as scaffolding and material piles) change often. Personalized evacuation routes are based on BIM grid modeling (e.g., the specific grid side length of an L×L×L is 0.5m×0.5m×0.5m), and the grid status can be updated in real time (e.g., adding material piles marked as obstacle nodes), avoiding the "path failure" caused by traditional fixed paths due to site changes. It also adapts to complex spatial structures: For the diverse spatial characteristics of construction sites, including "main passages / narrow passages / stairs / high-altitude work surfaces," the path prioritizes main passages (W=1.0) and avoids narrow passages (W=1.5) through a traffic efficiency weight W. Simultaneously, it adapts to changes in Z-axis height (e.g., going up and down stairs) through 3D grid modeling, solving the problem that "planar paths cannot cover three-dimensional construction spaces." Adaptable to sudden risk scenarios: Construction sites are prone to instantaneous risks, such as local flammable gas leaks, local high temperatures caused by welding sparks, and temporary gatherings of people. Personalized evacuation routes are dynamically adjusted through real-time data (such as temperature field distribution data and UWB personnel information statistics). For example, when a sudden congestion occurs in a certain area, detour routes are immediately planned for subsequent personnel to avoid the risk of "secondary stampedes".

[0046] More specifically, personalized evacuation routes also have the following advantages: Prioritizing the avoidance of fatal risks: the cost of fire risk g in the algorithm. fire Highest weight (k) fire =10), during path generation, the path is forced to stay away from the center of the danger zone and high-temperature areas (T≥60℃), ensuring that the path is always within the "safe temperature zone and non-fire data spread range," thus fundamentally preventing people from accidentally entering the danger zone. Reduce the risk of congestion and stampedes: through the congestion risk cost g crowdQuantifying the number of people in a single grid and prioritizing low-congestion nodes (C(n) < 5) for path selection, while dispersing evacuation paths for different personnel (e.g., avoiding multiple people crowding into the same narrow passage), solves the pain point of "evacuation congestion" caused by dense personnel and limited passages at construction sites. Ensuring evacuation reliability in special scenarios: Combined with infrared thermal imaging units, evacuation paths can still be accurately planned based on temperature field data in low-light and dense smoke environments (unaffected by light / smoke obstruction); combined with temporary power outage logic, paths avoid dangerous passages around the power outage area, ensuring environmental safety throughout the evacuation process. Simplifying emergency decision-making costs: Through "directional voice broadcasting" of UWB tags (e.g., "walk 3 meters west along the current passage → turn left to the main passage"), complex spatial paths are transformed into concise "direction + distance" instructions. Construction personnel do not need to remember the site layout or assess risks; they can directly evacuate according to instructions, significantly reducing the understanding cost in emergency situations and shortening evacuation time. Optimizing safety exit resource allocation: When generating paths, the "high priority + shortest distance" safety exits (main exit P) are automatically selected. exit =1.0, Backup outlet P exit =1.2), avoiding "exit congestion" caused by multiple people rushing to the same exit, and achieving balanced utilization of exit resources. Connecting off-site rescue coordination: The starting point (personnel location), ending point (safe exit), and path node data of personalized evacuation routes can be pushed to the fire rescue vehicle navigation system after being superimposed on the BIM model. Rescue personnel can know the evacuation trajectory and location of trapped personnel on the way, plan rescue routes in advance, and shorten rescue response time.

[0047] In some preferred embodiments, the operation and maintenance module further includes a cloud management platform submodule and a local touch screen submodule; The cloud management platform sub-module connects to the construction site BIM model. When a fire occurs, it automatically overlays fire data and personnel information onto the BIM model. The fire data includes the BIM 3D coordinates of the fire center point, the set of grid coordinates of the fire area boundary, the comprehensive risk value R, and the fire level. The personnel information includes the real-time BIM coordinates of personnel, personnel identity and tag number binding information, and the personalized evacuation route corresponding to the personnel. The local touch screen submodule generates a visual emergency command interface, which is simultaneously pushed to the fire rescue vehicle navigation system. The visual emergency command interface includes a BIM model top view and a core data panel. The BIM model top view identifies the dangerous area where the fire has occurred and the distribution of people in the surrounding area. The core data panel displays the fire level, comprehensive risk value R, number of people in the dangerous area, and operating status of fire-fighting equipment in real time.

[0048] Specifically, the operation and maintenance module also includes a cloud management platform sub-module and a local touch screen sub-module. The cloud management platform sub-module connects to the construction site BIM model and automatically overlays fire data and personnel information onto the BIM model when a fire occurs. By introducing the BIM model (Building Information Model) as a data visualization carrier, it solves the pain point of traditional fire protection systems where fire data and personnel information are scattered and independent, and cannot be linked to spatial locations. It transforms abstract smoke concentration and personnel coordinates into dangerous area markings and personnel location distribution in a three-dimensional model, intuitively presenting the risk situation. It connects fire data and personnel information to ensure the continuity of data sources. At the same time, it achieves real-time correlation between data and space through automatic overlay, providing dual decision-making basis for emergency command based on space and data (such as determining whether fire data has spread to critical equipment areas or whether personnel are trapped in dangerous areas). It strengthens the advantages of cloud management, adapts to the needs of centralized management of multiple construction zones and multiple projects, and allows managers to remotely monitor the on-site dynamics in real time through the BIM model, avoiding the lag of traditional methods that require on-site visits to understand the situation, and improving the efficiency of remote operation and maintenance and emergency command.

[0049] The local touchscreen submodule generates a visual emergency command interface, which is simultaneously pushed to the fire and rescue vehicle navigation system. The local touchscreen submodule transforms the BIM-overlaid related data into a simple and intuitive emergency command interface. The visual emergency command interface includes a top view of the BIM model and a core data panel. The top view of the BIM model marks the dangerous area where the fire has occurred and the distribution of people in the surrounding area. The core data panel displays the fire level, comprehensive risk value R, number of people in the dangerous area, and operating status of fire-fighting equipment in real time, which facilitates on-site management personnel to quickly obtain key information without relying on complex operations, improves the efficiency of on-site emergency response, and meets the needs of real-time command at the construction site. By simultaneously pushing the information to the fire and rescue vehicle navigation system, it solves the core pain point of traditional rescuers not knowing the site layout, fire data details, and personnel location before arriving at the scene. Rescuers can obtain accurate information including BIM coordinates while en route, plan the optimal rescue route in advance, shorten the rescue response time, and break down the information barriers between on-site emergency response and off-site rescue.

[0050] In some preferred embodiments, the intelligent linkage module is linked with the temporary power management system at the construction site. When a fire occurs, the intelligent linkage module controls the priority cut-off of temporary power circuits within 10m of the dangerous area, while preserving the power supply circuits for emergency lighting and fire-fighting equipment. The intelligent linkage module also includes a power failure linkage sub-module, which links and controls the temporary power management system at the construction site. When the intelligent linkage module determines that the fire level is ≥ medium fire, the power failure linkage submodule will trigger a fire alarm based on the coordinates of the fire center point (X). F Y F Z FThe system uses BIM grid modeling data (L×L×L, where L represents the grid side length) to delineate a 10m three-dimensional spatial range around the hazardous area and automatically sends a power outage command to the temporary power management system at the construction site. The power outage command limits the scope of cutting off non-emergency temporary power circuits within a 10m three-dimensional space around the danger zone, including power circuits for construction machinery, temporary lighting, and office power; at the same time, it retains emergency lighting power circuits and dedicated power circuits for fire-fighting equipment, including power circuits for fire extinguisher solenoid valves, sprinkler systems, and ultrasonic ranging submodules. After the temporary power supply management system at the construction site executes the power outage command, it sends the execution result back to the power outage linkage submodule. If the execution fails, the power outage linkage submodule will resend the power outage command and synchronize the execution result to the local touch screen submodule for alarm display.

[0051] Specifically, the intelligent linkage module is linked with the temporary power management system at the construction site: it addresses the safety pain points at the construction site, such as messy temporary power lines and large load fluctuations, which can easily lead to fires due to short circuits and overloads. Furthermore, failure to cut off power in time after a fire can result in electric shock or the fire spreading. The linkage control can interrupt the risk chain in advance.

[0052] When a fire occurs, the intelligent linkage module prioritizes the cutting off of temporary power circuits within a 10m radius of the danger zone. Triggered by the fire, the power-off linkage submodule acts as the decision-making center, issuing power-off commands and quantifying the power-off range (within 10m). This ensures precise and controllable power-off operations, avoiding the drawbacks of traditional "full-area power outages" that lead to construction stoppages and equipment data loss in non-dangerous areas, or "incomplete local power outages" that fail to block electrical fire sources. Emphasizing "priority cutting off," the module places power-off before or simultaneously with firefighting and alarm actions, quickly eliminating electrical risks in dangerous areas (such as short-circuit ignition and electric shock hazards), and providing a safe environment for subsequent tiered firefighting and evacuation operations.

[0053] Simultaneously, the power supply circuits for emergency lighting and fire-fighting equipment are retained: This ensures the rigor of the power outage logic, avoids secondary risks caused by a "one-size-fits-all" power outage, and maintains the emergency lighting circuit to ensure clear visibility during personnel evacuation, preventing congestion and trampling; it also maintains the power supply circuits for fire-fighting equipment to ensure the normal operation of core fire-fighting equipment such as sprinklers, alarms, and emergency broadcasts, without affecting fire data processing; it meets the requirement of "safety first, practicality second," blocking the source of risk while ensuring the basic conditions for emergency response and evacuation, unlike traditional power outage schemes that "ignore emergency power supply needs"; and it connects the hierarchical linkage logic to provide power support for various emergency actions under primary, intermediate, and advanced fire situations, ensuring the effective implementation of hierarchical linkage commands and strengthening the coherence and reliability of the system's functional chain.

[0054] In some preferred embodiments, the intelligent linkage module further includes a fire prediction submodule and a dynamic adaptation submodule; When the intelligent linkage module determines that the fire level is a high-level fire, the fire prediction submodule triggers the start of fire spread trend prediction, and the dynamic adaptation submodule triggers the start of dynamic strategy adaptation mode. The fire prediction submodule loads a spatiotemporal multidimensional fire spread prediction model. The spatiotemporal multidimensional fire spread prediction model uses BIM raster modeling data (L×L×L, where L is the grid side length) as the spatial carrier and is trained and calibrated by synchronously inputting construction site environmental data collected by the fire prediction submodule, real-time fire data collected by the multi-source sensing module, and historical fire data. The construction site environmental data includes wind speed, wind direction, and humidity at the construction site. The spatiotemporal multidimensional fire spread prediction model outputs the fire spread trend prediction results for the next 3 minutes at a frequency of 1 time / second, including the spread direction, spread speed, three-dimensional boundaries of the fire field at different time nodes, and the threatened sequence of high-risk areas. The high-risk areas include at least areas with dense combustible materials and areas with dense populations. The formula for calculating the fire spread rate using the spatiotemporal multidimensional fire spread prediction model is as follows: ; Where v fire (t) represents the rate of fire spread at time t in the future; k v This represents the wind direction correction factor, with a fixed value, and the wind spreads in the same direction (k). v Take 1.2, vertical spread k v Take 0.8, back propagation k v =0.5; v represents the real-time wind speed at the construction site; T core T represents the real-time core temperature of the fire; T0 represents the critical threshold of fire risk; T max The value represents the upper limit of the flame core temperature; h represents the real-time humidity at the construction site; the effect of humidity on the decay of the spread rate is quantified by 1-0.01h. The prediction formula for calculating the three-dimensional boundary coordinates of a fire scene using a spatiotemporal multidimensional fire spread prediction model is as follows: ; Among them (X) Fb (t), Y Fb (t), Z Fb (t) represents the BIM coordinates of any point on the three-dimensional boundary of the fire at time t in the future (forming a boundary set, corresponding to the three-dimensional boundary of the fire at different time nodes); (X) F0 Y F0 Z F0) represents the current BIM coordinates of the fire center (provided by real-time fire data collected by the multi-source sensing module); β represents the real-time wind direction at the construction site (with true north as 0°, provided by environmental data of the construction site collected by the fire prediction submodule); (cosβ, sinβ) correspond to the horizontal spread components; k z This represents the vertical creep correction factor (in construction scenarios, the vertical creep rate is 0.3 times that of the horizontal creep rate, k). z =0.3), adapting to the three-dimensional spatial characteristics of building floors and high-altitude work surfaces; grid association: coordinate values ​​are all based on BIM gridded modeling data (L×L×L, where L is the grid side length), ensuring accurate matching with the actual location on the construction site; The dynamic adaptation submodule dynamically updates personalized evacuation routes based on prediction results (correcting fire risk costs g). fire (n), avoid high-risk areas in advance and push the information to the UWB personnel positioning tag submodule in time, optimize the scheduling of fire-fighting equipment (adjust the fire-fighting mobile nozzles to key interception points in advance according to the predicted spread trajectory, and preset the pitch angle and spray pressure), and adjust the temporary power supply control range (cut off non-emergency power circuits within 15m of the predicted spread path in advance to prevent the fire from igniting the electrical equipment). The formula for calculating the corrected fire risk cost by the dynamic adaptation submodule is as follows: ; Where g fire (t) represents the corrected fire risk cost of the grid node at time t in the future; k fire Represents the fire risk weighting coefficient; d(n, F) b (t) represents the three-dimensional Euclidean distance from a grid node to the fire boundary at time t in the future, calculated using the following formula: ; This represents the fault tolerance coefficient, with a fixed value. =0.1; k speed (t) represents the propagation speed weighting coefficient, k speed (t) = 1 + 0.5·v fire (t), the faster the spread, the higher the weight, amplifying the risk and cost of rapidly spreading areas and ensuring that the path is avoided in advance; other parameters T core T0, T max All data is publicly available as mentioned above; The preset formula for calculating the pitch angle of the fire sprinkler head in the dynamic adaptation submodule is as follows: ; Where α(n, t) represents the preset pitch angle of the fire-fighting mobile sprinkler relative to the fire boundary at time t in the future; Z n The Z-axis BIM coordinates representing the installation location of the portable fire sprinkler head; d xy (n, F)b (t) represents the horizontal distance between the nozzle and the fire boundary at time t in the future, and is calculated using the following formula: ;k α The compensation angle is represented by a fixed value k. α =3°; The formula for calculating the temporary power cut-off range in the dynamic adaptation submodule is as follows: d(n, F) path (t))≤15m; Where d(n, F) path (t) represents the shortest distance from grid node n to the fire spread path at time t in the future; 15m represents the range value for early cut-off to avoid the fire igniting electrical equipment; when node n satisfies the formula, the dynamic adaptation submodule issues an instruction to cut off the non-emergency power circuit corresponding to that node; The fire prediction submodule verifies the prediction accuracy using real-time fire data and personnel information. When the deviation between the predicted fire boundary and the actual fire boundary is greater than 0.8m, it automatically corrects the parameters of the spatiotemporal multidimensional fire spread prediction model and re-predicts. At the same time, it synchronizes the prediction results, adaptation strategies, and verification data to the cloud management platform submodule for archiving, the local touch screen submodule for visualization display, and pushes them to the fire and rescue vehicle navigation system to assist in rescue route planning. The calibration formula for calculating the prediction accuracy of the fire prediction submodule is as follows: ; Where △d represents the three-dimensional deviation between the predicted fire boundary and the actual fire boundary at time t; (X Freal (t), Y Freal (t), Z Freal (t) represents the actual fire boundary BIM coordinates (real-time fire data) collected by the multi-source sensing module at time t; the correction trigger condition is △d > 0.8m, and the model parameters (k) are automatically adjusted after triggering. v k z ), and regenerate the prediction results.

[0055] The temporary fire-fighting linkage control system for construction sites provided in this application, based on the above technical solutions, has the following advantages: ① Enhanced Sensing Precision: Reduces false alarms and missed alarms, covering risks across all scenarios. Adopting a multi-source sensing architecture, it overcomes the limitations of traditional single sensors, combining an infrared thermal imaging unit to achieve accurate identification of fire data in low-light and dense smoke environments. It effectively filters interference from construction dust, welding sparks, etc., significantly reducing the false alarm rate. Quantifying sensing thresholds (e.g., methane ≥0.5% VOL, temperature ≥60℃) and feature matching degree (P2∈[0,1]) makes fire data and personnel information collection quantifiable and verifiable, providing accurate data support for subsequent decision-making.

[0056] ② Intelligent Linkage: Precise Tiered Response and Improved Response Efficiency. A secondary verification mechanism of "initial confidence P1 + flame matching degree P2" is established. Through weighted fusion and calibration algorithms (P=α×P1+(1-α)×P2), the false triggering of a single signal is avoided, improving the accuracy of fire situation judgment. Based on the tiered linkage logic of the comprehensive risk value R, differentiated emergency actions are matched according to "primary-intermediate-advanced" (such as fire extinguishing within 5m → power cut-off within 10m → evacuation of the entire site). This avoids the waste of resources caused by "over-response" and prevents the expansion of fire caused by "insufficient response", shortening the response time to the second level. Cross-system linkage design prioritizes cutting off temporary power circuits within 10m when a fire occurs while retaining emergency power supply, blocking electrical fire sources and electric shock risks from the source and ensuring the safe implementation of emergency actions.

[0057] ③ Personalized evacuation: Ensuring personnel safety and mitigating secondary risks. Based on improved A The algorithm's personalized evacuation routes integrate fire risk (gfire), congestion risk (gcrowd), and traffic efficiency (W) to achieve precise "one person, one route" planning, prioritizing the avoidance of high-temperature areas and congested nodes, significantly reducing the risk of stampedes. The combination of path command conversion (Pathfinal=Convert(...)) and UWB tag-based targeted broadcasting transforms complex spatial paths into concise voice commands of "direction + distance," adapting to rapid understanding in emergency situations and improving evacuation efficiency by more than 30%. 3D grid modeling and BIM coordinate adaptation cover three-dimensional spaces such as construction site floors and high-altitude work surfaces, solving the problem that traditional planar paths cannot adapt to complex construction environments.

[0058] ④ Collaborative Rescue: Breaking down information barriers and improving rescue efficiency. The operation and maintenance module features a dual-end design. A cloud-based BIM model overlays fire data and personnel information data, while a local touchscreen generates a visual command interface, which is simultaneously pushed to the fire and rescue vehicle navigation system. Rescuers can grasp the situation on the ground and plan the optimal rescue route while en route. In the case of advanced fires, personnel location data and comprehensive risk value R are pushed, achieving seamless collaboration between "on-site evacuation and off-site rescue," shortening rescue response time and increasing the probability of trapped personnel being rescued.

[0059] ⑤Scenario Adaptability: The system is stable and reliable, perfectly suited to temporary construction needs. The multi-source sensing module's "zonal deployment and mobility" design dynamically adjusts the detection range according to the construction progress, adapting to scenarios such as changes in temporary work areas and material relocation. The fault self-diagnosis function monitors for sensor offlineness and equipment malfunctions in real time, accurately marking fault locations and issuing warnings, ensuring long-term stable system operation and preventing equipment failure during emergencies. Ultrasonic ranging and directional water spraying achieve precise fire suppression within ±0.3m in hazardous areas, reducing water loss to equipment and materials in non-hazardous areas, balancing firefighting effectiveness with the protection of construction property.

[0060] ⑥ Application Value: Cost Reduction and Efficiency Improvement, Enhanced Safety Management. The integrated intelligent architecture replaces traditional independent equipment splicing, reducing manual inspection and operation costs and improving maintenance efficiency by 50%. Full-process quantitative indicators (confidence level P, risk value R, positioning accuracy ≤0.5m) shift fire management from "experience-based" to "data-driven," contributing to safety and compliance at construction sites. Its flexible design adapts to temporary construction scenarios, allowing for rapid deployment at various construction sites, including residential, municipal, and industrial sites, demonstrating strong versatility and broad application prospects.

[0061] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A temporary fire-fighting linkage control system for construction sites, characterized in that, It includes a multi-source sensing module, an intelligent linkage module, an execution module, and an operation and maintenance module, with each module communicating and connecting collaboratively. The multi-source sensing module is deployed in the construction zone and is mobile, and collects fire data and personnel information in real time. The fire data includes smoke concentration data, temperature data, combustible gas concentration data and flame information, and the personnel information includes personnel location data. The intelligent linkage module performs secondary verification of the fire situation based on fire data and personnel information and outputs hierarchical linkage instructions. The execution module executes hierarchical linkage instructions; The operation and maintenance module generates a visual emergency command interface based on fire data, personnel information, and hierarchical linkage instructions, and pushes it to the fire and rescue vehicle navigation system.

2. The temporary fire-fighting linkage control system for construction sites according to claim 1, characterized in that, The multi-source sensing module includes a smoke detector submodule, an infrared temperature detector submodule, a combustible gas sensor submodule, a video flame recognition submodule equipped with the YOLO target detection algorithm, and a UWB personnel positioning tag submodule. The smoke detector submodule collects smoke concentration data in the construction zone in real time. When the smoke concentration data exceeds the preset safety upper limit threshold, it triggers the generation of a smoke detection anomaly signal and transmits the smoke concentration data and the smoke detection anomaly signal to the intelligent linkage module. The infrared temperature sensor submodule is used to perform non-contact temperature monitoring of equipment, materials and work surfaces in the construction area, collect temperature data of the environment and target objects in real time, trigger the generation of abnormal high temperature signal when the temperature data is ≥60℃ and record the temperature change trend, and transmit the temperature data and abnormal high temperature signal to the intelligent linkage module. The combustible gas sensor submodule is used to detect combustible gas concentration data at the construction site. It presets safety thresholds for different gases. When the gas concentration reaches the threshold, it triggers a combustible gas exceeding the standard signal and simultaneously transmits the real-time combustible gas concentration data and the combustible gas exceeding the standard signal to the intelligent linkage module. The video flame recognition submodule, equipped with the YOLO target detection algorithm, acquires video footage of the construction area in real time. It quickly identifies suspected flame areas in the footage using the YOLO target detection algorithm, extracts flame information, compares it with the built-in flame feature library, generates flame recognition results, and transmits the flame recognition results and video footage to the intelligent linkage module to assist in secondary verification of fire data. The flame information includes the shape, color, and dynamic characteristics of the flame. The UWB personnel positioning tag submodule is used to collect positioning sensor data attached to the safety helmets of construction workers, with a positioning accuracy of ≤0.5m, and collects personnel information in real time and transmits it to the intelligent linkage module.

3. A temporary fire-fighting linkage control system for construction sites according to claim 2, characterized in that, The intelligent linkage module includes a multi-source data weighted fusion submodule and a logical judgment submodule; When a single submodule in the multi-source sensing module is triggered, that submodule uses the deviation rate between the real-time acquired value S and the preset threshold S0 as the basic parameter. The multi-source data weighted fusion submodule calculates the "initial fire confidence level P1" based on the basic parameter, using the following formula: ； Where i = 1, 2, 3, corresponding to the smoke detector submodule, infrared temperature detector submodule, and combustible gas sensor submodule, respectively; W i S represents the weights of each sub-module in the multi-source sensing module; i This represents the real-time acquired value of this submodule; S 0i This indicates the security threshold of the corresponding submodule; This represents the deviation rate, with a value range of [0, 1]. The closer it is to 0, the higher the risk. The logic judgment submodule, in conjunction with the video flame recognition submodule, retrieves the real-time footage captured by the video flame recognition submodule. It then outputs the flame feature matching degree P2 using the YOLO target detection algorithm. Based on P1, P2 is calibrated to obtain the "final fire confidence level P," calculated using the following formula: P=α×P1+(1-α)×P2; Where α represents the calibration coefficient; The logic judgment submodule performs a secondary verification of the fire situation determination: if P≥0.7, it is determined to be a suspected fire, triggering the subsequent tiered linkage command output; If P < 0.5, the device self-test is triggered; if 0.5 ≤ P < 0.7, the delayed verification mechanism is activated: data is collected and the P value is calculated repeatedly at 3-second intervals. If the P value calculated twice consecutively is ≥ 0.7, it is determined to be a suspected fire.

4. A temporary fire-fighting linkage control system for construction sites according to claim 3, characterized in that, The intelligent linkage module also includes a fire comprehensive risk value calculation submodule and a hierarchical decision-making submodule; The fire risk calculation submodule calculates the comprehensive risk value R based on fire data and personnel information, using the following formula: R=A×(β1×A1+β2×A2)+B×(γ1×B1+γ2×B2)+C×δ; Where R represents the comprehensive risk value; A represents the fire intensity coefficient; A1 represents the quantified flame area value; A2 represents the quantified core temperature value, which is collected by the infrared temperature sensor submodule; β1 represents the flame area weight; β2 represents the core temperature weight; B represents the personnel risk coefficient; B1 represents the quantified number of personnel in the danger zone; B2 represents the quantified distance between personnel and the fire center point; γ1 represents the weight of the number of personnel in the danger zone; γ2 represents the weight of the distance between personnel and the fire center point; C represents the environmental risk coefficient; and δ represents the quantified value of the combustible gas concentration exceeding the standard multiple. The hierarchical decision-making submodule classifies the fire level according to the range of the comprehensive risk value R and outputs the corresponding hierarchical linkage instructions: When R∈[0.3, 0.5], the hierarchical decision-making submodule classifies it as a primary fire and outputs a primary command, which only controls the audible and visual alarms in the danger zone and the solenoid valves of fire extinguishers within 5m of the center of the danger zone. When R∈[0.5, 0.8], the hierarchical decision-making submodule classifies it as a medium-level fire and outputs a medium-level command on the basis of the primary command execution, controlling the start of the zone sprinkler system and the temporary power circuit controller within 10m around the danger zone; When R≥0.8, the hierarchical decision-making submodule classifies it as a high-level fire and outputs additional high-level instructions on the basis of the execution of primary and intermediate instructions. It controls and triggers the whole-site emergency broadcast and evacuation indicator lights, and pushes the coordinates of the dangerous area, personnel location data and comprehensive risk value R data to the fire command platform.

5. A temporary fire-fighting linkage control system for construction sites according to claim 4, characterized in that, The video flame recognition submodule integrates an infrared thermal imaging unit and a temperature data fusion unit. The infrared thermal imaging unit collects temperature field distribution data of the construction area, and the temperature data fusion unit fuses the temperature field distribution data with the temperature data collected by the infrared temperature sensor submodule and transmits it to the intelligent linkage module to achieve accurate identification of fire data in low light and dense smoke environments.

6. A temporary fire-fighting linkage control system for construction sites according to claim 5, characterized in that, The execution module also includes an ultrasonic ranging submodule, which is linked with the intelligent linkage module to coordinate and control the pitch angle and spray pressure of the fire mobile sprinkler head to implement directional water spraying within a range of ±0.3m in the dangerous area; The intelligent linkage module determines a suspected fire through secondary verification, calculates the comprehensive risk value R and the fire coordinates, and then controls the linkage with the ultrasonic ranging submodule. The intelligent linkage module pushes the coordinates of the fire center point, the fire level, and the current coordinates of the mobile fire sprinkler head to the ultrasonic ranging submodule. After receiving the fire coordinates, the ultrasonic ranging submodule determines the fire location based on the current coordinates (X) of the mobile fire sprinkler head. S Y S Z S ) and the coordinates of the fire center point (X F Y F Z F Automatically adjusts the initial orientation of the ranging probe (horizontally aligned with X). F -Y F Align vertically with Z F ), to achieve directional ranging; The ultrasonic ranging submodule controls the ranging probe to emit ultrasonic signals toward the fire area, and after receiving the reflected signals, calculates the straight-line distance D from the fire sprinkler head to the nearest boundary of the fire area. real The measurement accuracy is ≤ ±0.1m, which meets the ±0.3m water spraying accuracy requirement; The ultrasonic ranging submodule calculates the theoretical distance from the fire sprinkler head to the center of the fire using the following formula: ； The ultrasonic ranging submodule will measure the actual distance D real The theoretical distance D from the fire hose reel to the center of the fire theory Compare, if |D real -D theory If | ≤ 0.2m, the distance measurement is considered valid; if |D real -D theory | > 0.2m, trigger re-distance measurement; The intelligent linkage module receives the valid D from the ultrasonic ranging submodule. real Then, based on the fire data stored in its own database, the pitch angle and spray pressure of the mobile fire sprinkler head are calculated. The intelligent linkage module is based on the installation height H of the fire-fighting mobile sprinkler head. S Measured distance D real The height of the fire's epicenter, Z F The vertical pitch angle θ and horizontal rotation angle φ are calculated using trigonometric functions to ensure that the water spraying direction is aimed at the center of the fire area; The formula for calculating the vertical pitch angle θ is: ； The horizontal distance ; The formula for calculating the horizontal rotation angle φ is: ； The formula for calculating spray pressure P is: P=P0×k 校准 ; Where P is the spraying pressure, P0 is the base spraying pressure, and k 校准 This is the distance calibration coefficient; The intelligent linkage module converts the calculated pitch angle θ, horizontal rotation angle φ, and spray pressure P into control commands and sends them to the ultrasonic ranging submodule. The ultrasonic ranging submodule controls the fire-fighting mobile sprinkler head to adjust to the calculated pitch angle, horizontal rotation angle, and spray pressure, and implements directional water spraying within a range of ±0.3m in the hazardous area.

7. A temporary fire-fighting linkage control system for construction sites according to claim 6, characterized in that, The UWB personnel positioning tag submodule also includes an emergency alarm button unit, a voice interaction unit, and a speaker unit. The emergency alarm button unit receives trigger operations from construction personnel, who can generate an alarm signal by pressing the alarm button on their safety helmet. The voice interaction unit is used to provide feedback on the scene through real-time voice interaction with personnel, and the microphone is located on the safety helmet. The speaker unit is used to transmit personalized evacuation route broadcast information to personnel, and the speaker is located on the construction personnel's safety helmet. The intelligent linkage module also includes a personalized evacuation route submodule, which calculates and generates personalized evacuation routes based on fire data and personnel information.

8. A temporary fire-fighting linkage control system for construction sites according to claim 7, characterized in that, The operation and maintenance module also includes a cloud management platform sub-module and a local touch screen sub-module; The cloud management platform sub-module connects to the construction site BIM model. When a fire occurs, it automatically overlays fire data and personnel information onto the BIM model. The fire data includes the BIM 3D coordinates of the fire center point, the set of grid coordinates of the fire area boundary, the comprehensive risk value R, and the fire level. The personnel information includes the real-time BIM coordinates of personnel, personnel identity and tag number binding information, and the personalized evacuation route corresponding to the personnel. The local touch screen submodule generates a visual emergency command interface, which is simultaneously pushed to the fire rescue vehicle navigation system. The visual emergency command interface includes a BIM model top view and a core data panel. The BIM model top view identifies the dangerous area where the fire has occurred and the distribution of people in the surrounding area. The core data panel displays the fire level, comprehensive risk value R, number of people in the dangerous area, and operating status of fire-fighting equipment in real time.

9. A temporary fire-fighting linkage control system for construction sites according to claim 8, characterized in that, The intelligent linkage module is linked with the temporary power management system at the construction site. In the event of a fire, the intelligent linkage module will prioritize cutting off the temporary power circuits within 10m of the dangerous area, while preserving the power supply circuits for emergency lighting and fire-fighting equipment. The intelligent linkage module also includes a power failure linkage sub-module, which links and controls the temporary power management system at the construction site. When the intelligent linkage module determines that the fire level is ≥ medium fire, the power failure linkage submodule will trigger a fire alarm based on the coordinates of the fire center point (X). F Y F Z F The system uses BIM grid modeling data (L×L×L, where L represents the grid side length) to delineate a 10m three-dimensional spatial range around the hazardous area and automatically sends a power outage command to the temporary power management system at the construction site. The power outage command limits the scope of cutting off non-emergency temporary power circuits within a 10m three-dimensional space around the danger zone, including power circuits for construction machinery, temporary lighting, and office power; at the same time, it retains emergency lighting power circuits and dedicated power circuits for fire-fighting equipment, including power circuits for fire extinguisher solenoid valves, sprinkler systems, and ultrasonic ranging submodules. After the temporary power supply management system at the construction site executes the power outage command, it sends the execution result back to the power outage linkage submodule. If the execution fails, the power outage linkage submodule will resend the power outage command and synchronize the execution result to the local touch screen submodule for alarm display.

10. A temporary fire-fighting linkage control system for construction sites according to claim 9, characterized in that, The intelligent linkage module also includes a fire prediction submodule and a dynamic adaptation submodule; When the intelligent linkage module determines that the fire level is a high-level fire, the fire prediction submodule triggers the start of fire spread trend prediction. The fire prediction submodule loads a spatiotemporal multidimensional fire spread prediction model. The spatiotemporal multidimensional fire spread prediction model uses BIM raster modeling data (L×L×L, where L is the grid side length) as the spatial carrier and is trained and calibrated by synchronously inputting construction site environmental data collected by the fire prediction submodule, real-time fire data collected by the multi-source sensing module, and historical fire data. The construction site environmental data includes wind speed, wind direction, and humidity at the construction site. The spatiotemporal multidimensional fire spread prediction model outputs the fire spread trend prediction results for the next 3 minutes at a frequency of 1 time / second, including the spread direction, spread speed, three-dimensional boundaries of the fire field at different time nodes, and the threatened sequence of high-risk areas. The high-risk areas include at least areas with dense combustible materials and areas with dense populations. The dynamic adaptation submodule dynamically updates personalized evacuation routes, optimizes fire equipment scheduling, and adjusts the scope of temporary power control based on the prediction results. The fire prediction submodule verifies the prediction accuracy using real-time fire data and personnel information. When the deviation between the predicted fire boundary and the actual fire boundary is greater than 0.8m, it automatically corrects the parameters of the spatiotemporal multidimensional fire spread prediction model and re-predicts. At the same time, it synchronizes the prediction results, adaptation strategies, and verification data to the cloud management platform submodule for archiving, the local touch screen submodule for visualization display, and pushes them to the fire and rescue vehicle navigation system to assist in rescue route planning.

Images