Building safety fire control method and system

Abstract

The application discloses a kind of building fire-fighting collaborative control method and system based on three-party game model.The application models fire evolution body, personnel group to be evacuated, fire-fighting facility group as first, second, third game party respectively, constructs the dynamic game model of three parties;Real-time acquisition fire parameter and personnel distribution parameter as game state;By constructing virtual game environment and using iterative strategy optimization method to solve stable equilibrium strategy, obtain three-party joint action plan;According to joint action plan, generate fire-fighting facility control instruction and evacuation guide instruction and execute in coordination;While monitoring the deviation of actual personnel behavior and strategy, trigger dynamic re-planning.The application overcomes the defects that fire, personnel and facilities are treated separately in the prior art, realizes active prevention and control and strategy cooperation under the antagonistic situation, can significantly improve the evacuation efficiency and survival probability of personnel in building fire, and has high industrial practical value.

Description

Technical Field

[0001] This invention belongs to the field of fire protection system technology, specifically relating to a building safety fire protection control method and system. Background Technology

[0002] Building fires pose a significant threat to public safety. An effective fire control system needs to quickly detect fires, guide evacuation, and activate fire-fighting facilities to minimize casualties and property damage.

[0003] Existing building fire control systems generally adopt a "sensing-decision-execution" architecture, with typical technical solutions including: fire detection methods based on multi-sensor fusion, optimal evacuation route planning methods based on static weighted graphs, and fire protection facility linkage control methods based on preset time sequences. However, these existing technologies have the following fundamental drawbacks: First, fire evolution and human behavior are treated separately: Current technologies typically view fires as external environmental disturbances and people as rational individuals who unconditionally obey system commands. However, in real fire scenarios, human behavior exhibits typical characteristics of bounded rationality—incomplete information, panic, and herd mentality can lead individual decisions to deviate from the system's preset "optimal path," resulting in secondary disasters such as exit congestion and stampedes. Current technologies fail to model and predict this behavioral game.

[0004] Second, the control of fire protection facilities lacks strategic coordination: facilities such as fire-resistant roller shutters, smoke exhaust fans, and evacuation signs mostly adopt independent triggering or preset time-series linkage, failing to achieve dynamic coordination with the fire spread and personnel distribution within a unified framework. For example, the start and stop of smoke exhaust fans can change the direction of fire spread, but existing technologies do not incorporate this causal relationship into control optimization, which may lead to a situation where the more control is applied, the worse the situation becomes.

[0005] Third, it cannot cope with the adversarial evolutionary characteristics of fire: Fire has nonlinear, time-varying, and dynamic characteristics that are sensitive to control actions. Existing technologies treat fire as a passive environmental state and lack a control framework that treats it as an "adversarial" evolving subject, resulting in delayed system response and even control actions that accelerate the spread of fire.

[0006] In summary, there is an urgent need for a fire control method and system that can uniformly describe the game relationship between fire, personnel, and facilities, and solve for the global equilibrium strategy. Summary of the Invention

[0007] The purpose of this invention is to provide a building safety fire control method and system to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a building safety fire protection control method, comprising the following steps: The fire evolution, the group of people to be evacuated, and the group of fire-fighting facilities are modeled as the first player, the second player, and the third player, respectively, to construct a three-way non-zero-sum dynamic game model. Real-time data collection of fire parameters and personnel distribution parameters within the building serves as the current state of the game theory model. By constructing a multi-party dynamic game environment that includes a fire evolution model, a personnel evacuation behavior model, and a fire-fighting facility action model, and using an iterative optimization algorithm to solve the stable equilibrium strategy of the game model, a joint action plan for the three parties is obtained. Firefighting facility control instructions are generated based on the strategy of the third party in the joint action plan, and evacuation guidance instructions are generated based on the strategy of the second party, and then executed in a coordinated manner. The deviation between actual personnel behavior and the joint action plan is monitored. If the deviation exceeds a preset threshold, the stable equilibrium strategy is re-solved.

[0009] Preferably, the utility objective of the first player is defined as maximizing the cumulative time of personnel exposure to toxic gases and high-temperature environments and the coverage of dangerous areas; the strategy space of the first player is generated by a fire dynamics proxy model, which is obtained by learning from historical fire evolution data or simulation data.

[0010] Preferably, the strategy solution for the second player adopts a layered game approach: the building is divided into multiple game zones, and the equilibrium distribution ratio of the non-cooperative game is solved for the exit selection problem of a finite number of evacuation exits in each zone; the local path selection downstream of the exit is handled by the continuous medium approximation method of crowd flow.

[0011] Preferably, the iterative strategy optimization method includes, but is not limited to: strategy iteration, value iteration, virtual game, multi-agent policy gradient method, and game strategy learning method based on experience replay.

[0012] Preferably, after solving the stable equilibrium strategy of the game model, the method further includes: calculating the Pareto boundary in the joint utility space including casualty losses, property losses, evacuation time and energy costs, and selecting a joint action plan as the actual implementation strategy from the Pareto boundary according to the social benefit weight vector corresponding to the building type.

[0013] A building safety fire protection control system, comprising: The environmental sensing module is used to collect real-time fire parameters and personnel distribution parameters within the building; A game modeling engine for building dynamic three-party game models based on building structure information and maintaining game states; Equilibrium strategy solver with built-in iterative strategy optimization algorithm, used to iteratively solve stable equilibrium of three-party strategies in a virtual game environment; The joint strategy output module is used to generate a joint action plan for the three parties based on the solved stable equilibrium. The collaborative control execution module is connected to the control interfaces of the smoke exhaust fan, fireproof roller shutter, dynamic evacuation sign and fire broadcast, and is used to execute collaborative actions according to the facility control instructions in the joint action plan. The feedback monitoring module is used to monitor the deviation between actual personnel behavior and personnel strategies in the joint action plan, and to trigger the equilibrium strategy solver to recalculate when the deviation exceeds the limit.

[0014] Preferably, the environmental sensing module includes one or more of the following: a multimodal smoke sensor, a temperature sensor, a toxic gas sensor, a flame radiation sensor, a thermal imaging personnel counting unit, and a wireless signal personnel positioning probe.

[0015] Preferably, the dynamic evacuation sign is an LED dot matrix sign with programmable display direction, and its display content is controlled by real-time instructions from the collaborative control execution module.

[0016] Preferably, the system further includes a Pareto evaluation module, used to calculate the Pareto frontier of different joint action schemes in the joint utility space, and select the actual implementation scheme according to a preset social benefit weight vector.

[0017] Preferably, the stable equilibrium strategy is Nash equilibrium, correlated equilibrium, or Pareto optimal equilibrium.

[0018] The technical effects and advantages of this invention are: it realizes the collaborative game control of fire, personnel and facilities, incorporates the three into a unified dynamic game framework, overcomes the defects of the fragmented processing of the prior art, and enables the system to formulate control strategies from the perspective of global equilibrium. By solving the non-cooperative game equilibrium of the second player, the natural game result of group dispersal is consistent with the global optimum, avoiding the congestion collapse caused by a single "optimal path" in traditional methods. Smoke extraction, roller shutters, evacuation signs, and other facilities can be jointly optimized within the same game framework to achieve better synergy and predict the reverse impact of facility actions on fire evolution. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the present invention. Detailed Implementation

[0020] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be noted that these descriptions are for the purpose of aiding understanding the present invention, but do not constitute a limitation thereof. Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0021] A building safety fire control method includes the following steps: Step S1: Construct a three-party dynamic game model In a building fire scenario, the fire evolution, the group of people to be evacuated, and the group of fire protection facilities are modeled as the first player, the second player, and the third player, respectively, and a dynamic game model of the three players is constructed.

[0022] in: The strategy space of the first player is generated by a fire dynamics proxy model, which learns from historical or simulation data of fire evolution and can quickly predict changes in the fire spread trend under different facility control actions. Its utility objective is to maximize the cumulative time of personnel exposure to toxic gases and high-temperature environments and the coverage of dangerous areas. The strategy space of the second player includes the probability of choosing each evacuation exit and the decision of turning around the path; its utility objective is to maximize the probability of successful escape and minimize the risk of exposure and the waiting time due to congestion. The strategy space of the third party includes the start / stop and speed of the smoke exhaust fan, the lifting and lowering position of the fireproof roller shutter, the direction of the dynamic evacuation signs, and the zoning and priority of the fire broadcast; its utility objective is to maximize safety benefits (reduce casualties and property losses) and minimize energy costs.

[0023] Step S2: Real-time collection of game state data Fire parameters, including but not limited to smoke concentration, temperature, toxic gas concentration, and flame radiation intensity, are collected in real time by environmental sensing modules deployed in various areas of the building; personnel density monitoring units obtain real-time personnel distribution density, movement speed, and exit personnel counts in each area; and facility status feedback modules obtain the current working status of each fire-fighting equipment. The above data constitute the current state of the game model.

[0024] Specifically, multimodal detectors, including smoke concentration sensors, temperature sensors, CO concentration sensors, and flame ultraviolet / infrared sensors, are installed on each floor according to fire compartments; thermal imaging binocular people counting cameras are installed at stairwell entrances and above main passageways; and Wi-Fi probes are deployed in public areas to assist in locating personnel density.

[0025] Step S3: Solve for the stable equilibrium strategy By constructing a virtual game environment that includes the strategic interactions of the first, second, and third players, and using an iterative strategy optimization method to solve for the stable equilibrium strategy of the game model, a joint action plan for the three parties is obtained.

[0026] Specifically, the model is trained using the following method, and the floor plans, evacuation routes, stairwell locations, fire compartment boundaries, and ventilation opening distribution are exported from the building's BIM model.

[0027] More than 300 computational fluid dynamics simulations of fires were conducted, covering different ignition points, different fire source powers, and different smoke extraction strategies. The simulation results (time series of temperature fields, smoke fields, and CO concentration fields) were used as training data to train a fire dynamics surrogate model. This model is a rapid prediction model that takes facility control actions as input and outputs the fire spread trend.

[0028] The iterative strategy optimization methods include, but are not limited to: strategy iteration, value iteration, virtual game, multi-agent policy gradient method, or game strategy learning method based on experience replay. The stable equilibrium strategy includes Nash equilibrium, relevant equilibrium, or Pareto optimal equilibrium.

[0029] The specific iterative strategy optimization method used is the multi-agent policy gradient method, which is pre-trained in a three-party virtual game environment. The environment state consists of building topology, fire agent model, and personnel behavior simulator (based on social force model and non-cooperative game exit selection model). The training continues until the policy converges, and an initial stable equilibrium policy library is obtained.

[0030] Step S4: Pareto Frontier Strategy Selection Within a combined utility space encompassing casualties, property damage, average evacuation time, and facility energy costs, Pareto boundaries for different joint action plans are calculated. Based on a pre-defined social benefit weight vector according to building type, the joint action plan that best matches the direction of this weight vector is selected from the Pareto boundaries as the actual implementation strategy.

[0031] Taking a commercial complex building as an example, the social benefit weight vector is set as follows: casualty loss weight 0.6, property loss weight 0.2, evacuation time weight 0.15, and energy consumption weight 0.05. The system selects the strategy point on the Pareto boundary that is closest to the direction of the weight vector, and fine-tunes the strategy of the third game player, slightly increasing the smoke exhaust power to prioritize personnel safety.

[0032] Step S5: Generation and execution of collaborative control instructions Based on the strategy of the third party in the joint action plan, commands for starting and stopping smoke exhaust fans and setting speeds, commands for raising and lowering fireproof roller shutters, commands for switching the direction of dynamic evacuation signs, and commands for broadcasting fire safety announcements in designated zones are generated. Based on the strategy of the second party, by adjusting the guidance direction of evacuation signs and the content of broadcast guidance, personnel behavior is encouraged to converge towards an equilibrium distribution. The above commands are executed in a coordinated manner.

[0033] Step S6: Feedback Monitoring and Replanning The actual distribution ratio of people or the movement trajectory of people at each exit is monitored in real time and compared with the equilibrium strategy of the second game player predicted in step S3. When the deviation exceeds the preset threshold, the current strategy is determined to be invalid, triggering the replanning in step S3 to solve the stable equilibrium strategy again.

[0034] like Figure 1 As shown, the overall architecture of this invention adopts a layered structure, consisting of an execution control layer, a platform decision layer, a data transmission layer, and an environment perception layer from top to bottom, with a feedback monitoring module set on the side as a cross-layer closed loop.

[0035] I. Environmental Perception Layer The environmental perception layer consists of various types of on-site data acquisition devices used to acquire real-time fire parameters and personnel distribution parameters within the building. Specifically, it includes: Multimodal smoke sensor: Detects smoke concentration Temperature sensor: detects ambient temperature and rate of temperature change. Toxic gas sensor: detects the concentration of toxic gases such as CO and CO2. Flame radiation sensor: detects ultraviolet / infrared flame radiation Thermal imaging people counting camera: captures thermal images of people and counts the number of people in an area. Wi-Fi / BLE positioning probes: assisting in locating personnel density by detecting wireless signals. The aforementioned devices are connected to the edge computing gateway of the data transmission layer via wired or wireless means.

[0036] II. Data Transmission Layer The data transmission layer is responsible for uploading the raw data collected by the environmental perception layer to the platform decision layer, and for issuing control commands from the platform decision layer to the execution control layer. This layer includes: Edge computing gateway: Deployed on each floor or in fire compartments, responsible for local data preprocessing, protocol conversion, and temporary storage. Redundant communication network: Employing a hybrid network of Ethernet, LoRaWAN, or 5G to ensure reliable and real-time data transmission. Data flow direction: Environment perception layer → Edge computing gateway → Platform decision layer; Instruction flow direction is the opposite.

[0037] III. Platform Decision-Making Level The platform decision layer is the core computing and decision-making unit of this system, typically deployed as a central server in the fire control room. Its internal modules include: Game Theory Modeling Engine: Constructs a three-party dynamic game model based on building BIM information and maintains the state transition function. Equilibrium Strategy Solver: Includes built-in iterative policy optimization algorithms (such as policy iteration, virtual game, etc.) for solving stable equilibrium strategies. Pareto Assessment Module: Calculates the Pareto frontier in the joint utility space and selects the optimal joint action plan based on the social benefit weight vector of building type. Joint Policy Output Module: Converts the solved equilibrium policy into specific joint action plans for use by the execution control layer. The data interaction between the modules is as follows: the game modeling engine inputs the current state into the equilibrium policy solver, the solution is sent to the Pareto evaluation module for optimization selection, and finally the joint policy output module generates executable instructions.

[0038] IV. Execution Control Layer The execution control layer receives control commands from the platform decision layer and drives field devices to perform coordinated actions. This layer includes: Smoke exhaust fan: Adjustable speed start / stop, used for smoke exhaust and altering airflow organization. Fire-resistant roller shutters: equipped with half-lowering and full-lowering controls, used to separate fire compartments. Dynamic evacuation signs: LED dot matrix programmable directional signs that can switch between arrow directions, crosses, or text prompts in real time. Fire broadcast system: Can broadcast customized voice guidance information by zone. Emergency lighting: activated in conjunction with other systems to assist evacuation; The aforementioned devices are driven by a unified collaborative control execution module, which is located between the platform's decision-making layer and execution control layer, and is represented by a connecting line in the diagram.

[0039] V. Feedback Monitoring Module The feedback monitoring module operates independently of the hierarchical structure, interacting with multiple levels to form a closed-loop control circuit. Input: Data such as actual exit pedestrian counts and pedestrian movement trajectories are obtained from the environmental perception layer. Processing: Compare the personnel balancing strategy predicted by the platform's decision-making level and calculate the deviation. Output: When the deviation exceeds the preset threshold, a replanning signal is triggered to the equilibrium strategy solver.

[0040] This invention also provides a building fire protection collaborative control system based on a three-party game model, comprising: Environmental perception module: Composed of a multimodal sensor array, a thermal imaging personnel counting unit, and a wireless signal personnel positioning probe, it is used to collect fire parameters and personnel distribution parameters in the building in real time.

[0041] Game modeling engine: It has a built-in building information parsing module and fire dynamics proxy model, which are used to build a dynamic game model of three parties based on building structural information and maintain the game state.

[0042] Equilibrium Strategy Solver: Built-in iterative strategy optimization algorithm for iteratively solving stable equilibrium of three-party strategies in a virtual game environment.

[0043] Joint Strategy Output Module: Used to generate a joint action plan for the three parties based on the solved stable equilibrium.

[0044] Collaborative control execution module: Connects to the control interfaces of smoke exhaust fan, fireproof roller shutter, dynamic evacuation sign and fire broadcast respectively, and is used to execute collaborative actions according to the facility control instructions in the joint action plan.

[0045] Feedback monitoring module: Used to monitor the deviation between actual personnel behavior and personnel strategies in the joint action plan, and to trigger the equilibrium strategy solver to recalculate when the deviation exceeds the limit.

[0046] Pareto Assessment Module: Used to calculate the Pareto frontier of different joint action schemes in the joint utility space, and select the actual implementation scheme based on the preset social benefit weight vector.

[0047] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A building safety fire protection control method, characterized in that, Includes the following steps: The fire evolution, the group of people to be evacuated, and the group of fire-fighting facilities are modeled as the first player, the second player, and the third player, respectively, to construct a three-way non-zero-sum dynamic game model. Real-time data collection of fire parameters and personnel distribution parameters within the building serves as the current state of the game theory model. By constructing a multi-party dynamic game environment that includes a fire evolution model, a personnel evacuation behavior model, and a fire-fighting facility action model, and using an iterative optimization algorithm to solve the stable equilibrium strategy of the game model, a joint action plan for the three parties is obtained. Firefighting facility control instructions are generated based on the strategy of the third party in the joint action plan, and evacuation guidance instructions are generated based on the strategy of the second party, and then executed in a coordinated manner. The deviation between actual personnel behavior and the joint action plan is monitored. If the deviation exceeds a preset threshold, the stable equilibrium strategy is re-solved.

2. The building safety fire protection control method according to claim 1, characterized in that: The utility objective of the first player is defined as maximizing the cumulative time of personnel exposure to toxic gases and high-temperature environments and the coverage of dangerous areas; the strategy space of the first player is generated by a fire dynamics proxy model, which is obtained by learning from historical fire evolution data or simulation data.

3. The building safety fire protection control method according to claim 1, characterized in that: The strategy solution for the second player adopts a layered game approach: the building is divided into multiple game zones, and the equilibrium distribution ratio of the non-cooperative game is solved for the exit selection problem of a finite number of evacuation exits in each zone. For the selection of local paths downstream of the outlet, the continuous medium approximation method of population flow is used.

4. The building safety fire protection control method according to claim 1, characterized in that: The iterative strategy optimization methods include, but are not limited to: strategy iteration, value iteration, virtual game, multi-agent policy gradient method, and game strategy learning method based on experience replay.

5. The building safety fire protection control method according to claim 1, characterized in that: After solving for the stable equilibrium strategy of the game model, the method further includes: calculating the Pareto boundary in the joint utility space that includes casualty losses, property losses, evacuation time and energy costs, and selecting a joint action plan as the actual implementation strategy from the Pareto boundary according to the social benefit weight vector corresponding to the building type.

6. A building safety fire protection control system, used to implement the method according to any one of claims 1 to 5, characterized in that, include: The environmental sensing module is used to collect real-time fire parameters and personnel distribution parameters within the building; A game modeling engine for building dynamic three-party game models based on building structure information and maintaining game states; Equilibrium strategy solver with built-in iterative strategy optimization algorithm, used to iteratively solve stable equilibrium of three-party strategies in a virtual game environment; The joint strategy output module is used to generate a joint action plan for the three parties based on the solved stable equilibrium. The collaborative control execution module is connected to the control interfaces of the smoke exhaust fan, fireproof roller shutter, dynamic evacuation sign and fire broadcast, and is used to execute collaborative actions according to the facility control instructions in the joint action plan. The feedback monitoring module is used to monitor the deviation between actual personnel behavior and personnel strategies in the joint action plan, and to trigger the equilibrium strategy solver to recalculate when the deviation exceeds the limit.

7. A building safety fire protection control method according to claim 6, characterized in that: The environmental perception module includes one or more of the following: a multimodal smoke sensor, a temperature sensor, a toxic gas sensor, a flame radiation sensor, a thermal imaging personnel counting unit, and a wireless signal personnel positioning probe.

8. A building safety fire protection control method according to claim 6, characterized in that: The dynamic evacuation sign is an LED dot matrix sign with programmable display direction, and its display content is controlled by real-time instructions from the collaborative control execution module.

9. A building safety fire protection control method according to claim 6, characterized in that: The system also includes a Pareto evaluation module, which is used to calculate the Pareto frontier of different joint action schemes in the joint utility space and select the actual implementation scheme according to the preset social benefit weight vector.

10. A building safety fire protection control method according to claim 6, characterized in that: The stable equilibrium strategy is Nash equilibrium, correlated equilibrium, or Pareto optimal equilibrium.

Images