Building fire multi-failure recovery intelligent management and control method and system based on physical hypergraph

Abstract

The present application relates to a physical hypergraph-based building fire multi-fault recovery intelligent management and control method and system, wherein the method constructs a fire safety function hypergraph based on multi-source static configuration data, realizes initialization of the fire safety function hypergraph based on real-time physical state data, extracts all scene subgraphs with a fire scene node as a root node from the initialized fire safety function hypergraph, takes a device node in each scene subgraph as a starting point, propagates the device state of the device node along the topological order of the directed hyperedge inside the scene subgraph step by step until the root node, obtains the scene protection capability of each scene subgraph, screens out weak scene subgraphs and corresponding fault node sets and fault directed hyperedge sets, dynamically deduces the generated recovery scheme, and screens out the optimal recovery scheme for intelligent recovery of the weak scene subgraph. Thus, the present application improves the fault identification accuracy and intelligent recovery management and control level of the building fire system.

Description

Technical Field

[0001] This invention relates to the field of building fire protection technology, and in particular to a method and system for intelligent management and control of multi-fault recovery in building fire protection based on physical hypergraphs. Background Technology

[0002] Traditional fire protection maintenance typically revolves around equipment ledgers and fixed schedules, with equipment being inspected item by item according to monthly, quarterly, or annual plans. Many systems convert a detector going offline, a backup pump malfunctioning, an alarm control panel failing, or insufficient fire hydrant supply into similar fault records or work orders, which are then manually categorized as urgent, important, or general. However, building fire protection systems consist of multiple subsystems with complex interconnected control logic. This method is prone to situations where critical faults are buried under ordinary work orders, or partial repairs fail to restore the entire interconnected chain.

[0003] In recent years, information technologies such as digital twins, knowledge graphs, and artificial intelligence have been widely used in the fire protection field. However, there are still the following shortcomings in fault tracing and function recovery strategies under multi-fault scenarios: (I) Currently, most building fire protection systems rely on traditional knowledge graphs and ordinary topology networks for modeling, primarily employing either single static topology modeling or pure data-driven modeling. Single static topology modeling uses simple point-to-point connections to solidify the physical connections between storage devices, failing to reflect the actual operating conditions of equipment on-site. Pure data-driven modeling, on the other hand, relies entirely on real-time data iteration, resulting in chaotic topology association logic. Furthermore, neither of these modeling methods can represent the complex operational mechanisms between multiple subsystems within a building fire protection system.

[0004] (II) Existing technologies mostly employ a multi-index macro-unified evaluation system for fault analysis, obtaining the overall system evaluation result through global unified weighted calculation. This model is prone to masking local functional defects and hidden safety hazards, and cannot truly reflect the actual level of fire safety protection. Especially under conditions of multiple concurrent faults, traditional macro-evaluation mechanisms struggle to accurately locate the core fault sources that lead to the degradation of the overall system's protection function. This results in problems such as a wide scope of fault investigation, a large amount of ineffective investigation, long location time, and low accuracy, severely restricting the efficiency of fire fault handling and the precision of hazard mitigation.

[0005] Therefore, there is an urgent need for a technical solution that can accurately model building fire protection systems, truly reflect the level of fire safety protection, and accurately trace the source of multiple faults. Summary of the Invention

[0006] The technical problem to be solved by the present invention is: the present invention provides a method and system for intelligent management and control of multi-fault recovery in building fire protection based on physical hypergraph, so as to realize accurate modeling of building fire protection system, truly reflect the level of fire safety protection, realize accurate source tracing of multiple faults, and improve the fault identification accuracy, hazard management integrity and intelligent recovery and control level of building fire protection system.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a building fire protection multi-fault recovery intelligent management and control method based on a physical hypergraph, comprising: The system acquires multi-source static configuration data and real-time physical status data of the building fire protection system. Based on the multi-source static configuration data, it constructs a fire safety functional hypergraph consisting of N types of nodes and M types of directed hyperedges. Based on the real-time physical status data, it initializes the physical parameters of each directed hyperedge in the fire safety functional hypergraph and assigns equipment status values ​​to each device node to initialize the fire safety functional hypergraph and obtain the initialized fire safety functional hypergraph. The N types of nodes include fire scene nodes and device nodes. Extract all scene subgraphs with fire scene nodes as root nodes from the initialized fire safety function hypergraph. For each scene subgraph, starting from the device nodes within the scene subgraph, propagate the device status of the device nodes along the topological order of the directed hyperedges inside the scene subgraph until the root node, so as to output the scene protection capability of each scene subgraph. Based on the scene protection capability, weak scene subgraphs are selected, and the set of fault nodes and the set of fault directed hyperedges are identified from the weak scene subgraphs. Based on the set of faulty nodes and the set of faulty directed superedges, L sets of recovery schemes are generated. Each set of recovery schemes is dynamically simulated using digital twin technology, and the dynamic recovery simulation results of each set of recovery schemes are generated. The dynamic recovery simulation results of the L sets of recovery schemes are evaluated using a multi-objective evaluation function to select the optimal recovery scheme for intelligent recovery of the weak scene subgraph.

[0008] The beneficial effects of this invention are as follows: A fire safety functional hypergraph is constructed based on multi-source static configuration data, and initialized based on real-time physical state data. This employs a dual-layer modeling approach combining static data modeling and dynamic data-driven methods, which differs from traditional single static modeling and pure data-driven modes. This approach ensures the stability and standardization of the directed hyperedges and nodes of the constructed fire safety functional hypergraph, while also adapting to dynamically changing equipment states. Initializing physical parameters using real-time physical state data enables the representation of the actual operating mechanism. Furthermore, by splitting the scene subgraphs corresponding to fire scene nodes and creating independent positive... The system propagates the protection capabilities of the scene to identify vulnerable scene subgraphs, achieving decoupled assessment of multiple fire scenarios in building fire protection. This avoids the problems of vulnerable scenes being masked and their location being ambiguous due to globally unified calculations, enabling refined calculation of scene protection capabilities and accurate identification of vulnerable scene subgraphs. It identifies sets of fault nodes and directed hyperedges from the vulnerable scene subgraphs, achieving fault location while reducing the fault location range and improving location efficiency. Combining digital twin technology and multi-objective evaluation functions, it selects the optimal recovery scheme, ensuring the feasibility and scientific validity of the selected optimal recovery scheme and improving the effectiveness of intelligent recovery.

[0009] Optionally, the multi-source static configuration data includes building space topology data, pipeline medium connectivity data, fire equipment deployment data, control logic data, and scene partitioning data. The construction of a fire safety functional hypergraph based on the multi-source static configuration data, consisting of N types of nodes and M types of directed hyperedges, and the initialization of the physical parameters of each directed hyperedge in the fire safety functional hypergraph based on the real-time physical state data, while simultaneously assigning equipment state values ​​to each device node, to achieve the initialization of the fire safety functional hypergraph includes: Fire protection space nodes are generated based on building space topology data, spatial connectivity and spatial isolation relationships between fire protection space nodes are generated based on scene partition data, and fire scene nodes are generated based on scene partition data to establish spatial ownership relationships between fire scene nodes and fire protection space nodes. Based on pipeline medium connectivity data, energy medium nodes and energy flow relationships between energy medium nodes are generated; based on fire equipment deployment data, equipment nodes and spatial deployment relationships between equipment nodes are generated; based on control logic data, control nodes and control association relationships between control nodes and equipment nodes are generated. Based on the building space topology data, the pipeline medium connectivity data, the control logic data, and the spatial affiliation relationship between the fire protection space nodes and the fire scene nodes, fire protection function nodes are generated, as well as the energy supply relationship between the fire protection function nodes and the corresponding equipment nodes and the fire protection service relationship between the corresponding fire protection space nodes. Directed superedges are generated based on the fire protection service relationship, the energy supply relationship, the spatial connectivity relationship, the spatial isolation relationship, the spatial affiliation relationship, the energy circulation relationship, the spatial layout relationship, and the control association relationship. A corresponding logical operator is bound to each directed superedge according to its relationship type. The relationship types of the directed superedges include series superedges, parallel superedges, capacity superedges, time-series superedges, and common-cause directed superedges. The logical operators include series minimum operator, parallel replacement operator, capacity supply-demand ratio operator, time-series gating operator, and common-cause fault gating operator. The corresponding fire protection space nodes, fire scene nodes, energy medium nodes, equipment nodes, control nodes, and fire protection function nodes are connected according to the directed hyperedge to generate a fire safety function hypergraph. Based on the real-time physical state data, the device state value is assigned to each device node in the fire safety function hypergraph, and the physical parameters of each directed hyperedge are initialized to achieve the initialization of the fire safety function hypergraph.

[0010] As described above, fire protection space nodes, fire scene nodes, energy medium nodes, equipment nodes, control nodes, and fire protection function nodes are generated based on multi-source static configuration data. This fully replicates the physical architecture, media links, control logic, and spatial attributes of the fire safety system, achieving a digital replication of the fire safety system. Directed hyperedges are generated based on the fire protection service relationships, energy supply relationships, spatial connectivity relationships, spatial isolation relationships, spatial ownership relationships, energy circulation relationships, spatial layout relationships, and control association relationships between the nodes. Corresponding logical operators are bound to the relationship type of each directed hyperedge, transforming the static directed hyperedges with topological connections into computable dynamic directed hyperedges, which can dynamically reflect the complex working mechanism of collaborative linkage between multiple nodes.

[0011] Optionally, the initialization of the physical parameters of each directed hyperedge in the fire safety function hypergraph based on the real-time physical state data, and the assignment of device status values ​​to each device node, to achieve the initialization of the fire safety function hypergraph, includes: Obtain the relationship type of each directed hyperedge in the fire safety function hypergraph, and initialize the physical parameters of the corresponding directed hyperedge based on the relationship type of the directed hyperedge and the preset equipment design table; When the relationship type of the directed superedge is a series superedge or a parallel superedge, the physical parameters of the directed superedge are assigned null values. When the relationship type of the directed superedge is a capacity superedge, the design fire water consumption, design smoke exhaust volume, and design power supply capacity are obtained from the equipment design table, and the design fire water consumption, design smoke exhaust volume, and design power supply capacity are used as demand capacity to initialize the physical parameters of the corresponding capacity superedge. When the relationship type of the directed superedge is a timing superedge, the design permitted action delay interval is obtained from the equipment design table, and the physical parameters of the corresponding timing superedge are initialized according to the design permitted action delay interval. When the relationship type of the directed superedge is a common cause directed superedge, the design common voltage threshold, design common water source threshold, and design common communication threshold are obtained from the equipment design table, and the design common voltage threshold, design common water source threshold, and design common communication threshold are used as common cause triggering conditions to initialize the corresponding common cause directed superedge, thus obtaining all the initialized directed superedges. Based on the directed hyperedge associated with each device node after initialization and combined with the real-time physical state data, the controllability, observability, action completion capability, output capability, action latency, and reliability of the device node are calculated. Based on the controllability, observability, action completion capability, output capability, action latency, and reliability, the corresponding device node is assigned a device state value to achieve the initialization of the fire safety function hypergraph.

[0012] As described above, a differentiated initialization assignment mechanism is implemented for physical parameters based on different relation types of directed hyperedges. This accurately adapts to the fire protection physical logic of different relation types, avoiding distortion in the initialization assignment from affecting the accuracy of subsequent scenario protection capability calculations. First, the physical constants of the directed hyperedges are initialized using a preset equipment design table. Then, the equipment nodes are assigned equipment status values ​​in conjunction with real-time physical status data. This achieves dual calibration of theoretical design and actual status, improving the authenticity and reliability of equipment status assignment. Furthermore, the multi-dimensional equipment status assignment breaks the limitations of a single equipment status assignment, achieving comprehensive coverage of equipment status.

[0013] Optionally, the N types of nodes include fire protection function nodes, and the extraction of all scene subgraphs rooted at fire scene nodes from the initialized fire safety function hypergraph includes: Taking each fire scene node as the root node, a multi-layer reverse backtracking strategy is used to backtrack along the directed super edges associated with the root node, extracting all first directed super edges pointing to the root node and the corresponding leaf node set, and simultaneously extracting all second directed super edges pointing to the leaf node set and the corresponding child node set, until the reverse backtracking reaches the device node. The extracted first directed super edges and the corresponding leaf node set, as well as all second directed super edges and the corresponding child node set, are combined to form the scene subgraph of the root node.

[0014] As described above, taking the fire scene node as the root node, a multi-layer reverse backtracking strategy is adopted. This strategy extracts all first directed superedges and leaf node sets pointing to the root node, as well as all second directed superedges and child node sets pointing to the leaf node sets, until the reverse backtracking reaches the device node. This completely preserves all relevant directed superedges and nodes from the device node to the fire scene node, i.e., the complete propagation link, thus improving the integrity and comprehensiveness of the obtained scene subgraph.

[0015] Optionally, the N types of nodes include fire protection function nodes. For each scene subgraph, starting with the device nodes within the scene subgraph, the device status of the device nodes is propagated sequentially along the topological order of the directed hyperedges within the scene subgraph until the root node, to output the scene protection capability of each scene subgraph. Based on the scene protection capability, weak scene subgraphs are selected, and the set of faulty nodes and the set of faulty directed hyperedges are identified from the weak scene subgraphs, including: For each scene subgraph, starting from the device node within the scene subgraph, the device state of the device node is propagated step by step along the topological order of the directed hyperedges within the scene subgraph. During the propagation process, the capability state value of the current node is calculated and output edge by edge through the logical operators bound to the directed hyperedges until the propagation reaches the root node, thus completing the capability propagation and obtaining the scene subgraph after the capability propagation is completed. Extract the corresponding fire protection function node set from the scene subgraph after the capability propagation is completed, obtain the capability status value of each fire protection function node in the fire protection function node set, take the fire protection function node with the smallest capability status value as the weak link node, and take the capability status value of the weak link node as the scene protection capability of the corresponding scene subgraph after the capability propagation is completed. The scene protection capability of each scene subgraph is compared with the preset protection threshold, and the scene subgraphs with scene protection capabilities lower than the preset protection threshold are identified as weak scene subgraphs. For each weak scenario subgraph, starting from the shortest node inside the weak scenario subgraph, backtracking in the reverse direction along the directed superedges associated with the shortest node is performed to extract all shortest superedges pointing to the shortest node and the corresponding set of short child nodes. The capability status value of each short child node in the set of short child nodes is obtained. The capability status value of each short child node is compared with the fault threshold. Short child nodes with capability status values ​​lower than the fault threshold are identified as fault nodes, and the directed superedges associated with the fault nodes are identified as fault directed superedges, thus obtaining the set of fault nodes and the set of fault directed superedges.

[0016] As described above, starting from the device node, the system propagates topologically, calculating the capability status value of the current node edge by edge using logical operators bound to directed hyperedges, and then propagating to the root node. This aligns with the actual working logic of a building fire protection system, from equipment power supply and functional implementation to scene protection, avoiding distortion caused by general global calculations and improving the accuracy of capability propagation. The fire protection functional node with the smallest capability status value is designated as the weakest node, and its capability status value is used as the scene protection capability, accurately restoring the actual protection level of the scene subgraph. This avoids the safety hazards of local defects being masked and scene protection capabilities being overestimated due to traditional averaging calculations. For weak scene subgraphs, the system traces back in the reverse direction from the weakest node to extract faulty nodes and their corresponding directed hyperedges, achieving accurate tracing of the fault root cause and improving the efficiency and accuracy of fault identification.

[0017] Optionally, the step of designating scene subgraphs with scene protection capabilities below a preset protection threshold as weak scene subgraphs includes: Scene subgraphs with scene protection capabilities below a preset protection threshold are designated as weak scene subgraphs. It is determined whether the number of weak scene subgraphs is greater than 0. If so, the scene protection capability of each scene subgraph is input into the scene risk formula and the scene resilience formula respectively to perform scene risk status assessment and scene resilience assessment, thereby obtaining the scene risk status assessment result and scene resilience assessment result corresponding to each scene subgraph. Based on the scenario risk status assessment results and scenario resilience assessment results corresponding to each vulnerable scenario subgraph, priority ranking is performed using a priority strategy to obtain the priority-ranked vulnerable scenario subgraphs. The priority strategy is to use the scenario risk status assessment result as the first priority, and when the scenario risk status assessment results are the same, the scenario resilience assessment result is used as the second priority.

[0018] As described above, when there are multiple vulnerable scene subgraphs, the corresponding scene risk status assessment result and scene resilience assessment result are calculated based on the scene protection capability of each vulnerable scene subgraph. These results are then prioritized to ensure that the most dangerous vulnerable scene subgraphs are processed first during subsequent intelligent recovery, thereby reducing the overall safety risk of the building fire protection system and improving the scientific nature of multi-fault collaborative management.

[0019] Optionally, the N types of nodes include energy medium nodes, and the generation of L sets of recovery schemes based on the set of faulty nodes and the set of faulty directed superedges includes: Calculate the first impact degree of the decrease in scene protection capability of the corresponding weak scene subgraph for each fault node in the fault node set, and the second impact degree of the increase in scene protection capability of the corresponding weak scene subgraph after each fault node is restored, and use the first impact degree and the second impact degree as the weight of the corresponding fault node. The difference between the current output value and the target output value of each faulty directed superedge in the set of faulty directed superedges is used as the weight of the faulty directed superedge. Using the fire scene nodes of the weak scene subgraph as the sink and the equipment nodes and energy medium nodes in the fault node set as the source set, a minimum fault cut set solution model is constructed based on the weights of all fault nodes and the weights of all fault directed super edges. The minimum fault cut set solution model is iteratively solved based on the maximum flow minimum cut theorem to obtain the minimum fault cut set, which is the combination of the fault node with the minimum total weight and the fault directed super edge. Using the scenario protection capability of the weak scenario subgraph recovering to a preset safety threshold as a constraint, all intervention schemes that can repair the minimum fault cut set are enumerated to generate L groups of recovery schemes.

[0020] As described above, by using the fire scene nodes of the weak scene subgraph as the sink and the equipment nodes and energy medium nodes in the fault node set as the source set, the complex multi-fault coupling problem is transformed into a network flow problem. The difference between the current output value and the target output value of the fault directed hyperedge is used as the weight of the fault directed hyperedge. The first influence degree of the fault node on the decrease of scene protection capability and the second influence degree of the recovery on the improvement of scene protection capability are used as the weight of the fault node to construct a minimum fault cut set solution model adapted to building fire protection scenarios. Iterative solution based on the maximum flow minimum cut theorem can quickly locate the combination of fault node and fault directed hyperedge with the minimum total weight, improving the efficiency of iterative solution. The constraint condition of the scene protection capability of the weak scene subgraph recovering to the preset safety threshold is used to ensure that the actual generated recovery scheme is feasible and can solve practical problems.

[0021] Optionally, the step of dynamically reconstructing each set of recovery schemes using digital twin technology to generate dynamic recovery simulation results for each set of recovery schemes includes: Construct L physically isolated digital twin sandbox simulation branches, with each digital twin sandbox simulation branch corresponding to a set of recovery solutions; For each recovery scheme, dynamic recovery simulation is performed in an independent digital twin sandbox simulation branch based on the initialized fire safety functional hypergraph. Starting from the node being intervened, incremental propagation is performed along the affected directed hyperedges. The logical operators bound to each directed hyperedge are executed sequentially to obtain the capability state value after intervention. During the dynamic recovery simulation, the capability state value of the corresponding node is triggered sequentially according to the expected start time and expected completion time of the intervention action in each recovery scheme. After each capability state value change, incremental propagation is re-executed to update the scene protection capability. The scene protection capability after each update is recorded in chronological order to generate a recovery trajectory corresponding to the recovery scheme. The recovery trajectory is used as the result of the dynamic recovery simulation.

[0022] As described above, each digital twin sandbox simulation branch corresponds to a set of recovery schemes. This ensures that each set of recovery schemes is not disturbed during dynamic recovery simulations based on the initialized fire safety function hypergraph, guaranteeing the independence and stability of the dynamic recovery simulations. During the dynamic recovery simulation, the capability status values ​​of the corresponding nodes are triggered sequentially according to the expected start and completion times of the intervention actions in each set of recovery schemes. Incremental propagation is re-executed after each capability status value change, fully recording the gradual recovery trajectory of the scene protection capabilities and improving the continuity and completeness of the dynamic recovery simulation results.

[0023] Optionally, the step of evaluating the dynamic recovery simulation results of the L groups of recovery schemes through a multi-objective evaluation function to select the optimal recovery scheme for intelligent recovery of the vulnerable scene subgraph includes: The area of ​​loss of life safety capabilities and the recovery time of critical functions are calculated based on the dynamic recovery simulation results of each recovery plan. At the same time, the marginal resilience gain of a single intervention action and the synergistic recovery gain of a combination of multiple intervention actions in each recovery plan are calculated. The life safety capability loss area, critical function recovery time, marginal resilience gain, and collaborative recovery gain corresponding to each recovery scheme are input into a multi-objective evaluation function for evaluation. The evaluation results corresponding to each recovery scheme are obtained, and the recovery scheme corresponding to the smallest evaluation result is selected as the optimal recovery scheme to intelligently restore the weak scene subgraph.

[0024] As described above, four indicators are introduced: the area of ​​loss of life safety capability, the recovery time of key functions, the marginal resilience gain of a single intervention action, and the synergistic recovery gain of a combination of multiple intervention actions. These indicators are then input into a multi-objective evaluation function for assessment, which avoids the one-sidedness of a single indicator and improves the accuracy of the evaluation results.

[0025] Secondly, the present invention provides a building fire protection multi-fault recovery intelligent management and control system based on physical hypergraph, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the building fire protection multi-fault recovery intelligent management and control method based on physical hypergraph described in the first aspect.

[0026] The technical effects of the intelligent management and control system for multi-fault recovery of building fire protection based on physical hypergraph provided in the second aspect are the same as those of the intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph provided in the first aspect. Attached Figure Description

[0027] Figure 1 This is a flowchart of the intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph provided in this embodiment; Figure 2 This is a schematic diagram of the intelligent management and control system for multi-fault recovery of building fire protection based on physical hypergraph provided in this embodiment.

[0028] [Explanation of Labels in the Attached Image] 1. Intelligent management and control system for multi-fault recovery of building fire protection based on physical hypermap; 2. Processor; 3. Memory. Detailed Implementation

[0029] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0030] Example 1 Please refer to Figure 1 This invention provides a smart management and control method for multi-fault recovery in building fire protection based on a physical hypergraph, comprising the following steps: S1. Obtain multi-source static configuration data and real-time physical status data of the building fire protection system. Based on the multi-source static configuration data, construct a fire safety functional hypergraph composed of N types of nodes and M types of directed hyperedges. Based on the real-time physical status data, initialize the physical parameters of each directed hyperedge in the fire safety functional hypergraph and assign equipment status values ​​to each device node to initialize the fire safety functional hypergraph and obtain the initialized fire safety functional hypergraph. The N types of nodes include fire scene nodes and device nodes. In this embodiment, as Figure 1 As shown, multi-source static configuration data and real-time physical status data of the building fire protection system are obtained. The multi-source static configuration data includes building space topology data, pipeline media connectivity data, fire equipment layout data, control logic data, and scene zoning data. Building space topology data includes the floor plan layout of each floor, fire compartment boundaries, and evacuation corridor locations. Pipeline media connectivity data includes the direction, diameter, and valve locations of the fire water supply network, as well as the path of the smoke exhaust duct. Fire equipment layout data includes the installation location and numbering of equipment such as fire pumps, smoke exhaust fans, fire dampers, alarm control panels, detectors, and sprinklers. Control logic data includes the programming logic of the linkage controller, the control trigger conditions and action sequence between various devices. Scene zoning data includes fire scenarios divided according to different fire locations, different personnel distributions, and different building usage states. Real-time physical status data includes real-time acquisition from various sensors deployed within the building, including at least: pressure and flow sensors at the fire pump outlet, air volume or air pressure sensors at the smoke exhaust fan outlet, level sensors in fire water tanks and elevated water tanks, current or voltage acquisition units in fire distribution cabinets, valve position feedback devices for various valves, operating status and fault signals of alarm control panels, and linkage outputs and feedback signals from controllers. The acquired multi-source static configuration data and real-time physical status data undergo preprocessing such as protocol parsing, timestamp alignment, and data cleaning via an edge gateway.

[0031] A fire safety functional hypergraph consisting of N types of nodes and M types of directed hyperedges is constructed based on multi-source static configuration data. That is, firstly, the skeleton of the fire safety functional hypergraph is built using multi-source static configuration data, and then the physical parameters of each directed hyperedge in the fire safety functional hypergraph are initialized and assigned values ​​using real-time physical state data. At the same time, the device status is assigned to each device node. In other words, the fire safety functional hypergraph is transformed into a computable, propagable, and inferable hypergraph model using real-time physical state data.

[0032] At this point, the multi-source static configuration data mentioned in step S2 includes building space topology data, pipeline medium connectivity data, fire equipment deployment data, control logic data, and scene partitioning data. The construction of a fire safety functional hypergraph based on the multi-source static configuration data, consisting of N types of nodes and M types of directed hyperedges, and the initialization of the physical parameters of each directed hyperedge in the fire safety functional hypergraph based on the real-time physical state data, along with the assignment of equipment status values ​​to each device node, to achieve the initialization of the fire safety functional hypergraph includes: S21. Based on building space topology data, generate fire protection space nodes; based on scene partition data, generate spatial connectivity and spatial isolation relationships between fire protection space nodes; and based on scene partition data, generate fire scene nodes and establish spatial ownership relationships between fire scene nodes and fire protection space nodes. S22. Generate energy medium nodes and energy flow relationships between energy medium nodes based on pipeline medium connectivity data; generate equipment nodes and spatial layout relationships between equipment nodes based on fire equipment layout data; generate control nodes and control association relationships between control nodes and equipment nodes based on control logic data. S23. Based on the building space topology data, the pipeline medium connectivity data, the control logic data, and the spatial affiliation relationship between the fire protection space node and the fire scene node, generate fire protection function nodes, as well as the power supply relationship between the fire protection function nodes and the corresponding equipment nodes and the fire protection service relationship between the corresponding fire protection space nodes. S24. Generate directed superedges based on the fire protection service relationship, the energy supply relationship, the spatial connectivity relationship, the spatial isolation relationship, the spatial affiliation relationship, the energy flow relationship, the spatial layout relationship, and the control association relationship. Bind a corresponding logical operator to each directed superedge according to the relationship type of each directed superedge. The relationship types of the directed superedges include series superedges, parallel superedges, capacity superedges, time-series superedges, and common-cause directed superedges. The logical operators include series minimum operator, parallel replacement operator, capacity supply-demand ratio operator, time-series gating operator, and common-cause fault gating operator. S25. Connect the corresponding fire protection space nodes, fire scene nodes, energy medium nodes, equipment nodes, control nodes and fire protection function nodes according to the directed hyperedge to generate a fire safety function hypergraph. S26. Based on the real-time physical state data, assign equipment state values ​​to each device node in the fire safety function hypergraph, and initialize the physical parameters of each directed hyperedge to achieve the initialization of the fire safety function hypergraph.

[0033] In this embodiment, fire-resistant space nodes are generated based on building space topology data. Each fire compartment corresponds to one fire-resistant space node, which includes at least the spatial coordinates, area, volume, designed capacity, and fire compartment number of that fire compartment. Spatial connectivity and isolation relationships between fire-resistant space nodes are generated based on scene partitioning data. Fire scene nodes are also generated based on scene partitioning data, and these fire scenes include at least the fire location, personnel distribution, and building usage status. Spatial attribution relationships are established between fire scene nodes and fire-resistant space nodes, indicating which fire-resistant space nodes a fire scene node covers. Energy medium nodes and energy flow relationships between them are generated based on pipeline medium connectivity data. Energy medium nodes include at least: fire water tank nodes, water supply network nodes, smoke exhaust duct nodes, and fire power supply nodes. Energy flow relationships represent the flow direction, flow distribution, and pressure transmission relationships of energy media in pipelines or lines. Based on fire equipment deployment data, equipment nodes and their spatial layout relationships are generated. Equipment nodes include at least fire pump nodes, smoke exhaust fan nodes, fire damper nodes, alarm control panel nodes, distribution cabinet nodes, detector nodes, and sprinkler head nodes. Based on control logic data, control nodes and their control relationships with equipment nodes are generated. Control nodes include at least alarm control panel linkage output point nodes, PLC controller output module nodes, and manual control panel control button nodes. Control relationships represent the logical paths by which control nodes send start, stop, open, and close control commands to equipment nodes. Based on building space topology data, pipeline medium connectivity data, control logic data, and the spatial affiliation relationships between fire protection space nodes and fire scene nodes, fire protection function nodes are generated. Fire protection function nodes include at least fire detection function nodes, alarm confirmation function nodes, fire water supply function nodes, smoke control and exhaust function nodes, fire separation function nodes, and evacuation guidance function nodes. The power supply relationships between fire protection function nodes and their corresponding equipment nodes represent the physical supply paths by which equipment nodes provide capacity and power to fire protection function nodes. The fire protection service relationship between fire protection functional nodes and their corresponding fire protection space nodes represents the spatial coverage relationship in which fire protection functional nodes provide protection services to fire protection space nodes.

[0034] Directed hyperedges are generated based on fire protection service relationships, energy supply relationships, spatial connectivity relationships, spatial isolation relationships, spatial affiliation relationships, energy flow relationships, spatial layout relationships, and control association relationships. Each directed hyperedge is then bound to a corresponding logical operator based on its relationship type. The relationship types for directed hyperedges include series hyperedges, parallel hyperedges, capacity hyperedges, time-series hyperedges, and common-cause directed hyperedges. Logical operators include the series minimum operator, the parallel replacement operator, the capacity supply-demand ratio operator, the time-series gating operator, and the common-cause fault gating operator. Specifically, a series hyperedge requires all tail nodes to simultaneously meet certain conditions for the head node's capacity value to be effectively output; if any tail node's capacity is insufficient, the head node's capacity value is reset to zero. The series minimum operator takes the minimum capacity of all tail nodes as the output of the hyperedge. A parallel hyperedge requires any one of its tail nodes to be available to maintain the head node's function, with redundancy between multiple alternative paths. The series replacement operator takes the maximum capacity of all tail nodes as the output of the hyperedge. Capacity excess edge refers to the combined effective capacity of multiple tail node devices, compared with the scenario's required capacity to determine if the demand is met. The capacity supply-demand ratio operator refers to the minimum effective capacity after accumulation. Timing excess edge refers to the head node's output capacity depending not only on the tail node's capacity but also on whether the actual action latency falls within the allowable latency range. The timing gating operator is represented as: .in, Represents the timing gate operator. Indicates the baseline capability value. This represents the minimum value of the allowable delay interval. The maximum value of the allowed delay interval, represents the measured delay, and 'I' represents the indicator function, which takes the value 1 when the condition is met and 0 when the condition is not met. A common-cause directed superedge refers to a situation where multiple tail nodes or downstream devices jointly depend on the same common resource. When this common resource fails, all devices that depend on it simultaneously lose functionality. The common-cause fault gating operator is represented as: .in, This represents the common-cause fault gating operator. This represents the baseline output value when no common-cause fault is triggered. This represents the common cause trigger flag, which is 1 when the condition is met and 0 when the condition is not met.

[0035] Connecting the corresponding fire protection space nodes, fire scene nodes, energy medium nodes, equipment nodes, control nodes, and fire protection function nodes using directed hyperedges generates a fire safety function hypergraph; the fire safety function hypergraph... Figure 1The constructed topology remains unchanged before building renovations or system expansion. The topology includes: the types and number of nodes, the types and number of directed hyperedges, and the logical operators bound to the directed hyperedges. Based on real-time physical state data, device status values ​​are assigned to each device node in the fire safety functional hypergraph, and the physical parameters of each directed hyperedge are initialized to achieve the initialization of the fire safety functional hypergraph. That is, real-time physical state data does not change the topology.

[0036] At this point, step S26 includes: S261. Obtain the relationship type of each directed hyperedge in the fire safety function hypergraph, and initialize the physical parameters of the corresponding directed hyperedge based on the relationship type of the directed hyperedge and the preset equipment design table. S262. When the relationship type of the directed superedge is a series superedge or a parallel superedge, the physical parameters of the directed superedge are assigned null values. When the relationship type of the directed superedge is a capacity superedge, the design fire water consumption, design smoke exhaust volume, and design power supply capacity are obtained from the equipment design table, and the design fire water consumption, design smoke exhaust volume, and design power supply capacity are used as demand capacity to initialize the physical parameters of the corresponding capacity superedge. When the relationship type of the directed superedge is a timing superedge, the design permitted action delay interval is obtained from the equipment design table, and the physical parameters of the corresponding timing superedge are initialized according to the design permitted action delay interval. When the relationship type of the directed superedge is a common cause directed superedge, the design common voltage threshold, design common water source threshold, and design common communication threshold are obtained from the equipment design table, and the design common voltage threshold, design common water source threshold, and design common communication threshold are used as common cause triggering conditions to initialize the corresponding common cause directed superedge, thus obtaining all the initialized directed superedges. S263. Based on the directed hyperedge associated with each device node after initialization and combined with the real-time physical state data, calculate the controllability, observability, action completion capability, output capability, action latency, and reliability of the device node. Assign device state values ​​to the corresponding device node based on the controllability, observability, action completion capability, output capability, action latency, and reliability to achieve the initialization of the fire safety function hypergraph.

[0037] In this embodiment, the relationship type of each directed hyperedge in the fire safety functional hypergraph is obtained. Based on the relationship type of the directed hyperedge and the preset equipment design table, the physical parameters of the corresponding directed hyperedge are initialized. The preset equipment design table includes parameter information such as design fire water volume, design smoke exhaust volume, design power supply capacity, design permissible action delay interval, design common voltage threshold, design common water source threshold, and design common communication threshold, and establishes a mapping relationship between the parameter information and the type of the corresponding directed hyperedge. Based on the initialized directed hyperedge associated with each device node and real-time physical state data, the controllability, observability, action completion capability, output capability, action delay, and reliability of the device node are calculated, thereby realizing the initialization of the fire safety functional hypergraph. The directed hyperedges in the initialized fire safety functional hypergraph are not only used to represent the topological connection relationship between nodes, but also to carry the equipment capability transmission, resource supply constraints, action timing constraints, and common resource failure propagation rules in the building fire protection system, so that the directed hyperedge can serve as a calculation unit for equipment state propagation. Controllability is determined by a combination of control loop reachability, access control status, and local interlocking status; observability is determined by a combination of the effective proportion of the necessary observation channels for the device node, timestamp freshness, and communication quality; action completion capability is directly taken from the result of the most recent active function verification. Active function verification refers to the process of performing actual action tests and verifying whether its physical output meets the standards; output capability is determined by normalizing the current pressure, flow rate, air volume, or electrical power relative to the required capacity or equipment rating in the capacity superedge connected to the device node; action latency is determined by the difference between the command timestamp and the timestamp of reaching the physical threshold; credibility is obtained by aggregating the credibility of controllability, observability, action completion capability, output capability, and action latency across different dimensions.

[0038] S2. Extract all scene subgraphs with fire scene nodes as root nodes from the initialized fire safety function supergraph. For each scene subgraph, starting from the device nodes within the scene subgraph, propagate the device status of the device nodes along the topological order of the directed super edges inside the scene subgraph until the root node, so as to output the scene protection capability of each scene subgraph. Based on the scene protection capability, select weak scene subgraphs and identify the set of fault nodes and the set of fault directed super edges from the weak scene subgraphs. At this point, step S2, which involves extracting all scene subgraphs rooted at the fire scene node from the initialized fire safety function hypergraph, includes: S21. Taking each fire scene node as the root node, a multi-layer reverse backtracking strategy is adopted to backtrack along the directed super edges associated with the root node, extracting all first directed super edges pointing to the root node and the corresponding leaf node set, and simultaneously extracting all second directed super edges pointing to the leaf node set and the corresponding child node set, until the reverse backtracking reaches the device node, and forming the scene subgraph of the root node by combining all the extracted first directed super edges and the corresponding leaf node set, as well as all the second directed super edges and the corresponding child node set.

[0039] In this embodiment, each fire scene node is taken as the root node, and a multi-layer reverse backtracking strategy is used to extract the corresponding scene subgraph. The first layer: backtracking in the reverse direction along the directed superedges associated with the root node to extract all first directed superedges pointing to the root node and the corresponding leaf node set; the second layer: extracting all second directed superedges pointing to the leaf node set and the corresponding child node set, and continuing to backtrack in the reverse direction until the reverse backtracking reaches the device node. The extracted first directed superedges and the corresponding leaf node set, as well as all second directed superedges and the corresponding child node set, are combined to form the scene subgraph, thereby obtaining the exclusive scene subgraph corresponding to each fire scene node.

[0040] At this point, the N types of nodes mentioned in step S2 include fire protection function nodes. For each scene subgraph, starting with the device nodes within the scene subgraph, the device status of the device nodes is propagated step-by-step along the topological order of the directed hyperedges within the scene subgraph until the root node, to output the scene protection capability of each scene subgraph. Based on the scene protection capability, weak scene subgraphs are selected, and the set of faulty nodes and the set of faulty directed hyperedges are identified from the weak scene subgraphs, including: S22. For each scene subgraph, starting from the device node within the scene subgraph, propagate the device state of the device node along the topological order of the directed hyperedges within the scene subgraph. During the propagation process, calculate and output the capability state value of the current node along each edge using the logical operator bound to the directed hyperedge until the propagation reaches the root node, thereby completing the capability propagation and obtaining the scene subgraph after the capability propagation is completed. S23. Extract the corresponding fire protection function node set from the scene subgraph after the capability propagation is completed, obtain the capability status value of each fire protection function node in the fire protection function node set, take the fire protection function node with the smallest capability status value as the weak link node, and take the capability status value of the weak link node as the scene protection capability of the corresponding scene subgraph after the capability propagation is completed. S24. Compare the scene protection capability of each scene subgraph with the preset protection threshold, and designate the scene subgraphs with scene protection capabilities lower than the preset protection threshold as weak scene subgraphs. In this embodiment, the scene protection capability of each scene subgraph is calculated individually. Starting from a device node within the scene subgraph, the device state of the device node is propagated sequentially along the topological order of the directed hyperedges within the scene subgraph. During the propagation process, the capability state value of the current node is calculated and output edge by edge through the logical operators bound to the directed hyperedges until it propagates to the root node. That is, the capability state value of the current node is used as the input value of the next node, and the next node reads the input value according to the bound logical operator. This process is repeated to complete the capability propagation, resulting in the scene subgraph after capability propagation. The corresponding set of fire protection function nodes is extracted from the scene subgraph after capability propagation. The fire protection function node with the smallest capability state value is designated as the weakest node, and the capability state value of the weakest node is used as the scene protection capability of the corresponding scene subgraph after capability propagation. That is, the barrel effect strategy is adopted. The scene protection capability of each scene subgraph is compared with a preset protection threshold, and the scene subgraph with a scene protection capability lower than the preset protection threshold is designated as a weak scene subgraph.

[0041] At this point, step S24, which defines scene subgraphs with scene protection capabilities below a preset protection threshold as weak scene subgraphs, includes: S241. Take the scene sub-graphs whose scene protection capability is lower than the preset protection threshold as weak scene sub-graphs, determine whether the number of weak scene sub-graphs is greater than 0, if so, input the scene protection capability of each scene sub-graph into the scene risk formula and the scene resilience formula respectively to perform scene risk status assessment and scene resilience assessment, and obtain the scene risk status assessment result and scene resilience assessment result corresponding to each scene sub-graph. S242. Based on the scenario risk status assessment results and scenario resilience assessment results corresponding to each vulnerable scenario sub-graph, and combined with the priority strategy, priority ranking is performed to obtain the priority-ranked vulnerable scenario sub-graphs. The priority strategy is to use the scenario risk status assessment result as the first priority, and when the scenario risk status assessment results are the same, the scenario resilience assessment result is used as the second priority.

[0042] In this embodiment, if there is more than one weak scene subgraph, the scene risk status assessment result and scene resilience assessment result corresponding to each scene subgraph are calculated and prioritized. During subsequent intelligent recovery, the weak scene subgraph with higher priority is processed first.

[0043] The formula for scenario risk is as follows: ; in, This represents the scenario risk status assessment result corresponding to the vulnerable scenario subgraph s at time t. This indicates the probability of a weak scenario subgraph s occurring at time t. This represents the number of people exposed to the vulnerable subgraph s at time t. This represents the potential consequence coefficient of the weak scene subgraph s at time t. This represents the scene protection capability of the weak scene subgraph s at time t. This represents the state correction coefficient of the weak scene subgraph s at time t.

[0044] Scene resilience formula: ; in, This represents the scene resilience assessment result corresponding to the weak scene subgraph s at time t. This represents the proportion of redundant paths in the weak scene subgraph s at time t. This represents the normalization capability that temporary alternatives can provide for the weak scenario subgraph s at time t. This represents the estimated recovery time of the weak scene subgraph s at time t. This indicates the reference recovery time. This represents the proportion of qualified personnel, spare parts, and instruments available in the vulnerable scenario subgraph s at time t. This represents the scene protection capability of the weak scene subgraph s at time t. This represents the maximum uncertainty of device nodes in the weak scenario subgraph s at time t. , , , , as well as These represent the first resilience weight, the second resilience weight, the third resilience weight, the fourth resilience weight, the fifth resilience weight, and the sixth resilience weight, respectively.

[0045] The scenario risk status assessment result is used as the first priority. The higher the scenario risk status assessment result, the higher the priority. When the scenario risk status assessment results are the same, the scenario resilience assessment result is used as the second priority. The lower the scenario resilience assessment result, the higher the priority.

[0046] S25. For each weak scenario subgraph, starting from the short board node inside the weak scenario subgraph, backtrack in the reverse direction along the directed superedge associated with the short board node to extract all short board superedges pointing to the short board node and the corresponding set of short child nodes. Obtain the capability status value of each short child node in the set of short child nodes. Compare the capability status value of each short child node with the fault threshold. Short child nodes with capability status values ​​lower than the fault threshold are identified as fault nodes. The directed superedges associated with the fault nodes are identified as fault directed superedges to obtain the set of fault nodes and the set of fault directed superedges.

[0047] For each weak scenario subgraph, starting from the short board node inside the weak scenario subgraph, backtracking in the reverse direction along the directed superedges associated with the short board node is performed to extract all short board superedges pointing to the short board node and the corresponding set of short child nodes. The capability state value of each short child node in the set of short child nodes is compared with the fault threshold. Short child nodes with capability state values ​​lower than the fault threshold are identified as fault nodes, and the directed superedges associated with the fault nodes are identified as fault directed superedges, thus obtaining the set of fault nodes and the set of fault directed superedges.

[0048] Furthermore, in this embodiment, if the faulty node is a fire protection function node, the fault will be traced back in the opposite direction along the directed superedge pointing to the fire protection function node until the device node is reached.

[0049] S3. Based on the set of faulty nodes and the set of faulty directed superedges, generate L groups of recovery schemes. Use digital twin technology to perform dynamic recovery simulation on each group of recovery schemes, generate dynamic recovery simulation results for each group of recovery schemes, and evaluate the dynamic recovery simulation results of the L groups of recovery schemes using a multi-objective evaluation function to select the optimal recovery scheme for intelligent recovery of the weak scene subgraph.

[0050] At this point, the N types of nodes mentioned in step S3 include energy medium nodes, and the generation of L sets of recovery schemes based on the set of faulty nodes and the set of faulty directed superedges includes: S31. Calculate the first influence degree of the decrease in scene protection capability of each fault node in the fault node set on the corresponding weak scene subgraph and the second influence degree of the increase in scene protection capability of the corresponding weak scene subgraph after each fault node is restored, and use the first influence degree and the second influence degree as the weight of the corresponding fault node. S32. The difference between the current output value and the target output value of each faulty directed superedge in the set of faulty directed superedges is used as the weight of the faulty directed superedge. S33. Using the fire scene nodes of the weak scene subgraph as the sink and the equipment nodes and energy medium nodes in the fault node set as the source set, construct a minimum fault cut set solution model based on the weights of all fault nodes and the weights of all fault directed super edges. S34. Based on the maximum flow minimum cut theorem, the minimum fault cut set solution model is iteratively solved to obtain the minimum fault cut set, which is the combination of the fault node with the minimum total weight and the fault directed super edge. S35. Using the scenario protection capability of the weak scenario subgraph as a constraint to recover to a preset safety threshold, enumerate all intervention schemes that can repair the minimum fault cut set and generate L groups of recovery schemes.

[0051] In this embodiment, the first influence degree of each fault node in the fault node set on the decrease in scene protection capability of the corresponding vulnerable scene subgraph and the second influence degree of each fault node after restoration on the improvement of scene protection capability of the corresponding vulnerable scene subgraph are calculated. The first influence degree and the second influence degree are used as the weights of the corresponding fault nodes. The first influence degree is calculated as the difference between the scene protection capability of the corresponding vulnerable scene subgraph after the fault node is removed and before removal; the second influence degree is calculated as the difference between the scene protection capability of the corresponding vulnerable scene subgraph after restoration and before restoration. The weight of the directed hyperedge of the fault is the difference between the current output value and the target output value. Using the fire scene node of the vulnerable scene subgraph as the sink and the equipment node and energy medium node in the fault node set as the source set, all directed hyperedges originating from the source nodes in the source set eventually converge to the sink. A minimum fault cut set solution model is constructed based on the weights of all fault nodes and the weights of all fault directed hyperedges. A larger weight indicates a more severe loss and is more likely to be selected in the minimum fault cut set solution.

[0052] The minimum fault cut set solution model is iteratively solved based on the maximum flow minimum cut theorem. The maximum flow minimum cut theorem states that in any directed network flow model, the maximum flow from the source to the sink is equal to the capacity of the minimum cut of the network. Therefore, by solving for the maximum flow, the corresponding minimum fault cut set is obtained. Then, the minimum weighted directed superedge set required to disconnect the source and sink is calculated. This minimum weighted directed superedge set is the minimum fault cut set. With the scenario protection capability of the weak scenario subgraph recovering to a preset safety threshold as a constraint, all intervention schemes capable of repairing the minimum fault cut set are enumerated. The enumeration strategy for intervention schemes generates corresponding intervention actions by querying an action primitive table. The combination of intervention actions constitutes the recovery scheme. The action primitive table is a predefined mapping table that maps fault types to specific intervention actions. The corresponding fault type can be obtained based on the fault node and the faulted directed superedge.

[0053] At this point, step S3, which involves using digital twin technology to dynamically simulate the recovery of each recovery scheme and generating the dynamic recovery simulation results for each recovery scheme, includes: S36. Construct L physically isolated digital twin sandbox simulation branches, with each digital twin sandbox simulation branch corresponding to a set of recovery solutions; S37. For each recovery scheme, dynamic recovery simulation is performed in an independent digital twin sandbox simulation branch based on the initialized fire safety function hypergraph. Starting from the node being intervened, incremental propagation is performed along the affected directed hyperedges. The logical operators bound to each directed hyperedge are executed sequentially to obtain the capability state value after intervention. During the dynamic recovery simulation, the capability state value of the corresponding node is triggered sequentially according to the expected start time and expected completion time of the intervention action in each recovery scheme. After each capability state value change, incremental propagation is re-executed to update the scene protection capability. The scene protection capability after each update is recorded in chronological order to generate a recovery trajectory corresponding to the recovery scheme. The recovery trajectory is used as the result of the dynamic recovery simulation.

[0054] In this embodiment, L physically isolated digital twin sandbox simulation branches are constructed. Each digital twin sandbox simulation branch corresponds to a set of recovery schemes. Each digital twin sandbox simulation branch is a complete copy of the current digital twin state of the building fire protection system and runs independently in memory. The various digital twin sandbox simulation branches are isolated from each other and do not affect each other. Any intervention action performed in any digital twin sandbox simulation branch will not affect other digital twin sandbox simulation branches. For each recovery scheme, a dynamic recovery simulation is performed in a separate digital twin sandbox branch based on the initialized fire safety functional hypergraph. Starting from the intervened node, incremental propagation is performed along the affected directed hyperedges, executing the logical operators bound to each directed hyperedge in sequence to obtain the post-intervention capability state value. During the dynamic recovery simulation, the capability state value of the corresponding node is triggered sequentially according to the expected start and completion times of the intervention actions in each recovery scheme. After each capability state value change, incremental propagation is re-executed to update the scene protection capability. The updated scene protection capability is recorded in chronological order to generate a recovery trajectory corresponding to the recovery scheme, and the recovery trajectory is used as the result of the dynamic recovery simulation. After the dynamic recovery simulation of all digital twin sandbox branches is completed, the system is released. The dynamic recovery simulation results are not directly written back to the real equipment status or automatically issued operation permits.

[0055] At this point, step S3, which involves evaluating the dynamic recovery simulation results of the L groups of recovery schemes using a multi-objective evaluation function to select the optimal recovery scheme for intelligent recovery of the weak scene sub-graph, includes: S36. Calculate the area of ​​loss of life safety capability and the recovery time of key functions based on the dynamic recovery simulation results of each recovery plan. At the same time, calculate the marginal resilience gain of a single intervention action and the synergistic recovery gain of a combination of multiple intervention actions in each recovery plan. S37. Input the area of ​​life safety capability loss, key function recovery time, marginal resilience gain and collaborative recovery gain corresponding to each recovery scheme into a multi-objective evaluation function for evaluation, obtain the evaluation results corresponding to each recovery scheme, and select the recovery scheme corresponding to the smallest evaluation result as the optimal recovery scheme to intelligently restore the weak scene sub-graph.

[0056] In this embodiment, the area of ​​loss of life safety capability and the recovery time of critical functions are calculated based on the dynamic recovery simulation results of each recovery plan. Simultaneously, the marginal resilience gain of a single intervention action and the synergistic recovery gain of a combination of multiple intervention actions in each recovery plan are calculated. The area of ​​loss of life safety capability is calculated as follows: ; in, The weak scene subgraph s represents the recovery scheme. The area of ​​loss of life safety capacity Indicates the starting point for restoration. This represents the time domain of the deduction, i.e., the recovery scheme. The interval between the expected start and expected completion times of the intervention actions. This represents the target protection capability of the vulnerable subgraph s. Indicates recovery plan The dynamic recovery simulation results generated for the weak scene subgraph s This means that losses are only counted when the target's protection capability is greater than the result of the dynamic recovery simulation.

[0057] The critical function recovery time is calculated as follows: ; in, The weak scene subgraph s represents the recovery scheme. The critical function recovery time is now available. This represents the minimum security protection threshold for the vulnerable subgraph s. This indicates the earliest time when the condition is met.

[0058] The marginal resilience gain of a single intervention action is calculated as follows: ; in, This represents the marginal resilience gain of a single intervention action j. This represents the scene protection capability of the weak scene subgraph s. This represents the scene protection capability of the vulnerable scene subgraph s after intervention action j is completed. This indicates that it belongs to the recovery plan. Intervention action j, This represents the expected value function calculated based on the success probability of an action.

[0059] The synergistic recovery gain of multiple intervention combinations is calculated as follows: ; in, This represents the synergistic recovery gain of intervention action combination B. This indicates that it belongs to the recovery plan. Intervention action combination B.

[0060] The area of ​​life safety loss, critical function recovery time, marginal resilience gain, and collaborative recovery gain corresponding to each recovery plan are input into a multi-objective evaluation function for evaluation. The multi-objective evaluation function is as follows: ; in, Represents the recovery scheme for the weak scene subgraph s The corresponding evaluation results , , , These represent the first evaluation weight, the second evaluation weight, the third evaluation weight, and the fourth evaluation weight, respectively.

[0061] The optimal recovery plan is selected based on the minimum assessment result, enabling intelligent recovery of vulnerable scenario subgraphs. During intelligent recovery, the building fire protection system outputs the optimal recovery plan to the on-site execution terminal. On-site maintenance personnel execute specific operations according to the intervention targets, intervention actions, estimated execution time, and resource requirements outlined in the recovery plan. Before executing tasks that alter the power supply status of fire protection equipment, media connectivity, automatic control logic, valve positions, or loop shielding status, the system generates a digital operation permission token based on the safety degradation envelope. This token is then verified by an edge operation permission interlocking device, which either allows or blocks the corresponding physical operation. After on-site execution, the system verifies whether the function has truly recovered using a three-element physical causal lock. Upon successful verification, the system updates the equipment capability status, scenario protection capability, risk status, and resilience status, triggering the next round of rolling optimization. The three-element physical causal lock includes: a first causal lock: verification of the time delay from the action command to the actuator electrical signal; a second causal lock: verification of the slope of the physical parameters of the energy medium node; and a third causal lock: verification of the response sequence of upstream and downstream equipment nodes. A true functional recovery is considered complete only if all three causal locks pass.

[0062] Example 2 Please refer to Figure 2 The present invention provides a building fire protection multi-fault recovery intelligent management and control system 1 based on physical hypergraph, including a memory 3, a processor 2 and a computer program stored on the memory 3 and run on the processor 2. When the processor 2 executes the computer program, it implements the steps in Embodiment 1.

[0063] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0064] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0066] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0067] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0068] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0069] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A smart management and control method for multi-fault recovery of building fire protection based on physical hypergraph, characterized in that, include: The system acquires multi-source static configuration data and real-time physical status data of the building fire protection system. Based on the multi-source static configuration data, it constructs a fire safety functional hypergraph consisting of N types of nodes and M types of directed hyperedges. Based on the real-time physical status data, it initializes the physical parameters of each directed hyperedge in the fire safety functional hypergraph and assigns equipment status values ​​to each device node to initialize the fire safety functional hypergraph and obtain the initialized fire safety functional hypergraph. The N types of nodes include fire scene nodes and device nodes. Extract all scene subgraphs with fire scene nodes as root nodes from the initialized fire safety function hypergraph. For each scene subgraph, starting from the device nodes within the scene subgraph, propagate the device status of the device nodes along the topological order of the directed hyperedges inside the scene subgraph until the root node, so as to output the scene protection capability of each scene subgraph. Based on the scene protection capability, weak scene subgraphs are selected, and the set of fault nodes and the set of fault directed hyperedges are identified from the weak scene subgraphs. Based on the set of faulty nodes and the set of faulty directed superedges, L sets of recovery schemes are generated. Each set of recovery schemes is dynamically simulated using digital twin technology, and the dynamic recovery simulation results of each set of recovery schemes are generated. The dynamic recovery simulation results of the L sets of recovery schemes are evaluated using a multi-objective evaluation function to select the optimal recovery scheme for intelligent recovery of the weak scene subgraph.

2. The intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph as described in claim 1, characterized in that, The multi-source static configuration data includes building space topology data, pipeline medium connectivity data, fire equipment deployment data, control logic data, and scene partitioning data. The construction of a fire safety functional hypergraph based on the multi-source static configuration data, consisting of N types of nodes and M types of directed hyperedges, involves initializing the physical parameters of each directed hyperedge in the fire safety functional hypergraph based on the real-time physical state data, and simultaneously assigning equipment state values ​​to each device node. This initialization of the fire safety functional hypergraph includes: Fire protection space nodes are generated based on building space topology data, spatial connectivity and spatial isolation relationships between fire protection space nodes are generated based on scene partition data, and fire scene nodes are generated based on scene partition data to establish spatial ownership relationships between fire scene nodes and fire protection space nodes. Based on pipeline medium connectivity data, energy medium nodes and energy flow relationships between energy medium nodes are generated; based on fire equipment deployment data, equipment nodes and spatial deployment relationships between equipment nodes are generated; based on control logic data, control nodes and control association relationships between control nodes and equipment nodes are generated. Based on the building space topology data, the pipeline medium connectivity data, the control logic data, and the spatial affiliation relationship between the fire protection space nodes and the fire scene nodes, fire protection function nodes are generated, as well as the energy supply relationship between the fire protection function nodes and the corresponding equipment nodes and the fire protection service relationship between the corresponding fire protection space nodes. Directed superedges are generated based on the fire protection service relationship, the energy supply relationship, the spatial connectivity relationship, the spatial isolation relationship, the spatial affiliation relationship, the energy circulation relationship, the spatial layout relationship, and the control association relationship. A corresponding logical operator is bound to each directed superedge according to its relationship type. The relationship types of the directed superedges include series superedges, parallel superedges, capacity superedges, time-series superedges, and common-cause directed superedges. The logical operators include series minimum operator, parallel replacement operator, capacity supply-demand ratio operator, time-series gating operator, and common-cause fault gating operator. The corresponding fire protection space nodes, fire scene nodes, energy medium nodes, equipment nodes, control nodes, and fire protection function nodes are connected according to the directed hyperedge to generate a fire safety function hypergraph. Based on the real-time physical state data, the device state value is assigned to each device node in the fire safety function hypergraph, and the physical parameters of each directed hyperedge are initialized to achieve the initialization of the fire safety function hypergraph.

3. The intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph as described in claim 2, characterized in that, The initialization of the fire safety function hypergraph involves assigning physical parameters to each directed hyperedge based on the real-time physical state data, and simultaneously assigning device state values ​​to each device node. Obtain the relationship type of each directed hyperedge in the fire safety function hypergraph, and initialize the physical parameters of the corresponding directed hyperedge based on the relationship type of the directed hyperedge and the preset equipment design table; When the relationship type of the directed superedge is a series superedge or a parallel superedge, the physical parameters of the directed superedge are assigned null values. When the relationship type of the directed superedge is a capacity superedge, the design fire water consumption, design smoke exhaust volume, and design power supply capacity are obtained from the equipment design table, and the design fire water consumption, design smoke exhaust volume, and design power supply capacity are used as demand capacity to initialize the physical parameters of the corresponding capacity superedge. When the relationship type of the directed superedge is a timing superedge, the design permitted action delay interval is obtained from the equipment design table, and the physical parameters of the corresponding timing superedge are initialized according to the design permitted action delay interval. When the relationship type of the directed superedge is a common cause directed superedge, the design common voltage threshold, design common water source threshold, and design common communication threshold are obtained from the equipment design table, and the design common voltage threshold, design common water source threshold, and design common communication threshold are used as common cause triggering conditions to initialize the corresponding common cause directed superedge, thus obtaining all the initialized directed superedges. Based on the directed hyperedge associated with each device node after initialization and combined with the real-time physical state data, the controllability, observability, action completion capability, output capability, action latency, and reliability of the device node are calculated. Based on the controllability, observability, action completion capability, output capability, action latency, and reliability, the corresponding device node is assigned a device state value to achieve the initialization of the fire safety function hypergraph.

4. The intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph as described in claim 1, characterized in that, The step of extracting all scene subgraphs rooted at the fire scene node from the initialized fire safety functional hypergraph includes: Taking each fire scene node as the root node, a multi-layer reverse backtracking strategy is used to backtrack along the directed super edges associated with the root node, extracting all first directed super edges pointing to the root node and the corresponding leaf node set, and simultaneously extracting all second directed super edges pointing to the leaf node set and the corresponding child node set, until the reverse backtracking reaches the device node. The extracted first directed super edges and the corresponding leaf node set, as well as all second directed super edges and the corresponding child node set, are combined to form the scene subgraph of the root node.

5. The intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph as described in claim 1, characterized in that, The N types of nodes include fire protection function nodes. For each scene subgraph, starting with the device nodes within the scene subgraph, the device status of the device nodes is propagated sequentially along the topological order of the directed hyperedges within the scene subgraph until the root node, to output the scene protection capability of each scene subgraph. Based on the scene protection capability, weak scene subgraphs are selected, and the set of faulty nodes and the set of faulty directed hyperedges are identified from the weak scene subgraphs, including: For each scene subgraph, starting from the device node within the scene subgraph, the device state of the device node is propagated step by step along the topological order of the directed hyperedges within the scene subgraph. During the propagation process, the capability state value of the current node is calculated and output edge by edge through the logical operators bound to the directed hyperedges until the propagation reaches the root node, thus completing the capability propagation and obtaining the scene subgraph after the capability propagation is completed. Extract the corresponding fire protection function node set from the scene subgraph after the capability propagation is completed, obtain the capability status value of each fire protection function node in the fire protection function node set, take the fire protection function node with the smallest capability status value as the weak link node, and take the capability status value of the weak link node as the scene protection capability of the corresponding scene subgraph after the capability propagation is completed. The scene protection capability of each scene subgraph is compared with the preset protection threshold, and the scene subgraphs with scene protection capabilities lower than the preset protection threshold are identified as weak scene subgraphs. For each weak scenario subgraph, starting from the shortest node inside the weak scenario subgraph, backtracking in the reverse direction along the directed superedges associated with the shortest node is performed to extract all shortest superedges pointing to the shortest node and the corresponding set of short child nodes. The capability status value of each short child node in the set of short child nodes is obtained. The capability status value of each short child node is compared with the fault threshold. Short child nodes with capability status values ​​lower than the fault threshold are identified as fault nodes, and the directed superedges associated with the fault nodes are identified as fault directed superedges, thus obtaining the set of fault nodes and the set of fault directed superedges.

6. The intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph as described in claim 5, characterized in that, The step of designating scene sub-graphs with scene protection capabilities below a preset protection threshold as weak scene sub-graphs includes: Scene subgraphs with scene protection capabilities below a preset protection threshold are designated as weak scene subgraphs. It is determined whether the number of weak scene subgraphs is greater than 0. If so, the scene protection capability of each scene subgraph is input into the scene risk formula and the scene resilience formula respectively to perform scene risk status assessment and scene resilience assessment, thereby obtaining the scene risk status assessment result and scene resilience assessment result corresponding to each scene subgraph. Based on the scenario risk status assessment results and scenario resilience assessment results corresponding to each vulnerable scenario subgraph, priority ranking is performed using a priority strategy to obtain the priority-ranked vulnerable scenario subgraphs. The priority strategy is to use the scenario risk status assessment result as the first priority, and when the scenario risk status assessment results are the same, the scenario resilience assessment result is used as the second priority.

7. The intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph as described in claim 1, characterized in that, The N types of nodes include energy medium nodes, and the generation of L sets of recovery schemes based on the set of faulty nodes and the set of faulty directed superedges includes: Calculate the first impact degree of the decrease in scene protection capability of the corresponding weak scene subgraph for each fault node in the fault node set, and the second impact degree of the increase in scene protection capability of the corresponding weak scene subgraph after each fault node is restored, and use the first impact degree and the second impact degree as the weight of the corresponding fault node. The difference between the current output value and the target output value of each faulty directed superedge in the set of faulty directed superedges is used as the weight of the faulty directed superedge. Using the fire scene nodes of the weak scene subgraph as the sink and the equipment nodes and energy medium nodes in the fault node set as the source set, a minimum fault cut set solution model is constructed based on the weights of all fault nodes and the weights of all fault directed super edges. The minimum fault cut set solution model is iteratively solved based on the maximum flow minimum cut theorem to obtain the minimum fault cut set, which is the combination of the fault node with the minimum total weight and the fault directed super edge. Using the scenario protection capability of the weak scenario subgraph recovering to a preset safety threshold as a constraint, all intervention schemes that can repair the minimum fault cut set are enumerated to generate L groups of recovery schemes.

8. The intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph as described in claim 1, characterized in that, The process of dynamically reconstructing each recovery plan using digital twin technology, and generating dynamic recovery simulation results for each recovery plan, includes: Construct L physically isolated digital twin sandbox simulation branches, with each digital twin sandbox simulation branch corresponding to a set of recovery solutions; For each recovery scheme, dynamic recovery simulation is performed in an independent digital twin sandbox simulation branch based on the initialized fire safety functional hypergraph. Starting from the node being intervened, incremental propagation is performed along the affected directed hyperedges. The logical operators bound to each directed hyperedge are executed sequentially to obtain the capability state value after intervention. During the dynamic recovery simulation, the capability state value of the corresponding node is triggered sequentially according to the expected start time and expected completion time of the intervention action in each recovery scheme. After each capability state value change, incremental propagation is re-executed to update the scene protection capability. The scene protection capability after each update is recorded in chronological order to generate a recovery trajectory corresponding to the recovery scheme. The recovery trajectory is used as the result of the dynamic recovery simulation.

9. The intelligent management and control method for multi-fault recovery of building fire protection based on physical hypergraph as described in claim 1, characterized in that, The step of evaluating the dynamic recovery simulation results of L groups of recovery schemes through a multi-objective evaluation function to select the optimal recovery scheme for intelligent recovery of the weak scene subgraph includes: The area of ​​loss of life safety capabilities and the recovery time of critical functions are calculated based on the dynamic recovery simulation results of each recovery plan. At the same time, the marginal resilience gain of a single intervention action and the synergistic recovery gain of a combination of multiple intervention actions in each recovery plan are calculated. The life safety capability loss area, critical function recovery time, marginal resilience gain, and collaborative recovery gain corresponding to each recovery scheme are input into a multi-objective evaluation function for evaluation. The evaluation results corresponding to each recovery scheme are obtained, and the recovery scheme corresponding to the smallest evaluation result is selected as the optimal recovery scheme to intelligently restore the weak scene subgraph.

10. A building fire protection multi-fault recovery intelligent management and control system based on a physical hypergraph, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 8.

Images