An emergency plan deduction verification method, system, device and medium

Abstract

The application relates to the technical field of emergency management business data processing, and discloses an emergency plan deduction verification method, system, device and medium, the method comprising the following steps: acquiring state data of physical entities in an emergency scene and constructing a digital twin model; updating the digital twin model according to the state data to obtain a twin state snapshot; acquiring a candidate emergency plan comprising a plurality of plan operations; converting the plan operations into graph transformation operations acting on the digital twin model based on the twin state snapshot, and generating a deduction state sequence; performing coupling propagation according to the coupling relationship between different physical fields in the deduction state sequence generation process to obtain a deduction result; determining a verification index and a verification result according to the deduction result; and the application can form a quantitative verification basis before the candidate emergency plan is executed.

Description

Technical Field

[0001] This application relates to the field of emergency management business data processing technology, specifically to an emergency plan simulation and verification method, system, equipment, and medium. Background Technology

[0002] In scenarios such as emergency management, urban operation management, and public transportation operation management, emergency plan databases are typically established in advance. On-site operational data is collected through dispatch systems, monitoring systems, or business management platforms to assist in emergency response, resource allocation, and decision support after an incident occurs. Existing emergency plan management systems can generally handle business data processing tasks such as plan querying, plan matching, process display, personnel notification, material statistics, and record retention, thus improving the informatization level of emergency response to a certain extent.

[0003] However, existing technologies still largely rely on textual rules, historical experience, or manual consultation to verify candidate emergency plans before execution. For emergency scenarios involving multiple physical entities and operational states that interact, simply determining whether a plan is "applicable" is usually insufficient to determine whether its execution will achieve the desired effect. For example, in a scenario of power failure in the contact rail of urban rail transit, a single emergency plan may simultaneously affect the power supply section, train operation, platform passenger flow, communication command issuance, and emergency resource allocation. If the system only retrieves plans based on event type without assessing the cascading impact of power supply changes on train operation, passenger flow, and communication before execution, dispatchers will find it difficult to determine whether the plan will lead to subsequent problems such as passenger congestion, excessively long response times, or excessive resource investment.

[0004] Furthermore, status data in existing emergency response systems is typically scattered across different systems such as power supply monitoring, train dispatching, passenger flow statistics, communication monitoring, and resource management. While each system can provide operational data for its respective domain, insufficient connection between status data and operational plans can easily arise during contingency plan simulation and verification. On the one hand, operational steps in candidate emergency plans are often represented by textual or tabular flowcharts, making it difficult to directly reflect their impact on different equipment, areas, vehicles, personnel flow nodes, or communication nodes. On the other hand, the status transmission relationships between different operational domains are not fully incorporated into the verification process, resulting in the inability to quantify the potential cross-domain impacts after plan execution in advance. Therefore, even if existing systems can output multiple candidate emergency plans, it is difficult to compare response time, resource consumption, status recovery level, and risk changes under a unified standard.

[0005] The aforementioned problems become even more pronounced when multiple candidate contingency plans coexist. Existing systems typically only provide a list of plans or recommend plans according to preset levels, lacking quantitative verification criteria for emergency operational management and dispatch decisions. Dispatchers need to manually compare the handling steps, resource requirements, and potential impacts of multiple candidate plans, which can easily lead to unclear selection criteria, inconsistent verification indicators, and unpredictable handling outcomes. Especially when emergencies are constantly changing, the on-site situation has already changed, but the plan verification remains at the initial state or static text level, easily causing the verification results to become disconnected from the actual operational status.

[0006] Therefore, the technical problem that existing technologies need to solve is: in complex emergency scenarios, how to effectively verify the execution process of the emergency plan and its cross-domain impact based on emergency scenario status data before the implementation of the candidate emergency plan, and form quantitative verification results that can be used for emergency management, resource scheduling, operational decision-making and plan selection, so as to reduce the shortcomings of relying solely on human experience to judge the feasibility of candidate emergency plans. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this application aims to provide an emergency response plan simulation and verification method, system, device, and medium. This addresses the technical problem that in current emergency management business data processing, candidate emergency response plans often rely on textual rules, historical experience, or manual consultation for pre-execution judgment. This makes it difficult to uniformly quantify and verify the execution process, cross-physical field impact, response time, resource consumption, and state recovery degree of the plan operations based on emergency scenario state data. Consequently, emergency management, resource scheduling, operational decisions, and plan selection lack dynamic simulation basis and quantitative evaluation results. This application constructs and updates a digital twin model corresponding to the emergency scenario, converting the plan operations in candidate emergency response plans into graph transformation operations acting on the digital twin model. During the generation of the simulation state sequence, coupling propagation between different physical fields is performed. Then, based on the simulation results, verification indicators and verification results are determined, enabling candidate emergency response plans to form a data-driven verification basis for emergency management and scheduling decisions before execution.

[0008] To achieve the above objectives, this application adopts the following technical solution:

[0009] Firstly, this application provides a method for emergency response plan simulation and verification, including:

[0010] Acquire the status data of physical entities in emergency scenarios, and construct a digital twin model corresponding to the emergency scenario based on the status data;

[0011] The digital twin model is updated based on the state data to obtain a snapshot of the twin state at the current moment;

[0012] Obtain candidate emergency response plans, which include multiple plan operations;

[0013] Based on the twin state snapshot and candidate emergency plans, the plan operations are converted into graph transformation operations that act on the digital twin model, and the inferred state sequence corresponding to the candidate emergency plans is generated according to the graph transformation operations.

[0014] In the process of generating the simulated state sequence, the state changes of different physical fields are coupled and propagated according to the coupling relationship between different physical fields in the digital twin model to obtain the simulation results corresponding to the candidate emergency plan;

[0015] Based on the simulation results, the verification indicators for the candidate emergency response plans are determined, and the verification results corresponding to the candidate emergency response plans are determined based on the verification indicators.

[0016] Preferably, a digital twin model corresponding to the emergency scenario is constructed based on the status data, including:

[0017] Nodes are generated based on physical entities;

[0018] Directed edges are generated based on the influence relationships and influence strengths between physical entities;

[0019] Generate subgraphs corresponding to different physical fields based on nodes and directed edges;

[0020] Generate coupling edges based on the influence relationships between different physical fields;

[0021] The digital twin model is obtained based on the subgraph and coupling edges.

[0022] Preferably, the digital twin model is updated based on the state data to obtain a snapshot of the twin state at the current moment, including:

[0023] The state data is fused to obtain the state estimation results;

[0024] The state change is determined based on the state estimation results and the node states in the digital twin model;

[0025] When the state change meets the preset update conditions, the node state, directed edge or coupled edge is updated to obtain a twin state snapshot.

[0026] Preferably, obtaining candidate emergency response plans includes:

[0027] Obtain the operation type, target entity, operation parameters, and planned execution time for each operation in the candidate emergency response plans;

[0028] Multiple simulation moments are determined based on the planned execution time and the preset simulation step size, and an event queue is generated based on the planned execution time.

[0029] The corresponding contingency plan operation is determined based on the event queue at each simulation moment.

[0030] Preferably, the pre-planned operation is converted into a graph transformation operation applied to the digital twin model, including:

[0031] Based on the operation type, target entity, and parameters corresponding to the pre-planned operation, determine the node state assignment operation, node state increment operation, edge weight modification operation, graph structure modification operation, or coupling propagation operation as the graph transformation operation.

[0032] Preferably, the deduced state sequence corresponding to the candidate emergency response plan is generated according to the graph transformation operation, including:

[0033] The simulation time is advanced according to the preset simulation step size;

[0034] Perform the corresponding graph transformation operation at each simulation moment;

[0035] Perform state propagation on the digital twin model after performing graph transformation operations;

[0036] Determine the state change between two adjacent state propagations;

[0037] When the change in state is less than the set convergence threshold, the simulation state corresponding to that simulation moment is generated.

[0038] Arrange the various simulation states according to the simulation time to obtain the simulation state sequence.

[0039] Preferably, the process of generating the deduced state sequence further includes:

[0040] Secondary events are determined based on the comparison between the node states in the simulation and the set safety thresholds.

[0041] Add the graph transformation operation corresponding to the secondary event to the event queue to obtain the event queue after addition;

[0042] The deduced state sequence continues to be generated based on the added event queue.

[0043] Preferably, based on the coupling relationship between different physical fields in the digital twin model, the state changes of different physical fields are coupled and propagated, including:

[0044] The coupling edge determines two physical fields that are coupled.

[0045] Based on the coupling strength between two physical fields, the node state changes in one physical field are propagated to the other physical field, thus obtaining the node state changes in the other physical field.

[0046] Preferably, the method for determining the coupling strength includes:

[0047] Obtain the correlation between two physical fields, the distance between physical entities in the two physical fields, and the state of the connecting medium between the physical entities;

[0048] The coupling strength is determined based on the correlation, distance, and state of the connecting medium.

[0049] Preferably, the verification metrics include entropy change metrics, graph edit distance metrics, response time metrics, and resource consumption metrics;

[0050] The entropy change index is determined based on the graph entropy change of the digital twin model in the deduced state sequence;

[0051] The graph editing distance index is determined based on the graph editing distance between the simulation results and the set target safety status.

[0052] The response time metric is determined based on the time required for the simulated states in the simulated state sequence to reach a set safety threshold.

[0053] Resource consumption indicators are determined based on the cumulative value of resource consumption during the planned operations.

[0054] Preferably, the verification results corresponding to the candidate emergency response plan are determined based on the verification indicators, including:

[0055] Based on the verification indicators and the corresponding reference benchmark values, each verification indicator is dimensionless to obtain the dimensionless verification indicators.

[0056] The comprehensive verification score of the candidate emergency response plan is determined based on the dimensionless verification indicators and their corresponding weights.

[0057] The verification results are determined based on the comprehensive verification score.

[0058] Preferably, there are multiple candidate emergency response plans;

[0059] Based on the comprehensive verification scores corresponding to each candidate emergency plan, the multiple candidate emergency plans are ranked to obtain the ranking results;

[0060] The sorting results, or the sorting results and the candidate emergency plans in the sorting results that meet the preset selection conditions, are determined as the verification results.

[0061] Preferably, the process of updating the digital twin model based on the state data is defined as real-time state synchronization;

[0062] The process of generating a deduced state sequence based on graph transformation operations is defined as the contingency plan deduction.

[0063] When the deviation between the projected state in the projected state sequence and the twin state snapshot updated based on the new state data meets the preset coordination conditions, the twin state snapshot updated based on the new state data is determined as the new twin state snapshot, and the projected state sequence is regenerated based on the new twin state snapshot.

[0064] Preferably, updating the digital twin model based on the state data further includes:

[0065] Obtain the amount of data change between the current state data and the previous state data;

[0066] When the amount of data change does not meet the set incremental update conditions, no update is performed on the corresponding physical entity;

[0067] When the data change meets the set incremental update conditions, the corresponding physical entity is added to the incremental update queue, and the digital twin model is updated based on the incremental update queue.

[0068] Preferably, before generating the deduced state sequence, the method further includes:

[0069] Obtain the propagation path index, coupling effect lookup table, or state template corresponding to the emergency scenario;

[0070] When generating the deduced state sequence, the state propagation result or coupling propagation result is determined based on the propagation path index, coupling effect lookup table, or state template.

[0071] Preferably, when there are multiple candidate emergency plans, the simulation tasks corresponding to multiple candidate emergency plans are merged into a batch simulation task;

[0072] The batch simulation tasks are executed in parallel to generate the simulation state sequence corresponding to each candidate emergency plan.

[0073] Preferably, the emergency scenario is an urban rail transit emergency scenario;

[0074] Status data includes power supply status data, vehicle status data, pedestrian flow status data, and communication status data;

[0075] The candidate emergency response plans include the contact rail power failure response plan, the train failure response plan, and the turnout failure response plan.

[0076] Secondly, this application provides an emergency response plan simulation and verification system for implementing the emergency response plan simulation and verification method described in any one of the first aspects, comprising:

[0077] The twin modeling module is used to acquire the state data of physical entities in emergency scenarios and construct a digital twin model corresponding to the emergency scenario based on the state data.

[0078] The state update module is used to update the digital twin model based on the state data to obtain a snapshot of the twin state at the current moment.

[0079] The contingency plan acquisition module is used to acquire candidate emergency plans, which include multiple contingency plan operations.

[0080] The deduction generation module is used to convert the plan operations into graph transformation operations that act on the digital twin model based on the twin state snapshot and candidate emergency plans, and generate the deduction state sequence corresponding to the candidate emergency plans according to the graph transformation operations.

[0081] The coupling propagation module is used to perform coupling propagation of the state changes of different physical fields in the digital twin model according to the coupling relationship between different physical fields in the process of generating the deduced state sequence, so as to obtain the deduced results corresponding to the candidate emergency plan.

[0082] The verification module is used to determine the verification indicators of candidate emergency plans based on the simulation results, and to determine the corresponding verification results of candidate emergency plans based on the verification indicators.

[0083] Thirdly, this application provides an electronic device, including a processor and a memory, wherein a computer program is stored in the memory, and when the processor loads and executes the computer program, it implements the emergency response plan simulation and verification method described in any one of the first aspects.

[0084] Fourthly, this application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the emergency response plan simulation and verification method described in any one of the first aspects.

[0085] Compared with the prior art, the beneficial effects achieved by this application are as follows:

[0086] This application addresses the pre-execution verification needs of emergency management business data processing, resource scheduling, operational decision-making, and contingency plan selection. First, it acquires the state data of physical entities in the emergency scenario and constructs a digital twin model corresponding to the emergency scenario based on this state data. This allows dispersed on-site states to be organized into model-based inputs that can be used for inference and calculation. Then, it updates the digital twin model based on the state data to obtain a snapshot of the twin state at the current moment, ensuring that the starting point for inference of candidate emergency plans corresponds to the current operational state and reducing verification bias caused by the disconnect between static plan text and real-time emergency status. Subsequently, it acquires candidate emergency plans including multiple plan operations and converts these operations into graph transformation operations acting on the digital twin model based on the twin state snapshot. This transforms textual or procedural plan operations into computable and disseminated model operations. Based on this, a deduced state sequence is generated according to graph transformation operations. During the generation of the deduced state sequence, the coupling propagation of state changes is performed according to the coupling relationship between different physical fields in the digital twin model. This allows the chain effect of the contingency plan operation on different physical fields to be expressed during the deduction process, avoiding the omission of cross-domain impacts caused by judging candidate emergency plans based solely on a single event type or a single business rule. Finally, the verification indicators of the candidate emergency plans are determined based on the deduction results, and the verification results are determined based on the verification indicators. This transforms the pre-execution verification of candidate emergency plans from experience-based judgment to a data processing process based on deduction results and quantitative indicators. This provides clearer verification basis for emergency management, resource scheduling, operational decision-making, and plan selection, and improves the comparability, reliability, and decision support value of candidate emergency plan verification results in complex emergency scenarios. Attached Figure Description

[0087] Figure 1 A flowchart illustrating an emergency response plan simulation and verification method provided in this application embodiment;

[0088] Figure 2 A flowchart illustrating the construction process of the digital twin model provided in this application embodiment;

[0089] Figure 3 A flowchart illustrating the update process of the digital twin model provided in this application embodiment;

[0090] Figure 4 A flowchart illustrating the process of obtaining candidate emergency response plans provided in this application embodiment;

[0091] Figure 5 A flowchart illustrating the process of determining the graph transformation operation provided in an embodiment of this application;

[0092] Figure 6 A flowchart illustrating the generation process of the deduced state sequence provided in this application embodiment;

[0093] Figure 7 A flowchart illustrating the secondary event detection process provided in an embodiment of this application;

[0094] Figure 8 A flowchart illustrating the coupling propagation process provided in an embodiment of this application;

[0095] Figure 9 A flowchart illustrating the process of determining the verification indicators provided in the embodiments of this application;

[0096] Figure 10 A flowchart illustrating the process of determining the verification result provided in the embodiments of this application;

[0097] Figure 11 A flowchart illustrating the coordination process between real-time status synchronization and contingency plan simulation provided in the embodiments of this application;

[0098] Figure 12 A flowchart illustrating the incremental update process provided in this application embodiment;

[0099] Figure 13 A flowchart illustrating the pre-calculation process provided in an embodiment of this application;

[0100] Figure 14 A flowchart illustrating the execution process of a batch simulation task provided in this application embodiment;

[0101] Figure 15 A schematic diagram of an emergency scenario for urban rail transit provided in the embodiments of this application;

[0102] Figure 16 A schematic diagram of the structure of an emergency response plan simulation and verification system provided in this application embodiment;

[0103] Figure 17 A schematic diagram of the overall implementation environment provided for the embodiments of this application. Detailed Implementation

[0104] To make the objectives, technical solutions, and technical effects of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are used to illustrate the technical solutions of this application and do not limit the scope of protection of this application. Those skilled in the art can make adaptive adjustments to the execution subject, data source, model construction method, deduction method, verification indicator setting method, equipment deployment form, or parameter configuration method in the embodiments without departing from the technical concept of this application.

[0105] It is understood that the sequential terms such as "first" and "second" in the embodiments of this application are only used to distinguish different objects, different data, different steps or different results, and do not constitute a sequential order, degree of importance or quantity limitation; "at least one" means one or more; "multiple" means two or more; "each" means that each individual object in the set performs the corresponding processing.

[0106] It should be noted that the state data, physical entities, candidate emergency plans, plan operations, digital twin models, graph transformation operations, deduced state sequences, deduced results, verification indicators, and verification results involved in the embodiments of this application may originate from sensors, business systems, scheduling systems, control systems, historical operation records, plan libraries, configuration libraries, or authorized data interfaces in the emergency scenario; where business data or operational data is involved, the executing device shall complete authorization verification, desensitization processing, access permission verification, or localization processing before acquisition and processing.

[0107] In this application embodiment, the emergency scenario is used to represent an operational scenario in which a response plan for an emergency needs to be simulated and the response effect verified. The emergency scenario can be an urban rail transit operation scenario, a public transportation operation scenario, a park operation scenario, an energy facility operation scenario, or other scenarios with needs for equipment operation, regional control, personnel flow, communication support, or resource scheduling. The emergency scenario is used to limit the scope of status data collection, the scope of application of candidate emergency plans, and the output objects of verification results.

[0108] In this application embodiment, physical entities are used to represent objects in an emergency scenario that can generate status data or be affected by pre-planned operations. Physical entities can be equipment, facilities, areas, vehicles, personnel flow nodes, communication nodes, power nodes, control nodes, or scheduling nodes. Each physical entity has at least an entity identifier, an entity type, an entity location, and an entity status field. The entity identifier is used to distinguish different physical entities, the entity type is used to determine the physical field to which the physical entity belongs, the entity location is used to determine the distance or connection relationship between physical entities, and the entity status field is used to record the operating status of the physical entity at the current moment.

[0109] The state data in this embodiment is used to represent the state record of a physical entity in an emergency scenario. The state data can be obtained from sensors, scheduling systems, control systems, monitoring systems, business systems, historical operation records, or authorized data interfaces. The state data includes at least an entity identifier, collection time, and a state field. The state field can include operating status, equipment status, area status, communication status, passenger flow status, power supply status, vehicle operation status, or resource occupancy status. The state data is used to construct a digital twin model, update the digital twin model, and generate a snapshot of the twin state at the current moment.

[0110] The digital twin model in this embodiment is used to represent the state mapping model of an emergency scenario in a computing environment. The digital twin model includes nodes, directed edges, subgraphs, and coupling edges. Nodes are generated based on physical entities and are used to record the node states corresponding to the physical entities. Directed edges are generated based on the influence relationships and influence strengths between physical entities and are used to record the state propagation direction and state propagation strength between nodes. Subgraphs are generated based on nodes and directed edges and are used to represent the state structure in a physical field. Coupling edges are generated based on the influence relationships between different physical fields and are used to record the state propagation relationships between different physical fields.

[0111] In this embodiment, the candidate emergency response plan represents an emergency response plan to be simulated and verified. The candidate emergency response plan includes multiple plan operations. Each plan operation includes at least an operation type, an operation target entity, operation parameters, and a planned execution time. The operation type is used to determine the processing category corresponding to the plan operation. The operation target entity is used to determine the physical entity affected by the plan operation. The operation parameters are used to determine the state value, state increment, connection relationship change, or structural change corresponding to the plan operation. The planned execution time is used to write the plan operation into the event queue and to determine the simulation time corresponding to the plan operation.

[0112] In this embodiment, the graph transformation operation represents the computational operation obtained from the pre-planned operation and applied to the digital twin model. The graph transformation operation includes node state assignment operation, node state increment operation, edge weight modification operation, graph structure modification operation, or coupling propagation operation. The node state assignment operation is used to set the node state corresponding to the operation target entity to the state specified by the operation parameters. The node state increment operation is used to adjust the node state corresponding to the operation target entity according to the operation parameters. The edge weight modification operation is used to adjust the influence intensity corresponding to the directed edge. The graph structure modification operation is used to add, delete, or replace nodes, directed edges, or coupled edges. The coupling propagation operation is used to transmit node state changes between different physical fields.

[0113] In this embodiment, the deduction state sequence is used to represent a set of states arranged according to multiple deduction moments; each deduction moment corresponds to a deduction state; the deduction state is used to record the node state, directed edge, subgraph state, coupled edge state, and secondary event state in the digital twin model at the corresponding deduction moment; the deduction state sequence is generated after performing the corresponding graph transformation operation and coupling propagation operation at each deduction moment, using the twin state snapshot as the initial state; the deduction state sequence is used to determine the deduction results and the verification indicators of the candidate emergency response plan.

[0114] The simulation results in this application embodiment are used to represent the execution results of the candidate emergency response plan obtained from the simulation state sequence. The simulation results may include the final simulation state in the simulation state sequence, the state change process at each simulation moment, the coupling propagation results between different physical fields, the secondary event detection results, and the time to reach the safety recovery threshold or enter the safety state range. The simulation results are used to calculate the verification indicators corresponding to the candidate emergency response plan and to output the state change record of the candidate emergency response plan during the simulation process to the emergency command terminal.

[0115] The verification metrics in this application embodiment are used to represent the quantitative results of evaluating the simulation effect of candidate emergency plans; the verification metrics include entropy change metrics, graph edit distance metrics, response time metrics, and resource consumption metrics; the entropy change metrics are used to represent the graph entropy change of the digital twin model in the simulation state sequence; the graph edit distance metrics are used to represent the graph structure difference between the simulation result and the set target safety state; the response time metrics are used to represent the time when the simulation state in the simulation state sequence reaches the safety recovery threshold or enters the safety state range; the resource consumption metrics are used to represent the cumulative value of resource consumption of the plan operation.

[0116] The verification results in this application embodiment are used to represent the verification output of candidate emergency plans obtained according to the verification indicators; the verification results may include verification indicators, comprehensive verification scores, ranking results, and candidate emergency plans that meet the preset selection conditions in the ranking results; the comprehensive verification score is determined according to the verification indicators, reference benchmark values, and corresponding weights; the ranking results are generated according to the comprehensive verification scores corresponding to multiple candidate emergency plans; the verification results are used for emergency command terminals to view, compare, or select candidate emergency plans.

[0117] See Figure 17 The implementation environment of this application embodiment includes an emergency plan simulation and verification system, a state data source, an emergency plan library, a configuration library, a digital twin model storage unit, and an emergency command terminal. The emergency plan simulation and verification system is communicatively connected to the state data source, the emergency plan library, the configuration library, the digital twin model storage unit, and the emergency command terminal. The state data source is used to provide state data of physical entities in the emergency scenario. The emergency plan library is used to provide candidate emergency plans. The configuration library is used to provide preset update conditions, preset simulation step size, set convergence threshold, set safety threshold, reference benchmark value, and weight. The digital twin model storage unit is used to store the digital twin model, twin state snapshots, and simulation state sequences. The emergency command terminal is used to receive verification results.

[0118] The emergency response plan simulation and verification system is used to acquire state data from a state data source, construct and update a digital twin model based on the state data, and obtain a snapshot of the twin state at the current moment; retrieve candidate emergency response plans from the emergency response plan library, and convert the plan operations in the candidate emergency response plans into graph transformation operations; generate a simulation state sequence based on the graph transformation operations, and couple and propagate the state changes of different physical fields during the generation of the simulation state sequence; determine verification indicators based on the simulation results, and determine the verification results based on the verification indicators; the emergency response plan simulation and verification system sends the verification results to the emergency command terminal.

[0119] The emergency response plan simulation and verification system can be deployed in four layers: physical layer, data layer, model layer, and application layer. The physical layer is used to collect the state data of physical entities in the emergency scenario. The data layer is used to fuse and distribute the state data incrementally. The model layer is used to perform digital twin model construction, digital twin model update, graph transformation simulation, and coupling propagation. The application layer is used to display verification indicators, comprehensive verification scores, ranking results, and verification results, and to provide decision support to the emergency command center.

[0120] Example 1:

[0121] See Figure 1 This embodiment provides a method for emergency response plan simulation and verification. This embodiment can be executed by emergency response plan simulation and verification equipment, or collaboratively by a processor, twin modeling module, state update module, plan acquisition module, simulation generation module, coupling propagation module, and verification module within an emergency response plan simulation and verification system. This embodiment includes the following steps.

[0122] Step S101: Obtain the status data of physical entities in the emergency scenario, and construct a digital twin model corresponding to the emergency scenario based on the status data.

[0123] Status data can originate from sensors, scheduling systems, control systems, monitoring systems, business systems, historical operation records, or authorized data interfaces. Status data includes at least entity identifier, acquisition time, status field, status value, and status unit. The entity identifier is used to correspond to the physical entity, the acquisition time is used to determine the time location corresponding to the status data, the status field is used to distinguish the type of status data, the status value is used to record the status content of the physical entity at the acquisition time, and the status unit is used to determine the dimension of the status value.

[0124] Physical entities can include equipment, facilities, areas, vehicles, personnel flow nodes, communication nodes, power nodes, control nodes, or dispatch nodes; each physical entity has at least an entity identifier, entity type, entity location, and entity status field; the entity type is used to determine the physical field to which the physical entity belongs, the entity location is used to determine the distance or connection relationship between physical entities, and the entity status field is used to record the operating status of the physical entity in the emergency scenario.

[0125] Digital twin models are used to characterize physical entities, the influence relationships between physical entities, and the coupling relationships between different physical fields in emergency scenarios in a computing environment. Digital twin models include nodes, directed edges, subgraphs, and coupling edges. Nodes are generated based on physical entities, directed edges are generated based on the influence relationships and influence strength between physical entities, subgraphs are generated based on nodes and directed edges, and coupling edges are generated based on the influence relationships between different physical fields.

[0126] When constructing a digital twin model, the corresponding physical entity is first determined based on the entity identifier in the state data; then, nodes are generated based on the physical entity, and the entity state field of the physical entity is written into the corresponding node; then, directed edges are generated based on the power supply impact relationship, operation impact relationship, personnel flow impact relationship, communication impact relationship, or resource scheduling impact relationship between physical entities; finally, a subgraph is generated based on the physical field to which the physical entity belongs, and coupling edges are generated based on the impact relationship between different physical fields to obtain the digital twin model.

[0127] It should be noted that in step S101, during the model construction phase, the entity identifier, entity type, entity location, and status field in the status data are mainly read, and the structure of the digital twin model is generated by combining the influence relationship configuration record and the coupling relationship configuration record; if the status value has been written during the first construction, the status value is used as the initial node status.

[0128] Step S102: Update the digital twin model based on the state data to obtain a snapshot of the twin state at the current moment.

[0129] When updating the digital twin model, the node corresponding to the entity identifier in the state data is found based on the state data acquired at the current moment; when the state field in the state data corresponds to the entity state field in the node, the state value in the state data is written into the node state of the corresponding node; when the state data reflects a change in the influence relationship between physical entities, the corresponding directed edge is updated; when the state data reflects a change in the influence relationship between different physical fields, the corresponding coupling edge is updated.

[0130] The twin state snapshot at the current moment is used to represent the model state of the digital twin model at the current moment; the twin state snapshot includes at least the node state, directed edge, subgraph state and coupling edge state at the current moment; the twin state snapshot serves as the initial state for candidate emergency plan simulation and as the input for generating subsequent simulation state sequences.

[0131] Step S103: Obtain candidate emergency plans, which include multiple plan operations.

[0132] Candidate emergency response plans can be sourced from emergency response plan databases, emergency command terminals, historical plan records, manual input records, or automatic recommendation results; candidate emergency response plans represent emergency response schemes that need to be simulated and verified; candidate emergency response plans correspond to emergency scenarios, physical entity types, emergency event types, or emergency response objectives.

[0133] Each contingency plan operation includes an operation type, an operation target entity, operation parameters, and a planned execution time. The operation type is used to determine the processing category corresponding to the contingency plan operation. The operation target entity is used to determine the physical entity affected by the contingency plan operation. The operation parameters are used to determine the state value, state increment, connection relationship change, or structural change corresponding to the contingency plan operation. The planned execution time is used to determine the execution time position of the contingency plan operation during the simulation process.

[0134] Step S104: Based on the twin state snapshot and candidate emergency plans, the plan operation is converted into a graph transformation operation that acts on the digital twin model, and the deduced state sequence corresponding to the candidate emergency plan is generated according to the graph transformation operation.

[0135] When generating graph transformation operations, the initial state is a twin state snapshot, and the operations of each plan in the candidate emergency plan are read. The operation category of the graph transformation operation is determined according to the operation type in the plan operation. The node, directed edge, subgraph or coupling edge affected by the graph transformation operation is determined according to the operation target entity in the plan operation. The state adjustment content corresponding to the graph transformation operation is determined according to the operation parameters in the plan operation. The simulation time corresponding to the graph transformation operation is determined according to the planned execution time in the plan operation.

[0136] When generating the simulation state sequence, the corresponding graph transformation operations are executed sequentially according to the simulation time. After each graph transformation operation corresponding to a simulation time is executed, the state of the digital twin model is propagated, and the simulation state corresponding to the simulation time is generated. The simulation states corresponding to each simulation time are arranged according to the simulation time to obtain the simulation state sequence corresponding to the candidate emergency plan.

[0137] Step S105: In the process of generating the deduced state sequence, the state changes of different physical fields are coupled and propagated according to the coupling relationship between different physical fields in the digital twin model to obtain the deduced results corresponding to the candidate emergency plan.

[0138] Different physical fields are used to represent sets of states with different operating rules or different state propagation methods in emergency scenarios; different physical fields include at least two physical fields; each physical field corresponds to a subgraph in the digital twin model; the coupling relationship between different physical fields is represented by coupling edges, which are used to record the influence path of node state changes in one physical field on node state changes in another physical field.

[0139] During coupling propagation, two physical fields with a coupling relationship are identified based on the coupling edge. The coupling propagation conditions are given by the coupling propagation configuration record, which includes at least the source physical field identifier, target physical field identifier, source state field, target state field, minimum source node state change, lower limit of coupling strength, and applicable scenario. When the source node state change is greater than or equal to the minimum source node state change and the coupling strength is greater than or equal to the lower limit of coupling strength, the coupling propagation conditions are satisfied. When the node state change in one physical field satisfies the coupling propagation conditions, the node state change is propagated along the coupling edge to the other physical field, and the node state of the corresponding node in the other physical field is updated. The graph transformation operation, state propagation, and coupling propagation are repeated at each simulation time point to obtain the simulation results corresponding to the candidate emergency plan.

[0140] The simulation results may include the simulation state sequence, the final simulation state in the simulation state sequence, the state change process at each simulation moment, the coupling propagation results between different physical fields, and the secondary event detection results; the simulation results are used to represent the execution process and execution results of the candidate emergency plan in the digital twin model.

[0141] Step S106: Determine the verification indicators of the candidate emergency response plan based on the simulation results, and determine the verification results corresponding to the candidate emergency response plan based on the verification indicators.

[0142] When determining verification indicators, state change indicators are determined based on the state changes of the digital twin model in the simulated state sequence; target state proximity indicators are determined based on the difference between the simulation results and the set target safety state; response time indicators are determined based on the time when the simulated state in the simulated state sequence reaches the safety recovery threshold or enters the safety state range; and resource consumption indicators are determined based on the resource consumption records corresponding to the contingency plan operations.

[0143] The verification results may include the verification indicators of the candidate emergency response plans, the comprehensive verification score, the ranking results, or the candidate emergency response plans that meet the preset selection conditions in the ranking results. When there is only one candidate emergency response plan, the verification results corresponding to the candidate emergency response plan are generated based on the verification indicators. When there are multiple candidate emergency response plans, the comprehensive verification score is determined based on the verification indicators corresponding to each candidate emergency response plan, and the ranking results are generated based on the comprehensive verification scores.

[0144] Through the above steps, this embodiment performs simulation verification of candidate emergency plans based on the current twin state snapshot; during the simulation, the plan operation is converted into a graph transformation operation acting on the digital twin model, and the state changes during the emergency response process are reflected through the coupling propagation between different physical fields; then, the verification indicators and verification results are determined based on the simulation results, so that the candidate emergency plan can obtain the corresponding simulation verification results before execution.

[0145] Example 2:

[0146] This embodiment, based on Embodiment 1, provides a more detailed explanation of the process in step S101 of constructing a digital twin model corresponding to the emergency scenario based on the state data. See also... Figure 2 The process of building a digital twin model includes the following steps.

[0147] First, nodes are generated based on physical entities. When generating nodes, the entity identifier, entity type, entity location, and entity status fields are read from the status data. The entity identifier is used as the node identifier; the entity type is used as the basis for dividing the physical field to which the node belongs; the entity location is used as the basis for distance calculation and connection relationship determination; the node status is generated based on the entity status field. When a physical entity includes multiple entity status fields that need to be deduced separately, a node is generated for each entity status field, and the entity identifier and entity status field are written into the node identifier.

[0148] The node status is determined based on the status value corresponding to the entity status field. When the status value is numeric data, the status value is written to the node status according to the corresponding status unit; when the status value is enumerated data, the status value is converted into a status code according to the status coding table, and the status code is written to the node status. The status coding table is provided by the configuration library and includes at least the status field, status value, status code, and applicable entity type.

[0149] Subsequently, directed edges are generated based on the influence relationships and influence strengths between physical entities. When generating directed edges, connection records, operation association records, control association records, or state propagation records in the emergency scenario are read; the node corresponding to the starting physical entity of the influence relationship is determined as the starting node; the node corresponding to the target physical entity of the influence relationship is determined as the ending node; directed edges are generated from the starting node to the ending node; and the influence strength is written into the edge weight field of the directed edge.

[0150] Impact strength represents the degree to which a change in the state of the starting node affects a change in the state of the ending node. Impact strength is given by the impact relationship configuration record in the configuration library. This record includes at least the starting entity type, ending entity type, impact relationship type, impact strength, and applicable scenario. The emergency response simulation and verification equipment uses the starting entity type, ending entity type, and impact relationship type as query conditions to read the corresponding impact strength and writes it to the edge weight field of the directed edge.

[0151] Subsequently, subgraphs corresponding to different physics fields are generated based on the nodes and directed edges. When generating subgraphs, the physics field corresponding to a node is determined according to the entity type in the node; nodes belonging to the same physics field are written into the same subgraph; directed edges where both the starting node and the ending node belong to the same physics field are written into the same subgraph; each subgraph includes at least a subgraph identifier, a set of nodes, a set of directed edges, an edge weight field, and a subgraph state.

[0152] The subgraph state is determined based on the node states and directed edges in the subgraph. When generating the subgraph state, all node states in the subgraph are read and arranged according to node identifiers to form node state records; all directed edges in the subgraph are read and arranged according to the starting node identifier, ending node identifier, and edge weight fields to form directed edge records; the node state records and directed edge records are combined to determine the subgraph state.

[0153] Subsequently, coupling edges are generated based on the influence relationships between different physical fields. When generating coupling edges, the influence relationship records between different physical fields are read; the node corresponding to the source physical field in the influence relationship record is determined as the source node; the node corresponding to the target physical field in the influence relationship record is determined as the target node; a coupling edge is generated pointing from the source node to the target node; and the coupling edge is written between the subgraph corresponding to the source physical field and the subgraph corresponding to the target physical field.

[0154] Coupled edges are used to represent the state propagation relationship between different physical fields. A coupled edge includes at least the source physical field identifier, the target physical field identifier, the source node identifier, the target node identifier, and the coupling strength field. The coupling strength field is given by the coupling relationship configuration record in the configuration library. The coupling relationship configuration record includes at least the source physical field identifier, the target physical field identifier, the source entity type, the target entity type, the coupling strength, and the applicable scenario.

[0155] Finally, a digital twin model is obtained based on the subgraphs and coupling edges. The digital twin model includes multiple subgraphs and coupling edges between them; the subgraphs represent the node states and directed edges within the same physical field; the coupling edges represent the state propagation relationships between different physical fields; the digital twin model is used in subsequent steps to receive state data updates, perform graph transformation operations, generate inferred state sequences, and determine the inference results.

[0156] In some implementations, the digital twin model is represented as:

[0157] ;

[0158] In the formula, express A digital twin model of time; Represents a set of physical fields; Represents a physical field in a set of physical fields; express physical field at any moment Corresponding subgraph; express The set of coupling edges between physical fields at different times.

[0159] In some implementations, physical fields The corresponding subgraph is represented as:

[0160] ;

[0161] In the formula, express physical field at any moment The corresponding set of nodes; express physical field at any moment The corresponding set of directed edges; express physical field at any moment The corresponding set of edge weights; express physical field at any moment The corresponding set of node states.

[0162] In some implementations, the set of coupling edges is represented as:

[0163] ;

[0164] In the formula, express From the physical field at all times Nodes in Pointing to physical field Nodes in The coupling edge; Represents the source physical field; Represents the target physical field; Indicates the source node; Indicates the target node.

[0165] Example 3:

[0166] This embodiment, based on Embodiment 2, provides a more detailed explanation of step S102, which involves updating the digital twin model based on state data to obtain a snapshot of the twin's state at the current moment. See [link to previous embodiment]. Figure 3 The update process of a digital twin model includes the following steps.

[0167] First, the state data is fused to obtain the state estimation result. During the fusion process, the entity identifier, acquisition time, state field, state value, and state unit are read from the state data. Then, the state data is grouped according to the entity identifier and state field. When the same entity identifier and the same state field correspond to multiple state data entries, these multiple state data entries are written into the same state data group.

[0168] Perform time alignment processing on the status data group. The time alignment processing uses the current time as the alignment benchmark and reads the acquisition time of each status data in the status data group; when the time interval between the acquisition time and the current time is not greater than the set time alignment threshold, the corresponding status data is retained; when the time interval between the acquisition time and the current time is greater than the set time alignment threshold, the corresponding status data is deleted; the set time alignment threshold is given by the configuration library, which includes at least entity type, status field, time alignment threshold and applicable scenario.

[0169] Field matching is performed on the time-aligned state data group. The field matching process reads the entity identifier and state field from the state data group and searches for nodes with the same entity identifier and the same state field in the digital twin model. When a corresponding node is found, a data mapping relationship is established between the state data group and the corresponding node. When no corresponding node is found, an unmatched state data record is generated and written to the review record database.

[0170] Anomaly handling is performed on the status data group that has completed field matching. Anomaly handling reads the status value range record from the configuration library. The status value range record includes at least the entity type, status field, minimum status value, maximum status value, and applicable scenario. When the status value is between the minimum and maximum status values, the corresponding status data is retained. When the status value is not between the minimum and maximum status values, the corresponding status data is written to the abnormal status data record, but the corresponding status data is not written to the status estimation result.

[0171] A fusion process is performed on the state data group after outlier handling. When there is only one state data in the state data group, the state value in the state data is determined as the state estimation result; when there are multiple state data in the state data group, the state values ​​are weighted according to the data quality weights corresponding to each state data to obtain the state estimation result; the data quality weights are given by the data quality configuration record, which includes at least the data source, entity type, state field, data quality weights, and applicable scenarios.

[0172] In some implementations, the state estimation result is determined according to the following formula:

[0173] ;

[0174] In the formula, Indicates the first Each node The state estimation results at time 1; Indicates the relationship with the first The number of status data corresponding to each node; Indicates the first Status data in The state value at any given time; Indicates the first The data quality weights corresponding to each state data point are determined by the data quality configuration record. The data quality weights are non-negative. When at least one data quality weight in the state data group is greater than 0, the state estimation result is determined according to the above formula. When all data quality weights are 0, weighted fusion is not performed, and the corresponding state data group is written into the fusion review record, or the state value is selected as the state estimation result according to the order of the backup data sources in the data quality configuration record.

[0175] In other implementations, when the state data comes from the same sensor sequence and needs to be corrected using the node state from the previous time step, the state estimation result is determined according to the following formula:

[0176] ;

[0177] In the formula, Indicates the first Each node The state estimation results at time 1; Indicates the first The node state of each node at the previous time step; Indicates the first Each node The state value corresponding to the given moment; Indicates the first Each node The correction factor at any given time; the correction factor is given by the configuration library according to the entity type and status field.

[0178] Subsequently, the state change amount is determined based on the state estimation results and the node states in the digital twin model. To determine the state change amount, the nodes in the digital twin model are first located based on the entity identifier and state field in the state estimation results; then, the node state before the update is read; next, the difference between the state estimation results and the node state before the update is calculated; finally, the absolute value of the difference is determined as the state change amount.

[0179] In some implementations, the change in state is determined according to the following formula:

[0180] ;

[0181] In the formula, Indicates the first Each node Changes in state at any given time; Indicates the first Each node The state estimation results at time 1; Indicates the first The node state at the previous time step; the state change is used to determine whether to update the digital twin model.

[0182] When the state change meets the preset update conditions, at least one of the node state, directed edges, or coupled edges is updated to obtain a twin state snapshot. The preset update conditions are given by the update condition configuration record in the configuration library. The update condition configuration record includes at least the entity type, state field, state change threshold, directed edge update flag, coupled edge update flag, and applicable scenario. Using the entity type and state field as query conditions, the corresponding state change threshold, directed edge update flag, and coupled edge update flag are read.

[0183] When the state change is greater than or equal to the state change threshold, it is determined that the state change meets the preset update condition, and the state estimation result is written into the node state of the corresponding node; when the directed edge update flag indicates that the directed edge needs to be updated, the directed edge is updated according to the influence relationship between the physical entities corresponding to the state estimation result; when the coupled edge update flag indicates that the coupled edge needs to be updated, the coupled edge is updated according to the influence relationship between the different physical fields corresponding to the state estimation result.

[0184] When updating a directed edge, read the starting entity type, ending entity type, influence relationship type, and influence strength from the influence relationship configuration record; find the corresponding influence relationship configuration record based on the entity type corresponding to the state estimation result; when the influence strength in the influence relationship configuration record is inconsistent with the edge weight field of the corresponding directed edge in the digital twin model, write the influence strength in the influence relationship configuration record into the edge weight field of the corresponding directed edge.

[0185] When updating the coupling edge, read the source physics field identifier, target physics field identifier, source entity type, target entity type, and coupling strength from the coupling relationship configuration record; find the corresponding coupling relationship configuration record based on the entity type and physics field corresponding to the state estimation result; when the coupling strength in the coupling relationship configuration record is inconsistent with the coupling strength field of the corresponding coupling edge in the digital twin model, write the coupling strength in the coupling relationship configuration record into the coupling strength field of the corresponding coupling edge.

[0186] When the state change does not meet the preset update conditions, the node state of the corresponding node remains unchanged, and the corresponding directed edges and corresponding coupling edges are not updated. For state estimation results where the state change does not meet the preset update conditions, the entity identifier, state field, state estimation result, state change, and current time are written to the unupdated state record; the unupdated state record is used for subsequent queries and does not participate in the generation of the inferred state sequence at the current time.

[0187] In some implementations, updating the digital twin model based on state data also includes incremental update processing. Incremental update processing first obtains the amount of data change between the current state data and the previous state data; if the amount of data change does not meet the set incremental update conditions, no update is performed on the corresponding physical entity; if the amount of data change meets the set incremental update conditions, the corresponding physical entity is added to the incremental update queue, and the digital twin model is updated based on the incremental update queue.

[0188] The incremental update conditions are set by the incremental update condition configuration records in the configuration library. The incremental update condition configuration records include at least the entity type, status field, data change threshold, update priority, and applicable scenario. The entity type and status field are used as query conditions to read the corresponding data change threshold and update priority. When the data change is greater than or equal to the data change threshold, the corresponding physical entity and update priority are written to the incremental update queue.

[0189] The incremental update queue is used to record the physical entities that need to be updated. Each record in the incremental update queue includes at least the entity identifier, entity type, state field, state estimation result, data change amount, update priority, and current time. When updating the digital twin model based on the incremental update queue, the records in the incremental update queue are read from front to back according to the update priority, and the node state of the corresponding node is updated according to the state estimation result in the record.

[0190] In some implementations, the amount of data change is determined according to the following formula:

[0191] ;

[0192] In the formula, Indicates the first A physical entity in The amount of data change at any given time; Indicates the first A physical entity in The state value in the state data at any given time; Indicates the first The state value of a physical entity in the state data of the previous time step; the data change is used to determine whether to add the corresponding physical entity to the incremental update queue.

[0193] A twin state snapshot is generated based on the updated node states, directed edges, and coupled edges. The twin state snapshot includes at least the current time, the set of node states, the set of directed edges, the set of edge weights, the set of subgraph states, the set of coupled edges, and the set of coupling strengths. The twin state snapshot is written into the digital twin model storage unit and serves as the initial state for subsequent candidate emergency response plan simulations.

[0194] Example 4:

[0195] This embodiment, based on Embodiment 1, provides a more detailed explanation of the process for obtaining candidate emergency plans in step S103. See [link to Embodiment 1]. Figure 4 The process of obtaining candidate emergency response plans includes the following steps.

[0196] First, obtain the operation type, target entity, operation parameters, and planned execution time for each operation in the candidate emergency response plans. Candidate emergency response plans can be obtained from the emergency response plan database, emergency command terminal, historical plan records, or manual entry records; each candidate emergency response plan should at least include a plan identifier, applicable emergency scenarios, a list of plan operations, and the plan source; each plan operation in the plan operation list should at least include an operation type, target entity, operation parameters, and planned execution time.

[0197] The operation type indicates the processing category corresponding to the planned operation. The operation type corresponds to subsequent graph transformation operations; operation types include node state assignment type, node state increment type, edge weight modification type, graph structure modification type, or coupling propagation type. The node state assignment type is used to determine the node state assignment operation in subsequent steps; the node state increment type is used to determine the node state increment operation in subsequent steps; the edge weight modification type is used to determine the edge weight modification operation in subsequent steps; the graph structure modification type is used to determine the graph structure modification operation in subsequent steps; and the coupling propagation type is used to determine the coupling propagation operation in subsequent steps.

[0198] The operation target entity represents the physical entity responsible for the planned operation. The operation target entity includes at least an entity identifier, an entity type, and a target state field. The entity identifier is used to locate the corresponding node in the digital twin model; the entity type is used to determine the physical field to which the corresponding node belongs; the target state field is used to determine the node state that the planned operation needs to process; when the operation target entity corresponds to the influence relationship between two physical entities, the corresponding directed edge is found based on the nodes corresponding to the two physical entities; when the operation target entity corresponds to the influence relationship between two physical fields, the corresponding coupling edge is found based on the nodes corresponding to the two physical fields.

[0199] Operational parameters are used to represent the processing content corresponding to the contingency plan operation. Operational parameters include at least parameter fields, parameter values, parameter units, and parameter sources; parameter fields are used to determine the node status, edge weight field, connection relationship field, or coupling strength field that needs to be adjusted; parameter values ​​are used to determine the specific adjustment value; parameter units are used to determine the dimensions of the parameter value; parameter sources are used to record whether the operational parameters come from the emergency plan database, emergency command terminal, historical plan records, or manual entry records.

[0200] The planned execution time indicates the execution time position of the contingency plan operation during the simulation process. The planned execution time can be a specific execution time or a time interval based on the simulation start time. When the planned execution time is a specific execution time, the simulation time corresponding to the contingency plan operation is determined based on the time difference between the specific execution time and the simulation start time. When the planned execution time is a time interval, the simulation time corresponding to the contingency plan operation is determined based on the time interval. The simulation start time is determined by the time corresponding to the current twin state snapshot.

[0201] Subsequently, multiple simulation moments are determined based on the planned execution time and the preset simulation step size, and an event queue is generated based on the planned execution time. The preset simulation step size is given by the simulation configuration record in the configuration library; the simulation configuration record includes at least the applicable emergency scenario, the total simulation duration, the preset simulation step size, and the applicable contingency plan type; using the applicable emergency scenario and contingency plan type corresponding to the candidate emergency plan as query conditions, the corresponding total simulation duration and preset simulation step size are retrieved.

[0202] When determining the simulation time, the simulation start time is used as the first simulation time; the preset simulation step size is used as the time interval between two adjacent simulation times; starting from the simulation start time, the time is increased sequentially according to the preset simulation step size until the total simulation duration is reached, resulting in multiple simulation times; each simulation time is used to carry out the contingency plan operation when the planned execution time falls into the corresponding time position.

[0203] When generating the event queue, the planned execution time corresponding to each contingency plan operation is read; the simulation time corresponding to the contingency plan operation is determined based on the planned execution time; the contingency plan operation is written into the event queue at the position corresponding to the simulation time; each record in the event queue includes at least the simulation time, contingency plan identifier, record type, contingency plan operation identifier, graph transformation operation identifier, operation type, operation target entity, and operation parameters; when multiple contingency plan operations correspond to the same simulation time, they are written into the event queue according to the order of their arrangement in the candidate emergency plans. The event queue is used to hold contingency plan operation records or graph transformation operation records obtained from contingency plan operations or secondary events; when the event queue record is a contingency plan operation record, it is first converted into a graph transformation operation and then executed at the corresponding simulation time; when the event queue record is a graph transformation operation record, it is executed directly at the corresponding simulation time.

[0204] When the planned execution time is the same as a certain simulation time, that simulation time is determined as the simulation time corresponding to the contingency plan operation; when the planned execution time is between two adjacent simulation times, the next simulation time that is no earlier than the planned execution time and is the closest is determined as the corresponding simulation time; when the planned execution time is earlier than the simulation start time, the simulation start time is determined as the corresponding simulation time; when the planned execution time is later than the simulation end time, the corresponding contingency plan operation is written into the record outside the simulation range and does not participate in the generation of the state sequence of this simulation.

[0205] Finally, the event queue records corresponding to each simulation moment are determined based on the event queue. When processing any simulation moment, all event queue records corresponding to that simulation moment are read from the event queue; when the event queue record is a pre-plan operation record, the corresponding pre-plan operation is determined as the operation to be converted for that simulation moment; when the event queue record is a graph transformation operation record, the corresponding graph transformation operation is determined as the operation to be executed for that simulation moment.

[0206] Through the above steps, the contingency plans in the candidate emergency plans are converted into event queues organized according to the simulation time. The event queues record the execution time, target, and processing content of the contingency plans during the simulation process. Subsequent steps can read the corresponding contingency plans at each simulation time based on the event queues and convert the contingency plans into graph transformation operations that act on the digital twin model.

[0207] Example 5:

[0208] This embodiment, based on Embodiment 4, provides a more detailed explanation of the process in step S104 of converting the pre-plan operation into a graph transformation operation applied to the digital twin model. See [link to previous embodiment]. Figure 5 The process of determining the graph transformation operation includes the following steps.

[0209] First, the operation type, target entity, and parameters corresponding to the planned operation are read. The operation type, target entity, and parameters are derived from the event queue in Example 4; the operation type is used to determine the category of the graph transformation operation; the target entity is used to determine the node, directed edge, subgraph, or coupled edge affected by the graph transformation operation; the operation parameters are used to determine the state value, state increment, edge weight change, structural change content, or coupling propagation content corresponding to the graph transformation operation.

[0210] Subsequently, based on the operation type, target entity, and operation parameters, the node state assignment operation, node state increment operation, edge weight modification operation, graph structure modification operation, or coupling propagation operation are determined as graph transformation operations. When the operation type is node state assignment, a node state assignment operation is determined; when the operation type is node state increment, a node state increment operation is determined; when the operation type is edge weight modification, an edge weight modification operation is determined; when the operation type is graph structure modification, a graph structure modification operation is determined; and when the operation type is coupling propagation, a coupling propagation operation is determined.

[0211] The node state assignment operation is used to set the node state corresponding to the target entity to the state value specified in the operation parameters. When performing the node state assignment operation, the node in the digital twin model is located based on the entity identifier and target state field in the target entity; the parameter value and unit are read from the operation parameters; when the parameter unit matches the node state's unit, the parameter value is written to the node state; when the parameter unit differs from the node state's unit, the parameter value is converted to the node state's unit according to the unit conversion record before being written to the node state.

[0212] The node state increment operation is used to adjust the node state corresponding to the target entity according to the operation parameters. When executing a node state increment operation, the node in the digital twin model is located based on the entity identifier and target state field in the target entity; the node state before the update is read; the state increment and parameter unit in the operation parameters are read; when the parameter unit is consistent with the node state unit, the node state before the update is added to the state increment to obtain the updated node state; when the parameter unit of the state increment is inconsistent with the node state unit, the state increment is first converted to the node state unit according to the unit conversion record, and then the node state before the update is added to the converted state increment to obtain the updated node state; the updated node state is then written to the corresponding node.

[0213] The edge weight modification operation is used to adjust the influence strength of directed edges. When performing an edge weight modification operation, the directed edges in the digital twin model are located based on the starting and ending entity identifiers in the target entity; the target edge weight value or edge weight change amount is read from the operation parameters; when the operation parameters include the target edge weight value, it is written to the edge weight field of the directed edge; when the operation parameters include the edge weight change amount, the edge weight field before the update is added to the edge weight change amount to obtain the updated edge weight field.

[0214] Graph structure modification operations are used to add, delete, or replace nodes, directed edges, or coupled edges in a digital twin model. When executing a graph structure modification operation, the operation parameters are read: the modification type and the modification object. When the modification type is "add," the modification object is written to the digital twin model. When the modification type is "delete," the modification object is deleted from the digital twin model. When the modification type is "replace," the original node, directed edge, or coupled edge corresponding to the modification object is first deleted, and then the new node, directed edge, or coupled edge corresponding to the modification object is written.

[0215] The coupling propagation operation is used to record the cross-physics propagation conditions triggered by the pre-planned operation. When executing the coupling propagation operation, the source physics field, target physics field, source node, target node, and corresponding coupling edge are determined based on the target entity. The source node state change, coupling edge identifier, and trigger deduction time are written into the coupling propagation record. This step does not repeatedly modify the target node's state. After reaching the corresponding deduction time, step S105 reads the coupling propagation record and performs a single cross-physics field coupling propagation operation according to the coupling edge and its coupling strength.

[0216] In some implementations, the node state assignment operation is represented as:

[0217] ;

[0218] In the formula, This indicates the nodes after the graph transformation operation is performed. The node state; Indicates the state value specified by the operation parameters; node The target entity is determined by its entity identifier and target status field; the status value is determined by the parameter value in the operation parameters.

[0219] In some implementations, the node state increment operation is represented as:

[0220] ;

[0221] In the formula, Nodes before graph transformation operation The node status; This indicates the nodes after the graph transformation operation is performed. The node status; This indicates the state increment specified by the operation parameters; the state increment is given by the state increment field in the operation parameters.

[0222] In some implementations, the edge weight modification operation is represented as:

[0223] ;

[0224] In the formula, This indicates that the graph transformation operation was performed on the nodes before the transformation was performed. Pointing to node The edge weight field corresponding to the directed edge; This indicates that after the graph transformation operation is performed, the node... Pointing to node The edge weight field corresponding to the directed edge; This indicates the amount of edge weight change specified in the operation parameters; the amount of edge weight change is given by the edge weight change field in the operation parameters.

[0225] In some implementations, the graph structure modification operation is represented as:

[0226] ;

[0227] In the formula, This represents the digital twin model before the graph transformation operation is performed. The digital twin model after the graph transformation operation is performed; Indicates the object whose structure is being modified; This indicates that nodes, directed edges, or coupled edges are added, deleted, or replaced according to the structure modification type.

[0228] In some implementations, the coupling propagation operation is represented as:

[0229] ;

[0230] In the formula, express Time Source Physics Field Source node in The amount of change in node state; express Target physics field at any time target node in The amount of change in node state; Indicates from the source physical field Source node in Pointing to the target physical field target node in The coupling strength field corresponding to the coupled edge; the coupling strength field is given by the coupling relationship configuration record.

[0231] In some implementations, one contingency plan operation corresponds to at least two basic graph transformation operations; the emergency contingency plan simulation and verification equipment arranges at least two basic graph transformation operations into composite graph transformation operations according to the execution order field in the contingency plan operation; each basic graph transformation operation in the composite graph transformation operation is executed sequentially according to the order in the event queue.

[0232] Through the above steps, the pre-planned operations in the event queue can be converted into graph transformation operations that act on the digital twin model; node state assignment and node state increment operations are used to change the node state; edge weight modification operations are used to change the influence intensity corresponding to the directed edge; graph structure modification operations are used to change the nodes, directed edges, or coupled edges in the digital twin model; and coupling propagation operations are used to transmit the node state changes in one physical field to another physical field.

[0233] Example 6:

[0234] This embodiment, based on Embodiment 5, provides a more detailed explanation of the process in step S104 of generating the deduced state sequence corresponding to the candidate emergency response plan according to the graph transformation operation. See [link to embodiment]. Figure 6 The process of generating the deduced state sequence includes the following steps.

[0235] First, the simulation time is advanced according to the preset simulation step size. The preset simulation step size is given by the simulation configuration record in the configuration library; the simulation configuration record includes at least the applicable emergency scenario, the plan type, the total simulation duration, the preset simulation step size, the step size unit, the set convergence threshold, and the set upper limit of propagation times; using the applicable emergency scenario and plan type corresponding to the candidate emergency plan as query conditions, the corresponding total simulation duration, preset simulation step size, set convergence threshold, and set upper limit of propagation times are read.

[0236] The simulation time is determined based on the simulation start time, the preset simulation step size, and the total simulation duration. The simulation start time is the time corresponding to the twin state snapshot of the current moment; the simulation start time is used as the first simulation time; each time, a preset simulation step size is added to obtain the next simulation time; when the time interval between the simulation time and the simulation start time reaches the total simulation duration, the generation of new simulation times stops.

[0237] Subsequently, the corresponding graph transformation operation is executed at each simulation moment. When processing any simulation moment, the event queue record corresponding to the simulation moment is first read from the event queue; when the event queue record is a pre-plan operation record, the pre-plan operation is converted into a graph transformation operation in accordance with the method in Example 5; when the event queue record is a graph transformation operation record, the graph transformation operation is read directly; when there are multiple graph transformation operations at the same simulation moment, the multiple graph transformation operations are executed in the order of arrangement in the event queue; when there are no graph transformation operations at the same simulation moment, the current digital twin model structure remains unchanged, and state propagation processing begins.

[0238] Subsequently, state propagation is performed on the digital twin model after the graph transformation operation. State propagation takes the digital twin model after the graph transformation operation as input; for each subgraph, the node state, directed edge, and edge weight fields in the subgraph are read; the node state change of the starting node is propagated to the ending node along the direction of the directed edge; the amount of node state change propagated to the ending node is determined according to the edge weight field; the updated node state of the ending node is obtained based on the node state before the end node update and the amount of node state change propagated to the ending node.

[0239] It should be noted that the state propagation in this step refers to the state propagation along the directed edges within the corresponding subgraph of the same physical field; the state change propagation between different physical fields is uniformly executed by step S105 to avoid the same state change being repeatedly propagated across physical fields at the same deduction time.

[0240] State propagation also includes state transfer between different physical fields. For two physical fields with a coupling edge, the source node, target node, and coupling strength fields in the coupling edge are read; based on the node state change of the source node and the coupling strength field, the node state change of the target node is determined; the node state change of the target node is written into the node state of the target node; when a target node receives multiple node state changes, the multiple node state changes are added together and written into the node state of the target node.

[0241] Subsequently, the state change between two adjacent state propagations is determined. After each state propagation, all node states after the current propagation are read, and all node states before the current propagation are also read; the two sets of node states are matched according to the same node identifier; based on the state normalization benchmark value or state value range corresponding to the state field, each node state is converted into a dimensionless node state; the dimensionless node state difference of each node before and after the current state propagation is calculated; the absolute values ​​of all dimensionless node state differences are added together to obtain the state change between two adjacent state propagations.

[0242] The state normalization baseline value and state value range are determined based on the state value range record corresponding to the state field. The state value range record includes at least the entity type, state field, state unit, minimum state value, maximum state value, and applicable scenario. The state normalization baseline value is the maximum of the absolute values ​​of the minimum and maximum states. When both the absolute values ​​of the minimum and maximum states are 0, the state normalization baseline value is set to 1. The state normalization method field in the deduction configuration record is used to determine whether to use the state normalization baseline value or the state value range for conversion.

[0243] When using a state normalization benchmark for conversion, the node state is divided by the state normalization benchmark to obtain the dimensionless node state. When using a state value range for conversion, the node state is first subtracted from the minimum state value, and then divided by the difference between the maximum and minimum state values ​​to obtain the dimensionless node state. When the maximum state value equals the minimum state value, the dimensionless node state is set to 0. When the node state is enumerated data, the state value is converted into a state code according to the state coding table, and the dimensionless state value corresponding to the state code is read. The dimensionless state value is given by the state coding table and ranges from 0 to 1.

[0244] When the state change is less than the set convergence threshold, the simulation state corresponding to the simulation time is generated. The set convergence threshold is given by the simulation configuration record; the set convergence threshold is used to determine whether the state propagation within this simulation time has ended; when the state change is less than the set convergence threshold, the node state, directed edge, edge weight field, subgraph state, coupled edge, and coupling strength field in the current digital twin model are read, and the reading results are written into the simulation state corresponding to the simulation time.

[0245] When the state change is greater than or equal to the set convergence threshold, the next state propagation continues. The next state propagation uses the digital twin model after the previous state propagation as input. When the number of state propagations reaches the set propagation limit, state propagation within the current simulation time is stopped, and the digital twin model after the last state propagation is written into the simulation state corresponding to the simulation time. The set propagation limit is given by the simulation configuration record and is used to limit the number of state propagations within a single simulation time.

[0246] Finally, the simulated states are arranged according to the simulation time to obtain the simulated state sequence. The simulated state sequence uses the twin state snapshot as the initial state; the simulated states corresponding to each simulation time are written into the sequence record in order from the simulation time to the end; the sequence record includes at least the candidate emergency plan identifier, simulation time, simulation state, state change amount, state propagation number, and graph transformation operation identifier; the simulated state sequence is written into the digital twin model storage unit and used as input for subsequent determination of simulation results and verification indicators.

[0247] When the number of nodes or the dimension of state fields are inconsistent in different physical fields, the node state of the source physical field is first converted into a state dimension that can be accepted by the target physical field according to the node mapping relationship and the state field mapping relationship. To from the physical field The set of nodes is mapped to the physical field The coupling matrix of the node set, whose matrix dimension is... and correspond; To from the physical field Mapping state field dimensions to physical fields The coupling propagation parameter matrix of the state field dimension.

[0248] In some implementations, state propagation within a physical field is performed according to the following formula:

[0249] ;

[0250] In the formula, Represents physical field During the simulation Next The set of node states before the next state propagation; Represents physical field During the simulation Next The set of node states after the next state propagation; Represents physical field During the simulation The normalized adjacency matrix is ​​obtained based on the directed edges and edge weight fields. Represents physical field The corresponding propagation parameter matrix; This represents the state mapping function; the propagation parameter matrix and the state mapping function are given by the propagation configuration record in the configuration library.

[0251] The normalized adjacency matrix is ​​determined based on the edge weight matrix and the degree matrix. The edge weight matrix is ​​based on the physical field. The corresponding directed edges and edge weight fields are generated; the degree matrix is ​​generated based on the sum of the edge weights corresponding to each node in the edge weight matrix; when maintaining the direction of directed edge propagation, the normalized adjacency matrix adopts the out-degree normalized form. When the propagation configuration record indicates that directed edges should be converted to bidirectional edges or a symmetric adjacency matrix, the normalized adjacency matrix is ​​adopted. ,in This represents the symmetric adjacency matrix.

[0252] In some implementations, state propagation involving coupled edges is performed according to the following formula:

[0253] ;

[0254] In the formula, Represents a set of physical fields; Represents physical field Different physical fields; Indicates the time of the deduction. From the physical field To the physical field The coupling matrix; This represents the coupling propagation parameter matrix; the coupling matrix is ​​generated based on the coupling edge and coupling strength fields; the coupling propagation parameter matrix is ​​given by the propagation configuration record in the configuration library.

[0255] In some implementations, the change in state is determined according to the following formula:

[0256] ;

[0257] In the formula, Indicates the time of the deduction. Next Next-state propagation and the first The amount of state change between sub-state propagations; Represents the set of nodes in a digital twin model; Represents the first node in the set of nodes. One node; Represents a node In the Dimensionless node states before the next state propagation; Represents a node In the The dimensionless node state after the next state propagation; the dimensionless node state is determined based on the state normalization reference value or the state value range corresponding to the state field.

[0258] In some implementations, convergence determination is performed according to the following formula:

[0259] ;

[0260] In the formula, Indicates the time of the deduction. Next Next-state propagation and the first The amount of state change between sub-state propagations; This represents the set convergence threshold; the set convergence threshold is given by the deduction configuration record; when the formula is true, the deduction time is generated. The corresponding deduced state.

[0261] Through the above steps, the graph transformation operation corresponding to the candidate emergency plan is executed according to the simulation time; the digital twin model completes state propagation and convergence determination at each simulation time; a corresponding simulation state is generated at each simulation time; multiple simulation states are arranged according to the simulation time to form a simulation state sequence.

[0262] Example 7:

[0263] This embodiment, based on Embodiment 6, provides a more detailed explanation of the secondary event detection process during the generation of the deduced state sequence. See also... Figure 7 The secondary event detection process includes the following steps.

[0264] First, secondary events are determined based on the comparison between the node states in the simulation and the set security thresholds. When processing the simulation state corresponding to any simulation moment, the node states in the simulation state are read; based on the node identifier and state field, the corresponding set security threshold is read from the security threshold configuration record; the node states are compared with the set security thresholds; when the comparison result meets the conditions for secondary event generation, a secondary event is generated. Event deduplication flags are set for the same simulation task, the same node identifier, the same state field, and the same secondary event type; if the same secondary event has already been written before the target simulation moment, it is not written again; the planned execution time is no earlier than the next simulation moment after the current simulation moment.

[0265] The set safety threshold is provided by the safety threshold configuration record in the configuration library. In this embodiment, the set safety threshold is the risk trigger threshold. The safety threshold configuration record includes at least the applicable emergency scenario, physical field identifier, entity type, status field, risk trigger threshold, comparison direction, secondary event type, and secondary event operation template; the emergency plan simulation and verification equipment uses the applicable emergency scenario, physical field identifier, entity type, and status field as query conditions to read the corresponding risk trigger threshold, comparison direction, secondary event type, and secondary event operation template.

[0266] The comparison direction is used to determine the comparison method between the node state and the safety threshold. When the comparison direction is an upper limit comparison, if the node state is greater than or equal to the safety threshold, the comparison result is determined to meet the conditions for the generation of a secondary event. When the comparison direction is a lower limit comparison, if the node state is less than or equal to the safety threshold, the comparison result is determined to meet the conditions for the generation of a secondary event. When the comparison direction is an interval comparison, if the node state is not located within the safety interval corresponding to the safety threshold, the comparison result is determined to meet the conditions for the generation of a secondary event.

[0267] Secondary events are used to represent subsequent events generated during the simulation process when a node's state exceeds a safety threshold. When a secondary event is generated, the simulation time, node identifier, entity type, state field, node state, safety threshold, comparison direction, and secondary event type are written into the secondary event record. The secondary event record is used to generate the graph transformation operation corresponding to the secondary event and is also used to write it into the event queue.

[0268] Subsequently, the graph transformation operation corresponding to the secondary event is added to the event queue, resulting in the added event queue. When generating the graph transformation operation corresponding to the secondary event, the secondary event type is read from the secondary event record; the secondary event operation template in the safety threshold configuration record is searched according to the secondary event type; the operation type, operation target entity, operation parameters, and planned execution time are generated according to the secondary event operation template; and the contingency plan operation corresponding to the secondary event is generated according to the operation type, operation target entity, operation parameters, and planned execution time.

[0269] The secondary event operation template includes at least the secondary event type, operation type, operation target entity generation rules, operation parameter generation rules, and planned execution time generation rules. The operation target entity generation rules determine the operation target entity based on the node identifier and entity type in the secondary event record; the operation parameter generation rules determine the operation parameters based on the node status, safety threshold, and comparison direction in the secondary event record; the planned execution time generation rules determine the planned execution time based on the simulation time corresponding to the secondary event and the preset simulation step size. The planned execution time is no earlier than the next simulation time after the current simulation time; when the planned execution time is earlier than the next simulation time after the current simulation time, the next simulation time after the current simulation time is determined as the planned execution time.

[0270] When converting the pre-planned operation corresponding to the secondary event into a graph transformation operation, the conversion method in Example 5 is followed. If the operation type corresponding to the secondary event is a node state assignment type, a node state assignment operation is generated; if the operation type corresponding to the secondary event is a node state increment type, a node state increment operation is generated; if the operation type corresponding to the secondary event is an edge weight modification type, an edge weight modification operation is generated; if the operation type corresponding to the secondary event is a graph structure modification type, a graph structure modification operation is generated; if the operation type corresponding to the secondary event is a coupling propagation type, a coupling propagation operation is generated.

[0271] When adding the graph transformation operation corresponding to the secondary event to the event queue, the target derivation time is determined based on the planned execution time of the secondary event. If the target derivation time already exists in the event queue, the graph transformation operation corresponding to the secondary event is written to the position corresponding to the target derivation time. If the target derivation time does not exist in the event queue, a new position corresponding to the target derivation time is created in the event queue, and the graph transformation operation corresponding to the secondary event is written to the position corresponding to the target derivation time. After writing is completed, the added event queue is obtained.

[0272] Finally, the derivation state sequence is generated based on the added event queue. While generating the derivation state sequence, the target derivation time and the graph transformation operations after the target derivation time are read from the added event queue. Following the method in Example 6, the corresponding graph transformation operation is executed at each derivation time, and state propagation and convergence determination are performed on the digital twin model after the graph transformation operation. The derivation state corresponding to each derivation time is generated, and each derivation state is written into the derivation state sequence.

[0273] In some implementations, the secondary event detection results are also written into the simulation state. The simulation state includes at least the simulation time, node state, directed edges, edge weight field, subgraph state, coupled edges, coupling strength field, and secondary event record; when no secondary event is generated at the simulation time, the secondary event record is empty; when a secondary event is generated at the simulation time, the secondary event record includes the secondary event type, triggering node identifier, triggering state field, node state, security threshold, and target simulation time.

[0274] In some implementations, the conditions for the generation of secondary events are expressed according to the following formula:

[0275] ;

[0276] In the formula, Represents a node During the simulation Whether to generate secondary events, This indicates the generation of secondary events; Represents a node During the simulation The node status; Represents a node The corresponding risk trigger threshold; the formula corresponds to the case where the comparison direction is the upper limit comparison.

[0277] In other implementations, when the comparison direction is a lower bound comparison, the secondary event generation condition is expressed by the following formula:

[0278] ;

[0279] In the formula, Represents a node During the simulation Whether to generate secondary events; Represents a node During the simulation The node status; Represents a node The corresponding set safety threshold; the formula corresponds to the case where the comparison direction is the lower limit comparison.

[0280] Through the above steps, the deduced state sequence generation process can compare the node state with a safety threshold at each deduced time and generate a secondary event when the node state meets the conditions for secondary event generation. The secondary event is converted into a graph transformation operation and written into the event queue, so that subsequent deduced time steps can continue to generate the deduced state sequence based on the added event queue.

[0281] Example 8:

[0282] This embodiment, based on Embodiments 2 and 6, elaborates on the process of coupling and propagating the state changes of different physical fields according to the coupling relationship between different physical fields in the digital twin model in step S105. See [link to previous embodiment]. Figure 8 The coupling propagation process includes the following steps.

[0283] First, the two physical fields with a coupling relationship are identified based on the coupling edges. When processing the simulation state at any simulation time, the set of coupling edges in the digital twin model is read; each coupling edge includes a source physical field identifier, a target physical field identifier, a source node identifier, a target node identifier, and a coupling strength field; the source physical field identifier corresponds to the source physical field, and the target physical field identifier corresponds to the target physical field; the source node identifier corresponds to the source node in the source physical field, and the target node identifier corresponds to the target node in the target physical field.

[0284] When two physical fields with a coupling relationship are identified, the source physical field identifier and the target physical field identifier in the coupling edge are read. When the source physical field identifier and the target physical field identifier are different, the source physical field and the target physical field are identified as two physical fields with a coupling relationship. The source node and the target node are identified as the coupling propagation objects between the two physical fields. The coupling strength field in the coupling edge is identified as the basis for the state propagation calculation between the two physical fields.

[0285] When determining whether to perform coupling propagation, read the coupling propagation configuration record. The coupling propagation configuration record includes at least the source physical field identifier, target physical field identifier, source state field, target state field, minimum source node state change, lower limit of coupling strength, and applicable scenario. Using the source physical field identifier, target physical field identifier, source state field, and target state field as query conditions, read the corresponding minimum source node state change and lower limit of coupling strength. When the source node state change is greater than or equal to the minimum source node state change and the coupling strength is greater than or equal to the lower limit of coupling strength, perform coupling propagation. When either condition is not met, do not perform coupling propagation for the corresponding coupled edge.

[0286] Subsequently, based on the coupling strength between the two physical fields, the node state changes in one physical field are propagated to the other physical field, resulting in the node state changes in the other physical field. During coupling propagation, the node state change of the source node at the current inference time is read; the coupling strength field between the source node and the target node is read; the node state change of the source node is multiplied by the coupling strength field to obtain the node state change of the target node; and the node state change of the target node is written into the node state of the target node.

[0287] When the target node receives node state changes from multiple source nodes, it calculates the node state change corresponding to each source node according to the coupling edges; it adds up the node state changes corresponding to each source node to obtain the total node state change of the target node; it adds the node state of the target node before the update to the total node state change to obtain the updated node state of the target node; and it writes the updated node state of the target node into the subgraph corresponding to the target physics field.

[0288] The coupling strength is determined based on the correlation between two physical fields, the distance between physical entities in the two physical fields, and the state of the connection medium between the physical entities. The correlation is given by the coupling relationship configuration record; the correlation indicates whether there is a state propagation relationship between the source physical field and the target physical field; the distance is determined based on the entity position of the physical entity corresponding to the source node and the entity position of the physical entity corresponding to the target node; the connection medium state is given by the state data or configuration library, and the connection medium state indicates the availability of the connection medium between the source node and the target node.

[0289] The coupling relationship configuration record includes at least the source physical field identifier, target physical field identifier, source entity type, target entity type, correlation degree, balance coefficient, characteristic attenuation length, and applicable scenario. Using the source physical field identifier, target physical field identifier, source entity type, and target entity type as query conditions, the corresponding correlation degree, balance coefficient, and characteristic attenuation length are retrieved. The balance coefficient is used to adjust the contribution of different types of coupling relationships to the coupling strength. The characteristic attenuation length is used to convert distance into an attenuation ratio; the characteristic attenuation length is provided by historical operation records or the configuration library according to the applicable scenario. The balance coefficient, correlation degree, and connection medium status are all non-negative numbers. The characteristic attenuation length is greater than 0. When the calculated coupling strength exceeds 1, the coupling strength is truncated to 1. When the connection medium status is 0, the coupling strength is 0.

[0290] The connection medium status can be given directly from status data or from connection medium status configuration records in the configuration library. A connection medium status configuration record includes at least the connection medium identifier, source entity type, target entity type, connection medium status, and applicable scenario. The connection medium status is recorded in numerical form. When the connection medium is unavailable, the connection medium status value is 0; when the connection medium is available, the connection medium status value is greater than 0 and not greater than 1. The connection medium status is used in coupling strength calculations.

[0291] In some implementations, the distance between physical entities in two physical fields is determined according to the following formula:

[0292] ;

[0293] In the formula, Indicates the source node Corresponding physical entities and target nodes The distance between corresponding physical entities; , and Indicates the source node The coordinates of the physical entity's location; , and Represents the target node The physical coordinates of the entity; the physical coordinates are provided by the status data or configuration library.

[0294] In some implementations, the coupling strength is determined according to the following formula:

[0295] ;

[0296] In the formula, express From the source physical field at all times Source node in Pointing to the target physical field target node in The coupling strength; Represents the source physical field With the target physical field The balance coefficient between them; Indicates the source node With the target node The degree of correlation between them; express Time source node With the target node The state of the connection medium between them; Indicates the source node Corresponding physical entities and target nodes The distance between corresponding physical entities; Represents the source physical field With the target physical field The characteristic attenuation length between them.

[0297] After obtaining the initial coupling strength according to the coupling strength formula, range processing is performed on the initial coupling strength; when the initial coupling strength is less than 0, the coupling strength is set to 0; when the initial coupling strength is greater than 1, the coupling strength is set to 1; when the connection medium state is 0, the coupling strength is set to 0; the coupling strength after range processing is written into the coupling strength field of the coupling edge.

[0298] Specifically, when the source node state field and the target node state field are inconsistent in type, unit, or state dimension, the source node state change is first converted into a dimensionless state change under the corresponding state dimension of the target node according to the state field mapping relationship and coupling propagation parameter matrix in Example 6. Then, the target node state change amount is determined according to the coupling strength. For enumerated target states, the source state change range, coupling strength range, and connection medium state are jointly mapped to the target state code according to the coupling effect lookup table.

[0299] In some implementations, coupling propagation is performed according to the following formula:

[0300] ;

[0301] In the formula, express Time Source Physics Field Source node in The amount of change in node state; express Target physics field at any time target node in The amount of change in node state; express From the source physical field at all times Source node in Pointing to the target physical field target node in The coupling strength.

[0302] When multiple source nodes propagate to the same target node in a coupled manner, the change in the target node's state is determined according to the following formula:

[0303] ;

[0304] In the formula, This indicates that the target node is pointed to through the coupling edge. The set of source nodes; express Target physics field at any time target node in The amount of change in node state; Indicates the source node To the target node The coupling strength; Indicates the source node The change in node state.

[0305] Through the above steps, the coupling edge establishes state propagation relationships between nodes in different physical fields; the coupling strength is determined based on the correlation degree, distance, and state of the connecting medium; changes in the state of nodes in the source physical field can be transmitted to the target physical field through the coupling edge, and changes in the state of nodes in the target physical field are obtained; the coupling propagation results are written into the deduced state sequence and used as input for subsequent determination of deduced results and verification indicators.

[0306] Example 9:

[0307] This embodiment, based on Embodiment 1, provides a more detailed explanation of the process in step S106 for determining the verification indicators of candidate emergency plans based on the simulation results. See [link to previous embodiment]. Figure 9 The process of determining the verification indicators includes the following steps.

[0308] The entropy change index is determined based on the graph entropy change of the digital twin model in the deduced state sequence. To determine the entropy change index, each deduced state in the deduced state sequence is read; for each deduced state, the node states in the digital twin model are read; the node states are converted into node state probabilities; the graph entropy corresponding to each deduced time step is calculated based on the node state probabilities; the graph entropy change is calculated based on the graph entropy corresponding to each deduced time step, and the graph entropy change is written into the entropy change index.

[0309] Node state probability represents the normalized proportion of a node state within the same inferred state. When node states are numerical data, the node state is first read; then, it is converted into a non-negative state value; finally, the non-negative state value is divided by the sum of all non-negative state values ​​corresponding to all nodes to obtain the node state probability. When node states are enumerated data, they are first converted into state codes according to the state coding table, and then converted into non-negative state values ​​according to the risk level mapping record corresponding to the state field; both the state coding table and the risk level mapping record are provided by the configuration library; the risk level mapping record includes at least the state field, state code, risk level value, non-negative state value, and applicable entity type. When the sum of all non-negative state values ​​corresponding to all nodes is 0, the node state probability is set to the same value, and the sum of all node state probabilities is 1.

[0310] In some implementations, the graph entropy is determined according to the following formula:

[0311] ;

[0312] In the formula, Indicates the time of the deduction. The graph entropy of the corresponding digital twin model; Represents the set of nodes in a digital twin model; Represents the first node in the set of nodes. One node; Represents a node During the simulation The corresponding node state probability; Using the natural logarithm; when the state probability of any node... When, corresponding item The summation is performed using the limit value of 0, and the logarithmic operation of 0 is not performed directly.

[0313] In some implementations, the node state probability is determined according to the following formula:

[0314] ;

[0315] In the formula, Represents a node During the simulation The corresponding non-negative dimensionless state values; numerical node states are obtained by converting state normalization baseline values ​​or state value ranges; enumerated node states are obtained by converting state coding tables and risk level mapping records; To prevent smoothing values ​​with a denominator of 0, the smoothing value is provided by the graph entropy configuration record; when no smoothing value is used and the sum of all non-negative dimensionless state values ​​is 0, the state probability of each node is taken as... ; This represents the j-th node in the node set V; Represents a node During the simulation The corresponding non-negative dimensionless state values; the denominator of the node state probability formula is used to normalize all non-negative dimensionless state values ​​in the node set V.

[0316] In some implementations, the entropy change index is determined according to the following formula:

[0317] ;

[0318] In the formula, Indicates candidate emergency response plan The corresponding entropy change index; Indicates the number of deduction steps in the deduction state sequence; Indicates the time of the deduction. The graph entropy of the corresponding digital twin model; Indicates the time of the deduction. The graph entropy of the corresponding digital twin model.

[0319] The graph editing distance index is determined based on the graph editing distance between the simulation results and the set target security state. When determining the graph editing distance index, the final simulation state is read from the simulation results; the set target security state is read; the node differences, directed edge differences, edge weight differences, and coupled edge differences between the final simulation state and the set target security state are compared; and the graph editing distance index is determined based on the node differences, directed edge differences, edge weight differences, and coupled edge differences.

[0320] The target security status is set by the target security status configuration record in the configuration library. The target security status configuration record includes at least the applicable emergency scenario, physical field identifier, entity type, status field, target node status, target directed edge status, target edge weight field, target coupled edge status, and applicable contingency plan type; the corresponding set target security status is retrieved by using the applicable emergency scenario and applicable contingency plan type as query conditions.

[0321] In some implementations, the graph editing distance index is determined according to the following formula:

[0322] ;

[0323] In the formula, Indicates candidate emergency response plan The corresponding graph edit distance metric; This represents the set of editing operation sequences required to transition from the final simulated state in the simulation results to the set target safety state; This represents an edit operation sequence in the set of edit operation sequences; Indicates the sequence of edit operations The first in One editing operation; Indicates the first The editing cost corresponding to each editing operation.

[0324] It should be noted that editing operations include at least node state modification, directed edge addition, directed edge deletion, edge weight modification, coupled edge addition, coupled edge deletion, and coupling strength modification; the cost of each editing operation is given by the graph editing cost configuration record. For a specific implementation, the cost of adding or deleting a node or edge is set to 1, and the cost of modifying the node state, edge weight, and coupling strength is the absolute value of the difference between the dimensionless values ​​before and after the modification.

[0325] The response time metric is determined based on the time taken for the simulated states in the simulated state sequence to reach the safety recovery threshold or the safety state range. When determining the response time metric, the simulated states are read in chronological order from beginning to end. For each simulated state, the states of the nodes participating in the safety state determination and their corresponding safety recovery thresholds or safety state ranges are read. When all node states participating in the safety state determination meet their corresponding safety recovery thresholds or safety state ranges, the time interval between the corresponding simulated state time and the simulated start time is determined as the response time metric.

[0326] In this application, the set security thresholds include risk trigger thresholds, security recovery thresholds, and security state ranges; when used for secondary event detection, the set security thresholds refer to risk trigger thresholds; when used for response time index calculations, the set security thresholds refer to security recovery thresholds or security state ranges.

[0327] The safety threshold configuration record distinguishes between risk trigger thresholds and safety recovery thresholds; secondary event detection uses risk trigger thresholds; response time indicators use safety recovery thresholds or safety status ranges; the safety threshold configuration record must include at least the applicable emergency scenario, physical field identifier, entity type, status field, risk trigger threshold, safety recovery threshold, safety status range, comparison direction, and safety status determination flag.

[0328] The nodes participating in the security status determination are provided by the security threshold configuration record. The security status determination flag indicates whether the corresponding node status is included in the response time index calculation; the comparison direction is used to determine the judgment method for whether the node status meets the set security threshold.

[0329] In some implementations, the response time metric is determined according to the following formula:

[0330] ;

[0331] In the formula, Indicates candidate emergency response plan The corresponding response time metrics; Indicates the start time of the simulation; Represents the first state in the deduced state sequence. One simulation moment; This represents the set of nodes that participate in determining the security status. Represents a node The corresponding safety status range; Represents a node During the simulation The corresponding node status.

[0332] It should be noted that when there is no simulation moment in which all nodes participating in the security status determination simultaneously meet the corresponding security recovery threshold or security status range within the total simulation time, the response time index will be set to the total simulation time plus a preset simulation step size.

[0333] The resource consumption index is determined based on the cumulative value of resource consumption for the planned operations. When determining the resource consumption index, the planned operations in the candidate emergency plans are read; based on the operation type, target entity, and parameters in the planned operations, the corresponding unit resource consumption is read from the resource consumption configuration record; the individual resource consumption is determined based on the execution time length corresponding to the planned operation; and the individual resource consumption corresponding to all planned operations is added together to obtain the resource consumption index.

[0334] Resource consumption configuration records are provided by the configuration database. Each resource consumption configuration record includes at least the applicable emergency scenario, operation type, entity type, operation parameter range, unit resource consumption, standard resource equivalent, and applicable contingency plan type. Using the applicable emergency scenario, operation type, entity type, and applicable contingency plan type as query conditions, the corresponding unit resource consumption and standard resource equivalent are retrieved. The standard resource equivalent is used to represent the unified unit of measurement corresponding to the quantity of manpower, equipment, vehicles, power supply capacity, communication capacity, or materials.

[0335] In some implementations, the resource consumption configuration record also includes the resource type, original unit of measurement, and standard resource conversion factor. For personnel, vehicles, equipment, power supply capacity, communication capacity, or material quantity, the actual quantity used is first multiplied by the corresponding standard resource conversion factor to obtain the standard resource equivalent under the same dimensionless caliber; for continuously occupied resources, it is then multiplied by the corresponding execution time length; for one-time consumable resources, the execution time length is not multiplied. The standard resource equivalents corresponding to each plan operation are added together to obtain the resource consumption index.

[0336] In some implementations, the resource consumption index is determined according to the following formula:

[0337] ;

[0338] In the formula, Indicates candidate emergency response plan Corresponding resource consumption indicators; Indicates candidate emergency response plan The first in Each contingency plan operation; Indicates contingency plan operation Corresponding unit resource consumption; Indicates contingency plan operation The corresponding execution time; the unit resource consumption and execution time are given by the resource consumption configuration record or the candidate contingency plan.

[0339] Through the above steps, the simulation results are converted into entropy change indicators, graph edit distance indicators, response time indicators, and resource consumption indicators. The entropy change indicator is used to record the graph entropy change in the simulation state sequence. The graph edit distance indicator is used to record the state difference between the simulation results and the set target safety state. The response time indicator is used to record the time to reach the safety recovery threshold or enter the safety state range. The resource consumption indicator is used to record the cumulative value of resource consumption corresponding to the contingency plan operation.

[0340] Example 10:

[0341] This embodiment, based on Embodiment 9, provides a more detailed explanation of the process for determining the verification results corresponding to candidate emergency plans based on verification indicators. See also... Figure 10 The process for determining the verification results includes the following steps.

[0342] First, based on the verification indicators and their corresponding reference benchmark values, each verification indicator is dimensionless, resulting in dimensionless verification indicators. The reference benchmark values ​​are provided by the verification scoring configuration records in the configuration library; these records include at least the applicable emergency scenario, applicable contingency plan type, entropy change reference benchmark value, graph edit distance reference benchmark value, response time reference benchmark value, resource consumption reference benchmark value, weight, and applicable conditions.

[0343] When retrieving reference benchmark values, the applicable emergency scenarios and applicable plan types corresponding to the candidate emergency plans are used as query conditions to retrieve the corresponding reference benchmark values ​​from the verification scoring configuration records. Reference benchmark values ​​are determined based on historical simulation records, target safety status configuration records, emergency response rules, or manual configuration records; these reference benchmark values ​​are used to convert verification indicators from different units into dimensionless verification indicators.

[0344] During dimensionless processing, the entropy change index is divided by the entropy change reference value to obtain the dimensionless entropy change index; the graph editing distance index is divided by the graph editing distance reference value to obtain the dimensionless graph editing distance index; the response time index is divided by the response time reference value to obtain the dimensionless response time index; and the resource consumption index is divided by the resource consumption reference value to obtain the dimensionless resource consumption index.

[0345] When any reference baseline value is 0, division calculation is not performed directly, and the corresponding backup reference baseline value is read from the verification score configuration record. The backup reference baseline value is provided by the configuration library and is greater than 0. When there is no backup reference baseline value in the verification score configuration record, a verification score review record is generated and the determination of the comprehensive verification score of the candidate emergency plan is stopped.

[0346] Subsequently, based on the dimensionless verification indicators and their corresponding weights, the comprehensive verification score of the candidate emergency response plan is determined. The weights are provided by the verification score configuration record; the weights include at least the entropy change indicator weight, graph edit distance indicator weight, response time indicator weight, and resource consumption indicator weight; the sum of the values ​​of each weight is 1; each weight is used to represent the proportion of each verification indicator in the comprehensive verification score.

[0347] The weighting method is as follows: the verification target is determined according to the applicable emergency scenario and the applicable plan type; when the verification target emphasizes state change control, the weight of the entropy change index is increased; when the verification target emphasizes reaching the target safety state, the weight of the graph editing distance index is increased; when the verification target emphasizes disposal time control, the weight of the response time index is increased; when the verification target emphasizes resource usage control, the weight of the resource consumption index is increased; the weights are written into the verification scoring configuration record and then invoked by the emergency plan simulation verification equipment.

[0348] The comprehensive validation score is used to represent the validation results of candidate emergency plans under multiple validation indicators. The calculation direction of the comprehensive validation score is as follows: the smaller the dimensionless entropy change indicator, the larger the corresponding score component; the smaller the dimensionless graph edit distance indicator, the larger the corresponding score component; the smaller the dimensionless response time indicator, the larger the corresponding score component; and the smaller the dimensionless resource consumption indicator, the larger the corresponding score component.

[0349] In some implementations, the comprehensive verification score is determined according to the following formula:

[0350] ;

[0351] In the formula, Indicates candidate emergency response plan The corresponding comprehensive verification score; Indicates candidate emergency response plan The corresponding entropy change index; Indicates candidate emergency response plan The corresponding graph edit distance metric; Indicates candidate emergency response plan The corresponding response time metrics; Indicates candidate emergency response plan Corresponding resource consumption indicators; This represents the reference value for entropy change. This indicates the reference baseline value for the distance to the image editing section. This indicates the reference value for response time; This represents a reference baseline value for resource consumption. Indicates the weight of the entropy change index; This indicates the weight of the graph editing distance indicator; Indicates the weight of the response time metric; This indicates the weight of resource consumption indicators.

[0352] Finally, the verification result is determined based on the comprehensive verification score. When there is only one candidate emergency plan, the verification indicators of the candidate emergency plan, the dimensionless verification indicators, the comprehensive verification score, and the candidate emergency plan identifier are all written into the verification result; the verification result is sent to the emergency command terminal and written into the verification result record database; the verification result record database is used for subsequent querying, review, or re-simulation.

[0353] When there are multiple candidate emergency response plans, they are ranked according to their comprehensive verification scores. The ranking process is performed in descending order of comprehensive verification scores; when two candidate emergency response plans have the same comprehensive verification score, they are ranked in ascending order of response time indicators; when response time indicators are the same, they are ranked in ascending order of resource consumption indicators.

[0354] The ranking results include at least the candidate emergency plan identifier, comprehensive verification score, ranking position, entropy change index, graph edit distance index, response time index, and resource consumption index. The ranking results are used to display the verification comparison results of multiple candidate emergency plans to the emergency command center; the ranking results are also used to determine the candidate emergency plans that meet preset selection criteria from among the multiple candidate emergency plans.

[0355] The preset selection criteria are provided by the verification selection configuration records in the configuration library. Each verification selection configuration record includes at least the applicable emergency scenario, applicable plan type, minimum comprehensive verification score, maximum allowable response time, maximum allowable resource consumption, and number of selections. When the comprehensive verification score of a candidate emergency plan is greater than or equal to the minimum comprehensive verification score, and the response time indicator is not greater than the maximum allowable response time, and the resource consumption indicator is not greater than the maximum allowable resource consumption, the candidate emergency plan is determined to meet the preset selection criteria. When the number of candidate emergency plans meeting the preset selection criteria exceeds the number of selections, candidate emergency plans are selected from the top to the bottom according to their sorting position, up to the specified number, and written into the verification results.

[0356] The ranking results, or the ranking results and candidate emergency plans that meet the preset selection criteria, are determined as the verification results. The verification results include at least the candidate emergency plan identifier, verification indicators, comprehensive verification score, ranking results, and selection flags; the selection flags are used to indicate whether the candidate emergency plan meets the preset selection criteria; the verification results are written to the verification result record database and sent to the emergency command terminal.

[0357] In some implementations, when no candidate emergency plan meets the preset selection criteria, a verification failure flag is generated. The verification failure flag includes at least a candidate emergency plan identifier, a comprehensive verification score, a reason for failure field, and a review prompt field. The reason for failure field is determined based on a comparison of the lowest comprehensive verification score, the maximum allowable response time, or the maximum allowable resource consumption. The review prompt field is used to prompt the emergency command center to reselect a candidate emergency plan or adjust the plan operation.

[0358] Through the above steps, entropy change index, graph edit distance index, response time index, and resource consumption index are converted into a comprehensive verification score under a unified scoring caliber; multiple candidate emergency plans can generate a ranking result according to the comprehensive verification score; the ranking result and the candidate emergency plans that meet the preset selection conditions together form the verification result, which is used to support the emergency command terminal in comparing and selecting plans.

[0359] Example 11:

[0360] This embodiment, based on Embodiment 1, explains the coordination process between real-time status synchronization and contingency plan simulation. See also... Figure 11 The coordination process between real-time status synchronization and contingency plan simulation includes the following steps.

[0361] When updating the digital twin model based on state data, the process of updating the digital twin model based on state data is defined as real-time state synchronization. Real-time state synchronization takes state data as input; after fusing the state data, it updates the node states, directed edges, or coupled edges in the digital twin model; after the update is completed, a snapshot of the twin state at the current moment is obtained; the snapshot of the twin state at the current moment is used to record the model state corresponding to the emergency scenario at the current moment.

[0362] After obtaining candidate emergency plans, the process of generating a deduced state sequence based on graph transformation operations is defined as emergency plan deduction. Emergency plan deduction takes a twin state snapshot and candidate emergency plans as input; it generates graph transformation operations based on the plan operations in the candidate emergency plans; it executes graph transformation operations, state propagation, and coupling propagation according to the deduction time; it generates the deduced state corresponding to each deduction time, and arranges the deduced states according to the deduction time to obtain the deduced state sequence.

[0363] Real-time status synchronization corresponds to passive mode; passive mode is used to update the digital twin model and generate a twin status snapshot when status data changes; contingency plan simulation corresponds to active mode; active mode is used to generate a simulation status sequence based on the twin status snapshot and graph transformation operation after obtaining candidate emergency plans; handling after deviations meet preset coordination conditions corresponds to coordination mode; coordination mode is used to obtain a new twin status snapshot based on the new status data and regenerate the simulation status sequence based on the new twin status snapshot.

[0364] During the contingency plan simulation, new status data continues to be acquired. The new status data is provided by the status data source after the contingency plan simulation begins; the new status data includes at least entity identifier, collection time, status field, status value, and status unit; the digital twin model is updated according to the new status data in the manner described in Example 3 to obtain a new twin status snapshot; the new twin status snapshot is used to represent the model status of the emergency scenario at the collection time corresponding to the new status data.

[0365] When the deviation between the projected state in the projected state sequence and the twin state snapshot updated based on the new state data meets the preset coordination conditions, the digital twin model is updated based on the new state data to obtain a new twin state snapshot, and the projected state sequence is regenerated based on the new twin state snapshot. The deviation is used to represent the difference between the model state obtained from the contingency plan at the same time position and the model state obtained from the update of the new state data.

[0366] When determining the deviation, first determine the corresponding simulation time based on the acquisition time of the new state data; then read the simulation state at the corresponding simulation time from the simulation state sequence; then read the node state, the edge weight field corresponding to the directed edge, and the coupling strength field corresponding to the coupled edge in the new twin state snapshot; finally, compare the differences in node state, edge weight, and coupling strength between the simulation state and the new twin state snapshot to obtain the deviation value.

[0367] The preset coordination conditions are given by the coordination configuration records in the configuration library. The coordination configuration records include at least the applicable emergency scenario, the applicable plan type, the node state deviation threshold, the edge weight deviation threshold, the coupling strength deviation threshold, the graph distance threshold, and the coordination processing method; using the applicable emergency scenario and applicable plan type corresponding to the candidate emergency plan as the query conditions, the corresponding node state deviation threshold, edge weight deviation threshold, coupling strength deviation threshold, graph distance threshold, and coordination processing method are read.

[0368] When the difference in node state is greater than or equal to the node state deviation threshold, the deviation is determined to meet the preset coordination condition; when the difference in edge weight is greater than or equal to the edge weight deviation threshold, the deviation is determined to meet the preset coordination condition; when the difference in coupling strength is greater than or equal to the coupling strength deviation threshold, the deviation is determined to meet the preset coordination condition; when the deviation value is greater than or equal to the graph distance threshold, the deviation is determined to meet the preset coordination condition.

[0369] When the deviation meets the preset coordination conditions, state coordination is performed according to the coordination handling method. State coordination includes stopping the subsequent generation of the current simulated state sequence; using the new twin state snapshot as the initial state for re-simulation; retaining the unexecuted contingency plan operations in the candidate emergency plans; and regenerating the simulated state sequence based on the new twin state snapshot, the unexecuted contingency plan operations, and the event queue.

[0370] When the deviation does not meet the preset coordination conditions, the generation of the current simulation state sequence is not stopped. A new twin state snapshot is written to the synchronization record library; the synchronization record library includes at least the acquisition time, entity identifier, state field, state value, deviation value, and corresponding simulation time; the synchronization record library is used for subsequent queries and does not change the current simulation state sequence.

[0371] In some implementations, the deviation value is determined according to the following formula:

[0372]

[0373] In the formula, This indicates the state in the deduced state sequence at the deduced time. Next node Dimensionless node states; This indicates the new twin state snapshot at the extrapolation time. Next node Dimensionless node states; , These represent the dimensionless values ​​of the corresponding directed edge weight fields; , These represent the dimensionless values ​​of the corresponding coupling strength field. Indicates the time of the deduction. The corresponding deviation value; V represents the set of nodes; E represents the set of directed edges; Represents the set of coupling edges; , and These represent the node state difference weight, the directed edge weight difference weight, and the coupling strength difference weight, respectively. , and All three are non-negative and their sum is 1; the node state difference weight, directed edge weight difference weight, and coupling strength difference weight are given by the coordination configuration record; when the coordination configuration record does not specify three weights, all three weights are set to 1 / 3; when the coordination configuration record gives three original weights, each original weight is divided by the sum of the three original weights to obtain the corresponding weight; when the sum of the three original weights is 0, all three weights are set to 1 / 3.

[0374] In some implementations, the preset coordination conditions are expressed according to the following formula:

[0375] ;

[0376] In the formula, Indicates the time of the deduction. The corresponding deviation value; The graph distance threshold is represented by the coordinate configuration record. When the formula holds, the digital twin model is updated based on the new state data to obtain a new twin state snapshot, and the inferred state sequence is regenerated based on the new twin state snapshot.

[0377] Through the above steps, real-time state synchronization is used to update the digital twin model based on state data; contingency plan simulation is used to generate a simulation state sequence based on graph transformation operations; when the deviation between the simulation state in the simulation state sequence and the new twin state snapshot meets the preset coordination conditions, the simulation state sequence is regenerated based on the new twin state snapshot, so that the simulation state sequence can be updated according to the model state corresponding to the new state data.

[0378] Example 12:

[0379] This embodiment, based on Embodiment 3, describes the incremental update process in updating the digital twin model according to state data. See also... Figure 12The incremental update process includes the following steps.

[0380] First, obtain the amount of data change between the current state data and the previous state data. When obtaining the amount of data change, read the current state data according to the entity identifier and the state field; then read the previous state data according to the same entity identifier and the same state field; when both the current state data and the previous state data exist, calculate the difference between the current state value and the previous state value, and determine the absolute value of the difference as the amount of data change.

[0381] When the current state data exists but the previous state data does not exist, the physical entity corresponding to the current state data is identified as the physical entity that needs to be added to the incremental update queue; when the current state data does not exist but the previous state data exists, the physical entity corresponding to the previous state data is identified as the physical entity that needs to be added to the incremental update queue; when neither the current state data nor the previous state data exists, no data change is generated.

[0382] In some implementations, the amount of data change is determined according to the following formula:

[0383] ;

[0384] In the formula, Indicates the first A physical entity in The amount of data change at any given time; Indicates the first A physical entity in The state value in the state data at any given time; Indicates the first The state value of each physical entity in the state data of the previous moment; the data change is used to determine whether to update the state value of the first physical entity. Each physical entity is added to the incremental update queue.

[0385] When the data change does not meet the set incremental update conditions, no update is performed on the corresponding physical entity. During processing, the node state of the corresponding physical entity in the digital twin model remains unchanged; the directed edges and coupling edges associated with the corresponding physical entity are not updated; the entity identifier, state field, data change amount, current time, and unupdated flag of the corresponding physical entity are written into the unupdated state record.

[0386] When the data change meets the set incremental update conditions, the corresponding physical entity is added to the incremental update queue, and the digital twin model is updated based on the incremental update queue. During processing, an incremental update queue record is first generated; the incremental update queue record includes at least the entity identifier, entity type, status field, current status value, previous status value, data change amount, update priority, and current time; then the incremental update queue record is written to the incremental update queue.

[0387] The incremental update conditions are set by the incremental update condition configuration records in the configuration library. The incremental update condition configuration records include at least the applicable emergency scenario, entity type, status field, data change threshold, update priority, directed edge update flag, coupled edge update flag, and applicable conditions; using the applicable emergency scenario, entity type, and status field as query conditions, the corresponding data change threshold, update priority, directed edge update flag, and coupled edge update flag are retrieved.

[0388] When the data change is greater than or equal to the data change threshold, it is determined that the data change meets the set incremental update conditions; when the data change is less than the data change threshold, it is determined that the data change does not meet the set incremental update conditions. The data change threshold is determined based on the acquisition accuracy of the status field, the unit of the status field, and the emergency scenario configuration record. After the data change threshold is written into the incremental update condition configuration record, it is called by the emergency plan simulation and verification equipment when updating the digital twin model.

[0389] When updating the digital twin model based on the incremental update queue, the incremental update queue record is read according to the update priority; the node in the digital twin model is found according to the entity identifier and status field in the incremental update queue record; the current status value is written to the node status of the corresponding node; when the directed edge update mark indicates that the directed edge needs to be updated, the influence relationship configuration record associated with the corresponding physical entity is read, and the edge weight field of the directed edge is updated according to the influence relationship configuration record.

[0390] When the coupling edge update flag indicates that the coupling edge needs to be updated, the coupling relationship configuration record associated with the corresponding physical entity is read, and the coupling strength field of the coupling edge is updated according to the coupling relationship configuration record. After completing the node state, directed edge, or coupling edge update, the processing status of the incremental update queue record is written as updated; when all incremental update queue records in the incremental update queue have been processed, a snapshot of the twin state at the current moment is generated according to the updated digital twin model.

[0391] In some implementations, the incremental update determination is made according to the following formula:

[0392] ;

[0393] In the formula, Indicates the first A physical entity in Whether to add to the incremental update queue at any time; Indicates the first Each physical entity is added to the incremental update queue; Indicates the first Each physical entity is not added to the incremental update queue; Indicates the first A physical entity in The amount of data change at any given time; Indicates the first The data change thresholds for each physical entity are provided by the incremental update condition configuration record.

[0394] In some implementations, when the status data includes multiple status fields, the amount of data change is determined according to the following formula:

[0395] ;

[0396] In the formula, Indicates the first A physical entity in Changes in multi-field data over time; Indicates the first A set of status fields corresponding to each physical entity; This represents a state field in the state field set; Indicates the first Status fields of physical entities exist The state value at any given time; Indicates the first Status fields of physical entities The state value at the previous moment; Represents the status field The corresponding field weights; field weights are given by the incremental update condition configuration record.

[0397] When using changes in multiple fields of data for judgment, the incremental update judgment is determined according to the following formula:

[0398] ;

[0399] In the formula, Indicates the first A physical entity in Whether to add to the incremental update queue at any time; Indicates the first A physical entity in Changes in multi-field data over time; Indicates the first The threshold for changes in multi-field data corresponding to each physical entity; the threshold for changes in multi-field data is given by the incremental update condition configuration record.

[0400] By following the steps above, when updating the digital twin model, the amount of data change is first determined based on the current state data and the previous state data; then, the physical entities that need to be updated are selected according to the set incremental update conditions; then, only the physical entities that meet the set incremental update conditions are added to the incremental update queue; finally, the node states, directed edges, or coupled edges in the digital twin model are updated based on the incremental update queue.

[0401] Example 13:

[0402] This embodiment, based on Embodiment 6, describes the pre-calculation processing before generating the deduced state sequence. See also... Figure 13 The pre-calculation process includes the following steps.

[0403] Before generating the simulated state sequence, obtain the propagation path index, coupling effect lookup table, or state template corresponding to the emergency scenario. The propagation path index, coupling effect lookup table, and state template can be provided by the configuration library, digital twin model storage unit, historical simulation records, or pre-calculation task records; the propagation path index, coupling effect lookup table, and state template all correspond to the applicable emergency scenario, digital twin model version, physical field identifier, and applicable contingency plan type.

[0404] The propagation path index is used to record the state propagation paths between nodes in the digital twin model. The propagation path index includes at least the applicable emergency scenario, digital twin model version, physical field identifier, starting node identifier, ending node identifier, path node sequence, path directed edge sequence, path length, and path propagation coefficient; the path propagation coefficient is determined based on the edge weight field in the path directed edge sequence; the propagation path index is used to determine the ending node that a node's state change can reach during state propagation.

[0405] When using the propagation path index, when a graph transformation operation is applied to a node, that node is used as the starting node; when a graph transformation operation is applied to a directed edge, the starting node of that directed edge is used as the starting node; when a graph transformation operation is applied to an object whose graph structure is modified, the starting node is used as the node that is newly added, replaced, or affected by changes in edge weights. The path record corresponding to the starting node is queried in the propagation path index; the node that needs to be updated is determined based on the endpoint node identifier in the path record; the endpoint node's state change is determined based on the path propagation coefficient and the node state change of the starting node; and a line-by-line search is no longer performed on all directed edges in the digital twin model.

[0406] The coupling effect lookup table is used to record the propagation results of state changes between different physical fields. The coupling effect lookup table includes at least the applicable emergency scenario, digital twin model version, source physical field identifier, target physical field identifier, source entity type, target entity type, source state change range, coupling strength range, connection medium state range, and target state change value; the coupling effect lookup table is used to determine the amount of node state change in the target physical field during coupling propagation.

[0407] When using the coupling effect lookup table, first read the node state change of the source node, the coupling strength field in the coupled edge, and the connection medium state; then query the coupling effect lookup table according to the source physical field identifier, target physical field identifier, source entity type, and target entity type; then match the node state change of the source node, the coupling strength field, and the connection medium state to the corresponding ranges; finally, read the corresponding target state change value and write the target state change value into the node state of the target node.

[0408] State templates are used to record the state propagation results under preset state combinations. A state template must include at least the applicable emergency scenario, digital twin model version, physical field identifier, operation type, entity type, state field, initial state range, operation parameter range, template deduction steps, and template state change results. State templates are used to directly read the template state change results when the node state and operation parameters at the deduction time fall within the corresponding range.

[0409] When using a state template, first read the operation type, target entity, operation parameters, and node status corresponding to the current simulation moment; then query the state template based on the operation type, entity type, and status field; then determine whether the node status is within the range of the initial state and whether the operation parameters are within the range of the operation parameters; when both determinations are true, read the template state change result and use the template state change result as the state propagation result for the current simulation moment.

[0410] When generating the deduced state sequence, the state propagation result or coupling propagation result is determined based on the propagation path index, coupling effect lookup table, or state template. For state propagation within the physical field, the propagation path index is queried first; when a corresponding path record exists in the propagation path index, the endpoint node and node state change amount are determined based on the path record; when no corresponding path record exists in the propagation path index, the state propagation method in Example 6 is followed.

[0411] For coupling propagation between different physical fields, the coupling effect lookup table is queried first. If a table record exists in the coupling effect lookup table corresponding to the source physical field identifier, target physical field identifier, source entity type, target entity type, source state change range, coupling strength range, and connection medium state range, the target state change value in the table record is read; if no corresponding table record exists in the coupling effect lookup table, the coupling propagation method in Example 8 is followed.

[0412] For deduction moments where template records already exist for state propagation results, the state template is queried first. If a template record corresponding to the operation type, entity type, state field, initial state range, and operation parameter range exists in the state template, the template state change result is read; if no corresponding template record exists in the state template, the state propagation method in Example 6 is followed.

[0413] In some implementations, the propagation path index is established as follows: For each physical field's corresponding subgraph, the node set, directed edge set, and edge weight set in the subgraph are read; taking each node as the starting node, a path is found that does not contain duplicate nodes and whose path length does not exceed a preset maximum propagation order; the preset maximum propagation order is given by the propagation path configuration record, which includes at least the applicable emergency scenario, physical field identifier, entity type, state field, preset maximum propagation order, and applicable contingency plan type; the node identifiers in the path are written into the path node sequence, the directed edge identifiers in the path are written into the path directed edge sequence, the number of directed edges in the path is written into the path length, and the product of the edge weight fields in the path is written into the path propagation coefficient.

[0414] The edge weight field involved in the path propagation coefficient calculation is first converted into non-negative normalized edge weights according to the propagation configuration record. The value range of the normalized edge weight is 0 to 1. When there are multiple paths from the same starting node to the same ending node, the path propagation coefficient is selected from largest to smallest. Merge the paths. Provided by the propagation path configuration record.

[0415] In some implementations, the path propagation coefficient is determined according to the following formula:

[0416] ;

[0417] In the formula, Representing a path The corresponding path propagation coefficient; This represents a path from the starting node to the ending node; Representing a path One of the directed edges in the middle; Indicates from node Pointing to node The edge weight field corresponds to the directed edge; the path propagation coefficient is used to determine the amount of node state change received by the endpoint node during state propagation.

[0418] When merging multiple paths, the change in the state of the endpoint node is the sum of the changes in the state of each path. When the merged result exceeds the upper limit of the target state field value, it is truncated according to the target state field value range.

[0419] In some implementations, the change in the state of the endpoint node is determined according to the propagation path index using the following formula:

[0420] ;

[0421] In the formula, Indicates the time of the deduction. Next endpoint node The amount of change in node state; Indicates the time of the deduction. Next starting node The amount of change in node state; Indicates the starting node To the destination node Corresponding path The path propagation coefficient.

[0422] In some implementations, the coupling effect lookup table is established as follows: The source physical field identifier, target physical field identifier, source entity type, target entity type, coupling strength, and connection medium status are read from the coupling relationship configuration record; a source state change range is generated according to a set state change range division rule; a coupling strength range is generated according to a set coupling strength range division rule; the state change range and coupling strength range are divided according to equal width division, historical quantile division, or a boundary specified by the configuration library; when equal width division is used, the representative value is the midpoint of the interval; when historical quantile division is used, the representative value is the mean of historical samples within the interval; the target state change value is calculated based on the source state change range, coupling strength range, and connection medium status, and written into the coupling effect lookup table.

[0423] In some implementations, the target state change value in the coupling effect lookup table is determined according to the following formula:

[0424] ;

[0425] In the formula, This represents the target state change value in the coupling effect lookup table; This indicates the range of coupling strengths corresponding to the representative coupling strengths. This represents the representative state change corresponding to the range of source state changes; the representative coupling strength and the representative state change are given by the coupling effect lookup table generation rules; the coupling effect lookup table generation rules include at least the partitioning method, the number of ranges, the range boundaries, the representative value calculation method, and the applicable scenarios; when the partitioning method is equal-width partitioning, the representative coupling strength and the representative state change are both taken from the midpoint of the corresponding interval; when the partitioning method is historical quantile partitioning, the representative coupling strength and the representative state change are both taken from the historical sample mean within the corresponding interval; when the partitioning method is partitioning by the boundary specified in the configuration library, the representative coupling strength and the representative state change are determined according to the representative value field in the configuration library.

[0426] In some implementations, the state template is established as follows: read simulation segments corresponding to the same applicable emergency scenario, the same physical field identifier, the same operation type, and the same entity type from historical simulation records; classify the starting node state in the simulation segment into the starting state range; classify the operation parameters in the simulation segment into the operation parameter range; write the node state change results output by the simulation segment into the template state change results; and write the template simulation steps, starting state range, operation parameter range, and template state change results together into the state template.

[0427] By following the above steps, before generating the deduced state sequence, the propagation path index, coupling effect lookup table, or state template can be obtained in advance. When generating the deduced state sequence, the state propagation result within the physical field can be determined according to the propagation path index, the coupling propagation result between different physical fields can be determined according to the coupling effect lookup table, and the template state change result at the corresponding deduced time can be determined according to the state template. When there is no corresponding record, the propagation method in Examples 6 and 8 is still executed.

[0428] Example 14:

[0429] This embodiment, based on Embodiment 6, describes the parallel execution process of simulation tasks corresponding to multiple candidate emergency plans. See also... Figure 14 The execution process of the batch simulation task includes the following steps.

[0430] When there are multiple candidate emergency plans, the simulation tasks corresponding to multiple candidate emergency plans are merged into a batch simulation task. During processing, the candidate emergency plan identifier, twin state snapshot, event queue, preset simulation step size, total simulation duration, and simulation configuration record corresponding to each candidate emergency plan are read; multiple simulation tasks under the same emergency scenario, using the same twin state snapshot, and using the same preset simulation step size are written into the same batch simulation task.

[0431] A batch simulation task should include at least a batch simulation task identifier, an emergency scenario identifier, a twin state snapshot identifier, a preset simulation step size, a total simulation duration, a number of candidate emergency plans, and a batch simulation task record. The batch simulation task identifier is used to distinguish different batch simulation tasks; the twin state snapshot identifier is used to determine the initial state used by multiple candidate emergency plans; and the number of candidate emergency plans is used to record the number of candidate emergency plans participating in the simulation in a batch simulation task.

[0432] Batch simulation task records are used to establish the correspondence between candidate emergency plans and simulation tasks. Each batch simulation task record includes at least the candidate emergency plan identifier, simulation task number, event queue identifier, graph transformation operation list, simulation state sequence identifier, and task status field; the candidate emergency plan identifier corresponds to one candidate emergency plan; the simulation task number corresponds to the simulation task of one candidate emergency plan; and the event queue identifier is used to read the event queue of the corresponding candidate emergency plan.

[0433] When generating batch simulation tasks, a simulation task number is generated for each candidate emergency plan. The simulation task number is generated according to the batch simulation task identifier and the order in which the candidate emergency plans are arranged in the batch simulation tasks; each simulation task number in the same batch simulation task corresponds to only one candidate emergency plan; the same candidate emergency plan corresponds to only one simulation task number in the same batch simulation task.

[0434] Subsequently, batch simulation tasks are executed in parallel to generate simulation state sequences corresponding to each candidate emergency plan. When executing batch simulation tasks, the batch simulation task records are read according to the simulation task number; for each simulation task number, the corresponding event queue and graph transformation operation list are read; using the same twin state snapshot as the initial state, the simulation state sequence of the corresponding candidate emergency plan is generated in accordance with the method in Example 6.

[0435] During parallel execution, the simulation state sequence corresponding to each simulation task number is stored separately. During storage, the candidate emergency plan identifier, simulation task number, simulation time, simulation state, state change amount, and state propagation number are written into the corresponding simulation state sequence record; the simulation state sequence record is bound to the candidate emergency plan identifier; when determining the verification indicators later, the corresponding simulation state sequence is read according to the candidate emergency plan identifier.

[0436] In some implementations, parallel execution can be achieved using multiprocessor resources, graph computing resources, or batch processing mechanisms; graph computing resources may include GPU computing resources. Before execution, the simulation task number allocated to each computing resource is determined based on the number of candidate contingency plans and the number of available computing resources in the batch simulation task; during execution, each computing resource reads the event queue under the corresponding simulation task number; after execution, each computing resource outputs the simulation status sequence of the corresponding candidate contingency plan.

[0437] Before generating the node state batch matrix and adjacency batch matrix, a unified node order and a unified matrix dimension are established according to the same digital twin model version. For a node or directed edge that does not exist in a candidate emergency plan, 0 is filled in the corresponding matrix position, and the valid nodes and valid edges of the candidate emergency plan are recorded through the task mask. When a graph structure modification operation results in the addition of a node, the new node is written into the batch expansion node set, and 0 is filled in the inference tasks that do not involve the new node. When a new node or a new directed edge is generated at a certain inference time, the node state batch matrix, adjacency batch matrix, and task mask are regenerated according to the batch expansion node set starting from the next state propagation at that inference time. The propagation results completed at the current inference time remain unchanged.

[0438] When performing batch simulation tasks in parallel, the node state sets corresponding to multiple candidate emergency plans are arranged according to the simulation task number to generate a node state batch matrix; the normalized adjacency matrices corresponding to multiple candidate emergency plans are arranged according to the simulation task number to generate an adjacency batch matrix; the GPU computing resources perform matrix propagation calculations on the node state batch matrix and the adjacency batch matrix to obtain the simulation state sequence corresponding to each candidate emergency plan; the simulation state sequence is written into the digital twin model storage unit according to the candidate emergency plan identifier.

[0439] In some implementations, when the number of candidate emergency plans exceeds the available computing resources, the batch simulation task is split into multiple execution rounds. Each execution round includes a simulation task number not exceeding the number of available computing resources; the simulation task number in the current execution round is executed first; after the simulation state sequence in the current execution round is generated, the simulation task number in the next execution round is executed; after all execution rounds are completed, the simulation state sequence corresponding to each candidate emergency plan is obtained.

[0440] In some implementations, the batch simulation task can be represented as In the formula, Indicates batch simulation tasks; Indicates the first Each simulation task number corresponds to a simulation task; This indicates the number of candidate emergency response plans in each batch of simulation tasks; each simulation task Each corresponds to a candidate emergency response plan.

[0441] In some implementations, the correspondence between simulation tasks and candidate emergency plans is expressed as follows: In the formula, Indicates the first Each simulation task number corresponds to a simulation task; Indicates the first There are 10 candidate emergency response plans; this correspondence is used to find the candidate emergency response plan based on the simulation task number, and to find the corresponding simulation state sequence based on the candidate emergency response plan identifier.

[0442] In some implementations, the output of the batch simulation task is represented as In the formula, This represents the set of simulation state sequences output by the batch simulation task; Indicates the first One candidate emergency plan The corresponding deduced state sequence; This indicates the number of candidate emergency response plans in the batch simulation task; the simulation state sequence set is used to subsequently determine the verification indicators corresponding to each candidate emergency response plan.

[0443] Through the above steps, the simulation tasks corresponding to multiple candidate emergency plans are merged into batch simulation tasks. The batch simulation tasks establish a correspondence between candidate emergency plans and simulation state sequences through simulation task numbers. After executing the batch simulation tasks in parallel, the simulation state sequences corresponding to each candidate emergency plan can be obtained. Each simulation state sequence serves as the input for subsequent determination of verification indicators, comprehensive verification scores, and ranking results.

[0444] Example 15:

[0445] This embodiment, based on Embodiment 1, describes the implementation method for an emergency scenario involving urban rail transit. See also... Figure 15 In this embodiment, the emergency scenario is an urban rail transit emergency scenario; the urban rail transit emergency scenario is used to represent the operational scenario in which the emergency plan needs to be simulated and verified after an abnormal event occurs in the station, section, line, vehicle, traction power supply equipment, communication equipment, turnout equipment, platform area or transfer area.

[0446] Status data includes power supply status data, train operation status data, pedestrian flow status data, and communication status data. Power supply status data originates from the traction power supply monitoring system, substation monitoring equipment, or contact rail monitoring equipment; power supply status data includes at least the power supply section identifier, contact rail section identifier, switch status, voltage value, current value, power failure status, fault location, and acquisition time; power supply status data is used to generate node states and directed edges in the power supply physical field.

[0447] Train operation status data comes from the automatic train monitoring system, train control system, or dispatching and command system; train operation status data includes at least train identification, track section identification, direction of travel, train speed, station status, section occupancy status, departure interval, turnout position status, and acquisition time; train operation status data is used to generate node status and directed edges in the train operation physical field, and is used to determine the impact of planned operations on train operation.

[0448] The pedestrian flow status data comes from the passenger flow statistics system, the turnstile system, the video passenger flow statistics equipment, or the platform area detection equipment; the pedestrian flow status data includes at least station signs, area signs, number of people entering the station, number of people exiting the station, number of people transferring, number of people in the area, population density, evacuation direction, and collection time; the pedestrian flow status data is used to generate the node status and directed edges in the pedestrian flow physical field, and is used to determine whether the inferred status meets the set safety threshold.

[0449] Communication status data originates from communication network monitoring systems, dispatch communication systems, vehicle-to-ground communication systems, or broadcasting systems. Communication status data includes at least communication node identifiers, communication link identifiers, link status, bandwidth occupancy, communication latency, device online status, broadcast coverage status, and acquisition time. Communication status data is used to generate node status and directed edges in the communication physical field and to determine whether emergency response instructions and evacuation information are ready for transmission.

[0450] The candidate emergency plans include the contact rail power failure handling plan, the train failure handling plan, and the turnout failure handling plan. The contact rail power failure handling plan is applicable to scenarios where the power supply to the contact rail section is interrupted, the voltage of the contact rail section is abnormal, or the contact rail section protection is activated. The planned operations in the contact rail power failure handling plan include disconnecting the faulty power supply section, adjusting the power supply switch status, restricting trains from entering the power failure section, organizing the evacuation of the section, or issuing dispatch communication instructions.

[0451] The train malfunction handling plan is applicable to scenarios where trains cannot be pulled, trains brake abnormally, trains are stuck in sections, or trains stop abnormally. The plan operations include setting the train running status, adjusting the train stopping status, adjusting the section occupancy status, modifying the running interval of subsequent trains, or triggering passenger evacuation operations. The train malfunction handling plan acts on the train operation physical field through graph transformation operations and affects the pedestrian flow physical field or communication physical field through coupling edges.

[0452] The turnout fault handling plan is applicable to scenarios where the turnout position is abnormal, the turnout locking status is abnormal, or the path where the turnout is located is unavailable. The plan operations include setting the turnout position status, deleting directed edges corresponding to unavailable paths, adding directed edges corresponding to alternative paths, or modifying the edge weight field between line segments. The turnout fault handling plan changes the line connection relationship or path propagation relationship in the digital twin model through graph structure modification operations.

[0453] In the emergency response plan for power loss of the contact rail, the status of the power supply node corresponding to the contact rail section is determined based on the power supply status data. When the power supply node status indicates a power loss state, the change in the power supply node status is propagated to the adjacent power supply node along the directed edge in the power supply physical field. Then, based on the coupling edge between the power supply physical field and the train physical field, the change in the power supply node status is propagated to the train node corresponding to the affected line section. The train node status is updated to a speed limit state, a stop state, or a prohibited entry state based on the coupling propagation result.

[0454] The emergency response plan for contact rail power failure also considers the impact of changes in train operation status on pedestrian flow status. When a train in the affected section enters a stopped or restricted state, the changes in train operation status are propagated to station area nodes or section evacuation nodes through the coupling edge between the train operation physical field and the pedestrian flow physical field. The station area nodes or section evacuation nodes update the number of people, population density, or evacuation direction in the area based on the coupling propagation results. The updated pedestrian flow node status is used to determine whether secondary events are generated.

[0455] The emergency response plan for power outage of the contact rail also determines the propagation conditions of dispatch communication instructions and evacuation information based on communication status data. When the communication node status indicates that the communication link is available, the broadcast instructions, dispatch instructions, or evacuation instructions corresponding to the emergency response plan for power outage of the contact rail are written into the communication nodes in the communication physical field; when the communication node status indicates that the communication link is unavailable, the communication node status affects the evacuation information receiving nodes in the pedestrian flow physical field through the coupling edge, and the communication restriction status is recorded in the simulation status.

[0456] In the train fault handling plan, the status of train nodes, section occupancy nodes, and station nodes are determined based on the train operation status data. When the train node status indicates a fault, the change in train node status is propagated along the directed edges in the train operation physical field to adjacent section nodes and subsequent train nodes. Then, through the coupling edge between the train operation physical field and the passenger flow physical field, the change in train operation node status is propagated to platform area nodes or carriage passenger flow nodes. The impact of the train fault on train operation organization and personnel distribution is simulated and recorded.

[0457] In the turnout failure handling plan, the affected path is determined based on the turnout position status and the connection relationship of the track section. When the turnout position status indicates that the turnout is unavailable, the directed edge corresponding to the affected path is deleted through graph structure modification operation, or the directed edge corresponding to the alternative path is added through graph structure modification operation. The updated digital twin model performs state propagation according to the new track connection relationship and generates the deduced state sequence corresponding to the candidate emergency plan.

[0458] In some implementations, candidate emergency plans also include fire response plans or large passenger congestion response plans; the actions in fire response plans include adjusting ventilation status, setting evacuation directions, restricting trains from entering affected areas, or issuing broadcast instructions; the actions in large passenger congestion response plans include setting flow control status, adjusting transfer routes, adding evacuation guidance nodes, or adjusting train stopping times.

[0459] Through the above steps, power supply status data, train operation status data, passenger flow status data, and communication status data in urban rail transit emergency scenarios can be mapped to a digital twin model; contact rail power failure handling plans, train fault handling plans, and turnout fault handling plans can be converted into graph transformation operations; coupling edges between different physical fields are used to represent the cross-field influence between power supply status, train operation status, passenger flow status, and communication status.

[0460] Example 16:

[0461] This embodiment provides another implementation method for emergency plan simulation and verification. Unlike Embodiment 1, in this embodiment, the digital twin model can be pre-built by the emergency plan simulation and verification device based on the emergency scenario configuration record and updated after acquiring state data; the digital twin model can also be built when the state data is acquired for the first time and updated when new state data is acquired subsequently; both construction methods are used to obtain a digital twin model corresponding to the emergency scenario and to generate a snapshot of the twin state at the current moment.

[0462] When the digital twin model is pre-built, the physical entity record, influence relationship record, and coupling relationship record in the emergency scenario configuration record are read first; then nodes are generated based on the physical entity record, directed edges are generated based on the influence relationship record, and coupled edges are generated based on the coupling relationship record; finally, the nodes, directed edges, and coupled edges are written into the digital twin model storage unit; after obtaining the state data, the node state, directed edges, or coupled edges in the digital twin model are updated based on the state data.

[0463] When acquiring status data for the first time and building a digital twin model, physical entities are first determined based on the acquired status data; then nodes are generated based on the physical entities, and node states are generated based on the entity identifier, entity type, entity location, and status field in the status data; then, the influence relationships between physical entities and the coupling relationships between different physical fields are supplemented based on the emergency scenario configuration records; after the construction is completed, the generated digital twin model is used as the basic model for subsequent status updates and contingency plan simulations.

[0464] In this embodiment, candidate emergency plans can be provided by the emergency command terminal, generated from the emergency plan database, or generated by the emergency plan simulation and verification equipment based on rule-based record matching. When candidate emergency plans are provided by the emergency command terminal, the system receives the candidate emergency plan identifier and plan operation list input by the emergency command terminal. When candidate emergency plans are generated from the emergency plan database, the system queries the emergency plan database based on the applicable emergency scenario, the type of emergency event, and the type of physical entity. When candidate emergency plans are generated from rule-based record matching, the system matches the corresponding plan operation based on the type of emergency event, the node status, and the set safety threshold.

[0465] In this embodiment, the verification metrics may include at least two of the following: entropy change metrics, graph edit distance metrics, response time metrics, and resource consumption metrics. Alternatively, entropy change metrics, graph edit distance metrics, response time metrics, resource consumption metrics, or other verification metrics corresponding to the emergency response objective may be selected based on the configuration records of the emergency scenario. The selection results of the verification metrics are written into the verification scoring configuration record; the subsequent comprehensive verification score is determined according to the verification metrics, reference benchmark values, and weights in the verification scoring configuration record.

[0466] The combination of verification indicators is determined based on the applicable emergency scenario and the applicable contingency plan type. The verification score configuration record shall include at least the applicable emergency scenario, the applicable contingency plan type, the indicator activation flag, the reference benchmark value, and the weight. When the indicator activation flag indicates that the corresponding verification indicator is enabled, the corresponding verification indicator shall be included in the calculation scope of the comprehensive verification score. When the indicator activation flag indicates that the corresponding verification indicator is not enabled, the corresponding verification indicator shall not be included in the calculation scope of the comprehensive verification score.

[0467] This embodiment is applicable to emergency scenarios where a basic digital twin model already exists, as well as emergency scenarios where a digital twin model needs to be temporarily constructed based on the initially acquired state data. Compared with Embodiment 1, the difference in this embodiment lies in the timing of digital twin model construction, the method of providing candidate emergency plans, and the configuration method of verification indicator combinations. This embodiment still updates the digital twin model based on state data, generates a deduced state sequence based on the twin state snapshot and candidate emergency plans, and determines the verification result based on the deduced result.

[0468] Through the above methods, the emergency response plan simulation and verification method can be applied to both pre-modeled and initial modeled scenarios; candidate emergency response plans can be provided by the emergency command terminal, emergency response plan library, or rule matching records; verification indicators can be selected according to the configuration records of the emergency scenario; different implementation methods do not change the processing relationship of state update, plan simulation, coupling propagation, and verification result determination in the emergency response plan simulation and verification method.

[0469] Example 17:

[0470] See Figure 16 This embodiment provides an emergency response plan simulation and verification system. The system is used to implement the emergency response plan simulation and verification methods of any one of Embodiments 1 to 16. The system includes a twin modeling module, a state update module, a plan acquisition module, a simulation generation module, a coupling propagation module, and a verification module.

[0471] The twin modeling module is used to acquire the state data of physical entities in emergency scenarios and construct digital twin models corresponding to the emergency scenarios based on the state data. The twin modeling module is also used to generate nodes based on physical entities, generate directed edges based on the influence relationships between physical entities, generate subgraphs corresponding to different physical fields based on the nodes and directed edges, and generate coupling edges based on the influence relationships between different physical fields.

[0472] The state update module is used to update the digital twin model based on the state data to obtain a snapshot of the twin state at the current moment. The state update module is also used to perform fusion processing on the state data to obtain the state estimation result; determine the state change amount based on the state estimation result and the node state in the digital twin model; when the state change amount meets the preset update conditions, update at least one of the node state, directed edge, or coupled edge.

[0473] The contingency plan acquisition module is used to acquire candidate emergency plans, which include multiple plan operations. The module also acquires the operation type, target entity, operation parameters, and planned execution time for each plan operation; determines multiple simulation times based on the planned execution time and preset simulation step size; and generates an event queue based on the planned execution time. Finally, it determines the plan operation corresponding to each simulation time based on the event queue.

[0474] The simulation generation module is used to convert emergency response plan operations into graph transformation operations applied to the digital twin model based on the twin state snapshot and candidate emergency response plans, and to generate simulation state sequences corresponding to the candidate emergency response plans based on the graph transformation operations. The simulation generation module is also used to advance the simulation time according to a preset simulation step size; execute the corresponding graph transformation operations at each simulation time; perform state propagation on the digital twin model after the graph transformation operations; and generate simulation state sequences based on the state propagation results.

[0475] The coupling propagation module is used to propagate the state changes of different physical fields during the generation of the deduced state sequence, based on the coupling relationships between different physical fields in the digital twin model, to obtain the deduced results corresponding to the candidate emergency response plan. The coupling propagation module is also used to determine two physical fields with a coupling relationship based on the coupling edge, and to propagate the node state changes in one physical field to the other physical field based on the coupling strength between the two physical fields.

[0476] The verification module is used to determine the verification indicators of candidate emergency plans based on the simulation results, and to determine the corresponding verification results of the candidate emergency plans based on the verification indicators. The verification module is also used to determine the comprehensive verification score based on the entropy change index, graph edit distance index, response time index, and resource consumption index; when there are multiple candidate emergency plans, a ranking result is generated based on the comprehensive verification score of each candidate emergency plan.

[0477] In some implementations, the emergency response plan simulation and verification system further includes a storage module, a configuration module, a display module, or a communication module. The storage module stores state data, digital twin models, twin state snapshots, event queues, simulation state sequences, and verification results; the configuration module stores preset update conditions, preset simulation step sizes, set safety thresholds, reference benchmarks, and weights; the display module displays the simulation state sequences, verification indicators, and verification results; and the communication module interacts with state data sources, emergency response plan libraries, configuration libraries, and emergency command terminals.

[0478] In some implementations, the emergency response plan simulation and verification system further includes a state coordination unit, an incremental update unit, a pre-calculation unit, and a batch simulation unit. The state coordination unit can be integrated into the state update module or the simulation generation module, and is used to regenerate the simulation state sequence based on the new twin state snapshot when the deviation between the simulation state in the simulation state sequence and the new twin state snapshot meets preset coordination conditions. The incremental update unit can be integrated into the state update module, and is used to determine whether to add the corresponding physical entity to the incremental update queue based on the data change amount and the set incremental update conditions. The pre-calculation unit can be integrated into the simulation generation module, and is used to obtain the propagation path index, coupling effect lookup table, or state template. The batch simulation unit can be integrated into the simulation generation module, and is used to merge the simulation tasks corresponding to multiple candidate emergency response plans into a batch simulation task and execute the batch simulation task in parallel.

[0479] In some implementations, the emergency response plan simulation and verification system is deployed in the physical layer, data layer, model layer, and application layer; the twin modeling module, state update module, simulation generation module, and coupling propagation module are deployed in the model layer, the configuration module and storage module are deployed in the data layer or model layer, and the display module and decision support interface are deployed in the application layer.

[0480] The emergency response plan simulation and verification system in this embodiment belongs to the same inventive concept as the emergency response plan simulation and verification methods in Embodiments 1 to 16. The processing procedures performed by each module in the emergency response plan simulation and verification system can be found in the corresponding method steps in Embodiments 1 to 16. The modules in the emergency response plan simulation and verification system can be integrated in the same device or distributed in different devices and interact with each other through wired communication, wireless communication, industrial Ethernet, fieldbus or service interface.

[0481] Through the above structure, the emergency response plan simulation and verification system can build and update digital twin models based on state data, generate simulation state sequences based on candidate emergency response plans, perform coupling propagation between different physical fields during the generation of simulation state sequences, and determine verification indicators and verification results based on simulation results.

[0482] Example 18:

[0483] This embodiment provides an electronic device. The electronic device includes a processor and a memory; the memory stores a computer program; when the processor loads and executes the computer program, it implements the emergency response plan simulation and verification method of any one of Embodiments 1 to 16.

[0484] The processor executes program instructions corresponding to the computer program to complete state data processing, digital twin model construction, digital twin model updating, twin state snapshot generation, candidate emergency plan acquisition, graph transformation operation determination, deduced state sequence generation, coupling propagation, verification index determination, and verification result determination. The processor can also be used to perform incremental updates, secondary event detection, pre-computation processing, parallel execution of batch deduction tasks, comprehensive verification score determination, and ranking result generation.

[0485] The memory is used to store data used or generated during the execution of computer programs and emergency response plan simulation and verification methods. It also stores state data, candidate emergency response plans, digital twin models, twin state snapshots, event queues, graph transformation operations, simulation state sequences, simulation results, verification indicators, comprehensive verification scores, ranking results, and verification results. Furthermore, the memory can store preset update conditions, preset simulation step sizes, set convergence thresholds, set safety thresholds, reference benchmarks, and weights.

[0486] Electronic devices can be servers, edge computing devices, computer equipment, scheduling platforms, or clusters of multiple computing devices. They can communicate with status data sources, emergency plan databases, configuration databases, digital twin model storage units, or emergency command terminals. The electronic devices load and execute computer programs stored in memory via their processors to implement the processing flow corresponding to the emergency plan simulation and verification methods.

[0487] Example 19:

[0488] This embodiment provides a computer-readable storage medium. The computer-readable storage medium stores a computer program; when the computer program is loaded and executed by a processor, it implements the emergency response plan simulation and verification method of any one of Embodiments 1 to 16.

[0489] Computer-readable storage media may include read-only memory, random access memory, magnetic disk, optical disk, flash memory, removable storage device, or other storage media capable of storing computer programs. The computer programs stored in the computer-readable storage media may include program instructions for performing state data acquisition, digital twin model construction, digital twin model updating, candidate contingency plan acquisition, graph transformation operation determination, deduced state sequence generation, coupling propagation, verification indicator determination, and verification result determination.

[0490] When a computer program is executed by a processor, it can acquire state data of physical entities in an emergency scenario and construct a digital twin model corresponding to the emergency scenario based on the state data; update the digital twin model based on the state data to obtain a snapshot of the twin state at the current moment; acquire candidate emergency plans; based on the twin state snapshot and candidate emergency plans, convert the plan operations into graph transformation operations acting on the digital twin model, and generate a deduced state sequence corresponding to the candidate emergency plans based on the graph transformation operations; in the process of generating the deduced state sequence, couple and propagate the state changes of different physical fields to obtain the deduced results; determine the verification indicators based on the deduced results, and determine the verification results based on the verification indicators.

[0491] The above description is merely a preferred embodiment of this application, and the scope of protection of this application shall be determined by the scope defined in the claims. Any improvements and modifications made by those skilled in the art without departing from the essence of the technical solution of this application shall fall within the scope of protection of this application. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of this application shall also be considered within the scope of protection of this application.

Claims

1. A method for simulating and verifying emergency response plans, characterized in that, include: Acquire the status data of physical entities in emergency scenarios, and construct a digital twin model corresponding to the emergency scenario based on the status data; The digital twin model is updated based on the state data to obtain a snapshot of the twin state at the current moment; Obtain candidate emergency response plans, which include multiple plan operations; Based on the twin state snapshot and candidate emergency plans, the plan operations are converted into graph transformation operations that act on the digital twin model, and the inferred state sequence corresponding to the candidate emergency plans is generated according to the graph transformation operations. In the process of generating the simulated state sequence, the state changes of different physical fields are coupled and propagated according to the coupling relationship between different physical fields in the digital twin model to obtain the simulation results corresponding to the candidate emergency plan; Based on the simulation results, the verification indicators for the candidate emergency response plans are determined, and the verification results corresponding to the candidate emergency response plans are determined based on the verification indicators.

2. The method according to claim 1, characterized in that, Based on the status data, a digital twin model corresponding to the emergency scenario is constructed, including: Nodes are generated based on physical entities; Directed edges are generated based on the influence relationships and influence strengths between physical entities; Generate subgraphs corresponding to different physical fields based on nodes and directed edges; Generate coupling edges based on the influence relationships between different physical fields; The digital twin model is obtained based on the subgraph and coupling edges.

3. The method according to claim 2, characterized in that, The digital twin model is updated based on the state data to obtain a snapshot of the twin state at the current moment, including: The state data is fused to obtain the state estimation results; The state change is determined based on the state estimation results and the node states in the digital twin model; When the state change meets the preset update conditions, the node state, directed edge, or coupled edge is updated to obtain a twin state snapshot.

4. The method according to claim 1, characterized in that, Obtain candidate emergency response plans, including: Obtain the operation type, target entity, operation parameters, and planned execution time for each operation in the candidate emergency response plans; Multiple simulation moments are determined based on the planned execution time and the preset simulation step size, and an event queue is generated based on the planned execution time. The corresponding contingency plan operation is determined based on the event queue at each simulation moment.

5. The method according to claim 4, characterized in that, The pre-planned operations are converted into graph transformation operations applied to the digital twin model, including: Based on the operation type, target entity, and parameters corresponding to the pre-planned operation, determine the node state assignment operation, node state increment operation, edge weight modification operation, graph structure modification operation, or coupling propagation operation as the graph transformation operation.

6. The method according to claim 5, characterized in that, The graph transformation operation generates a sequence of simulated states corresponding to the candidate emergency response plans, including: The simulation time is advanced according to the preset simulation step size; Perform the corresponding graph transformation operation at each simulation moment; Perform state propagation on the digital twin model after performing graph transformation operations; Determine the state change between two adjacent state propagations; When the change in state is less than the set convergence threshold, the simulation state corresponding to that simulation moment is generated. Arrange the various simulation states according to the simulation time to obtain the simulation state sequence.

7. The method according to claim 6, characterized in that, The process of generating the deduced state sequence also includes: Secondary events are determined based on the comparison between the node states in the simulation and the set safety thresholds. Add the graph transformation operation corresponding to the secondary event to the event queue to obtain the event queue after addition; The deduced state sequence continues to be generated based on the added event queue.

8. The method according to claim 2, characterized in that, Based on the coupling relationships between different physical fields in the digital twin model, the state changes of different physical fields are coupled and propagated, including: The coupling edge determines two physical fields that are coupled. Based on the coupling strength between two physical fields, the node state changes in one physical field are propagated to the other physical field, thus obtaining the node state changes in the other physical field.

9. The method according to claim 8, characterized in that, Methods for determining coupling strength include: Obtain the correlation between two physical fields, the distance between physical entities in the two physical fields, and the state of the connecting medium between the physical entities; The coupling strength is determined based on the correlation, distance, and state of the connecting medium.

10. The method according to claim 1, characterized in that, Validation metrics include entropy change metrics, graph edit distance metrics, response time metrics, and resource consumption metrics; The entropy change index is determined based on the graph entropy change of the digital twin model in the deduced state sequence; The graph editing distance index is determined based on the graph editing distance between the simulation results and the set target safety status. The response time metric is determined based on the time required for the simulated states in the simulated state sequence to reach a set safety threshold. Resource consumption indicators are determined based on the cumulative value of resource consumption during the planned operations.

11. The method according to claim 10, characterized in that, The verification results corresponding to the candidate emergency response plans are determined based on the verification indicators, including: Based on the verification indicators and the corresponding reference benchmark values, each verification indicator is dimensionless to obtain the dimensionless verification indicators. The comprehensive verification score of the candidate emergency response plan is determined based on the dimensionless verification indicators and their corresponding weights. The verification results are determined based on the comprehensive verification score.

12. The method according to claim 11, characterized in that, There are multiple candidate emergency response plans; Based on the comprehensive verification scores corresponding to each candidate emergency plan, the multiple candidate emergency plans are ranked to obtain the ranking results; The sorting results, or the sorting results and the candidate emergency plans in the sorting results that meet the preset selection conditions, are determined as the verification results.

13. The method according to claim 1, characterized in that, The process of updating the digital twin model based on state data is defined as real-time state synchronization; The process of generating a deduced state sequence based on graph transformation operations is defined as the contingency plan deduction. When the deviation between the projected state in the projected state sequence and the twin state snapshot updated based on the new state data meets the preset coordination conditions, the twin state snapshot updated based on the new state data is determined as the new twin state snapshot, and the projected state sequence is regenerated based on the new twin state snapshot.

14. The method according to claim 3, characterized in that, Updating the digital twin model based on state data also includes: Obtain the amount of data change between the current state data and the previous state data; When the amount of data change does not meet the set incremental update conditions, no update is performed on the corresponding physical entity; When the data change meets the set incremental update conditions, the corresponding physical entity is added to the incremental update queue, and the digital twin model is updated based on the incremental update queue.

15. The method according to claim 6, characterized in that, Before generating the deduced state sequence, the following is also included: Obtain the propagation path index, coupling effect lookup table, or state template corresponding to the emergency scenario; When generating the deduced state sequence, the state propagation result or coupling propagation result is determined based on the propagation path index, coupling effect lookup table, or state template.

16. The method according to claim 6, characterized in that, When there are multiple candidate emergency plans, the simulation tasks corresponding to multiple candidate emergency plans will be merged into a batch simulation task; The batch simulation tasks are executed in parallel to generate the simulation state sequence corresponding to each candidate emergency plan.

17. The method according to claim 1, characterized in that, The emergency scenario is an urban rail transit emergency scenario; Status data includes power supply status data, vehicle status data, pedestrian flow status data, and communication status data; The candidate emergency response plans include the contact rail power failure response plan, the train failure response plan, and the turnout failure response plan.

18. An emergency response plan simulation and verification system, used to implement the emergency response plan simulation and verification method according to any one of claims 1 to 17, characterized in that, include: The twin modeling module is used to acquire the state data of physical entities in emergency scenarios and construct a digital twin model corresponding to the emergency scenario based on the state data. The state update module is used to update the digital twin model based on the state data to obtain a snapshot of the twin state at the current moment. The contingency plan acquisition module is used to acquire candidate emergency plans, which include multiple contingency plan operations. The deduction generation module is used to convert the plan operations into graph transformation operations that act on the digital twin model based on the twin state snapshot and candidate emergency plans, and generate the deduction state sequence corresponding to the candidate emergency plans according to the graph transformation operations. The coupling propagation module is used to perform coupling propagation of the state changes of different physical fields in the digital twin model according to the coupling relationship between different physical fields in the process of generating the deduced state sequence, so as to obtain the deduced results corresponding to the candidate emergency plan. The verification module is used to determine the verification indicators of candidate emergency plans based on the simulation results, and to determine the corresponding verification results of candidate emergency plans based on the verification indicators.

19. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a computer program, and when the processor loads and executes the computer program, it implements the emergency plan simulation and verification method as described in any one of claims 1 to 17.

20. A computer-readable storage medium, characterized in that, A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the emergency response plan simulation and verification method as described in any one of claims 1 to 17.

Images