Emergency treatment measure dynamic deduction and evaluation method, system and electronic equipment

Abstract

The application provides an emergency treatment measure dynamic deduction and evaluation method, system and electronic equipment, fuses emergency information and real-time line network operation data to obtain initial event data packets; based on the initial event data packets, outputs multiple groups of treatment measure schemes; adopts a simulation tool to perform parallel deduction on the multiple groups of treatment measure schemes to output multiple groups of simulation result data sets; and based on the multiple groups of simulation result data sets, outputs a decision support report. The application forms a complete emergency treatment decision closed loop of multi-source data fusion, multiple scheme generation, parallel simulation deduction and decision support report output, comprehensively improves the overall treatment capability of the rail transit line network in response to emergencies, effectively reduces the influence of emergencies on line capacity and passenger flow organization, and guarantees the operation safety of the line network and the travel safety of passengers.

Description

Technical Field

[0001] This invention relates to the field of emergency response technology, and in particular to a method, system, and electronic equipment for dynamic simulation and evaluation of emergency response measures. Background Technology

[0002] With the exponential expansion of urban rail transit networks and the continued high growth in passenger flow, operational safety and emergency response capabilities face unprecedented challenges. Against this backdrop, traditional emergency command systems and existing decision support technologies consistently suffer from the following shortcomings: 1. Rigid System Architecture: Existing emergency command systems mostly adopt an architecture of stacked functional modules, which are essentially passive toolsets serving manual operations. They lack deep semantic understanding of event information and automated reasoning capabilities, making it difficult to generate response strategies that adapt to complex and ever-changing on-site situations, resulting in poor decision-making adaptability and slow response.

[0003] 2. Lack of key capabilities: The existing system relies on a static mapping logic of "event type - fixed contingency plan" or simple rule triggering. It lacks a professional, deep event analysis agent, making it difficult to predictively deduce the complete event impact chain, resulting in a lack of holistic and forward-looking response measures.

[0004] 3. Isolated Simulation Tools: Although specialized wireline simulation tools exist, they are traditionally isolated, offline analysis systems. These systems lack a dedicated simulation management system to handle the automatic orchestration, driving, and result collection of simulation tasks. In emergency scenarios, starting the simulation tool, configuring the scenario, inputting parameters, and interpreting results all require manual intervention by professionals. This process is cumbersome, time-consuming, and cannot be integrated into the critical real-time emergency decision-making loop. Summary of the Invention

[0005] In order to address the problems existing in the prior art, the present invention aims to provide a method, system and electronic equipment for dynamic simulation and evaluation of emergency response measures, which can improve the speed of decision-making.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for dynamic simulation and evaluation of emergency response measures, including: The initial event data packet is obtained by fusing information about emergencies with real-time network operation data. Based on the initial event data packet, output multiple sets of handling measures; Simulation tools were used to perform parallel simulations of multiple response measures, and multiple sets of simulation result datasets were output. Based on multiple sets of simulation result datasets, the optimal treatment plan and decision support report are output.

[0007] As a further improvement of the present invention, the step of fusing emergency event information with real-time network operation data to obtain an initial event data packet includes: After receiving and analyzing the emergency, emergency information is obtained. Obtain real-time network operation data; The initial event data packet is obtained by fusing and standardizing information on emergencies and real-time network operation data.

[0008] As a further improvement of the present invention, the step of outputting multiple sets of handling measures based on the initial event data packet includes: Based on the domain knowledge base, reason about the initial event data packet and output the influence chain object; Based on the impact chain object, multiple sets of disposal measures are generated.

[0009] As a further improvement of the present invention, the step of reasoning about the initial event data packet based on the domain knowledge base and outputting the influence chain object includes: The initial event data is parsed to obtain the event type and the location where the event occurred; Retrieve from the domain knowledge base based on event type and event location; Based on the retrieved knowledge, after sequentially performing direct impact analysis, traffic impact chain reasoning, passenger transport impact chain reasoning, and final risk point derivation, intermediate conclusions and final risk points are obtained. The intermediate conclusions and final risk points are assembled into an impact chain object according to a predefined structured format.

[0010] As a further improvement of the present invention, the step of performing passenger transport influence chain reasoning includes: Based on the intermediate conclusions obtained from the traffic impact chain reasoning, OD pairs containing fault sections are selected from all paths in the OD matrix, and the indirectly affected passenger flow is identified based on the OD pairs. Analyze the travel paths of indirectly affected passengers, and predict the location and intensity of passenger congestion based on these travel paths. By combining the initial passenger flow and the continuously increasing backlog of passengers at each station without malfunctions, the growth curve of the number of passengers on the platforms of each station without malfunctions is predicted. The growth curve is compared with the platform's safe capacity to obtain the risk trigger time. The continuously increasing backlog of passengers is determined by the rate and duration of the additional backlog of passengers. By analyzing the transfer passenger flow using the OD matrix and combining it with the transfer channel capacity of the station in the domain knowledge base, the risk level of transfer congestion and the location of congestion can be determined. Based on indirectly affected passenger flow, passenger backlog location and intensity, risk trigger time, congestion risk level and congestion occurrence point, a passenger transport impact chain is generated.

[0011] As a further improvement of the present invention, the step of outputting the optimal treatment plan and decision support report based on multiple sets of simulation result datasets includes: Calculate the quantitative metrics for each set of simulation result datasets; Based on preset decision weights, a multi-objective decision-making model is established, and the quantitative indicators are normalized and weighted for comprehensive scoring. The option with the highest overall score is selected as the optimal response plan, and a decision support report is generated.

[0012] As a further improvement of the present invention, the quantitative indicators include: the proportion of time the station platform is crowded beyond the limit, the proportion of time the train carriage is full of passengers beyond the limit, the total delay time of passengers across the network, the proportion of affected passengers, the increase rate of average travel time, the economic loss of passenger time delay, and the additional operating costs of enterprises.

[0013] This invention also provides a dynamic simulation and evaluation system for emergency response measures, used to implement the aforementioned dynamic simulation and evaluation method for emergency response measures, comprising: The sensing agent is used to fuse information about emergencies with real-time network operation data to obtain an initial event data packet; The event analysis agent outputs multiple sets of response measures based on the initial event data packet; The simulation management agent uses simulation tools to perform parallel simulations of multiple disposal measures and outputs multiple sets of simulation result datasets. The decision-making agent, based on multiple sets of simulation result datasets, outputs the optimal treatment plan and decision support report.

[0014] The present invention also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor is used to execute the programs stored in the memory to implement the above-mentioned method for dynamic simulation and evaluation of emergency response measures.

[0015] The beneficial effects of this invention are as follows: This invention integrates emergency event information with real-time network operation data to generate an initial event data package, breaking the siloed state of emergency event data and operation data, avoiding the one-sidedness of decision-making supported by a single data dimension, and providing a complete and accurate initial basis for the generation of subsequent response measures, which is in line with the actual operation status of the network. This improves the adaptability of emergency response plans from the source. By using simulation tools to recreate the actual operation scenario of the rail transit network, the implementation effects of multiple response plans can be simulated and deduced. There is no need to carry out trial and error of the plan on the actual operating line, which effectively avoids the secondary operation risks and passenger flow chaos that may be caused during the trial and error process. At the same time, it significantly reduces the actual trial and error cost of implementing emergency plans, forming a complete emergency response decision-making closed loop of multi-source data fusion → multi-plan generation → parallel simulation and deduction → decision support report output. It realizes the standardization and intelligent operation of the entire process of emergency response measures from plan generation to evaluation and decision-making, replacing the traditional fragmented response mode dominated by manual labor, comprehensively improving the overall response capability of the rail transit network to emergencies, effectively reducing the impact of emergencies on line capacity and passenger flow organization, and ensuring the safety of network operation and passenger travel. Attached Figure Description

[0016] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0017] Figure 1 This is a flowchart of the dynamic simulation and evaluation method for emergency response measures described in this invention; Figure 2 This is a flowchart of step S2 of the present invention; Figure 3 This is a flowchart of step S3 of the present invention; Figure 4 This is a flowchart of step S4 of the present invention. Detailed Implementation

[0018] To make the technical problems solved by the present invention, the technical solutions adopted, and the technical effects achieved clearer, the technical solutions of the embodiments of the present invention will be further described in detail below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] This invention provides a method for dynamic simulation and evaluation of emergency response measures, such as... Figure 1 As shown, it includes: S1. Integrate emergency information with real-time network operation data to obtain initial event data packets, providing a unified and comprehensive data foundation for subsequent analysis.

[0020] Step S1 specifically includes: S11. After receiving and analyzing the emergency, the emergency information is obtained. The emergency information includes: event type, event level, event occurrence time, event location, and initial impact range of the event.

[0021] S12. Obtain real-time network operation data; real-time network operation data includes: real-time passenger flow data, predicted passenger flow data, train running location, equipment status information, external status information (such as weather), etc.

[0022] S13. Perform data fusion and standardization processing on emergency information and real-time network operation data to obtain the initial event data packet. , represented as:

[0023] ; in, For event type, Event level, The location where the event occurred. The time when the event occurred. This is the initial scope of the incident's impact. Affected trains assembled. For real-time passenger flow data, To predict passenger flow data, For device status information, This refers to external weather conditions. The affected train set is identified by the train's location.

[0024] S2, Based on initial event data packet Output multiple sets of disposal measures, such as Figure 2 As shown.

[0025] S21. Based on the domain knowledge base, process the initial event data packet. Perform inference and output the impact chain object, ImpactChain.

[0026] S211, Initial event data packet The event type and location of the event are obtained through analysis.

[0027] S212. Retrieve from the domain knowledge base based on event type and event location.

[0028] S213. Based on the retrieved knowledge, after sequentially performing direct impact analysis, traffic impact chain reasoning, passenger transport impact chain reasoning, and final risk point derivation, intermediate conclusions and final risk points are obtained.

[0029] Direct impact analysis: Based on the retrieved knowledge, intermediate conclusions about the direct impact are obtained.

[0030] Driving impact chain reasoning: Based on retrieved knowledge and direct impact intermediate conclusions, intermediate conclusions of the driving impact chain are obtained.

[0031] Passenger transport impact chain reasoning: Based on retrieved knowledge and intermediate conclusions of the traffic impact chain, the passenger transport impact chain is obtained. Specifically, the process of passenger transport impact chain reasoning is as follows: S2131. Based on the intermediate conclusions obtained from the traffic impact chain reasoning, select all OD pairs containing fault sections from the OD matrix, and identify the indirectly affected passenger flow based on the OD pairs.

[0032] S2132. Analyze the travel paths of indirectly affected passenger flows, and predict the location and intensity of passenger flow backlog based on the travel paths of indirectly affected passenger flows.

[0033] S2133. Combining the initial passenger flow and the continuously increasing backlog passenger flow of each station that has not experienced a failure, predict the growth curve of the number of passengers on the platform of each station that has not experienced a failure, compare the growth curve with the platform's safe capacity, and obtain the risk trigger time; among which, the continuously increasing backlog passenger flow is determined by the rate and time of the additional backlog passenger flow.

[0034] S2134. Analyze passenger flow through transfers using the OD matrix and combine it with the transfer channel capacity of the station in the domain knowledge base to determine the level of transfer congestion risk and the location of congestion.

[0035] S2135. Based on indirectly affected passenger flow, passenger backlog location and intensity, risk trigger time, congestion risk level and congestion occurrence point, generate passenger transport impact chain.

[0036] Final risk point derivation: Based on the intermediate conclusions drawn from the above direct impact analysis, traffic impact chain reasoning, and passenger transport impact chain reasoning, specific final risk points are identified.

[0037] S214. Assemble intermediate conclusions and final risk points into an impact chain object according to a predefined structured format.

[0038] S22. Based on the impact chain object, generate multiple sets of response measures. Specifically, through the strategy generation module, execute the measure generation function. Dynamically generate M groups (M≥2) that conform to preset operational rules. C Traffic and passenger transport organization and handling measures: ; The strategy generation module is preferably implemented using a large language model based on domain knowledge.

[0039] S3. Use simulation tools to perform parallel simulations of multiple response measures and output multiple sets of simulation result datasets.

[0040] S4. Based on multiple sets of simulation result datasets, output the optimal treatment measures and decision support report.

[0041] S41. Calculate the quantitative indicators for each set of simulation result datasets; the quantitative indicators include: the proportion of time the station platform congestion exceeds the limit, the proportion of time the train carriage full load rate exceeds the limit, the total delay time of passengers across the network, the proportion of affected passengers, the average travel time increase rate, the economic loss of passenger time delay, and the additional operating costs of the enterprise.

[0042] S42. Based on the preset decision weights, establish a multi-objective decision model, and normalize and weightedly score the quantitative indicators. S43. The option with the highest comprehensive score is determined as the optimal treatment plan, and a decision support report is generated.

[0043] The invention will now be further described in conjunction with a specific implementation process, as follows: S1. Integrate the emergency information with real-time network operation data to obtain the initial event data packet. : At 7:20:01, the alarm information "1B4 zone tripped, reclosing failed" was obtained from the power monitoring system. It was automatically identified as the alarm type "power supply switch fault trip". The event type was automatically identified as "power supply switch fault trip", the event location was "1B4 power supply zone", and the event time was "7:20:01".

[0044] Almost simultaneously, the ATS (Automatic Train Monitoring) system retrieved the train running positions in the faulty section (station A to station D, northbound), identified the affected train set as {0704, 2002, 0804}, a total of 3 trains, and obtained their precise train running positions and equipment status information (such as "0804 has lowered its pantograph"); the system also retrieved real-time and predicted passenger flow data for the current morning peak from the passenger flow forecasting system, identifying this period as the peak period with enormous passenger flow pressure.

[0045] All information is automatically merged to form a structured initial event data packet. This provides a unified data foundation for subsequent analysis. It is represented as: ; in, For event type, The location of the incident. For the time of occurrence, Affected trains assembled. Flux This is passenger flow data.

[0046] S2, Based on initial event data packet Output multiple sets of response measures: S21. Based on the domain knowledge base, process the initial event data packet. Perform inference and output the impact chain object: S211, Perception.

[0047] Input: Initial event data packet .

[0048] Action: Analyze the event type ("Power switch fault trip") and the location of the event ("1B4 zone"), and immediately retrieve it from its domain knowledge base. Automatic search is initiated in the middle.

[0049] S212, Knowledge Retrieval.

[0050] Search terms: The answer to "Power supply range of 1B4 power supply zone" is: "Uplink contact network of station A to station D".

[0051] "Direct impact of power outage in the power supply zone" → resulting in the rule: "All trains within the jurisdiction will be forced to stop due to loss of power."

[0052] "During the morning rush hour, the percentage of transport capacity in the upward direction between stations A and D" → obtain the parameter The estimation method.

[0053] S213, Multi-step Chain Reasoning Based on initial perception and retrieved knowledge, and with the assistance of task memory, automated reasoning unfolds in the following steps: 1. Direct Impact Analysis Reasoning: "Power outage in zone 1B4" + "It supplies power to the upstream section from station A to station D" → Conclusion: "Power outage in the upstream section from station A to station D".

[0054] Reasoning: "Power outage in the section" + "Trains in the section" + "Train operation depends on overhead contact line power supply" → Conclusion: "Trains 0704, 2002, and 0804 were forced to stop in the section".

[0055] Status recording: Store the intermediate conclusions that directly affect the task into the task memory.

[0056] Direct impact analysis can achieve three key objectives: first, to establish the causal relationship between power outage and forced stop, providing a logical starting point for subsequent train operation impact chain reasoning; second, to achieve fault tolerance by independently deriving and reverse-verifying the integrity of external report data; and third, to dynamically maintain train status changes (such as from "forced stop" to "passengers need to be evacuated") through task memory and state manager, for use by subsequent stages.

[0057] 2. Reasoning about the impact chain of traffic Reasoning: "Three trains were forced to stop in the section" + "Trains physically occupied the tracks" → Conclusion: "The up line was physically blocked at the fault point."

[0058] Reasoning: "Upstream line congestion" + "Subsequent trains continue to run" → Conclusion: "Subsequent upstream trains will begin to back up at upstream stations (station H, station I)".

[0059] Reasoning: "Trains are backed up on the platform" + "Platform side track capacity is limited" → Conclusion: "The backlog effect will spread upstream and may eventually affect station L."

[0060] Reasoning: "Congestion in the upstream section between Station A and Station D" + "This section is an important part of the upstream trunk line" → Conclusion: "Overall capacity in the upstream direction will decrease by approximately..." (e.g., 55%).

[0061] State Record: Generate a summary of intermediate conclusions of the driving impact chain and store it in the task memory.

[0062] 3. Passenger transport impact chain reasoning Input: Output of the traffic impact chain (capacity reduction) (Area of ​​impact), and the morning peak passenger flow OD matrix retrieved in real time from the domain knowledge base. .

[0063] Reasoning process: S2131. Accurately identify affected passenger flow: First, based on the fault location (station A to station D, upstream) and the affected area, from the OD matrix... The process involves filtering out all origin-destination (OD) pairs that contain the faulty section (or paths that are impassable due to the interruption). For example, all passenger flows originating from upstream stations such as L, K, and I, and destined for downstream stations such as D, E, and F, are identified as directly affected passenger flows. Meanwhile, the overcrowding in carriages due to reduced capacity will also affect passengers on other origin-destination (OD) pairs; this portion is identified as indirectly affected passenger flow. .

[0064] S2132. Predicting the location and intensity of passenger congestion: Analysis of indirectly affected passenger flow The travel routes of these passengers are affected. These passengers were originally supposed to pass through the affected section, but due to the disruption, they will be congested at stations on the uphill direction, especially stations H and I near the affected section, and the line's terminus L. Estimates can be made for each station... Additional backlog passenger flow rate This is proportional to the sum of all passenger flows in the OD matrix that originate from that station and whose paths are blocked.

[0065] Formulas to aid understanding: ; in This is a set of destinations that cannot be reached due to interruption.

[0066] S2133, Platform Safety Risk Quantitative Assessment: Combined with the initial passenger flow of each station and the continued growth of backlogged passenger traffic The intelligent agent can more accurately predict the number of people on the platform. Growth curve: .

[0067] By Platform safety capacity In comparison, the agent can quantify the risk trigger time. For example, "Based on the current OD passenger flow, the up platform of station I is expected to exceed the safe capacity in 12 minutes," which is more accurate than the previous qualitative estimate ("15-20 minutes").

[0068] S2134. Analysis of passenger flow impact during transfers: OD matrix analysis is used to analyze transfer passenger flow. For example, it identifies passenger flow originally planning to transfer from Line M (adjacent line) to Line N (eastbound) at station D. Due to the interruption of Line N's northbound service, these passengers will be largely congested in the transfer hall or passageways of station D. Simultaneously, passenger flow transferring from Line N's southbound service to Line M (adjacent line) may also be affected by platform congestion. By querying the transfer passageway capacity of the station in the knowledge base, the agent can determine the risk level and occurrence point of transfer congestion.

[0069] S2135, Passenger Transport Impact Chain Output: Reduced capacity → Accurately identify affected OD passenger flow → Predict the location and intensity of backlog (Station I / Station H platform) → Quantify risk trigger time (e.g., Station I: 12min) → Specific transfer passenger flow impact (Station D, M line to N line transfer) → Sharp increase in travel time.

[0070] 4. Derivation of final risk points Reasoning: Based on all the above conclusions, the specific final risk points can be summarized as follows: risk of stampede due to overcrowding on the platforms of stations I and H; risk of congestion in the transfer passage of station D; and risk of passenger evacuation for trains forced to stop within the fault section.

[0071] S214, Structured Output Synthesis Extract all intermediate conclusions and final risk points from the task memory, assemble them according to a predefined structured format, and generate the final impact chain object. .

[0072] S22, Generation of Disposal Measures Plan Based on the influence chain object The strategy generation module executes the measure generation function. Dynamically generate M groups (M≥2) that conform to preset operational rules. C Traffic and passenger transport organization and handling measures: ; Constraint C includes turnaround point capacity, passenger clearance rules, etc. This example generates three sets of handling measures: 1) Handling measures plan ① (rapid restoration priority): clear customers from the fault area and carry out emergency repairs on a small-scale power outage.

[0073] 2) Response Plan ② (Ensuring Service in the Core Area): Based on Plan ①, immediately start the operation of the "Station I to Station G" short-route service and organize single-line bidirectional operation.

[0074] 3) Response Plan ③ (maximizing passenger travel protection): Based on Plan ②, bus connections will be initiated simultaneously, and passengers on upstream trains will be pre-cleared.

[0075] S3. A wire mesh simulation tool is used to perform parallel simulations of multiple treatment measures, outputting multiple sets of simulation result datasets. These multiple treatment measures are used as structured simulation input data, such as... Figure 3 As shown, the MCP Server automatically translates the three proposed solutions into executable instructions for the wireline simulation tool. This process can be viewed as a protocol conversion function. : ; in, It is a wired network topology. This is the current running graph.

[0076] The wire mesh simulation tool was called three times in parallel to execute the simulation function. Simulate the network status after implementing disposal measures ①, ②, and ③ within the next T hours (e.g., T=2): ; in, FluxOD ( t This is a dynamic passenger flow demand matrix. The simulation tool outputs three sets of simulation result datasets containing massive amounts of spatiotemporal data. .

[0077] S4. Based on multiple sets of simulation result datasets, output the optimal treatment plan and decision support report: This step is performed by a scheme evaluation and optimization algorithm, which transforms the massive simulation result dataset into a set of quantifiable, multi-level decision indicators, and then automates the comparison and selection through a target decision model. Input: M sets of simulation result datasets. Output the recommended optimal solution. And its detailed assessment report. For example... Figure 4 As shown, it specifically includes the following steps: S41. Extraction and Calculation of Multi-Dimensional Quantitative Indicators The algorithm automatically extracts raw data from each set of simulation result datasets and calculates sub-indicators across three dimensions: "safety, passenger service, and economic and social impact." 1. Security Dimension Indicators

[0078] This dimension assesses the effectiveness of response measures in preventing safety incidents and has veto power.

[0079] S411, Percentage of Station Platform Crowding Exceeding Limits ( ) Factor: For each station j The number of people on its platform in real time Platform safety capacity .

[0080] Calculation: Statistical simulation period T Inside, the number of people on the platform exceeds the safety threshold (e.g.) The cumulative time of each station is calculated, and its ratio to the total duration is determined. Finally, the average is calculated for all key stations.

[0081] formula: ,in, J For the set of key stations included in the assessment (such as station H, station I, station D, etc. in this event), 1 is the indicator function.

[0082] S412, Proportion of train carriages exceeding the full load rate limit for the duration ( ) Factor: For each train Real-time load factor .

[0083] Calculation: Statistical simulation period Within this range, the cumulative percentage of time during which the train's load factor exceeds a safety threshold (e.g., 120%) is calculated, and this percentage is averaged across all affected lines.

[0084] formula: ; Safety dimension indicators: Assign weights to the two core safety indicators mentioned above (e.g., platform safety has a higher weight), calculate the weighted sum, and the smaller the value, the better. ; 2. Passenger service dimension indicators ( ) This dimension assesses the impact of the proposed measures on the passenger travel experience.

[0085] S413, Total Delay Time for Passengers Across the Network ( ) Factor: per passenger actual arrival time With the planned arrival time .

[0086] Calculation: Sum the delay times for all affected passengers. This is a core metric for measuring the severity of service disruptions.

[0087] formula: , where p is the total number of passengers on the entire network during the simulation period.

[0088] S414, Percentage of Passengers Affected

[0089] Factor: Trip delays exceeding a certain threshold Number of passengers (e.g., 5 minutes) .

[0090] Calculation: This indicator reflects the breadth of the event's impact.

[0091] formula: ; S415, Average travel time increase rate ( ) Factor: Passengers Actual travel time Compared to normal travel time .

[0092] Calculation: Reflects the degree of decline in travel efficiency felt by each passenger on average.

[0093] formula: ; Comprehensive indicators for passenger service: The above three indicators are normalized and then weighted and summed. The smaller the value, the better. .

[0094] in, This represents the value after min-max normalization.

[0095] 3. Economic and social impact indicators ( ) This dimension assesses the total social and corporate costs of the proposed solutions from a more macro perspective.

[0096] S416, Economic losses due to passenger time delays ( ) Factor: Total delay time for all passengers across the network (Unit: person-hour), value of a unit of time (Unit: Yuan / person-hour, which can be converted based on local GDP and average working hours per person).

[0097] Calculation: Converting passenger time delays into monetary losses is the core of assessing social benefits.

[0098] formula: . S417, Additional operating costs of the enterprise ( ) factor: Number of bus shuttle services Cost per vehicle .

[0099] Additional increase in total train travel kilometers Unit energy consumption cost .

[0100] Ticket revenue loss (It can be estimated by the difference between the actual passenger volume and the planned passenger volume in the simulation).

[0101] calculate: ; Comprehensive indicator for economic and social impact: After normalizing economic losses, the smaller the value, the better. ; S4.2 Multi-objective decision analysis and optimal solution selection 1. Indicator Normalization: Since the dimensions of each indicator are different, the comprehensive indicator of the three dimensions of each scheme m is first normalized. Normalization is performed to map it to the interval [0,1], where 0 represents the best performance in this dimension and 1 represents the worst performance. ; 2. Weighted Overall Score: The system pre-calculates the weights of each dimension based on the operational strategy. (For example, under a security-first strategy,) It can be set to 0.5. and (Each is 0.25). Calculate the overall score for each option: .

[0102] It should be noted that since all indicators are negative, the smaller the value, the better. Therefore, (1 - normalized value) is used to convert them into positive scores.

[0103] 3. Solution Recommendation: The algorithm selects the solution with the highest overall score as the final recommended solution. .

[0104] S43. Generate decision support reports Automatically generate a structured report, including: Recommended solution: Clearly provide a solution. And its overall score.

[0105] Indicator Comparison Table: Clearly displays the values ​​and rankings of all participating schemes across various levels of indicators (from underlying factors to comprehensive dimensions).

[0106] Advantages and disadvantages analysis: Based on the data, point out the advantages of the recommended solution and the potential risks of other solutions.

[0107] In this case, the invention is applied to quantitatively evaluate treatment plan ①, treatment plan ②, and treatment plan ③ using the above method. The results show that: Solution 1: The value is extremely high (the platform is overcrowded for a long time), although and The recovery time was relatively short, but the overall score was the lowest.

[0108] Solution 3: Excellent value, but The value is relatively high due to the high costs of public transportation connections and passenger clearing. The total passenger delay may not be optimal due to connection efficiency.

[0109] Solution 2: In They performed well in and The best balance has been achieved, therefore the overall score is... It was the highest score and was recommended.

[0110] Using the traditional manual decision-making process, it takes approximately 60 minutes from the occurrence of an event to the full implementation of core measures (short-haul operation and single-track bidirectional operation). However, under the same conditions, the decision-making cycle can be shortened to less than 10 minutes using this invention, improving decision-making efficiency by over 80%. Based on this calculation, the method of this invention can reduce the maximum waiting interval on the line caused by this event from 58 minutes to less than 25 minutes, reducing the total delay time for passengers across the entire network by approximately 40%.

[0111] Based on the same inventive concept, this invention also provides an intelligent agent for implementing the above-described method for dynamic simulation and evaluation of emergency response measures, comprising: a perception intelligent agent, an event analysis intelligent agent, a simulation management intelligent agent, and a decision optimization intelligent agent. The perception intelligent agent integrates emergency event information with real-time network operation data to obtain an initial event data packet; the event analysis intelligent agent outputs multiple sets of response measure schemes based on the initial event data packet; the simulation management intelligent agent uses simulation tools to perform parallel simulations of the multiple sets of response measure schemes, outputting multiple sets of simulation result datasets; and the decision optimization intelligent agent outputs a decision support report based on the multiple sets of simulation result datasets.

[0112] The event analysis agent is configured with: a perception interface, a domain knowledge base, an inference engine, and a task memory and state manager.

[0113] The core of intelligent agent implementation is based on the input initial event data packet, through autonomous knowledge retrieval and multi-step reasoning, to generate a deep, structured operational impact analysis. Specifically: Perception Interface: Responsible for receiving and parsing initial event data packets from the perceptual agent. This completes the structured extraction of basic information about the event.

[0114] Domain Knowledge Base (K_rail): Provides a structured domain knowledge base built on knowledge graphs, storing machine-readable expert knowledge such as device associations, topology, operation rules, and influence propagation patterns, which serves as the basis for intelligent agents to make professional judgments.

[0115] Reasoning Engine: Built on a large language model pre-trained and fine-tuned based on rail transit professional corpus, it has powerful natural language understanding and logical reasoning capabilities and is the core of the intelligent agent's thinking.

[0116] Working Memory & State Manager: During the reasoning process, it dynamically stores initial events, intermediate conclusions, and final results, maintains the contextual consistency of reasoning, and enables multi-step chained reasoning.

[0117] Furthermore, the present invention also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store computer programs; and the processor is used to execute the programs stored in the memory to implement the above-mentioned dynamic simulation and evaluation method for emergency response measures.

[0118] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0119] The communication interface is used for communication between the aforementioned terminal and other devices.

[0120] The memory may include random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0121] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0122] In summary, this invention has the following technological innovations: 1. A dynamic simulation and evaluation system architecture for rail transit emergency response measures oriented towards emergency decision-making was constructed.

[0123] This invention designs a system consisting of an event-aware intelligent agent, an event-analyzing intelligent agent, a simulation-management intelligent agent, and a decision-making optimization intelligent agent. Through the workflow of "perception-analysis-deduction-decision", it achieves autonomous collaboration and jointly constructs a complete intelligent decision-making closed loop, realizing a paradigm shift from passive response to proactive decision-making.

[0124] 2. An event analysis agent with deep reasoning capabilities was constructed: This invention creates an event analysis agent that integrates a knowledge base and inference engine in the rail transit field. It possesses multi-step chain-like reasoning capabilities and can autonomously complete the entire derivation process from direct impact analysis to train operation impact chains and passenger transport impact chains. It innovatively incorporates spatiotemporal distribution knowledge, such as passenger flow OD matrices, into the reasoning process, enabling accurate identification of the impact range and quantitative risk prediction.

[0125] 3. A simulation management agent based on the MCP protocol (Model Context Protocol) was constructed: This invention creates a simulation management agent based on a model context protocol, which has the ability to automatically convert AI-generated response measures into simulation tool instructions, supports parallel simulation and deduction of multiple schemes, and solves the "availability" problem of professional simulation tools in emergency scenarios.

[0126] 4. A multi-dimensional quantitative decision-making intelligent agent was constructed: This invention creates a decision-making agent that integrates a multi-objective evaluation algorithm. It can automatically extract quantitative indicators of three dimensions—safety, service, and economy—from simulation results. It also innovatively establishes a target decision-making model based on a weighted scoring method, enabling automatic ranking of multiple options and recommendation of the optimal option.

[0127] 5. A standardized interaction protocol for efficient collaboration among intelligent agents has been established: This invention defines a standardized data structure and interface protocol for interaction between intelligent agents. Specifically, in step S1, this invention forms a structured initial event data packet. The system completes the structured extraction of basic event information, outputs structured response measures in step S2, and translates the three response measures into executable instructions for the wire network simulation tool in step S3, ensuring unambiguous and efficient information transmission. More importantly, it introduces task memory and a state manager to persistently store event context, intermediate inference results, and deduction data throughout the decision-making process. This allows subsequent agents to operate based on a complete context, avoiding information fragmentation and ensuring that the entire chain from event perception to final decision-making is logically rigorous and state-coherent, which is a key guarantee for achieving minute-level emergency decision-making.

[0128] Compared to traditional modular emergency command systems, this invention has the following advantages: 1. Decision-Making Autonomy: A Paradigm Shift from "Tool-Assisted" to "Agent-Driven" Traditional systems, acting as passive tools, rely entirely on manual intervention at every stage. In contrast, this invention's three core intelligent agents—event analysis, simulation management, and decision optimization—form an autonomous decision-making chain. This allows for the autonomous completion of the entire closed loop from event recognition to optimal response measures, freeing dispatchers from arduous information processing and solution searching, enabling them to focus on supervising and executing the final decision. This redefines and optimizes the human-machine roles.

[0129] 2. Response Timeliness: Intelligent agent collaboration significantly improves decision-making speed. Traditional systems rely on manual analysis, simulation configuration, and result comparison, a sequential process that takes hours. In this invention, intelligent agents collaborate asynchronously and in parallel via standardized interfaces. The event analysis agent triggers the simulation management agent immediately after completing its reasoning; while the simulation management agent is performing its deduction, the decision-making optimization agent is already ready. This efficient collaboration reduces the "analysis-deduction-decision" cycle from hours to minutes, gaining a crucial time window for emergency response.

[0130] 3. Global Optimality of the Solution: Complementary Agent Capabilities Ensure Scientific Decision-Making. Traditional systems, limited by human cognitive load, struggle to make global trade-offs among multiple solutions and objectives. The event analysis agent in this invention ensures the diversity and rationality of solutions; the simulation management agent provides accurate quantitative prediction data; and the decision-making optimization agent performs unbiased quantitative trade-offs among multi-dimensional objectives. The combined capabilities of multiple agents guarantee that the output solution is an approximate optimal solution from a global perspective, rather than a locally feasible solution.

[0131] 4. System Evolution: The intelligent agent possesses the inherent ability for continuous optimization. Traditional systems have static knowledge bases and rule bases that are outdated. The system in this invention has the ability to self-optimize parameters and update the knowledge base based on historical data. The deduction results and actual effects of each action can be used as feedback data, which can be used to optimize the reasoning accuracy of the event analysis agent, correct the model parameters of the simulation management agent, and adjust the evaluation weights of the decision-making optimization agent.

[0132] This invention can bring significant and comprehensive benefits to rail transit operations: 1. Safety Benefit Analysis By leveraging the event analysis intelligence agent's precise predictions and decision-making to prioritize safety considerations, risks such as platform congestion can be proactively avoided. It is projected that the number of passenger safety incidents caused by delayed responses can be reduced by more than 50%. This establishes a proactive safety assurance system for emergency response management, encompassing "pre-event prediction and early warning, and in-event precise prevention and control."

[0133] 2. Operational Benefit Analysis In terms of improving decision-making efficiency, the average emergency decision-making cycle has been shortened from hours to minutes, resulting in an efficiency improvement of over 80%. Regarding cost savings on delays, by recommending the globally optimal solution, the scope of the situation's impact can be effectively controlled, and it is expected that the total delay time for passengers across the network can be reduced. This reduces costs by 30%-50%; in terms of resource utilization optimization, simulations are used to avoid unnecessary resource waste such as bus connections, thereby reducing direct operating costs. This significantly improves the resilience and reliability of the network operation.

[0134] 3. Economic Benefit Analysis based on The model demonstrates that reduced passenger delays directly translate into significant socioeconomic savings. By optimizing resource allocation and lowering the direct operating costs of emergency response, it provides stable and reliable public transportation support for the normal operation of urban economic activities.

[0135] 4. Social Benefit Analysis By providing more reliable and faster emergency services, the number of passenger complaints and negative public opinion can be significantly reduced. Taking the above incident as an example, it is estimated that the number of related public opinions can be reduced by more than 30%, which will greatly improve the passenger travel experience and enhance the public's satisfaction and trust in rail transit services.

[0136] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style of the specification is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0137] The technical principles of the present invention have been described above with reference to specific embodiments. These descriptions are merely for explaining the principles of the invention and should not be construed as limiting the scope of protection of the invention in any way. Based on this explanation, those skilled in the art can readily conceive of other specific embodiments of the invention without inventive effort, and these embodiments will all fall within the scope of protection of the present invention.

Claims

1. A method for dynamic simulation and evaluation of emergency response measures, characterized in that, include: The initial event data packet is obtained by fusing information about emergencies with real-time network operation data. Based on the initial event data packet, output multiple sets of handling measures; Simulation tools were used to perform parallel simulations of multiple response measures, and multiple sets of simulation result datasets were output. Based on multiple sets of simulation result datasets, the optimal treatment plan and decision support report are output.

2. The method for dynamic simulation and evaluation of emergency response measures according to claim 1, characterized in that, The step of fusing emergency information with real-time network operation data to obtain an initial event data packet includes: After receiving and analyzing the emergency, emergency information is obtained. Obtain real-time network operation data; The initial event data packet is obtained by fusing and standardizing information on emergencies and real-time network operation data.

3. The method for dynamic simulation and evaluation of emergency response measures according to claim 1, characterized in that, The step of outputting multiple sets of handling measures based on the initial event data packet includes: Based on the domain knowledge base, reason about the initial event data packet and output the influence chain object; Based on the impact chain object, multiple sets of disposal measures are generated.

4. The method for dynamic simulation and evaluation of emergency response measures according to claim 3, characterized in that, The steps for reasoning about the initial event data packet and outputting the impact chain objects based on the domain knowledge base include: The initial event data is parsed to obtain the event type and the location where the event occurred; Retrieve from the domain knowledge base based on event type and event location; Based on the retrieved knowledge, after sequentially performing direct impact analysis, traffic impact chain reasoning, passenger transport impact chain reasoning, and final risk point derivation, intermediate conclusions and final risk points are obtained. The intermediate conclusions and final risk points are assembled into an impact chain object according to a predefined structured format.

5. The method for dynamic simulation and evaluation of emergency response measures according to claim 4, characterized in that, The steps for conducting passenger transport impact chain reasoning include: Based on the intermediate conclusions obtained from the traffic impact chain reasoning, OD pairs containing fault sections are selected from all paths in the OD matrix, and the indirectly affected passenger flow is identified based on the OD pairs. Analyze the travel paths of indirectly affected passengers, and predict the location and intensity of passenger congestion based on these travel paths. By combining the initial passenger flow and the continuously increasing backlog of passengers at each station without malfunctions, the growth curve of the number of passengers on the platforms of each station without malfunctions is predicted. The growth curve is compared with the platform's safe capacity to obtain the risk trigger time. The continuously increasing backlog of passengers is determined by the rate and duration of the additional backlog of passengers. By analyzing the transfer passenger flow using the OD matrix and combining it with the transfer channel capacity of the station in the domain knowledge base, the risk level of transfer congestion and the location of congestion can be determined. Based on indirectly affected passenger flow, passenger backlog location and intensity, risk trigger time, congestion risk level and congestion occurrence point, a passenger transport impact chain is generated.

6. The method for dynamic simulation and evaluation of emergency response measures according to claim 1, characterized in that, The steps for outputting the optimal treatment plan and decision support report based on multiple sets of simulation result datasets include: Calculate the quantitative metrics for each set of simulation result datasets; Based on preset decision weights, a multi-objective decision-making model is established, and the quantitative indicators are normalized and weighted for comprehensive scoring. The option with the highest overall score is selected as the optimal response plan, and a decision support report is generated.

7. The method for dynamic simulation and evaluation of emergency response measures according to claim 6, characterized in that, The quantitative indicators include: the proportion of time the station platform is overcrowded, the proportion of time the train carriage is overloaded, the total delay time of passengers across the network, the proportion of passengers affected, the increase rate of average travel time, the economic losses of passenger delays, and the additional operating costs of enterprises.

8. A dynamic simulation and evaluation system for emergency response measures, characterized in that, The method for dynamically simulating and evaluating emergency response measures as described in any one of claims 1 to 7 includes: The sensing agent is used to fuse information about emergencies with real-time network operation data to obtain an initial event data packet; The event analysis agent outputs multiple sets of response measures based on the initial event data packet; The simulation management agent uses simulation tools to perform parallel simulations of multiple disposal measures and outputs multiple sets of simulation result datasets. The decision-making agent, based on multiple sets of simulation result datasets, outputs the optimal treatment plan and decision support report.

9. An electronic device, characterized in that, The system includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; the memory is used to store computer programs; and the processor, when executing the programs stored in the memory, implements the dynamic simulation and evaluation method for emergency response measures as described in any one of claims 1 to 7.

Images