Task-driven multi-agent emergency decision support method and device

Abstract

The application relates to a task-driven multi-agent emergency decision support method and device, wherein the method comprises the following steps: constructing a subtask for a risk guidance decision support model, splitting a complex task into simple subtasks, so that a more accurate emergency decision support process is realized; constructing an interactive agent for each subtask, constructing corresponding prompt engineering content according to definition, so that an analysis agent and a test agent for different tasks are constructed; collecting safety analysis report data, and establishing a sample database for emergency operation procedure decision support; model training, giving the sample database to the agent, and constructing a successful case library and a failed case library; inputting super-reference original event information, and generating a decision support procedure for the described original event. Therefore, the problems that an existing emergency plan decision method is extremely dependent on expert knowledge, is limited by human cognition, and it is difficult for on-site personnel to make a quick decision on an event beyond the emergency plan are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis and decision support, in particular to a task-driven multi-agent emergency decision support method and device. BACKGROUND

[0002] Emergency operation procedures refer to work plans prepared in advance by governments at all levels and their departments, grassroots organizations, enterprises and institutions, and social groups, etc. to respond to emergencies in a lawful, rapid, scientific and orderly manner and minimize the damage caused by emergencies.

[0003] With global warming, extreme weather events occur frequently, and many accidents beyond the design benchmark continue to emerge. The existing emergency plan is extremely dependent on expert knowledge and is limited by human cognition. Therefore, it is difficult for on-site personnel to make quick decisions on events beyond the emergency plan, which needs to be addressed urgently. SUMMARY

[0004] The present application provides a task-driven multi-agent emergency decision support method and device to solve the problem that the existing emergency plan decision method is extremely dependent on expert knowledge and is limited by human cognition, making it difficult for on-site personnel to make quick decisions on events beyond the emergency plan.

[0005] The first aspect of the present application provides a task-driven multi-agent emergency decision support method, comprising the following steps: constructing a plurality of sub-tasks of a preset risk guidance decision support model, and establishing an interactive agent corresponding to each sub-task in the plurality of sub-tasks; collecting safety analysis report data of a target safety field to establish an emergency decision support sample database through the safety analysis report data, and training the interactive agent using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database; obtaining super-benchmark originating event information of a target user, and obtaining question and answer pair data in the success case library and the failure case library that meet a preset content similarity condition with the super-benchmark originating event information through the interactive agent, to generate a decision support procedure corresponding to the super-benchmark originating event information according to the question and answer pair data.

[0006] Optionally, in an embodiment of the present application, the plurality of sub-tasks of constructing the preset risk guidance decision support model are established, and the corresponding interactive agent of each sub-task in the plurality of sub-tasks is established, comprising: establishing a plurality of sub-tasks of the risk guidance decision support model, wherein the plurality of sub-tasks include an originating event sub-event analysis and identification task, a header event analysis task of an event tree, and a decision suggestion support task; constructing a first task interactive agent corresponding to the originating event sub-event analysis and identification task, wherein the first task interactive agent includes an originating event sub-event analysis agent and an originating event sub-event verification agent; establishing a second task interactive agent corresponding to the header event analysis task of the event tree, and constructing a third task interactive agent corresponding to the decision suggestion support task, wherein the second task interactive agent includes an event tree header event analysis agent and an event tree header event verification agent, and the third task interactive agent includes a strategy formulation agent and a strategy verification agent.

[0007] Optionally, in an embodiment of the present application, the safety analysis report data of the target safety field is collected to establish an emergency decision support sample database through the safety analysis report data, and the interactive agent is trained by using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database, comprising: collecting safety analysis report data corresponding to each sub-task in the target safety field, and sending the safety analysis report data to the corresponding interactive agent to determine whether the safety analysis report data meets a preset successful decision requirement through the interactive agent; if the safety analysis report data meets the preset successful decision requirement, a success case corresponding to the safety analysis report data is generated, the success case is compiled, and the compiled success case is stored in the success case library; if the safety analysis report data does not meet the preset successful decision requirement, a failure case corresponding to the safety analysis report data is generated, the failure case is compiled, and the compiled failure case is stored in the failure case library.

[0008] Optionally, in an embodiment of the present application, the obtaining, by the interactive agent, the question and answer pair data in the success case library and the failure case library that meets the preset content similarity condition with the super-reference originating event information, to generate the decision support procedure corresponding to the super-reference originating event information according to the question and answer pair data, comprises: inputting the super-reference originating event information into the first task interactive agent, to query the first question and answer pair data that meets the preset content similarity condition with the super-reference originating event information in the success case library and the failure case library corresponding to the first task interactive agent; adding the first question and answer pair data to the first prompt engineering content corresponding to the first task interactive agent, and inputting the first question and answer pair data into the second task interactive agent, to query the second question and answer pair data that meets the preset content similarity condition with the first question and answer pair data in the success case library and the failure case library corresponding to the second task interactive agent; adding the second question and answer pair data to the second prompt engineering content corresponding to the second task interactive agent, and inputting the second question and answer pair data into the third task interactive agent, to obtain the decision support procedure corresponding to the super-reference originating event information; adding the decision support procedure to the third prompt engineering content corresponding to the third task interactive agent, and optimizing the risk guidance decision support model through the first prompt engineering content, the second prompt engineering content and the third prompt engineering content, to optimize the decision support procedure corresponding to the super-reference originating event information based on the optimized risk guidance decision support model.

[0009] The second aspect embodiment of the present application provides a task-driven multi-agent emergency decision support device, comprising: a establishing module configured to construct a plurality of sub-tasks of a preset risk guidance decision support model, and establish an interactive agent corresponding to each sub-task in the plurality of sub-tasks; a training module configured to collect safety analysis report data of a target safety field, to establish an emergency decision support sample database through the safety analysis report data, and train the interactive agent by using the emergency decision support sample database, to generate a success case library and a failure case library corresponding to the emergency decision support sample database; and an emergency decision support module configured to obtain super-reference originating event information of a target user, and obtain question and answer pair data in the success case library and the failure case library that meets a preset content similarity condition with the super-reference originating event information through the interactive agent, to generate a decision support procedure corresponding to the super-reference originating event information according to the question and answer pair data.

[0010] Optionally, in an embodiment of the present application, the establishing module comprises: a first constructing unit configured to establish a plurality of sub-tasks of the risk guidance decision support model, wherein the plurality of sub-tasks comprise an originating event sub-event analysis and identification task, a header event analysis task of an event tree, and a decision suggestion support task; a second constructing unit configured to construct a first task interaction intelligent agent corresponding to the originating event sub-event analysis and identification task, wherein the first task interaction intelligent agent comprises an originating event sub-event analysis intelligent agent and an originating event sub-event verification intelligent agent; and a third constructing unit configured to establish a second task interaction intelligent agent corresponding to the header event analysis task of the event tree, and to construct a third task interaction intelligent agent corresponding to the decision suggestion support task, wherein the second task interaction intelligent agent comprises an event tree header event analysis intelligent agent and an event tree header event verification intelligent agent, and the third task interaction intelligent agent comprises a strategy formulation intelligent agent and a strategy verification intelligent agent.

[0011] Optionally, in an embodiment of the present application, the training module comprises: an acquisition unit configured to acquire safety analysis report data corresponding to each sub-task in the target safety field, and to send the safety analysis report data to a corresponding interaction intelligent agent, so as to determine, by the interaction intelligent agent, whether the safety analysis report data satisfies a preset successful decision requirement; a first generation unit configured to generate a successful case corresponding to the safety analysis report data if the safety analysis report data satisfies the preset successful decision requirement, to compile the successful case, and to store the compiled successful case in the successful case library; and a second generation unit configured to generate a failure case corresponding to the safety analysis report data if the safety analysis report data does not satisfy the preset successful decision requirement, to compile the failure case, and to store the compiled failure case in the failure case library.

[0012] Optionally, in an embodiment of the present application, the emergency decision support module comprises: a query unit configured to input the super-reference originating event information into the first task interaction agent to query first question-answer pair data that meets the preset content similarity condition with the super-reference originating event information in the success case library and the failure case library corresponding to the first task interaction agent; a first adding unit configured to add the first question-answer pair data into first prompt engineering content corresponding to the first task interaction agent, and input the first question-answer pair data into the second task interaction agent to query second question-answer pair data that meets the preset content similarity condition with the first question-answer pair data in the success case library and the failure case library corresponding to the second task interaction agent; a second adding unit configured to add the second question-answer pair data into second prompt engineering content corresponding to the second task interaction agent, and input the second question-answer pair data into the third task interaction agent to obtain a decision support procedure corresponding to the super-reference originating event information; and an optimization unit configured to add the decision support procedure into third prompt engineering content corresponding to the third task interaction agent, and optimize the risk guidance decision support model through the first prompt engineering content, the second prompt engineering content and the third prompt engineering content, so as to optimize the decision support procedure corresponding to the super-reference originating event information based on the optimized risk guidance decision support model.

[0013] The third aspect of the embodiments of the present application provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the task-driven multi-agent emergency decision support method as described in the above embodiments.

[0014] The fourth aspect of the embodiments of the present application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the task-driven multi-agent emergency decision support method as described above.

[0015] The fifth aspect of the embodiments of the present application provides a computer program product, comprising a computer program, and the computer program is executed to implement the task-driven multi-agent emergency decision support method as described above.

[0016] Therefore, the embodiments of the present application have the following beneficial effects:

[0017] Embodiments of the present application can build a plurality of sub-tasks of a preset risk guidance decision support model, and establish an interactive agent corresponding to each of the plurality of sub-tasks; collect safety analysis report data of a target safety field, to establish an emergency decision support sample database through the safety analysis report data, and train the interactive agent using the emergency decision support sample database, to generate a success case library and a failure case library corresponding to the emergency decision support sample database; obtain super-reference originating event information of a target user, and obtain question and answer pair data in the success case library and the failure case library that satisfy a preset content similarity condition with the super-reference originating event information through the interactive agent, to generate a decision support procedure corresponding to the super-reference originating event information according to the question and answer pair data, so that decision suggestions can be quickly obtained without too much expert knowledge, the work efficiency of emergency decision is improved, and the impact of accident consequences is reduced. Thus, the problem that the existing emergency plan decision method is extremely dependent on expert knowledge, is limited by human cognition, and it is difficult for on-site personnel to quickly make decisions on events that exceed the emergency plan is solved.

[0018] Additional aspects and advantages of the present application will be made apparent by the following description and the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0019] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description, taken in conjunction with the accompanying drawings, in which:

[0020] Figure 1 A flowchart of a task-driven multi-agent emergency decision support method according to an embodiment of the present application;

[0021] Figure 2 An execution logic schematic diagram of a task-driven multi-agent emergency decision support method according to an embodiment of the present application;

[0022] Figure 3 An example diagram of a task-driven multi-agent emergency decision support device according to an embodiment of the present application;

[0023] Figure 4 A structural schematic diagram of an electronic device according to an embodiment of the present application.

[0024] Among them, 10-task-driven multi-agent emergency decision support device; 100-establishing module, 200-training module, 300-emergency decision support module; 401-memory, 402-processor, 403-communication interface. DETAILED DESCRIPTION

[0025] Embodiments of the present application are described below in detail, examples of which are shown in the accompanying drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0026] The task-driven multi-agent emergency decision support method and device of the embodiments of the present application are described below with reference to the accompanying drawings. In view of the problems mentioned in the background art, the present application provides a task-driven multi-agent emergency decision support method, in which a plurality of sub-tasks of a preset risk guidance decision support model are constructed, and an interactive agent corresponding to each of the plurality of sub-tasks is established; safety analysis report data of a target safety field is collected to establish an emergency decision support sample database through the safety analysis report data, and the interactive agent is trained using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database; super-reference originating event information of a target user is obtained, and question and answer pair data in the success case library and the failure case library that satisfies a preset content similarity condition with the super-reference originating event information is obtained through the interactive agent, so as to generate a decision support procedure corresponding to the super-reference originating event information according to the question and answer pair data, thereby quickly obtaining decision suggestions without too much expert knowledge, improving the work efficiency of emergency decision, and reducing the impact of accident consequences. Thus, the problem that the existing emergency plan decision method is extremely dependent on expert knowledge, is limited by human cognition, and makes it difficult for on-site personnel to quickly make decisions on events beyond the emergency plan is solved.

[0027] Specifically, Figure 1 A flowchart of a task-driven multi-agent emergency decision support method provided by an embodiment of the present application is shown in FIG. 1.

[0028] As Figure 1 shown, the task-driven multi-agent emergency decision support method includes the following steps:

[0029] In step S101, a plurality of sub-tasks of a preset risk guidance decision support model are constructed, and an interactive agent corresponding to each of the plurality of sub-tasks is established.

[0030] The embodiment of the present application splits complex tasks into simple sub-tasks by constructing sub-tasks for the risk guidance decision support model, so as to realize a more accurate emergency decision support process; then, the embodiment of the present application also needs to construct an interactive agent for each sub-task, define and construct corresponding prompt engineering content, and realize the construction of analysis agents and inspection agents for different tasks.

[0031] Optionally, in an embodiment of the present application, a plurality of sub-tasks of the preset risk guidance decision support model are constructed, and an interactive agent corresponding to each of the plurality of sub-tasks is established, including: establishing a plurality of sub-tasks of the risk guidance decision support model, wherein the plurality of sub-tasks include an originating event sub-event analysis and identification task, a header event analysis task of an event tree, and a decision suggestion support task; constructing a first task interactive agent corresponding to the originating event sub-event analysis and identification task, wherein the first task interactive agent includes an originating event sub-event analysis agent and an originating event sub-event verification agent; establishing a second task interactive agent corresponding to the header event analysis task of the event tree, and constructing a third task interactive agent corresponding to the decision suggestion support task, wherein the second task interactive agent includes an event tree header event analysis agent and an event tree header event verification agent, and the third task interactive agent includes a strategy formulation agent and a strategy verification agent.

[0032] It should be noted that in the process of constructing sub-tasks for the risk guidance decision support model, the embodiments of the present application mainly refer to the core idea of the event tree, and realize the emergency decision process by combining probability risk analysis, which divides the entire emergency decision process into the following three sub-tasks:

[0033] Task 1: Originating event sub-event analysis and identification (i.e., originating event sub-event analysis and identification task).

[0034] This task is the basis for the analysis of subsequent task 2 and task 3, and the input is the originating event name and description, and the output is the sub-event of the originating event.

[0035] Task 2: Header event analysis of event tree, including identification and listing of possible header events that may lead to accidents in order (i.e., header event analysis task of event tree).

[0036] This task systematically analyzes the subsequent events of the originating event, and the input is the originating event, the originating event description, the sub-event (generated by task 1), the event process and system response, and the output is the header event of the event tree.

[0037] Task 3: Decision suggestion (i.e., decision suggestion support task)

[0038] After identifying and analyzing potential sub-events and header events, specific operational decisions are formulated and provided to cope with and mitigate adverse effects, and the input is the originating event, the originating event description, the sub-event (generated by task 1), the event process and system response, and the header event of the event tree (generated by task 2), and the output is the operator's action suggestion.

[0039] In addition, the interactive intelligent agent designed in the embodiments of the present application includes a first task interactive intelligent agent corresponding to task one, i.e., an initiating event sub-event analysis intelligent agent and an initiating event sub-event verification intelligent agent; a second task interactive intelligent agent corresponding to task two, i.e., an event tree question head event analysis intelligent agent and an event tree question head event verification intelligent agent; and a third task interactive intelligent agent corresponding to task three, i.e., a strategy formulation intelligent agent and a strategy verification intelligent agent.

[0040] The corresponding information of the interactive intelligent agent corresponding to each task is as follows:

[0041] 1. Initiating event sub-event analysis intelligent agent:

[0042] The initiating event sub-event analysis intelligent agent is responsible for identifying the sub-events of the initiating event. The main task of this intelligent agent is to examine detailed descriptions and cases, including correct and incorrect cases. For incorrect cases, the intelligent agent must summarize feedback to prevent the error from occurring again. For correct cases, the intelligent agent should refer to the logic applied. When there are no correct or incorrect cases, the intelligent agent needs to analyze the sub-events of the initiating event.

[0043] 2. Initiating event sub-event verification intelligent agent:

[0044] The initiating event sub-event verification intelligent agent is responsible for verifying the accuracy of the sub-events identified by the analysis intelligent agent. Based on the initiating event and its detailed description, the intelligent agent determines whether the identified sub-events match the actual sub-events, with judgment criteria including multiple key aspects to ensure the comprehensiveness and accuracy of the verification process.

[0045] 3. Event tree question head event analysis intelligent agent:

[0046] The event tree question head event analysis intelligent agent is specifically responsible for identifying and listing the question head events in the event tree. This intelligent agent also needs to summarize feedback from incorrect cases to avoid repeating errors and refer to the logic used in correct cases. By using feedback from incorrect cases, the intelligent agent can continuously improve future responses.

[0047] 4. Event tree question head event verification intelligent agent:

[0048] The event tree question head event verification intelligent agent is responsible for verifying the accuracy of the question head events provided by the question head event analysis intelligent agent. This intelligent agent ensures that the identified question head events match the actual event progress of the nuclear power plant. The verification process involves multiple key steps to maintain the reliability and integrity of the analysis, ensuring the accuracy and safety of operations.

[0049] 5. Strategy formulation intelligent agent:

[0050] The strategy-making agent of the nuclear power plant is responsible for determining the specific operations that the operator should perform according to the initiating event, event description, sub-event, event sequence, and examples of correct and incorrect responses; the agent must ensure that these operations are detailed and accurate. For error cases, the agent needs to summarize feedback to prevent the error from recurring, and analyze it by referring to the logic in the correct case.

[0051] 6、Strategy verification agent:

[0052] The strategy verification agent is responsible for evaluating the recommendations provided by the nuclear power plant strategy consultant, and the agent determines whether the recommendations are consistent with the actual operator's recommendations; by accurately verifying the strategy recommendations, the agent ensures the safety and reliability of the nuclear power plant operation.

[0053] It should be noted that the agents of the three tasks are driven by large language models, mainly using prompt engineering, and the prompt engineering content of each agent is as follows:

[0054] 1、For the prompt engineering content of the initiating event sub-event analysis agent, it is as follows:

[0055] "As a nuclear power plant initiating event sub-event analyst, you are responsible for analyzing and identifying the sub-events of the initiating event based on the given detailed description, correct cases, and error cases; for error cases, you need to summarize feedback to prevent recurrence; for correct cases, you should refer to their logic to guide your analysis; if there are no correct or error cases, you must analyze the sub-events of the initiating event based on your ability and the detailed description of the initiating event; please note that an initiating event may have only one sub-event, and the initiating event may cause serious consequences; the output format of the sub-event list is as follows: "Nuclear power plant initiating event sub-event analyst: <1. Sub-event 1 2. Sub-event 2 3. Sub-event 3> ". Make sure to only provide the sub-event list without additional output, and pay attention to the correctness of the units. In addition, try to specify the dimensions in your analysis."

[0056] 2、The prompt engineering content of the initiating event sub-event verification agent is as follows:

[0057] "You are the initial event sub-event verification officer of the nuclear power plant, according to the given initial event and its detailed description, your task is to determine whether the sub-events provided by the initial event analysis personnel are consistent with the actual sub-events, the judgment criteria are as follows:

[0058] 1) If the sub-events provided by the analysis personnel are more detailed than the actual sub-events, but the described phenomena are roughly or essentially the same, return "Yes"; otherwise, return "No";

[0059] 2) If the sub-events provided by the analyst generally cover the content of the actual event, and the size information of the dimensional sub-events is consistent, return "Yes"; otherwise, return "No";

[0060] 3) If the actual sub-event contains size information, but the event provided by the analyst lacks size information or the size information is incorrect, return "No";

[0061] 4) If the event provided by the analyst contains sub-events that do not belong to the initial event, return "No".

[0062] Please note that you should provide reasons regardless of your judgment, "Yes" or "No".

[0063] 3. The prompt engineering content of the event tree topic event analysis agent is as follows:

[0064] "You are an expert in nuclear power plant event tree. According to the provided initial event, sub-event, event progress and system response of the nuclear power plant, please analyze the topic event of the event tree; the specific requirements are as follows: only provide the topic event, ensure the correct order; the topic event should reflect the equipment or function, not the fault; the order of the topic event is crucial, it must be arranged according to the event progress, fully and accurately covering all important events; for incorrect cases, summarize feedback to avoid repeating errors; for correct cases, refer to their logic to help your analysis; use the reasons in the error cases to improve your response. The output format should be as follows: Nuclear power plant event tree topic event analyst: <1. Topic event 1\n 2. Topic event 2\n 3. Topic event 3... >"

[0065] 4. The prompt engineering content of the event tree topic event verification agent is as follows:

[0066] "According to the provided information, determine whether the topic event provided by the nuclear power plant event tree analyst is correct. If the topic event provided by the analyst is consistent with the actual event progress, output "Yes"; otherwise, output "No" and provide reasons; if the topic event provided by the analyst is more than the actual event, but covers all actual events, also return "Yes"; if the topic event provided by the analyst is broader, but contains the actual topic event, also return "Yes"; please note that the response must contain "Yes" or "No"; now, based on these rules and the actual topic event, determine whether the following event tree topic event analysis result is correct; please note that the output does not need to be exactly the same as the actual topic event; if the content or direction is roughly the same, it is considered correct."

[0067] 5. The prompt engineering content of the strategy-making agent is as follows:

[0068] "As a nuclear power plant expert, you can determine the specific actions that operators should perform based on the initiating event, its description, sub-events, event sequences, and examples of correct and incorrect responses; the actions must be detailed and accurate; for the wrong cases, you need to summarize the feedback to prevent the error from happening again; for the right cases, you should refer to its logic to guide your analysis; if no correct or incorrect cases are provided, you must analyze and determine the specific actions that operators should take using your professional knowledge and the given information."

[0069] 6. The prompt engineering content of the policy verification agent is as follows:

[0070] "Based on the provided information, evaluate the recommendations provided by the nuclear power plant policy proposer. Determine whether these recommendations are consistent with the actual operator's recommendations; if the content is roughly consistent and the operator's recommendations are more specific, but follow the same direction, output "Yes"; otherwise, output "No" and provide a reason. Ensure that the response contains "Yes" or "No"; note that since the nuclear power plant policy consultant lacks procedural knowledge, if their recommendations can prevent accidents from occurring, the response should be "Yes"; in addition, if their recommendations only lack references to specific accident handling procedures and detailed descriptions, the response should also be "Yes"; in addition, the response should focus on procedures, emphasizing the implementation of specific measures based on specific accident procedures (such as Procedure A and Procedure A1.1); focus on switching the operation mode of the injection system according to the real-time state of the reactor. Now, please determine whether the following operator recommendations are consistent according to these rules: output in the following format: "Nuclear power plant policy consultant: <action to action plan 1>".

[0071] Therefore, the embodiments of the present application provide reliable technical support for the implementation of multi-agent emergency decision support by constructing sub-tasks and interactive agents for risk guidance decision support models.

[0072] In step S102, safety analysis report data of a target safety field is collected to establish an emergency decision support sample database through the safety analysis report data, and an interactive agent is trained using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database.

[0073] Further, the embodiments of the present application also need to collect different system components and corresponding initiating events, initiating event descriptions, sub-events, event processes and system responses, event tree header events, and operator action recommendations from safety analysis reports in related fields, to obtain safety analysis report data and construct an emergency operation procedure decision support sample database (i.e., an emergency decision support sample database).

[0074] Afterwards, the embodiment of the present application can send the emergency decision support sample database to the interactive agent, and construct a successful case library and a failure case library. The specific construction process of the successful case library and the failure case library is as follows:

[0075] 1. Successful case library construction:

[0076] In the process of completing the task, it is beneficial for the operator to refer to the previously verified successful cases, which contain rich knowledge and show accurate and appropriate response logic to various emergency situations; therefore, the embodiment of the present application proposes to construct a successful case library for the agent to improve its ability; the successful case library is constructed in the form of question and answer pair, in which the question describes the basic situation of the original event that needs to complete the task, and the answer contains the verified positive answer of the agent; for each answer generated by the agent, if the answer is correct, the question and answer pair will be added to the successful case.

[0077] 2. Failure case library expansion:

[0078] Learning from mistakes is important for the growth of the operator. The professional agent based on the large language model can reflect on these mistakes, extract relevant principles (experience), and ensure correct completion of the task when encountering similar problems in the future; if the answer is wrong, the question and answer pair will be added to the failure case library.

[0079] Optionally, in an embodiment of the present application, safety analysis report data in a target safety field is collected to establish an emergency decision support sample database through the safety analysis report data, and an interactive agent is trained using the emergency decision support sample database to generate a successful case library and a failure case library corresponding to the emergency decision support sample database, including: collecting safety analysis report data corresponding to each sub-task in the target safety field, and sending the safety analysis report data to the corresponding interactive agent to determine whether the safety analysis report data meets the preset successful decision requirement through the interactive agent; if the safety analysis report data meets the preset successful decision requirement, a successful case corresponding to the safety analysis report data is generated, the successful case is compiled, and the compiled successful case is stored in the successful case library; if the safety analysis report data does not meet the preset successful decision requirement, a failure case corresponding to the safety analysis report data is generated, the failure case is compiled, and the compiled failure case is stored in the failure case library.

[0080] In actual execution, the embodiment of the present application obtains a comprehensive data set distribution (i.e., safety analysis report data) by collecting report data as shown in Table 1:

[0081] Table 1

[0082] Initiating event Abbreviation Number Loss of coolant accident LOCA 5 Loss of heat source accident LOHSA 3 Loss of feedwater event LOFW 2 Loss of offsite power event LOOP 1 Anticipated transient without reactor trip ATWS 7 Main feed line break MFLB 2 Steam generator tube rupture SGTR 6 Main steam line break MSLB 3 Loss of direct current power accident LODC 2 Main steam line break and steam generator tube rupture combined accident MSLB+SGTR 5 Secondary loop transient event SLTE 2

[0083] Table 1 shows that this dataset contains 11 different types of initiation events commonly encountered in nuclear power plants: Loss of Coolant Accident (LOCA) occurred 5 times, Loss of Heat Source Accident (LOHSA) occurred 3 times, Loss of Feedwater Event (LOFW) occurred 2 times, and Loss of External Power Event (LOOP) occurred 1 time. In addition, the most common event is the Expected Transient Without Emergency Reactor Shutdown (ATWS), which occurred 7 times. The dataset also includes 2 instances of Main Feedwater Line Breach (MFLB) and Steam Generator Tube Breach (SGTR), 6 instances of Main Steam Line Breach (MSLB), and 3 instances of Loss of Direct Current (LODC). Furthermore, the combined Main Steam Line Breach and Steam Generator Tube Breach (MSLB+SGTR) event occurred 5 times. Finally, 2 Secondary Loop Transient Events (SLTE) were recorded.

[0084] For Task 1, there are a total of 13 data points, of which 10 are used for training and 3 for testing. For Tasks 2 and 3, there are a total of 38 data points, of which 31 are used for training and 7 for testing.

[0085] Subsequently, this application may send the aforementioned emergency decision support sample database to the interactive intelligent agent for training, thereby constructing a success case library and a failure case library.

[0086] As one possible approach, the training method in this application embodiment is a parameter-free strategy. For each task, the task executor and the task verifier complete the task through continuous interaction and iteration. These three tasks are interconnected through a task flow, jointly achieving the decision support function.

[0087] This strategy has two important modules: a success case library and a failure case library. Successful cases are compiled and stored in the success case library as a reference for future results. For cases that fail, the results are added to the failure case library.

[0088] In the simulation of the three tasks, the embodiments of this application can use a dense searcher to retrieve relevant success case libraries and failure case libraries to help the agent provide better results; as success and failure cases accumulate, they will be actively applied, and the success and failure case libraries will be continuously updated; in addition, whether the answer is correct or not is also completely determined autonomously by the agent.

[0089] In step S103, the super-benchmark originating event information of the target user is obtained, and question-answer pairs that meet the preset content similarity conditions with the super-benchmark originating event information are obtained from the success case library and the failure case library respectively through the interactive intelligent agent, so as to generate the decision support procedure corresponding to the super-benchmark originating event information based on the question-answer pairs.

[0090] Further, the embodiment of the present application also needs to input the originating event information exceeding the design benchmark, and enhance the prompt information of the interactive agent based on the introduced success case library and failure case library; the embodiment of the present application compares the existing content in the success case library and the failure case library with the currently input originating event information, acquires the most similar content to the currently queried originating event information, selects the success and failure question and answer pairs, and adds them to the prompt engineering content, to generate the decision support procedure of the described originating event, and enhance the performance of the model.

[0091] Optionally, in an embodiment of the present application, the success case library and the failure case library are respectively acquired by the interactive agent to obtain the question and answer pair data satisfying the preset content similarity condition with the super-benchmark originating event information, to generate the decision support procedure corresponding to the super-benchmark originating event information according to the question and answer pair data, including: inputting the super-benchmark originating event information into the first task interactive agent, to query the first question and answer pair data satisfying the preset content similarity condition with the super-benchmark originating event information in the success case library and the failure case library corresponding to the first task interactive agent; adding the first question and answer pair data to the first prompt engineering content corresponding to the first task interactive agent, and inputting the first question and answer pair data into the second task interactive agent, to query the second question and answer pair data satisfying the preset content similarity condition with the first question and answer pair data in the success case library and the failure case library corresponding to the second task interactive agent; adding the second question and answer pair data to the second prompt engineering content corresponding to the second task interactive agent, and inputting the second question and answer pair data into the third task interactive agent, to obtain the decision support procedure corresponding to the super-benchmark originating event information; adding the decision support procedure to the third prompt engineering content corresponding to the third task interactive agent, and optimizing the risk guidance decision support model through the first prompt engineering content, the second prompt engineering content and the third prompt engineering content, to optimize the decision support procedure corresponding to the super-benchmark originating event information based on the optimized risk guidance decision support model.

[0092] It should be noted that the embodiment of the present application enhances the prompt information of the agent through the introduced success case library and failure case library. When there is a new query (i.e. super-benchmark originating event information), the system will search the relevant records in the library, and each task has its independent success case library and failure case library, to avoid using irrelevant records; by comparing the existing content in the success case library and the failure case library with the currently input originating event information, the most similar content to the currently queried originating event information is acquired.

[0093] After that, the embodiments of the present application can add the selected successful and failed question and answer pairs to the prompt engineering content to enhance the performance of the model; both the successful and failed search cases are calculated by similarity, and the calculation method is to use the "text-embedding-ada-002" model provided by OpenAI to embed the text into a vector space, and use cosine similarity to calculate the case similarity, and the mathematical expression of the cosine similarity is as follows:

[0094]

[0095] Wherein, Z1 and Z2 respectively represent the vector representation of the current input originating event and the vector representation of the existing content in the successful case library and the failed case library.

[0096] In the embodiments of the present application, the calculated similarity ranges from -1 to 1, wherein -1 indicates that the directions of the two vectors are exactly opposite, 1 indicates that the directions of the two vectors are exactly the same, 0 usually indicates that the two vectors are independent, and the values between them indicate the similarity or dissimilarity between the two vectors.

[0097] In the specific implementation process, the user can input various originating events to the model, and the embodiments of the present application take the originating event of the main steam line rupture accompanied by the steam generator pipe rupture accident as an example, which is not input as a case in the previous training process, so for the embodiments of the present application, it is an originating event beyond the design benchmark.

[0098] Therefore, the embodiments of the present application can first input the originating event information into the agent process of task one, and the agent inputs the content in the successful case library and the failed case library, and the output result is as follows:

[0099]

[0100] 1. Single pipe rupture

[0101] 2. Multi-pipe rupture

[0102] 3. Complete pipe rupture

[0103]

[0104] Secondly, the output of task one is input to task two, which is processed by the event tree head event analysis agent, and the input includes the originating event, the originating event description, the event description and the sub-event, and the final output event tree head event is as follows:

[0105]

[0106] 1. Reactor trip

[0107] 2. Main steam isolation valve automatic closure​​​

[0108] 3. Main feedwater system isolation

[0109] 4. Auxiliary feedwater system initiation

[0110] 5. Safety injection initiation

[0111] 6. Primary loop depressurization

[0112] 7. Primary loop cooling

[0113] 8. Normal coolant inventory restoration

[0114] ”

[0115] After that, these results are input to the strategy-making agent of task three, the input includes the initiating event, initiating event description, event description, sub-event, event tree head event, event process and system response, and the final output is the strategy support as follows:

[0116] “

[0117] 1. Event confirmation: Confirm that the emergency alarm system has been activated. Verify that the accident is a combination of main steam line rupture and steam generator pipe rupture.

[0118] 2. Emergency reactor shutdown: Confirm that the emergency shutdown signal has been activated, which may be triggered by the safety injection signal, low pressurizer pressure or extremely low steam generator water level signal. Ensure that the reactor is in a subcritical state and safely shut down.

[0119] 3. Confirm the activation of any of the following signals: low pressurizer pressure, high pressure difference between two steam lines, high steam line flow and low steam line pressure, or low primary loop average temperature. Start the safety injection system to supplement the primary loop water inventory. Ensure that the auxiliary feedwater system supplies water to the steam generator.

[0120] 4. Main steam isolation: Confirm that the main steam isolation valve has been closed, triggered by low steam pressure and high steam flow.

[0121] 5. Confirm that the high-pressure injection system injects coolant into the primary loop. Start the medium-pressure injection system as needed to maintain the primary loop water inventory. Ensure that the low-pressure injection system is started when necessary to provide coolant for the primary loop.

[0122] 6. Start the depressurization and cooling measures.

[0123] ”

[0124] To sum up, the embodiment of the application first constructs sub-tasks for risk guidance decision support model, including initial event sub-time analysis, event tree topic event analysis, and emergency decision suggestion; secondly, through prompt engineering, interactive agents for each sub-task are constructed based on large language model; then, data is collected to establish a sample database for emergency operation procedure decision support; the data is given to the agent for training iteration to construct a successful case and a failure case library; finally, the initial event information beyond the design benchmark is input to generate the decision support procedure of the described initial event beyond the design benchmark, so that industry personnel can quickly obtain decision suggestions without too much expert knowledge when facing accidents beyond the design benchmark, greatly improving the work efficiency of emergency decision.

[0125] The execution logic of the task-driven multi-agent emergency decision support method of the application is described below by combining the accompanying drawings.

[0126] Figure 2 The execution logic of the task-driven multi-agent emergency decision support method of the application is described below by combining the accompanying drawings. Figure 2 As shown in the figure, the execution process of the task-driven multi-agent emergency decision support method of the application is described as follows:

[0127] S201: Construct sub-tasks for risk guidance decision support model, and split complex tasks into simple sub-tasks to realize more accurate emergency decision support process;

[0128] S202: Construct interactive agents for each sub-task, and construct corresponding prompt engineering content according to definition to realize construction of analysis agents and inspection agents for different tasks;

[0129] S203: Collect safety analysis report data to establish a sample database for emergency operation procedure decision support;

[0130] S204: Model training, give the sample database to the agent to construct a successful case library and a failure case library;

[0131] S205: Input the initial event information beyond the design benchmark to generate the decision support procedure of the described initial event.

[0132] According to the task-driven multi-agent emergency decision support method provided in the embodiments of the present application, a plurality of sub-tasks of a preset risk guidance decision support model are constructed, and an interactive agent corresponding to each of the plurality of sub-tasks is established; safety analysis report data of a target safety field is collected to establish an emergency decision support sample database through the safety analysis report data, and the interactive agent is trained by using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database; super-reference originating event information of a target user is acquired, and question and answer pair data meeting a preset content similarity condition with the super-reference originating event information in the success case library and the failure case library is acquired through the interactive agent, so as to generate a decision support procedure corresponding to the super-reference originating event information according to the question and answer pair data, thereby quickly obtaining decision suggestions without too much expert knowledge, improving the work efficiency of emergency decision, and reducing the influence of accident consequences.

[0133] Secondly, the task-driven multi-agent emergency decision support device provided in the embodiments of the present application is described with reference to the drawings.

[0134] Figure 3 is a block schematic diagram of the task-driven multi-agent emergency decision support device of the embodiments of the present application.

[0135] As shown in Figure 3 , the task-driven multi-agent emergency decision support device 10 comprises an establishing module 100, a training module 200, and an emergency decision support module 300.

[0136] The establishing module 100 is configured to construct a plurality of sub-tasks of a preset risk guidance decision support model, and establish an interactive agent corresponding to each of the plurality of sub-tasks.

[0137] The training module 200 is configured to collect safety analysis report data of a target safety field to establish an emergency decision support sample database through the safety analysis report data, and train the interactive agent by using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database.

[0138] The emergency decision support module 300 is configured to acquire super-reference originating event information of a target user, and acquire question and answer pair data meeting a preset content similarity condition with the super-reference originating event information in the success case library and the failure case library through the interactive agent, so as to generate a decision support procedure corresponding to the super-reference originating event information according to the question and answer pair data.

[0139] Optionally, in an embodiment of the present application, the establishing module 100 comprises a first construction unit, a second construction unit, and a third construction unit.

[0140] The first construction unit is configured to establish a plurality of subtasks of the risk guidance decision support model, wherein the plurality of subtasks include an originating event subevent analysis and identification task, a header event analysis task of an event tree, and a decision suggestion support task.

[0141] The second construction unit is configured to construct a first task interaction intelligent agent corresponding to the originating event subevent analysis and identification task, wherein the first task interaction intelligent agent includes an originating event subevent analysis intelligent agent and an originating event subevent verification intelligent agent.

[0142] The third construction unit is configured to construct a second task interaction intelligent agent corresponding to the header event analysis task of the event tree and a third task interaction intelligent agent corresponding to the decision suggestion support task, wherein the second task interaction intelligent agent includes an event tree header event analysis intelligent agent and an event tree header event verification intelligent agent, and the third task interaction intelligent agent includes a strategy formulation intelligent agent and a strategy verification intelligent agent.

[0143] Optionally, in an embodiment of the present application, the training module 200 includes a collection unit, a first generation unit, and a second generation unit.

[0144] The collection unit is configured to collect safety analysis report data corresponding to each subtask in a target safety field, and send the safety analysis report data to a corresponding interaction intelligent agent, so as to determine whether the safety analysis report data meets a preset successful decision requirement through the interaction intelligent agent.

[0145] The first generation unit is configured to generate a successful case corresponding to the safety analysis report data if the safety analysis report data meets the preset successful decision requirement, compile the successful case, and store the compiled successful case in a successful case library.

[0146] The second generation unit is configured to generate a failure case corresponding to the safety analysis report data if the safety analysis report data does not meet the preset successful decision requirement, compile the failure case, and store the compiled failure case in a failure case library.

[0147] Optionally, in an embodiment of the present application, the emergency decision support module 300 includes a query unit, a first adding unit, a second adding unit, and an optimization unit.

[0148] The query unit is configured to input super-reference originating event information into the first task interaction intelligent agent, so as to query a first question and answer pair data meeting a preset content similarity condition from the super-reference originating event information in the successful case library and the failure case library corresponding to the first task interaction intelligent agent.

[0149] The first adding unit is configured to add the first question and answer pair data into the first prompt engineering content corresponding to the first task interaction agent, and input the first question and answer pair data into the second task interaction agent to query second question and answer pair data meeting a preset content similarity condition from the first question and answer pair data in the success case library and the failure case library corresponding to the second task interaction agent.

[0150] The second adding unit is configured to add the second question and answer pair data into the second prompt engineering content corresponding to the second task interaction agent, and input the second question and answer pair data into the third task interaction agent to obtain the decision support procedure corresponding to the super-reference originating event information.

[0151] The optimization unit is configured to add the decision support procedure into the third prompt engineering content corresponding to the third task interaction agent, and optimize the risk guidance decision support model through the first prompt engineering content, the second prompt engineering content and the third prompt engineering content, so as to optimize the decision support procedure corresponding to the super-reference originating event information based on the optimized risk guidance decision support model.

[0152] It should be noted that the foregoing explanation and description of the task-driven multi-agent emergency decision support method embodiment also apply to the task-driven multi-agent emergency decision support device of this embodiment, which will not be described here again.

[0153] The task-driven multi-agent emergency decision support device according to the embodiment of the present application comprises a establishing module configured to construct a plurality of sub-tasks of a preset risk guidance decision support model, and establish an interaction agent corresponding to each sub-task in the plurality of sub-tasks; a training module configured to collect safety analysis report data of a target safety field, to establish an emergency decision support sample database through the safety analysis report data, and to train the interaction agent by using the emergency decision support sample database, so as to generate a success case library and a failure case library corresponding to the emergency decision support sample database; and an emergency decision support module configured to obtain super-reference originating event information of a target user, and to obtain question and answer pair data meeting a preset content similarity condition from the super-reference originating event information in the success case library and the failure case library through the interaction agent, so as to generate a decision support procedure corresponding to the super-reference originating event information according to the question and answer pair data, thereby quickly obtaining decision suggestions without too much expert knowledge, improving the work efficiency of emergency decision, and reducing the impact of accident consequences.

[0154] Figure 4 A structural schematic diagram of an electronic device is provided for the embodiment of the present application. The electronic device can comprise:

[0155] The memory 401, the processor 402, and the computer program stored in the memory 401 and executable on the processor 402.

[0156] The processor 402 implements the task-driven multi-agent emergency decision support method provided in the above embodiments when executing a program.

[0157] Further, the electronic device further comprises:

[0158] The communication interface 403 is configured to communicate between the memory 401 and the processor 402.

[0159] The memory 401 is configured to store a computer program executable on the processor 402.

[0160] The memory 401 can include a high-speed RAM memory, and can further include a non-volatile memory, for example, at least one disk memory.

[0161] If the memory 401, the processor 402 and the communication interface 403 are implemented independently, the communication interface 403, the memory 401 and the processor 402 can be connected to each other through a bus and complete communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 In the figure, only one thick line is used to represent, but it does not mean that there is only one bus or one type of bus.

[0162] Optionally, in a specific implementation, if the memory 401, the processor 402 and the communication interface 403 are integrated on a chip, the memory 401, the processor 402 and the communication interface 403 can complete communication between each other through an internal interface.

[0163] The processor 402 can be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0164] The embodiments of the present application also provide a computer readable storage medium, having a computer program stored thereon, the program being executed by a processor to implement the task-driven multi-agent emergency decision support method as above.

[0165] The embodiment of the present application further provides a computer program product comprising a computer program, which, when executed, is configured to implement the task-driven multi-agent emergency decision support method.

[0166] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or N embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the present specification and the features of the different embodiments or examples, without contradiction.

[0167] In addition, the terms "first", "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "N" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0168] Any process or method descriptions in flow charts or otherwise described herein represent embodiments of examples that can be implemented as executable instructions stored in a computer readable medium, which can be executed by a processing unit or other computing component of a computer or computer system. In some embodiments, the flow diagrams can be understood as representing a state machine, in which each state subsumes the functionality of one or more state transitions described in the flow diagram. In some embodiments, the flow diagrams can be understood as representing a network of computing components, with each component corresponding to a machine accessible entity capable of executing one or more of the steps of the associated state transition.

[0169] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be considered as a sequence of instructions to implement logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- based system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a computer- readable storage medium or a computer-readable signal medium. The computer- readable storage medium can be, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include the following: an electrical connection having one or more wires (electrical connections), a portable computer diskette (a magnetic storage medium), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). Note that the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via, for example, optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

[0170] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware and in another embodiment, any of the following technologies, known in the art, or their combinations can be used: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0171] Those skilled in the art can understand that all or part of the steps carried out by the above-mentioned embodiment methods can be completed by programs instructing related hardware, and the programs can be stored in a computer-readable storage medium. When the programs are executed, they include one or a combination of the steps of the method embodiments.

[0172] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software functional module. When the integrated module is realized in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0173] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A task-driven multi-agent emergency decision support method, characterized in that, The method comprises the following steps: constructing a plurality of subtasks of a preset risk guidance decision support model and establishing an interactive agent corresponding to each of the plurality of subtasks; collecting safety analysis report data of a target safety field to establish an emergency decision support sample database through the safety analysis report data, and training the interactive agent using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database; obtaining super-reference originating event information of a target user, and obtaining question and answer pair data in the success case library and the failure case library that satisfy a preset content similarity condition with the super-reference originating event information through the interactive agent to generate a decision support procedure corresponding to the super-reference originating event information according to the question and answer pair data; wherein the constructing a plurality of subtasks of a preset risk guidance decision support model and establishing an interactive agent corresponding to each of the plurality of subtasks comprises: establishing a plurality of subtasks of the risk guidance decision support model, wherein the plurality of subtasks include an originating event sub-event analysis and identification task, a subject event analysis task of an event tree, and a decision suggestion support task; constructing a first task interactive agent corresponding to the originating event sub-event analysis and identification task, wherein the first task interactive agent includes an originating event sub-event analysis agent and an originating event sub-event verification agent; establishing a second task interactive agent corresponding to the subject event analysis task of the event tree, and constructing a third task interactive agent corresponding to the decision suggestion support task, wherein the second task interactive agent includes an event tree subject event analysis agent and an event tree subject event verification agent, and the third task interactive agent includes a strategy formulation agent and a strategy verification agent.

2. The method of claim 1, wherein, The collecting safety analysis report data of a target safety field to establish an emergency decision support sample database through the safety analysis report data, and training the interactive agent using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database comprises: collecting safety analysis report data corresponding to each of the subtasks in the target safety field, and sending the safety analysis report data to the corresponding interactive agent to determine whether the safety analysis report data satisfies a preset successful decision requirement through the interactive agent; if the safety analysis report data satisfies the preset successful decision requirement, generating a success case corresponding to the safety analysis report data, compiling the success case, and storing the compiled success case in the success case library; if the safety analysis report data does not satisfy the preset successful decision requirement, generating a failure case corresponding to the safety analysis report data, compiling the failure case, and storing the compiled failure case in the failure case library. ​ 3. The method of claim 2, wherein, The success case library and the failure case library are obtained by the interactive agent, and the question and answer pair data satisfying the preset content similarity condition is obtained, and the decision support procedure corresponding to the super-standard originating event information is generated according to the question and answer pair data. The super-standard originating event information is input into the first task interactive agent to query the first question and answer pair data satisfying the preset content similarity condition in the success case library and the failure case library corresponding to the first task interactive agent; The first question and answer pair data is added to the first prompt engineering content corresponding to the first task interactive agent, and the first question and answer pair data is input into the second task interactive agent to query the second question and answer pair data satisfying the preset content similarity condition in the success case library and the failure case library corresponding to the second task interactive agent; The second question and answer pair data is added to the second prompt engineering content corresponding to the second task interactive agent, and the second question and answer pair data is input into the third task interactive agent to obtain the decision support procedure corresponding to the super-standard originating event information; The decision support procedure is added to the third prompt engineering content corresponding to the third task interactive agent, and the risk guidance decision support model is optimized through the first prompt engineering content, the second prompt engineering content and the third prompt engineering content, so that the decision support procedure corresponding to the super-standard originating event information is optimized based on the optimized risk guidance decision support model.

4. A task-driven multi-agent emergency decision support apparatus, characterized by, Comprise: The establishment module is used for constructing a plurality of subtasks of a preset risk guidance decision support model, and establishing an interactive agent corresponding to each subtask in the plurality of subtasks; The training module is used for collecting safety analysis report data of a target safety field, establishing an emergency decision support sample database through the safety analysis report data, and training the interactive agent by using the emergency decision support sample database to generate a success case library and a failure case library corresponding to the emergency decision support sample database; The emergency decision support module is used for obtaining super-standard originating event information of a target user, and obtaining question and answer pair data satisfying a preset content similarity condition in the success case library and the failure case library through the interactive agent, so as to generate a decision support procedure corresponding to the super-standard originating event information according to the question and answer pair data; The establishment module comprises: The first construction unit is used for establishing a plurality of subtasks of the risk guidance decision support model, wherein the plurality of subtasks comprise originating event sub-event analysis and identification tasks, event tree title event analysis tasks and decision suggestion support tasks; The second construction unit is used for constructing a first task interactive agent corresponding to the originating event sub-event analysis and identification task, wherein the first task interactive agent comprises an originating event sub-event analysis agent and an originating event sub-event verification agent; The second construction unit is used for constructing a first task interactive agent corresponding to the originating event sub-event analysis and identification task, wherein the first task interactive agent comprises an originating event sub-event analysis agent and an originating event sub-event verification agent; A third construction unit is configured to establish a second task interaction intelligent agent corresponding to the analysis task of the header event of the event tree, and construct a third task interaction intelligent agent corresponding to the decision suggestion support task, wherein the second task interaction intelligent agent includes an event tree header event analysis intelligent agent and an event tree header event verification intelligent agent, and the third task interaction intelligent agent includes a strategy making intelligent agent and a strategy verification intelligent agent.

5. The apparatus of claim 4, wherein, The training module includes: The acquisition unit is configured to acquire safety analysis report data corresponding to each subtask in the target safety field, and send the safety analysis report data to the corresponding interaction intelligent agent, so as to determine whether the safety analysis report data meets a preset successful decision requirement through the interaction intelligent agent; The first generation unit is configured to generate a successful case corresponding to the safety analysis report data if the safety analysis report data meets the preset successful decision requirement, compile the successful case, and store the compiled successful case to the successful case library; The second generation unit is configured to generate a failed case corresponding to the safety analysis report data if the safety analysis report data does not meet the preset successful decision requirement, compile the failed case, and store the compiled failed case to the failed case library.

6. An electronic device, comprising: The computer program is executed by the processor to implement the task-driven multi-intelligent agent emergency decision support method according to any one of claims 1-3. The program is executed by the processor to implement the task-driven multi-intelligent agent emergency decision support method according to any one of claims 1-3.

7. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the task-driven multi-intelligent agent emergency decision support method according to any one of claims 1-3.

8. A computer program product comprising a computer program, characterized in that, ​

Images