A multi-agent based enterprise emergency drill script generation system

Abstract

The present application relates to the technical field of data processing, and in particular to a kind of enterprise emergency drill script generation system based on multi-agent, comprising: initial script generation unit, for generating multiple initial enterprise emergency drill scripts based on pre-configured multi-agent;Simulation unit is used to obtain multiple non-rational behavior data in historical accident, and based on multiple non-rational behavior data, make multi-agent simulate non-rational behavior;Non-rational behavior analysis unit is used to analyze the influence of different roles after non-rational behavior by the non-rational behavior simulated by multi-agent for each initial enterprise emergency drill script, determine the non-rational behavior tolerance of each initial enterprise emergency drill script to role;Target script determination unit is used to determine target enterprise emergency drill script from multiple initial enterprise emergency drill scripts according to the non-rational behavior tolerance of each initial enterprise emergency drill script to role.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to an enterprise emergency drill script generation system based on multiple agents. BACKGROUND

[0002] The enterprise emergency drill script refers to a full-process disposal guide designed in advance for sudden crisis events, which is used to guide personnel to carry out practical simulation drills, so that when a sudden crisis event occurs in the enterprise, employees and equipment can quickly respond and reduce unnecessary casualties and losses. With the gradual intelligentization of various equipment in enterprises, the proportion of agents in enterprises gradually increases. When facing various sudden crisis events, whether multiple agents can respond reasonably in time has a great influence on the safety of the enterprise.

[0003] An agent refers to an agent that can perceive the environment and take action to achieve a specific goal. It has strong autonomy, adaptability and interaction ability. By forming a computing system with multiple interacting agents, complex tasks can be completed through cooperation, competition or negotiation. By generating an enterprise emergency drill script through multiple agents, the risk resistance capability when facing a sudden crisis event can be greatly improved.

[0004] However, when generating an enterprise emergency drill script based on multiple agents, the existing method usually gives different agents different identity information, such as some agents for simulating environmental changes, some agents for simulating ordinary employees of the enterprise, and some agents for simulating enterprise managers. By collecting data on the reactions and actions of different agents after an accident occurs, and screening the better reactions and actions, a reasonable emergency drill script is generated. However, since agents usually cannot simulate the state of tension, fear and some irrational behavior of people when facing danger, it is easy to lead to an overly idealized emergency drill script. When the behavior of some personnel is inconsistent with the script behavior, it is easy to cause other accidents such as stampede and congestion. The lack of authenticity of the script will lead to a decrease in safety during the emergency drill process and the risk resistance capability when facing risks. SUMMARY

[0005] In order to solve the technical problem of how to improve the authenticity of the script, the safety during the emergency drill process and the risk resistance capability when facing risks, the purpose of the present application is to provide an enterprise emergency drill script generation system based on multiple agents, and the technical solution adopted is as follows:

[0006] In an embodiment of the present application, an enterprise emergency drill script generation system based on multiple agents is provided, which comprises:

[0007] The initial script generation unit is used to generate multiple initial enterprise emergency drill scripts based on pre-configured multi-agent systems.

[0008] The simulation unit is used to acquire various irrational behavior data from historical accidents, and based on the various irrational behavior data, enable the multi-agent to simulate irrational behavior.

[0009] The irrational behavior analysis unit is used to analyze the impact of irrational behavior on other roles after different roles exhibit irrational behavior through the irrational behavior simulated by the multi-agent system for each of the initial enterprise emergency drill scripts, and to determine the tolerance of each of the initial enterprise emergency drill scripts for the irrational behavior of the roles.

[0010] The target script determination unit is used to determine the target enterprise emergency drill script from the plurality of initial enterprise emergency drill scripts based on the tolerance of each initial enterprise emergency drill script for irrational behavior of the role.

[0011] In one embodiment of this application, the initial script generation unit includes:

[0012] The intelligent agent setting subunit is used to obtain the personnel composition information of the enterprise, and set intelligent agents corresponding to different roles according to the personnel composition information to obtain the multi-intelligent agent;

[0013] The process simulation subunit is used to simulate the occurrence process of a sudden crisis event based on the multi-agent system.

[0014] The iterative simulation subunit is used to iteratively simulate emergency drills based on preset rules and simulation methods for different roles during the occurrence of simulated sudden crisis events.

[0015] The initial script generation subunit is used to generate the multiple initial enterprise emergency drill scripts based on the decisions and behaviors of different agents in each emergency drill simulation.

[0016] In one embodiment of this application, the simulation unit includes a data acquisition subunit, the data acquisition subunit being used for:

[0017] Obtain official accident investigation reports, historical emergency drill records, and analysis results of historical disaster cases;

[0018] Extract various irrational behavior data from the official accident investigation report, the historical emergency drill records, and the historical disaster case analysis results. The various irrational behavior data include irrational behavior and information on sudden crisis events.

[0019] In one embodiment of this application, the irrational behavior analysis unit includes:

[0020] The probability analysis subunit is used to determine the probability of irrational behavior occurring in different roles based on the irrational behavior simulated by the multi-agent system.

[0021] The irrational impact intensity molecular unit is used to combine the initial enterprise emergency drill script and the probability of irrational behavior of different roles to analyze the impact of irrational behavior of different roles on other roles, and to obtain the degree of irrational impact of each role.

[0022] The irrational behavior tolerance analysis subunit is used to obtain the tolerance of each initial enterprise emergency drill script for the irrational behavior of each role based on the degree of irrational influence of each role.

[0023] In one embodiment of this application, determining the probability of irrational behavior occurring in different roles through the irrational behavior simulated by the multi-agent system includes:

[0024] The irrational behavior simulated by the multi-agent is analyzed to obtain the danger index of the sudden crisis event corresponding to the irrational behavior. Combined with the probability of each role appearing in the irrational behavior, the probability of each role exhibiting irrational behavior when the sudden crisis event occurs is obtained.

[0025] The average similarity of each role in the irrational behavior is obtained by comparing the role characteristics that are prone to irrational behavior with the role characteristics assigned to different intelligent agents.

[0026] By combining the probability of irrational behavior by each role during the aforementioned sudden crisis event, and the average similarity of the irrational behaviors among the roles, the probability of irrational behavior occurring in different roles during corporate emergency drills is obtained.

[0027] In one embodiment of this application, the danger index is obtained by processing the casualty coefficient, economic damage coefficient and environmental hazard coefficient in the analysis results of the irrational behavior, wherein the casualty coefficient is the proportion of the number of casualties to the total number of people, the economic damage coefficient is the proportion of the damaged value to the enterprise value, and the environmental hazard coefficient is the product of the scope of the event's impact and the intensity of the impact.

[0028] In one embodiment of this application, the step of combining the initial enterprise emergency drill script and the probability of irrational behavior of different roles to analyze the impact of irrational behavior of different roles on other roles, and obtaining the degree of irrational impact of each role, includes:

[0029] Based on the initial enterprise emergency drill script, the action flow of multiple agents in a sudden crisis event is determined;

[0030] The target moment of the multi-agent action in the action flow is set as an irrational behavior, and the distance between the multi-agents in the same space at the target moment is calculated.

[0031] The relative spatial density of the agent's location is determined based on the distance between the multiple agents in the same space at the target time.

[0032] By combining the progress of the action process with the progress of the environmental simulation, the urgency of the action is determined;

[0033] The behavioral difference index after the agent exhibits irrational behavior at the target time is analyzed, and combined with the relative spatial density, the urgency of the action, and the behavioral difference index, the influence intensity of the agent on other roles is obtained.

[0034] The average value of the agent's influence at different times in the action process is taken as the degree of irrational influence of each role.

[0035] In one embodiment of this application, the target script determination unit includes:

[0036] The exercise effectiveness analysis subunit is used to analyze the emergency exercise effectiveness of each initial enterprise emergency exercise script and obtain the theoretical emergency evaluation index.

[0037] The reasonable index determination sub-unit is used to combine the irrational behavior tolerance and the theoretical emergency evaluation index, and obtain the reasonable index of each of the initial enterprise emergency drill scripts after normalization.

[0038] The target script determination sub-unit is used to take the initial enterprise emergency drill script corresponding to the largest reasonable index as the target enterprise emergency drill script.

[0039] In one embodiment of this application, the step of analyzing the emergency drill effectiveness of each of the initial enterprise emergency drill scripts to obtain a theoretical emergency evaluation index includes:

[0040] The completeness of each initial enterprise emergency drill script is obtained by normalizing the product of the role coverage index, disaster scenario coverage index, and role behavior completeness index in the emergency drill effect of each initial enterprise emergency drill script.

[0041] By combining the emergency drill time for completing each initial enterprise emergency drill script with the completeness of each initial enterprise emergency drill script, a theoretical emergency evaluation index for each initial enterprise emergency drill script is obtained.

[0042] In one embodiment of this application, the system further includes a testing unit, the testing unit comprising:

[0043] The test data acquisition subunit is used to acquire test data when conducting real-world stress tests based on the target enterprise's emergency drill script.

[0044] The correction subunit is used to correct the emergency drill script of the target enterprise based on the test data, and generate the final enterprise emergency drill script.

[0045] The present invention has the following beneficial effects:

[0046] First, the initial script generation unit generates multiple initial enterprise emergency drill scripts based on pre-configured multi-agent systems. Then, the simulation unit acquires various irrational behavior data from historical accidents and, based on this data, enables the multi-agent systems to simulate irrational behaviors. Next, the irrational behavior analysis unit analyzes the impact of irrational behavior on other roles in each initial enterprise emergency drill script, determining the tolerance level of each initial enterprise emergency drill script for irrational behavior by different roles. Finally, the target script determination unit determines the target enterprise emergency drill script from the multiple initial enterprise emergency drill scripts based on the tolerance level of each initial enterprise emergency drill script for irrational behavior by different roles. In this invention, by setting up multiple agents, the behavior of different roles in emergency scenarios can be simulated, covering as many possible situations as possible, making the generated scripts closer to actual emergency drill needs. By incorporating data on irrational behaviors from historical accidents, multi-agent simulations of these behaviors enhance the script's realism by incorporating various complex scenarios that may arise in reality. This reduces accidents caused by inconsistencies between human behavior and the script, improving safety and resilience during emergency drills. Analyzing the impact of irrational behavior on other roles reveals the different scripts' responses to such behavior, identifying which scripts better maintain order and prevent chaos. This helps assess the script's feasibility and safety in actual emergency drills, further improving safety and resilience. Selecting scripts with high tolerance for irrational behavior—those that remain relatively stable and orderly even in the face of such behavior—as target scripts ensures greater reliability for the final emergency drills, improving safety and resilience while maintaining real-world realism by considering potential irrational behaviors. Attached Figure Description

[0047] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A schematic diagram of the implementation environment of a multi-agent-based enterprise emergency drill script generation system provided in one embodiment of the present invention;

[0049] Figure 2 A schematic diagram of the structure of a multi-agent-based enterprise emergency drill script generation system provided in one embodiment of the present invention;

[0050] Figure 3 This is a flowchart illustrating a method for generating enterprise emergency drill scripts based on multi-agent systems, as provided in one embodiment of the present invention. Detailed Implementation

[0051] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a multi-agent-based enterprise emergency drill script generation system proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0052] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0054] The following description, in conjunction with the accompanying drawings, details a specific solution for a multi-agent-based enterprise emergency drill script generation system provided by the present invention.

[0055] Please see Figure 1 , Figure 1 This is a schematic diagram illustrating the implementation environment of a multi-agent-based enterprise emergency drill script generation system, as provided in one embodiment of the present invention. Figure 1 As shown, the implementation environment includes a script generation terminal 101 and a data acquisition terminal 102. The script generation terminal 101 can be a terminal device configured with a target client, including but not limited to laptops, tablets, handheld computers, PADs, desktop computers, etc., with local computing capabilities. The target client can be a video client, instant messaging client, browser client, etc., that has installed a multi-agent-based enterprise emergency drill script generation system. The script generation terminal 101 can communicate with the data acquisition terminal 102 via a network, including but not limited to wired networks and wireless networks. The wired network includes local area networks (LANs), metropolitan area networks (MANs), and wide area networks (WANs). The wireless network includes Bluetooth, Wi-Fi, and other networks that enable wireless communication. The script generation terminal 101 may include, but is not limited to, a human-machine interface screen, a processor, and a memory. The human-machine interface screen may be used to display fault diagnosis results. The processor may be used, but is not limited to, to respond to human-machine interaction operations, execute corresponding operations, or generate corresponding instructions.

[0056] As an alternative, irrational behavior data can be collected through data acquisition terminal 102.

[0057] As an alternative, the script generation terminal 101 described above can also be a server. This server can be a single server, a server cluster consisting of multiple servers, or a cloud server. The above is merely an example, and this embodiment does not impose any limitations on it.

[0058] As an alternative approach, a multi-agent enterprise emergency drill script generation system can be mounted on the script generation terminal 101. The specific structure of this multi-agent enterprise emergency drill script generation system is as follows:

[0059] The initial script generation unit is used to generate multiple initial enterprise emergency drill scripts based on pre-configured multi-agent systems.

[0060] The simulation unit is used to acquire various irrational behavior data from historical accidents, and based on the various irrational behavior data, enable the multi-agent to simulate irrational behavior.

[0061] The irrational behavior analysis unit is used to analyze the impact of irrational behavior on other roles after different roles exhibit irrational behavior through the irrational behavior simulated by the multi-agent system for each of the initial enterprise emergency drill scripts, and to determine the tolerance of each of the initial enterprise emergency drill scripts for the irrational behavior of the roles.

[0062] The target script determination unit is used to determine the target enterprise emergency drill script from the plurality of initial enterprise emergency drill scripts based on the tolerance of each initial enterprise emergency drill script for irrational behavior of the role.

[0063] Based on the above approach, by setting up multiple agents, the behavior of different roles in emergency scenarios can be simulated, covering as many possible situations as possible, making the generated scripts closer to the needs of actual emergency drills. By introducing irrational behavior data from historical accidents, and having multiple agents simulate these behaviors, the scripts can consider various complex situations that may arise in reality, thereby improving the realism of the scripts. This reduces accidents caused by inconsistencies between personnel behavior and the script, improving safety during emergency drills and enhancing resilience against risks. By analyzing the impact of irrational behavior by different roles on other roles, we can understand the response capabilities of different scripts to irrational behavior, identifying which scripts are better at maintaining order and avoiding chaos. This helps assess the feasibility and safety of scripts in actual emergency drills, further improving safety during emergency drills and enhancing resilience against risks. By selecting scripts with high tolerance for irrational behavior—that is, scripts that can remain relatively stable and orderly when faced with irrational behavior from personnel—as target scripts, the scripts ultimately used for emergency drills become more reliable. This improves the safety of emergency drills and the ability to withstand risks. At the same time, by considering possible irrational behaviors in reality, the authenticity of the scripts is also guaranteed.

[0064] As an optional example, this embodiment does not limit the mounting subject of the above-mentioned multi-agent-based enterprise emergency drill script generation system. The above-mentioned multi-agent-based enterprise emergency drill script generation system can be mounted on the script generation terminal 101. For example, if the script generation terminal 101 is a desktop computer, some or all of the components of the above-mentioned multi-agent-based enterprise emergency drill script generation system can be mounted on the desktop computer.

[0065] Taking a fire at a company as an example, in the process of generating emergency drill scripts for companies based on multi-agent systems, the agents typically only mechanically perform corresponding actions according to their assigned roles. This fails to accurately simulate irrational human behavior in the face of sudden crises, leading to overly idealized scripts. Furthermore, irrational human behavior is highly contagious; when one person acts irrationally, it often affects those around them. For instance, during a sudden crisis, the fear and anxiety of some individuals can cause congestion, as a passageway designed for five people might be overcrowded. Therefore, a multi-agent-based emergency drill script generation system is needed to improve companies' resilience in the face of sudden crises.

[0066] In one embodiment of this application, a multi-agent-based enterprise emergency drill script generation system is provided. Figure 2 This is a schematic diagram of a multi-agent-based enterprise emergency drill script generation system according to an embodiment of the present invention. (See attached diagram.) Figure 2 The enterprise emergency drill script generation system based on multi-agent technology includes the following units as described in 210 to 240:

[0067] The initial script generation unit 210 is used to generate multiple initial enterprise emergency drill scripts based on pre-configured multi-agent systems.

[0068] In this context, an intelligent agent refers to a proxy capable of perceiving its environment and taking actions to achieve specific goals, possessing autonomy, adaptability, and interactive capabilities. Pre-configuration means setting up multiple intelligent agents before the system runs, based on specific needs and scenarios, including setting parameters such as the type, number, and functions of the agents.

[0069] This process involves using multiple pre-defined agents to simulate emergency drills according to specific rules and algorithms, generating multiple versions of emergency drill scripts to provide material for selecting the best script. For example, multiple agents are first set up based on the company's personnel composition. Then, an environmental agent simulates the spread of flames within the company during a fire. Based on pre-defined rules (decision trees) for different roles and existing artificial intelligence simulation algorithms (such as reinforcement learning algorithms), multiple iterative simulations are performed. Finally, the decisions and behaviors of different agents during each iteration are recorded to generate the emergency drill script.

[0070] The simulation unit 220 is used to acquire various irrational behavior data from historical accidents, and based on the various irrational behavior data, enable the multi-agent to simulate irrational behavior.

[0071] Among them, the data on various irrational behaviors in historical accidents are collected from records of various accidents that have occurred in the past. This data is about people’s irrational behavior in accident scenarios. This data can provide a reference for simulating real scenarios. The main objects of collection are enterprises. One irrational behavior data corresponds to all the contents of a sudden crisis event, including the irrational behavior of personnel and information about the sudden crisis event.

[0072] This approach utilizes data on irrational behaviors from historical accidents as a basis, allowing multi-agent systems to mimic these behaviors and making the simulated scenarios more closely resemble real-world accident situations. Irrational behavior data can include actions such as retrieving objects or being unable to move, enabling multi-agent systems to exhibit these irrational behaviors during simulated emergency drills, thus mimicking the true state of humans facing dangerous situations.

[0073] The irrational behavior analysis unit 230 is used to analyze the impact of irrational behavior of different roles on other roles by simulating irrational behavior through the multi-agent simulation of each initial enterprise emergency drill script, and to determine the tolerance of each initial enterprise emergency drill script for the irrational behavior of roles.

[0074] In a scenario involving multiple roles, the irrational behavior of one role has a significant impact on the actions and decisions of other roles, as well as the overall development of the scenario. When faced with dangerous situations, if one or more people in a group exhibit irrational behavior, others, influenced by environmental pressures, are also more likely to exhibit irrational behavior.

[0075] The tolerance of initial enterprise emergency drill scripts for irrational behavior is a measure of a script's ability to maintain normal drill operations without serious chaos or failure when faced with such behavior. The lower the impact of irrational behavior on different roles within an enterprise emergency drill script, the higher its tolerance. This tolerance is determined by analyzing the impact of irrational behavior on other roles.

[0076] This study utilizes multi-agent simulations of irrational behavior to investigate how the irrational behavior of different roles affects other roles, thereby evaluating the ability of each emergency drill script to address irrational behavior. It begins by acquiring data on irrational human behavior in dangerous situations to determine the probability of irrational behavior occurring in different roles. Then, combining this data with the emergency drill scripts and the probabilities of irrational behavior in different roles, the study analyzes the intensity of the irrational impact of different roles on other roles. Finally, it assesses the tolerance of the current enterprise emergency drill script for irrational behavior and obtains a reasonableness index for different emergency drill scripts.

[0077] The target script determination unit 240 is used to determine the target enterprise emergency drill script from the plurality of initial enterprise emergency drill scripts based on the tolerance of each initial enterprise emergency drill script for irrational behavior of the roles.

[0078] Among them, the target enterprise emergency drill script is the script that is most suitable for the actual emergency drill needs of the enterprise after a series of evaluations and screenings from multiple initially generated emergency drill scripts. It can be selected based on the tolerance of each initial enterprise emergency drill script for the irrational behavior of the roles and the emergency drill effect of the script (such as the theoretical emergency evaluation index). The initial enterprise emergency drill script with the highest reasonable index is selected. This script performs best in dealing with the irrational behavior of the roles and ensuring the effectiveness of the drill.

[0079] For example, a large factory has 5 senior executives, 50 key personnel (such as safety and medical staff), and 500 general employees. First, in the initial script generation unit, following a 1:1 ratio of senior executives to key personnel, a 1:5 ratio of general employees to key personnel, and a 1:10 ratio of general employees, 5 senior executive agents, 10 key personnel agents, and 50 general employee agents are configured. Using these pre-configured agents, a factory fire scenario is simulated. Multiple iterations are performed based on pre-defined rules for different roles and reinforcement learning algorithms to generate multiple initial enterprise emergency drill scripts. The simulation unit obtains data on various irrational behaviors from official accident investigation reports and historical emergency drill records, such as workers panicking and returning to retrieve important items, and managers being unable to issue correct instructions due to tension. The agents then simulate these irrational behaviors. The irrational behavior analysis unit analyzes the impact of an irrational behavior, such as a regular employee agent panicking and returning to retrieve an item, on other surrounding agents (e.g., nearby key personnel organizing rescue operations, and other regular employee agents evacuating). This analysis determines the tolerance level for irrational behavior in each initial enterprise emergency drill script. Finally, the target script determination unit identifies the target enterprise emergency drill script based on the tolerance levels for irrational behavior in each initial script.

[0080] Based on the above, the multi-agent enterprise emergency drill script generation system in this application embodiment, by setting up multiple agents, can simulate the behavior of different roles in emergency scenarios, covering as many possible situations as possible, making the generated scripts closer to actual emergency drill needs. By introducing irrational behavior data from historical accidents, allowing multiple agents to simulate these behaviors, the scripts can consider various complex situations that may occur in reality, thereby improving the realism of the scripts. This reduces accidents caused by inconsistencies between personnel behavior and the script, improving safety during emergency drills and enhancing resilience against risks. By analyzing the impact of irrational behavior by different roles on other roles, we can understand the response capabilities of different scripts to irrational behavior, identify which scripts are better at maintaining order and avoiding chaos, and help evaluate the feasibility and safety of scripts in actual emergency drills, thereby improving safety during emergency drills and enhancing resilience against risks. By selecting scripts with high tolerance for irrational behavior—that is, scripts that can remain relatively stable and orderly when faced with irrational behavior from personnel—as target scripts, the scripts ultimately used for emergency drills become more reliable. This improves the safety of emergency drills and the ability to withstand risks. At the same time, by considering possible irrational behaviors in reality, the authenticity of the scripts is also guaranteed.

[0081] In one embodiment of this application, the initial script generation unit includes:

[0082] The intelligent agent setting subunit is used to obtain the personnel composition information of the enterprise, and set intelligent agents corresponding to different roles according to the personnel composition information to obtain the multi-intelligent agent;

[0083] The process simulation subunit is used to simulate the occurrence process of a sudden crisis event based on the multi-agent system.

[0084] The iterative simulation subunit is used to iteratively simulate emergency drills based on preset rules and simulation methods for different roles during the occurrence of simulated sudden crisis events.

[0085] The initial script generation subunit is used to generate the multiple initial enterprise emergency drill scripts based on the decisions and behaviors of different agents in each emergency drill simulation.

[0086] The personnel composition information of an enterprise refers to the composition of its various personnel, including but not limited to the number of employees, job distribution, age groups, and gender ratio. This information reflects the overall structural characteristics of the enterprise's personnel. By specifying the number of employees in different positions, the importance of each position, and specific information such as the age and gender of employees, the intelligent agents can more accurately simulate the roles of various personnel in emergency drills. For example, based on the number of employees in different positions such as senior management, key personnel, and ordinary employees, as well as special personnel such as pregnant women and people with disabilities, appropriate proportions of intelligent agent roles can be set.

[0087] Based on the acquired data on the composition of the company's personnel, corresponding intelligent agents are created for different types of personnel roles. Each intelligent agent is assigned the attributes and behavioral characteristics of the corresponding role to represent that role in simulated scenarios. For example, a one-to-one intelligent agent role is set for senior management and special personnel, a one-to-five intelligent agent role is set for key personnel, and a one-to-ten intelligent agent role is set for ordinary employees. At the same time, the age and gender factors of the company's employees are taken into account so that the set intelligent agents can better simulate real personnel roles in all aspects.

[0088] This method utilizes multiple pre-defined intelligent agents, following specific logic and rules, to simulate the entire process of a sudden crisis from its inception to its development. This includes environmental changes and the actions of each role within the crisis, creating a virtual crisis scenario that closely resembles reality. Taking fire as an example, an environmental intelligent agent simulates the spread of flames within a company during a fire. Simultaneously, individual role-based intelligent agents perform corresponding actions based on their pre-defined behaviors, such as evacuation actions by ordinary employee agents and rescue actions by key personnel agents, comprehensively simulating the fire's occurrence.

[0089] Specifically, behavioral guidelines and simulation methods are pre-defined for different types of roles. These rules and methods determine the specific actions and decision-making logic of each role's intelligent agent during the simulation of a sudden crisis. For example, the senior management intelligent agent focuses on command and decision-making, the key personnel intelligent agent performs rescue and support actions based on their professional functions, and the ordinary employee intelligent agent mainly carries out evacuation actions. At the same time, existing artificial intelligence simulation algorithms (such as reinforcement learning algorithms) are combined to enable the intelligent agents to make reasonable decisions based on environmental changes and their own role characteristics during the simulation.

[0090] Iterative simulation involves repeatedly performing simulation operations during the simulation process, adjusting and optimizing subsequent simulations based on the results of each simulation, thereby continuously improving the accuracy and realism of the simulation and making the simulation results more consistent with actual situations. Existing artificial intelligence algorithms can be used to conduct multiple emergency drill simulations. After each simulation, the behavioral decisions of different agents are recorded and analyzed. Based on the analysis results, parameters and agent behaviors in subsequent simulations are adjusted, thus making the generated emergency drill scripts more complete and in line with actual needs.

[0091] For example, in an electronics manufacturing company, personnel information shows 3 senior executives, 30 key personnel (such as firefighters and technical support staff), and 300 ordinary employees. The majority of employees are between 25 and 45 years old, with a male-to-female ratio of approximately 3:2. Based on this information, the agent setting subunit sets up 3 senior executive agents, 6 key personnel agents, and 30 ordinary employee agents, assigning them according to age and gender ratios. The process simulation subunit, based on these multiple agents, simulates a sudden fire in the factory, with the environmental agent simulating the spread of flames from a specific area of ​​the production workshop. In the iterative simulation subunit, during this simulation, the senior executive agents make decisions using reinforcement learning algorithms based on preset command and coordination rules, such as determining evacuation directions and allocating resources; the key personnel agents act according to preset rescue and technical support rules, such as firefighters extinguishing the fire and technical support personnel ensuring power is cut off to critical equipment; and the ordinary employee agents act according to preset evacuation rules. After each simulation, the behavior and decisions of each agent are analyzed, such as whether there are ordinary employee agents whose evacuation routes are unreasonable and cause congestion. Then, the parameters or agent behavior rules for subsequent simulations are adjusted, and the simulation is carried out again. After multiple iterations of simulation, the initial script generation subunit generates multiple initial enterprise emergency drill scripts based on the decisions and behaviors of different agents in each simulation.

[0092] In this embodiment, by acquiring the personnel composition information of the enterprise and setting up intelligent agents corresponding to different roles, the simulated emergency drill scenarios are made more closely resemble the actual personnel situation of the enterprise, improving the realism of the simulation and thus enhancing the realism of the script, providing a more reliable foundation for subsequent emergency drills. Simulating the occurrence of sudden crisis events based on multiple intelligent agents comprehensively demonstrates various situations and the actions of each role in the crisis event, making the generated script more practical and helping to better cope with various situations in actual emergency drills, improving the safety and risk resistance of emergency drills. Iterative simulations are performed based on preset rules and simulation methods for different roles, continuously optimizing the simulation process, making the generated emergency drill script more complete and reasonable, better able to cope with various changes that may occur in actual emergency scenarios, further improving the safety of emergency drills and the ability to resist risks.

[0093] In one embodiment of this application, the simulation unit includes a data acquisition subunit, which is used for:

[0094] Obtain official accident investigation reports, historical emergency drill records, and analysis results of historical disaster cases;

[0095] Extract various irrational behavior data from the official accident investigation report, the historical emergency drill records, and the historical disaster case analysis results. The various irrational behavior data include irrational behavior and information on sudden crisis events.

[0096] Official accident investigation reports, compiled by relevant government departments or professional institutions after investigating a specific accident, contain detailed information such as the time, location, cause, process, losses, and determination of responsibility. They are authoritative and objective, serving as a crucial source of data for accident research. Studying official accident investigation reports allows us to obtain information on irrational behaviors and corresponding crisis events that occur when people face danger in real-life accidents, providing reliable data support for multi-agent simulations of irrational behavior. For example, an official report on a factory explosion might record the irrational behavior of some employees who, out of panic, ran against instructions.

[0097] Historical emergency drill records document the process and results of past emergency drills, covering scenario settings, participant actions, problems encountered, and lessons learned. These records reflect behavioral patterns and responses under simulated emergency conditions. Since emergency drills aim to simulate real crisis situations, the recorded behaviors of participants that deviate from conventional response logic—irrational behaviors—along with corresponding crisis event information, can provide practical reference for multi-agent simulations. For example, a historical emergency drill record might document the irrational behavior of some participants who, due to panic, forgot the correct escape route during a fire drill.

[0098] The historical disaster case analysis results are derived from in-depth analysis of various natural and man-made disasters throughout history. This includes research on the mechanisms of disaster occurrence, evolution, extent of damage, and human coping behaviors, providing rich material for understanding human behavior in disaster scenarios. By analyzing historical disaster cases, various irrational behaviors exhibited by people during real disasters, along with related disaster event information, are uncovered, making the irrational behaviors simulated by multi-agent systems more realistic. For example, in analyzing a specific earthquake disaster case, it might be discovered that some residents, out of fear, chose to hide under flimsy furniture instead of in open areas during an earthquake—an irrational behavior.

[0099] Irrational behavior data refers to data records of people's actions that do not conform to normal rational thinking and behavior patterns under specific dangerous or emergency situations. These behaviors may affect the effectiveness and safety of emergency response.

[0100] Irrational behavior refers to actions that do not conform to rational thinking. Sudden crisis event information includes the type of event (such as fire, earthquake, explosion, etc.), the environment in which it occurs, and its impact. Combining these two aspects provides a comprehensive description of human behavior and the state of events in a dangerous scenario. Each extracted piece of irrational behavior data corresponds to specific sudden crisis event information. For example, in a shopping mall fire, some customers panicked and rushed towards the exits, causing congestion. The corresponding sudden crisis event information includes the location of the fire in the mall, the cause of the fire, and the spread of the fire. This complete information provides a basis for accurate multi-agent simulation.

[0101] For example, the data acquisition subunit obtains an official accident investigation report on an explosion at a chemical industrial park. The report mentions that some employees, panicked after the explosion, did not escape along the planned route but instead ran around aimlessly – this is the extracted irrational behavior. The report also includes information about the time, location, and intensity of the explosion – information about the sudden crisis event. From historical emergency drill records, a previous fire emergency drill of the same chemical company is found. This drill recorded some employees forgetting how to use fire extinguishers due to nervousness – this is irrational behavior. The simulated fire area and size in the drill are information about the sudden crisis event. Furthermore, from historical disaster case analysis, an analysis of a similar chemical company's poisoning incident caused by a leak is obtained. This analysis mentions that some workers were afraid to approach rescue equipment – ​​this is irrational behavior. The type of leaked substance and the scope of its impact in this incident are information about the sudden crisis event. The data acquisition subunit extracts these various irrational behavior data from these materials, providing a data foundation for subsequent multi-agent simulations of irrational behavior.

[0102] In this embodiment, by acquiring official accident investigation reports, historical emergency drill records, and historical disaster case analysis results, and extracting irrational behavior data from them, a rich and realistic data source is provided for multi-agent simulation of irrational human behavior in dangerous situations. This makes the simulated scenarios closer to actual possible situations, greatly improving the realism of the scripts and helping to better respond to real-world irrational behavior in emergency drills. The extracted irrational behavior data includes information on irrational behavior and sudden crisis events; the combination of the two can comprehensively and accurately reflect the actual situation in dangerous scenarios. Multi-agent simulations based on this complete data can more accurately present various complex situations, making the generated emergency drill scripts more targeted and practical, thereby improving safety during emergency drills and the ability to withstand risks.

[0103] In one embodiment of this application, the irrational behavior analysis unit includes:

[0104] The probability analysis subunit is used to determine the probability of irrational behavior occurring in different roles based on the irrational behavior simulated by the multi-agent system.

[0105] The irrational impact intensity molecular unit is used to combine the initial enterprise emergency drill script and the probability of irrational behavior of different roles to analyze the impact of irrational behavior of different roles on other roles, and to obtain the degree of irrational impact of each role.

[0106] The irrational behavior tolerance analysis subunit is used to obtain the tolerance of each initial enterprise emergency drill script for the irrational behavior of each role based on the degree of irrational influence of each role.

[0107] The probability of irrational behavior refers to the likelihood of different roles exhibiting irrational behavior in a specific situation (such as an emergency drill). It is usually represented by a numerical value; the higher the value, the greater the likelihood of the role exhibiting irrational behavior. It is an indicator that measures a role's tendency to act irrationally when facing dangerous situations. For example, through analysis of historical accident data, it was determined that in a fire emergency drill scenario, the probability of an ordinary employee exhibiting irrational behavior (such as panic leading to slow action) is 30%, while the probability of a senior executive exhibiting similar irrational behavior is 10%.

[0108] In determining the probability of irrational behavior for different roles through multi-agent simulations of irrational behavior, relevant data can be collected and analyzed using the simulation process and results of the multi-agent simulations to calculate the likelihood of different roles exhibiting irrational behavior. Furthermore, by analyzing the data on irrational behavior simulated by the multi-agent simulations and combining it with factors such as the risk index of sudden crisis events and the probability of different roles exhibiting irrational behavior, the probability of different roles exhibiting irrational behavior during corporate emergency drills can be calculated.

[0109] The degree of irrational influence for each role refers to the severity or magnitude of the impact of irrational behavior on the actions, decisions, and outcomes of other roles. It measures the intensity of the interaction between irrational behaviors of different roles and can be determined through quantitative indicators or evaluation methods. Within the initial scenario of an enterprise emergency drill, the degree of impact of irrational behavior by one role on the actions of other roles in the script is analyzed, considering the probability of irrational behavior occurring in different roles.

[0110] This study uses a pre-generated initial enterprise emergency drill script as a background, considering the likelihood of irrational behavior by different roles. It investigates the impact of irrational behavior by one role on the behavior, decision-making, and overall emergency process of other roles within the script's emergency scenario, ultimately deriving quantitative or qualitative results to measure the degree of this impact. Specifically, the process involves: determining a multi-agent action flow based on the initial enterprise emergency drill script; assuming an agent exhibits irrational behavior at a certain moment during the action flow; calculating the distance between this agent and other agents, spatial density, urgency of action, and behavioral difference index; and combining these factors to determine the agent's influence on other roles. This process is repeated multiple times, and the average is taken to obtain the degree of irrational influence for each role. For example, in a fire emergency drill script, a scenario is set where an ordinary employee agent exhibits irrational behavior of panicking and running around during evacuation. The distance between this agent and other surrounding agents, spatial density, etc., are calculated to analyze the impact on the rescue actions of nearby key position agents and the evacuation actions of other ordinary employee agents. Multiple simulations are performed, and the average is taken to determine the degree of irrational influence of this ordinary employee role.

[0111] The ability of the initial corporate emergency drill script to maintain normal drill operations and avoid serious chaos or failure when faced with irrational behavior is judged by the degree to which the irrational behavior of different roles affects other roles. The smaller the degree of impact, the higher the script's tolerance for irrational behavior.

[0112] For example, in a shopping mall emergency drill script, the probability analysis subunit simulates irrational behavior in a shopping mall fire scenario using multiple agents. The analysis reveals a 40% probability of irrational behavior by customers (e.g., remaining motionless due to panic) and a 20% probability of irrational behavior by mall staff (e.g., forgetting to guide customers through evacuation routes). The irrational impact intensity analysis subunit, based on this initial emergency drill script, assumes that a customer agent exhibits irrational behavior such as panicking and remaining motionless at a certain moment during evacuation. It calculates the distance between this customer agent and other surrounding agents (e.g., nearby staff agents and other customers) to determine their relative spatial density. Combining this with the progress of the action flow and the simulated fire spread, it determines the urgency of the action. It analyzes the behavioral difference index after the customer agent exhibits irrational behavior, and synthesizes these factors to determine the intensity of the customer agent's influence on other agents. Multiple simulations are performed, and the average value is taken to obtain the degree of irrational influence of the customer agent. The same method is used to obtain the degree of irrational influence of the staff agent. The irrational behavior tolerance analysis subunit found that the degree of irrational influence of different roles in the script was low, indicating that the initial corporate emergency drill script had a high tolerance for the irrational behavior of the roles.

[0113] In this embodiment, by determining the probability of irrational behavior occurring in different roles, we can gain a deeper understanding of the likelihood of irrational behavior in emergency drills. This helps to more accurately assess the reliability of the script in practical applications, improving its realism and practicality. Analyzing the degree of irrational impact of each role clarifies the interaction relationships between irrational behaviors, enabling the design of emergency drill scripts to fully consider various possible chain reactions and formulate corresponding strategies in advance. This improves the safety of the emergency drill process and enhances the ability to withstand risks. Based on the degree of irrational impact of each role, we obtain the tolerance of the initial enterprise emergency drill script for irrational behavior. This helps to select scripts that are more adaptable to actual situations from multiple initial scripts. Scripts with high tolerance for irrational behavior can ensure that emergency drills can still be conducted relatively smoothly when faced with irrational behavior, further improving the quality and effectiveness of emergency drills and ensuring effective response in actual crises.

[0114] In one embodiment of this application, determining the probability of irrational behavior occurring for different roles through the irrational behavior simulated by the multi-agent system includes:

[0115] The irrational behavior simulated by the multi-agent is analyzed to obtain the danger index of the sudden crisis event corresponding to the irrational behavior. Combined with the probability of each role appearing in the irrational behavior, the probability of each role exhibiting irrational behavior when the sudden crisis event occurs is obtained.

[0116] The average similarity of each role in the irrational behavior is obtained by comparing the role characteristics that are prone to irrational behavior with the role characteristics assigned to different intelligent agents.

[0117] By combining the probability of irrational behavior by each role during the aforementioned sudden crisis event, and the average similarity of the irrational behaviors among the roles, the probability of irrational behavior occurring in different roles during corporate emergency drills is obtained.

[0118] Among them, the danger index corresponding to irrational behavior in sudden crisis events is a quantitative indicator used to measure the degree of danger of sudden crisis events associated with specific irrational behaviors. It comprehensively considers multiple factors of the event, such as the damage caused, the scope of impact, and the harm to people and the environment, and is calculated through a certain method. The higher the value, the more dangerous the event.

[0119] The probability of each role appearing in the irrational behavior refers to the proportion of frequency in which different roles participate in these irrational behaviors in a specific set of irrational behavior samples, reflecting the relative probability of each role appearing in the collected irrational behavior data.

[0120] Among them, the probability of each role exhibiting irrational behavior during a sudden crisis is an estimated probability of each role actually exhibiting irrational behavior during a sudden crisis, which is obtained by comprehensively considering the danger index and the probability of each role exhibiting irrational behavior. It combines the degree of danger of the event itself and the frequency of the role's participation in irrational behavior in similar events.

[0121] Among them, the average similarity of each role in irrational behavior is used to measure the similarity between the role characteristics in the collected irrational behavior data and the role characteristics represented by the intelligent agent in the current enterprise emergency drill. A comprehensive similarity index is obtained by averaging, which reflects the degree of fit between the two.

[0122] In this process, the characteristics of roles prone to irrational behavior are compared with the characteristics of roles assigned to different intelligent agents to obtain the average similarity of each role in the irrational behavior. This involves comparing the characteristics of roles prone to irrational behavior, derived from analysis of historical data and other sources, with the characteristics of roles set for different intelligent agents in the emergency drill system. The average similarity of these roles in terms of characteristics is obtained through a certain calculation method in order to more accurately simulate the actual situation.

[0123] In one embodiment of this application, the danger index is obtained by processing the casualty coefficient, economic damage coefficient and environmental hazard coefficient in the analysis results of the irrational behavior, wherein the casualty coefficient is the proportion of the number of casualties to the total number of people, the economic damage coefficient is the proportion of the damaged value to the enterprise value, and the environmental hazard coefficient is the product of the scope of the event's impact and the intensity of the impact.

[0124] For example, individuals who participate less in emergency drills or have poor psychological resilience are more likely to exhibit irrational behaviors when faced with dangerous situations, such as turning back to retrieve objects or becoming incapacitated. Analyzing the collected data on irrational behaviors can reveal the characteristics of roles prone to such behavior. Statistical analysis of the characteristics of individuals corresponding to these irrational behaviors can determine the likelihood of different characteristics leading to irrational behavior in dangerous situations. Furthermore, comparing these characteristics with the assigned roles of different agents can reveal the probability of irrational behavior occurring in different roles during corporate emergency drills.

[0125] First, regarding the first Analyze the danger index of each irrational behavior in relation to a sudden crisis event. The danger index of a sudden crisis event corresponding to an irrational behavior can be expressed as follows:

[0126]

[0127] in, Indicates the first Each irrational behavior corresponds to a risk index of a sudden crisis event; Indicates the first The casualty coefficient of a sudden crisis event corresponds to an irrational behavior, which is the proportion of casualties to the total number of people. The economic damage coefficient represents the proportion of the value of damaged equipment, goods, etc., to the total value of the enterprise. The environmental hazard coefficient is the product of the scope and intensity of an event's impact, such as the heat radiation intensity of a fire. This indicates normalization processing.

[0128] Subsequently, regarding the first Irrational behaviors are categorized based on the roles of all irrational individuals involved (e.g., senior management, key personnel, ordinary employees, and special personnel), and the results are statistically analyzed. Number of irrational characters in each role .

[0129] Furthermore, calculate the first... The first irrational behavior The probability of each character appearing That is, in this sudden crisis event, the first The percentage of individuals in each role who engage in irrational behavior;

[0130]

[0131] in, This indicates the number of irrational characters that appear.

[0132] Repeat the above steps to calculate the first [number]th [item] among other irrational behavior data. The probability of each character appearing .

[0133] Subsequently, combined with the risk index and probability of occurrence Calculate the first time when a sudden crisis event occurs. The possibility of each character exhibiting irrational behavior;

[0134]

[0135] in, When a sudden crisis occurs The possibility of each character exhibiting irrational behavior; This represents the total number of irrational behaviors.

[0136] At the same time, compared with the first The first of the irrational behaviors The role (exhibiting irrational behavior) and the current company's... The degree of similarity among roles. It should be noted that the roles referred to here are a group, so the similarity of roles here mainly compares the differences in the number of people in the enterprise group. The smaller the difference in the number of people, the more consistent the probability of the role exhibiting irrational behavior. In other words, the more people in a role, the higher the probability, and the fewer people in a role, the lower the probability.

[0137] The degree of similarity can be represented in the following ways:

[0138]

[0139] in, Indicates the first The first of the irrational behaviors The role (exhibiting irrational behavior) and the current company's... The degree of similarity between the characters; Indicates the first The first of the irrational behaviors The number of people in each role (total number of people, including those who act irrationally). Indicates the current enterprise's... The number of people in each role.

[0140] Furthermore, the calculation of the first among all irrational behaviors Average similarity of characters .

[0141] Finally, in conjunction with the first Average similarity of characters And the possibility that the character might exhibit irrational behavior during a sudden crisis. To obtain the current company's number The probability of irrational behavior occurring in each role during a corporate emergency drill. The probability of irrational behavior can be represented as follows:

[0142]

[0143] in, For the current enterprise The probability of irrational behavior occurring in each role during a corporate emergency drill.

[0144] In this embodiment, by calculating the danger index corresponding to sudden crisis events for irrational behavior, the degree of danger of different crisis events can be objectively reflected. This provides an important reference for determining the probability of roles exhibiting irrational behavior, making the assessment of the probability of irrational behavior more scientific and reasonable, thereby improving the realism of the script because the characteristics of the crisis event itself are considered in relation to the role's behavior. Statistical analysis of the probability of each role exhibiting irrational behavior, based on actual data on irrational behavior, provides a direct understanding of the frequency of different roles exhibiting irrational behavior in similar situations. This provides data support for accurately predicting role behavior in emergency drills, further enhancing the practicality and realism of the script. Calculating the probability of each role exhibiting irrational behavior when a sudden crisis event occurs, comprehensively considering both the danger index and the probability of role occurrence, more comprehensively reflects the actual situation. This helps in the rational allocation of resources and the formulation of response strategies in script design, improving the safety and effectiveness of emergency drills and enhancing the ability to withstand risks. By determining the average similarity of each role in irrational behavior and comparing the characteristics of actual irrational behavior roles with those of the intelligent agents in emergency drills, the simulation becomes more realistic, accurately reflecting the impact of the company's own characteristics on emergency drills. This further improves the script's realism and practicality, ensuring that emergency drills are truly tailored to the company's actual situation and enhancing its ability to respond to real crises. Comparing the characteristics of roles prone to irrational behavior with those of intelligent agents to obtain the average similarity provides a crucial basis for accurately calculating the probability of irrational behavior occurring in different roles during corporate emergency drills. This makes the probability calculation more aligned with the company's reality, thereby optimizing the emergency drill script, improving drill effectiveness, and better addressing irrational behavior in actual crises.

[0145] In one embodiment of this application, the step of combining the initial enterprise emergency drill script and the probability of irrational behavior of different roles to analyze the impact of irrational behavior of different roles on other roles, and obtaining the degree of irrational impact of each role, includes:

[0146] Based on the initial enterprise emergency drill script, the action flow of multiple agents in a sudden crisis event is determined;

[0147] The target moment of the multi-agent action in the action flow is set as an irrational behavior, and the distance between the multi-agents in the same space at the target moment is calculated.

[0148] The relative spatial density of the agent's location is determined based on the distance between the multiple agents in the same space at the target time.

[0149] By combining the progress of the action process with the progress of the environmental simulation, the urgency of the action is determined;

[0150] The behavioral difference index after the agent exhibits irrational behavior at the target time is analyzed, and combined with the relative spatial density, the urgency of the action, and the behavioral difference index, the influence intensity of the agent on other roles is obtained.

[0151] The average value of the agent's influence at different times in the action process is taken as the degree of irrational influence of each role.

[0152] In this context, the action flow of multiple agents in a sudden crisis refers to a series of orderly action steps performed by multiple agents according to their respective roles and rules when simulating a sudden crisis scenario. These action steps together constitute a complete crisis response process, covering all stages from the initial occurrence of the crisis to its subsequent handling. For example, in a fire emergency drill, the action flow of an ordinary employee agent might include sensing a fire, evacuating to a safe exit, and possibly assisting others during the evacuation; the action flow of a key personnel agent, such as a firefighter agent, might include receiving a fire alarm, rushing to the fire scene, and performing firefighting operations.

[0153] The target moment is a specific point in time selected within the multi-agent action process. It is used to analyze the impact on the entire system if a specific situation occurs at that moment (such as an agent exhibiting irrational behavior). It serves as a starting point for local analysis. For example, in the evacuation process of ordinary employee agents, the target moment can be selected when the evacuation is halfway complete. The impact of irrational behavior by the agents at this time (such as sudden panic and stagnation) can be observed.

[0154] Specifically, setting the target moment in the action flow for the multi-agent system to exhibit irrational behavior involves artificially inducing the agents to display irrational actions at selected target moments during the simulation of the multi-agent system following normal action flow. This is done to study the chain reactions and impacts caused by such irrational behavior throughout the entire action flow. For example, in a firefighter agent's firefighting action flow, at a target moment when the fire is under control to a certain extent, the firefighter agent is set to exhibit irrational behavior due to tension, resulting in incorrect operation of firefighting equipment.

[0155] Among them, the distance between multiple agents in the same space at the target time is a measure of the distance between various agents within the same spatial range at a set target time. This distance information is of great significance for analyzing the interaction and influence between agents, and can reflect the spatial density of the agents. For example, in a fire evacuation scenario, taking the target time when an ordinary employee agent panics and stops as the benchmark, the Euclidean distance between this agent and other ordinary employee agents and nearby key position agents in the same evacuation passage is calculated to understand their spatial relationship.

[0156] The relative spatial density of an agent's location is an indicator used to measure the density of an agent within its space. It is calculated by factors such as the distance between the agent and surrounding agents and reflects the crowding level of the agent's surrounding environment. It plays an important role in analyzing the impact of agent behavior.

[0157] Among them, the urgency of action indicates the degree of urgency of the agent's actions during the simulation of a sudden crisis. It takes into account factors such as the progress of the action process and the crisis development trend shown by the environmental simulation, and is used to measure how quickly the agent needs to complete the corresponding actions.

[0158] The urgency of action is derived by combining the progress of the action process with the progress of the environmental simulation. This involves combining the progress of the multiple agents in the action process with the progress of the crisis event environment simulation, and using a specific calculation method to obtain an index of the urgency of the agents' actions at that moment.

[0159] Among them, the behavior difference index is an indicator used to quantify the degree of difference between the agent's irrational behavior and normal behavior. It is derived by considering the deviation of the agent's behavior from normal behavior in terms of time, space, and operation mode, and reflects the significance of irrational behavior through a certain calculation method.

[0160] Among them, the behavioral difference index after the agent exhibits irrational behavior at the target time is calculated by analyzing the differences between the agent's behavior and normal behavior in terms of time, space, operation, etc., when the agent exhibits irrational behavior at the set target time. This behavioral difference index is used to measure the degree of deviation between the irrational behavior and normal behavior, and provides a quantitative basis for further analysis of its impact on other agents.

[0161] For example, when faced with a dangerous situation, if one or more people in a group exhibit irrational behavior, others in the group are also more likely to exhibit irrational behavior due to environmental pressures—a phenomenon known as contagion of group behavior. Furthermore, the intensity of the impact of irrational behavior on those around them is influenced by factors such as distance, environment, and behavioral differences. In the generated enterprise emergency drill script, by assigning irrational behaviors to different intelligent agent roles, the intensity of the impact of irrational behavior on other roles in the script can be determined.

[0162] First, based on the Extract the emergency drill script of an enterprise. The first of the roles The action process of an intelligent agent in the event of a sudden crisis.

[0163] Regarding the first The first intelligent agent in the action process The first action, setting this time as the target time, the target time is the [number]th action. When an agent exhibits irrational behavior, the target moment can be recorded as the [missing information]. That moment.

[0164] Subsequently, in the At the i-th moment, calculate the current i-th time. The first agent and other agents The distance between agents (Euclidean distance) It should be noted that this only applies to the first... At that moment, the first If other intelligent agents within the space where the first intelligent agent is located, and if they are related to the first intelligent agent... If agents are not in the same space, their distance is not calculated.

[0165] Based on the distance between multiple agents in the same space at the target time, calculate the current i-th The relative spatial density of the locations of each agent can be represented as follows:

[0166]

[0167] in, Indicates the first The relative spatial density of the locations of each agent; Indicates the first The number of other intelligent agents in the space where the intelligent agent is located.

[0168] At the same time, combined with the first At the moment of the first The progress of the action flow of each agent and the progress of the environmental simulation are analyzed at this time. The urgency of an agent's actions can be represented as follows:

[0169]

[0170] in, Indicates the first The urgency of an agent's actions; Indicates the first The time difference of the action at each moment, the larger the value, the greater the operation delay, and the smaller the value, the earlier the operation. This indicates the urgency of the operation; This indicates the progress of the fire spread; the higher the value, the greater the urgency of the situation.

[0171] Because roles have different functions during corporate emergency drills, and roles with stronger functions have a greater impact when exhibiting irrational behavior, corresponding weights are assigned based on the different roles. It should be noted that the weight can be set to 3 for senior management and special personnel, 2 for key personnel, and 1 for ordinary employees.

[0172] Further analysis of the first The moment A behavioral difference index is derived from the irrational behavior of an agent. It should be noted that the irrational behavioral differences include operation advance / delay, path change, and operation deviation. The behavioral difference index is obtained by quantifying these and calculating the normalized value of their product. Furthermore, since irrational behavior inevitably involves behavioral deviation, the behavioral difference index is further determined by combining the consistency of the agent's behavior with that of surrounding agents and the characteristics of the agent's role. The moment The behavioral difference index after an agent exhibits irrational behavior can be represented as follows:

[0173]

[0174] in, Indicates the first The moment A behavioral difference index after an agent exhibits irrational behavior; Indicates the first The consistency of behavior between an agent and other agents in the same space is determined by the ratio of the total number of agents in the space to the number of agents exhibiting consistent behavior. This is further assessed through role weights. To correct this, since there are fewer agents with larger weights and their behaviors are more distinctive, the larger the weight, the greater the behavioral difference index in the consistency analysis. Indicates the first The time difference of behavior at each moment; the greater the time difference, the greater the behavior difference index. Indicates the first The positional deviation of the agent at any given moment is such that the greater the positional deviation, the greater the behavioral difference index. This represents the normalization function with a range of [0,1].

[0175] In the At that moment, combined with the first Relative spatial density of individual agents Urgency of action and behavioral difference index , obtained the The strength of an agent's influence on other roles in the script can be represented in the following ways:

[0176]

[0177] in, For the first The strength of an agent's influence on other roles in the script.

[0178] Finally, repeat the above steps to obtain the first... The influence strength of an agent at different moments in the script's action flow Calculate their mean, denoted as the th The script in the first The intensity of an agent's irrational influence on other roles .

[0179] In this embodiment, by defining the action flow of multiple agents in a sudden crisis, a clear framework is provided for analyzing the behavior and mutual influence of agents. This makes the simulation and analysis of the entire emergency drill process more systematic and logical, helping to accurately assess the impact of different roles' behaviors on the overall drill, thereby improving the rationality and practicality of the emergency drill script and providing reliable guidance for actual emergencies. Setting the target moment of the multi-agent action flow as irrational behavior and analyzing its subsequent impact allows for targeted study of the role and consequences of irrational behavior in specific stages, accurately locating potential problem nodes, providing a basis for optimizing the emergency drill script, enhancing the drill's ability to respond to irrational behavior in actual crises, and improving the safety of emergency drills. Calculating the distance between multiple agents in the same space at the target moment, the relative spatial density of the agents' positions, the urgency of action, and the behavioral difference index, and comprehensively considering these factors to obtain the intensity of the agents' influence on other roles, comprehensively and meticulously considers multiple influencing factors, enabling more accurate quantification of the mutual influence between roles, making the design of the emergency drill script more scientific and reasonable, improving the simulation degree of the drill on actual situations, and thus enhancing the ability to resist risks. The average influence intensity of the agent at different moments in the action process is taken as the degree of irrational influence of each role. Through multiple analyses and comprehensive considerations, the one-sidedness of single-moment analysis is avoided, and the results are more representative and stable. This helps to assess the impact of irrational behavior of different roles on the entire emergency drill as a whole, further optimize the emergency drill script, and improve the drill effect and actual response capability.

[0180] In one embodiment of this application, the target script determination unit includes:

[0181] The exercise effectiveness analysis subunit is used to analyze the emergency exercise effectiveness of each initial enterprise emergency exercise script and obtain the theoretical emergency evaluation index.

[0182] The reasonable index determination sub-unit is used to combine the irrational behavior tolerance and the theoretical emergency evaluation index, and obtain the reasonable index of each of the initial enterprise emergency drill scripts after normalization.

[0183] The target script determination sub-unit is used to take the initial enterprise emergency drill script corresponding to the largest reasonable index as the target enterprise emergency drill script.

[0184] The effectiveness of emergency drills refers to the results achieved in realizing their predetermined goals. It encompasses multiple dimensions, such as the realism of the simulation of crisis scenarios, the ability of participants to respond to various situations, the rationality of resource allocation during the drill, and the guiding significance of the drill results for actual emergency response. It is a comprehensive reflection of the quality and value of emergency drills.

[0185] Among them, the theoretical emergency evaluation index is a quantitative numerical indicator, which is derived by analyzing and calculating various factors related to emergency drills. It is used to evaluate the quality and effectiveness of emergency drill scripts from a theoretical perspective, and provides a unified measurement standard for comparing the merits of different emergency drill scripts.

[0186] Among them, the analysis of the emergency drill effects of each initial enterprise emergency drill script is used to obtain a theoretical emergency evaluation index. This is to comprehensively evaluate the actual effects of each pre-generated initial enterprise emergency drill script in simulating emergency scenarios and guiding emergency actions, and to convert these effects into a quantitative theoretical emergency evaluation index through a specific calculation method, so as to compare and screen different scripts.

[0187] Normalization is a data processing method that uses specific mathematical transformations to map data to a specific interval (usually [0,1]), making data of different magnitudes or ranges comparable and eliminating the influence of different units or value ranges between data, so as to facilitate analysis and comparison on the same scale.

[0188] The rationality index of each initial enterprise emergency drill script is a comprehensive evaluation indicator. It is calculated by integrating multiple important factors related to the initial enterprise emergency drill script (such as tolerance for irrational behavior and theoretical emergency evaluation index). It reflects the overall rationality of the script in terms of considering the impact of irrational behavior and the effectiveness of emergency drills, and is used to select the optimal script from multiple initial scripts.

[0189] For example, the lower the impact of irrational behavior by different roles in a corporate emergency drill script, the higher the script's tolerance for such behavior. Conversely, the higher the script's tolerance for irrational behavior and the better its performance, the more reasonable it is. By analyzing both the tolerance for irrational behavior and the script's performance in corporate emergency drills, a reasonableness index can be determined, ultimately leading to the selection of the optimal corporate emergency drill script.

[0190] First, regarding the first Each script is based on the intensity of the irrational influence of all characters within it. , recorded as the number The script's tolerance for irrational behavior by its characters. It should be noted that the lower the impact of a character's irrational behavior on other characters in the script, the higher the script's tolerance for irrational behavior. The way a script can represent its tolerance for a character's irrational behavior can be:

[0191]

[0192] Subsequently, regarding the first The script is used to analyze the effectiveness of the emergency drill and calculate the results. Theoretical emergency evaluation index for each script .

[0193] Furthermore, in conjunction with the first The script's tolerance for the character's irrational behavior. and the Theoretical emergency evaluation index for each script Determine the reasonable index of the script ;

[0194]

[0195] Finally, repeat the above steps to obtain the reasonable index of all enterprise emergency drill scripts, filter out the maximum value, and use the initial enterprise emergency drill script corresponding to the largest reasonable index as the target enterprise emergency drill script.

[0196] In this embodiment, a theoretical emergency evaluation index is obtained by analyzing the emergency drill effects of various initial enterprise emergency drill scripts. The scripts are quantitatively evaluated from multiple key dimensions (such as script completeness and drill time), making the drill effects of different scripts comparable. This helps to select scripts that are theoretically more effective in guiding emergency drills, improving the scientific rigor and effectiveness of emergency drills, and providing enterprises with more reliable solutions for responding to actual crises. Normalization is performed to unify indicators of different natures and magnitudes (tolerance for irrational behavior and the theoretical emergency evaluation index) to the same scale, avoiding unreasonable comparisons due to data differences. This ensures that all factors are considered fairly and accurately when calculating a reasonable index, thereby more accurately assessing the overall rationality of the scripts and improving the accuracy and reliability of selecting target scripts. The rationality index of each initial enterprise emergency drill script is calculated, and the script corresponding to the maximum rationality index is taken as the target script. Taking into account two key aspects, namely the script's tolerance for irrational behavior and the effectiveness of emergency drills, the optimal script can be selected in actual application. This can both deal with irrational behavior of personnel and ensure the quality and effectiveness of drills, greatly improving the practicality and adaptability of emergency drill scripts and enhancing the enterprise's response capability and safety in actual emergencies.

[0197] In one embodiment of this application, the step of analyzing the emergency drill effectiveness of each of the initial enterprise emergency drill scripts to obtain a theoretical emergency evaluation index includes:

[0198] The completeness of each initial enterprise emergency drill script is obtained by normalizing the product of the role coverage index, disaster scenario coverage index, and role behavior completeness index in the emergency drill effect of each initial enterprise emergency drill script.

[0199] By combining the emergency drill time for completing each initial enterprise emergency drill script with the completeness of each initial enterprise emergency drill script, a theoretical emergency evaluation index for each initial enterprise emergency drill script is obtained.

[0200] The role coverage index is used to measure the proportion of the role types involved in the emergency drill script to all possible role types in reality. It reflects the comprehensiveness of the script in terms of roles. The higher the proportion, the wider the coverage of various roles in the script, and the more comprehensively it can simulate the personnel composition in the actual emergency scenario.

[0201] The disaster scenario coverage index refers to the ratio between the scope of disaster scenarios simulated by the emergency drill script and the scope of all possible disaster scenarios in reality. It reflects the degree to which the script simulates different disaster scenarios. The higher the ratio, the more comprehensive the script's coverage of various disaster scenarios, and the better it can cope with various disaster situations that may occur in actual emergencies.

[0202] Among them, the Role Behavior Completeness Index is used to evaluate the degree of consistency between the behavior of each role in the emergency drill script and the behavior of similar rational roles in historical data. It reflects the rationality and accuracy of the role behavior in the script. The higher the consistency, the higher the Role Behavior Completeness Index, indicating that the script simulates the behavior of the role in the emergency scenario more closely to the actual situation.

[0203] The completeness of each initial enterprise emergency drill script is a quantitative indicator that comprehensively considers the completeness of the emergency drill script in terms of roles, disaster scenarios, and role behaviors. It is obtained by normalizing the product of the role coverage index, disaster scenario coverage index, and role behavior completeness index, and fully reflects the comprehensiveness and accuracy of the script's simulation of actual emergency scenarios.

[0204] For example, the completeness of a script is evaluated, primarily by considering the character coverage index, disaster scenario coverage index, and character behavior completeness index. The normalized value of their product is calculated and denoted as the [missing value]. The completeness of each script. It should be noted that the completeness index of each character's behavior is determined by comparing it with data such as historical accident reports; that is, analyzing the consistency of its behavior with that of characters (rational characters) in historical data (obtaining behavioral similarity through artificial intelligence algorithms). The higher the consistency, the higher the completeness index. The completeness of a script can be represented as follows:

[0205]

[0206] in, Indicates the first The completeness of each script; , This indicates the number of included character types and the total number of character types. Indicates the role coverage index; , This indicates the number of included ignition points and the total number of possible ignition points. Indicates the disaster coverage index; Indicates the first The completeness index of each character's behavior.

[0207] Calculate the emergency drill time required to complete the emergency drill using this script. With the completeness of the script The product of , we get the first The theoretical emergency response evaluation index for each script indicates that the shorter the required time, the faster the response speed; and the higher the completeness of the script, the higher the safety factor of the exercise. The theoretical emergency evaluation index of a script can be represented as follows:

[0208]

[0209] in, Indicates the first The theoretical emergency evaluation index for each script.

[0210] In this embodiment, by calculating the role coverage index, it is ensured that the emergency drill script fully considers the participation of various roles in actual emergency scenarios, making the drill more realistic and improving the script's authenticity and practicality. Different roles assume different responsibilities during an emergency. Comprehensive role coverage helps to more accurately simulate the emergency response process, identify potential problems in collaboration between different roles in advance, thereby optimizing the script and enhancing the company's ability to cope with actual crises. Calculating the disaster scenario coverage index allows the emergency drill script to cover as many possible disaster scenarios as possible, improving the drill's adaptability to different disaster situations. Different disaster scenarios have different characteristics and response requirements. Comprehensive coverage allows participants to become familiar with emergency handling methods under various disaster scenarios, improving the company's overall emergency preparedness level. Determining the role behavior integrity index ensures the rationality and accuracy of role behavior in the script, making the drill more credible and instructive. By comparing with rational role behavior in historical data, it is possible to effectively avoid unreasonable or unrealistic role behavior in the script, ensuring that the drill process truly reflects the behavioral patterns of personnel in emergency scenarios. The completeness of each initial enterprise's emergency drill script was calculated, comprehensively considering multiple key factors such as roles, disaster scenarios, and role behaviors, to fully assess the script's simulation level of actual emergency scenarios. Higher completeness indicates a more accurate overall reflection of the actual emergency situation, providing a reliable foundation for subsequently calculating the theoretical emergency evaluation index based on the drill time. This helps in selecting higher-quality emergency drill scripts, improving the quality and effectiveness of emergency drills, and effectively enhancing enterprises' response capabilities in actual emergencies. For example, highly complete scripts can comprehensively test an enterprise's emergency response system during drills, promptly identifying and resolving potential problems.

[0211] In one embodiment of this application, the system further includes a testing unit, the testing unit comprising:

[0212] The test data acquisition subunit is used to acquire test data when conducting real-world stress tests based on the target enterprise's emergency drill script.

[0213] The correction subunit is used to correct the emergency drill script of the target enterprise based on the test data, and generate the final enterprise emergency drill script.

[0214] Among them, the practical stress test is a real-world test of the emergency drill script under conditions close to actual application scenarios and loads. It aims to verify the script's feasibility, stability, and effectiveness under high-intensity and complex conditions. By simulating various stress factors in real crisis events, such as time constraints, limited resources, and personnel shortages, the execution effect of the emergency drill script is observed and evaluated. For example, in a practical stress test simulating a factory fire, pressure conditions similar to a real fire, such as the rate of fire spread, smoke concentration, and limited escape routes, are set to observe how employees execute emergency actions according to the script.

[0215] The test data consists of various data collected during the actual stress test. This data reflects the performance of the emergency drill script in real-world execution, including but not limited to the speed and accuracy of personnel actions, resource utilization efficiency, completion time of each stage, and problems encountered. This data provides a basis for evaluating and improving the script. For example, in the actual stress test, the percentage of employees who successfully evacuated within the specified time is recorded as the evacuation rate; the percentage of employees who completed tasks according to the correct operating procedures is counted as the reasonable operation completion rate; and the control effect on disasters such as fires is counted as the disaster control rate. Feedback information, such as employee opinions and suggestions on the drill script, is also collected.

[0216] The process of revising the emergency drill script for the target enterprise based on the test data involves analyzing the problems and shortcomings of the script using test data obtained from real-world stress tests. Targeted adjustments, improvements, and optimizations are then made to enhance the script's quality and practicality, making it more aligned with actual emergency needs. For example, if test data shows a low evacuation rate, it may indicate an unreasonable evacuation route design or insufficient guidance measures; in this case, the evacuation routes and guidance plans in the script will be adjusted. Conversely, if the completion rate of reasonable operations is low, it may indicate overly complex procedures or inadequate training; in this case, the procedures will be simplified or relevant training content will be strengthened, resulting in a more comprehensive final emergency drill script for the enterprise.

[0217] For example, the organization can organize its employees to conduct corresponding drills based on the company's emergency drill script. During the drills, multiple evaluation dimensions such as personnel evacuation rate, reasonable operation completion rate, and disaster control rate can be analyzed. The script can be further adjusted based on the actual drill results and the overall characteristics of the company's employees (physical fitness, quality, etc.).

[0218] In this embodiment, by conducting real-world stress tests, the effectiveness and feasibility of the target company's emergency drill script can be verified in near-realistic scenarios. This allows for the early identification of potential problems in practical application, preventing serious consequences due to script deficiencies during real crises and improving the company's crisis response safety and reliability. Acquiring test data provides objective evidence for script evaluation. This data comprehensively reflects the script's performance during actual execution from multiple dimensions, making the evaluation more accurate and comprehensive. It helps to deeply understand the script's strengths and weaknesses, providing strong support for subsequent revisions. Revising the target company's emergency drill script based on the test data can specifically address existing problems, continuously optimize the script content, and make it more aligned with the company's actual emergency needs. This improves the quality and effectiveness of emergency drills, enhances the company's response capabilities in real emergencies, and ensures swift and effective action to minimize losses in the face of real crises. For example, revising evacuation routes and operational procedures based on test data can improve personnel evacuation efficiency and disaster control capabilities.

[0219] In one embodiment of this application, a method for generating enterprise emergency drill scripts based on multi-agent systems is also provided. Figure 3 This is a flowchart illustrating a method for generating enterprise emergency drill scripts based on multi-agent systems, as provided in one embodiment of the present invention. (See attached diagram.) Figure 3 The method includes the following steps:

[0220] S310: Based on pre-configured multi-agent systems, generate multiple initial enterprise emergency drill scripts;

[0221] S320. Obtain various irrational behavior data from historical accidents, and based on the various irrational behavior data, enable the multi-agent to simulate irrational behavior;

[0222] S330. For each of the initial enterprise emergency drill scripts, analyze the impact of irrational behavior of different roles on other roles through the irrational behavior simulated by the multi-agent system, and determine the tolerance of each of the initial enterprise emergency drill scripts for the irrational behavior of roles.

[0223] S340. Based on the tolerance of irrational behavior of roles in each of the initial enterprise emergency drill scripts, determine the target enterprise emergency drill script from the plurality of initial enterprise emergency drill scripts.

[0224] The multi-agent enterprise emergency drill script generation method in this application, by setting up multiple agents, can simulate the behavior of different roles in emergency scenarios, covering as many possible situations as possible, making the generated scripts closer to actual emergency drill needs. By introducing irrational behavior data from historical accidents, the multi-agent agents simulate these behaviors, enabling the scripts to consider various complex situations that may occur in reality, thereby improving the realism of the scripts. This reduces accidents caused by inconsistencies between personnel behavior and the script, improving safety during emergency drills and enhancing resilience against risks. By analyzing the impact of irrational behavior by different roles on other roles, we can understand the response capabilities of different scripts to irrational behavior, identify which scripts are better at maintaining order and avoiding chaos, and help evaluate the feasibility and safety of scripts in actual emergency drills, thereby improving safety during emergency drills and enhancing resilience against risks. By selecting scripts with high tolerance for irrational behavior—that is, scripts that can remain relatively stable and orderly when faced with irrational behavior from personnel—as target scripts, the scripts ultimately used for emergency drills become more reliable. This improves the safety of emergency drills and the ability to withstand risks. At the same time, by considering possible irrational behaviors in reality, the authenticity of the scripts is also guaranteed.

[0225] Specific embodiments of the enterprise emergency drill script generation method based on multi-agent systems in this application can be found in the examples shown in the above-described enterprise emergency drill script generation system based on multi-agent systems, and will not be repeated here.

[0226] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0227] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A multi-agent-based enterprise emergency drill script generation system, characterized in that, The system includes: The initial script generation unit is used to generate multiple initial enterprise emergency drill scripts based on pre-configured multi-agent systems. The simulation unit is used to acquire various irrational behavior data from historical accidents, and based on the various irrational behavior data, enable the multi-agent to simulate irrational behavior; wherein, the various irrational behavior data include irrational behavior and information on sudden crisis events; The irrational behavior analysis unit is used to analyze the impact of irrational behavior on other roles after different roles exhibit irrational behavior through the irrational behavior simulated by the multi-agent system for each of the initial enterprise emergency drill scripts, and to determine the tolerance of each of the initial enterprise emergency drill scripts for the irrational behavior of the roles. The irrational behavior analysis unit includes: a probability analysis subunit, used to determine the probability of irrational behavior occurring in different roles based on the irrational behavior simulated by the multi-agent system; an irrational impact intensity subunit, used to analyze the impact of irrational behavior on other roles by combining the initial enterprise emergency drill script and the probability of irrational behavior occurring in different roles, and to obtain the degree of irrational impact for each role; and an irrational behavior tolerance analysis subunit, used to obtain the tolerance of each initial enterprise emergency drill script for the irrational behavior of roles based on the degree of irrational impact of each role. Irrational behavior tolerance is an indicator that measures the ability of an emergency drill script to maintain normal drill operations without serious chaos or failure when faced with irrational behavior from roles. The smaller the impact intensity of irrational behavior from different roles in an enterprise emergency drill script, the higher the tolerance of the current script for irrational behavior from roles. The target script determination unit is used to determine the target enterprise emergency drill script from the plurality of initial enterprise emergency drill scripts based on the tolerance of each initial enterprise emergency drill script for irrational behavior of the role. The target script determination unit includes: a drill effect analysis subunit, used to analyze the emergency drill effect of each initial enterprise emergency drill script and obtain a theoretical emergency evaluation index; a rationality index determination subunit, used to combine the irrational behavior tolerance and the theoretical emergency evaluation index, and obtain a rationality index for each initial enterprise emergency drill script after normalization; and a target script determination subunit, used to select the initial enterprise emergency drill script corresponding to the largest rationality index as the target enterprise emergency drill script.

2. The enterprise emergency drill script generation system based on multi-agent technology as described in claim 1, characterized in that, The initial script generation unit includes: The intelligent agent setting subunit is used to obtain the personnel composition information of the enterprise, and set intelligent agents corresponding to different roles according to the personnel composition information to obtain the multi-intelligent agent; The process simulation subunit is used to simulate the occurrence process of a sudden crisis event based on the multi-agent system. The iterative simulation subunit is used to iteratively simulate emergency drills based on preset rules and simulation methods for different roles during the occurrence of simulated sudden crisis events. The initial script generation subunit is used to generate the multiple initial enterprise emergency drill scripts based on the decisions and behaviors of different agents in each emergency drill simulation.

3. The enterprise emergency drill script generation system based on multi-agent technology as described in claim 1, characterized in that, The simulation unit includes a data acquisition subunit, which is used for: Obtain official accident investigation reports, historical emergency drill records, and analysis results of historical disaster cases; Extract the data on various irrational behaviors from the official accident investigation report, the historical emergency drill records, and the analysis results of historical disaster cases.

4. The enterprise emergency drill script generation system based on multi-agent technology as described in claim 1, characterized in that, The determination of the probability of irrational behavior occurring for different roles through the simulation of irrational behavior by the multi-agent system includes: The irrational behavior simulated by the multi-agent is analyzed to obtain the danger index of the sudden crisis event corresponding to the irrational behavior. Combined with the probability of each role appearing in the irrational behavior, the probability of each role exhibiting irrational behavior when the sudden crisis event occurs is obtained. The average similarity of each role in the irrational behavior is obtained by comparing the role characteristics that are prone to irrational behavior with the role characteristics assigned to different intelligent agents. By combining the probability of irrational behavior by each role during the aforementioned sudden crisis event, and the average similarity of the irrational behaviors among the roles, the probability of irrational behavior occurring in different roles during corporate emergency drills is obtained.

5. The enterprise emergency drill script generation system based on multi-agent technology as described in claim 4, characterized in that: The danger index is obtained by processing the casualty coefficient, economic damage coefficient, and environmental hazard coefficient from the analysis results of the irrational behavior. The casualty coefficient is the proportion of casualties to the total number of people, the economic damage coefficient is the proportion of the damaged value to the enterprise value, and the environmental hazard coefficient is the product of the scope and intensity of the event's impact.

6. The enterprise emergency drill script generation system based on multi-agent technology as described in claim 1, characterized in that, The method combines the initial enterprise emergency drill script with the probability of irrational behavior occurring in different roles to analyze the impact of irrational behavior in different roles on other roles, thereby obtaining the degree of irrational impact for each role, including: Based on the initial enterprise emergency drill script, the action flow of multiple agents in a sudden crisis event is determined; The target moment of the multi-agent action in the action flow is set as an irrational behavior, and the distance between the multi-agents in the same space at the target moment is calculated. The relative spatial density of the agent's location is determined based on the distance between the multiple agents in the same space at the target time. By combining the progress of the action process with the progress of the environmental simulation, the urgency of the action is determined; The behavioral difference index after the agent exhibits irrational behavior at the target time is analyzed, and combined with the relative spatial density, the urgency of the action, and the behavioral difference index, the influence intensity of the agent on other roles is obtained. The average value of the agent's influence at different times in the action process is taken as the degree of irrational influence of each role.

7. The enterprise emergency drill script generation system based on multi-agent technology as described in claim 1, characterized in that, The analysis of the emergency drill effectiveness of each initial enterprise emergency drill script yields a theoretical emergency evaluation index, including: The completeness of each initial enterprise emergency drill script is obtained by normalizing the product of the role coverage index, disaster scenario coverage index, and role behavior completeness index in the emergency drill effect of each initial enterprise emergency drill script. By combining the emergency drill time for completing each initial enterprise emergency drill script with the completeness of each initial enterprise emergency drill script, a theoretical emergency evaluation index for each initial enterprise emergency drill script is obtained.

8. The enterprise emergency drill script generation system based on multi-agent technology as described in claim 1, characterized in that, The system also includes a testing unit, which includes: The test data acquisition subunit is used to acquire test data when conducting real-world stress tests based on the target enterprise's emergency drill script. The correction subunit is used to correct the emergency drill script of the target enterprise based on the test data, and generate the final enterprise emergency drill script.

Images