A multi-dimensional scenario construction and emergency plan decision method for chemical safety accidents

Abstract

This invention relates to the field of chemical safety emergency management technology, and discloses a method for constructing multi-dimensional scenarios of chemical safety accidents and making emergency response decisions. The key technical points are: constructing multi-dimensional scenarios of major and serious chemical safety accidents; constructing a scenario inference model based on Bayesian networks and fuzzy mathematics; and evaluating emergency response decisions based on the entropy weight method-TOPSIS. The invention employs knowledge element theory to construct multi-dimensional accident scenarios, combines fuzzy Bayesian networks to realize dynamic evolution and inference of scenarios, and uses the entropy weight-TOPSIS method to evaluate and optimize emergency response plans, thereby forming a systematic and quantitative method for accident scenario construction and decision support.

Description

Technical Field

[0001] This invention relates to the field of chemical safety emergency management technology, and more specifically, to a method for constructing multi-dimensional scenarios of chemical safety accidents and making emergency response decisions. Background Technology

[0002] In the field of chemical safety emergency management, the traditional "prediction-response" model and single-dimensional accident scenario analysis methods generally suffer from problems such as incomplete scenario description, insufficient accuracy of evolution and deduction, and lack of unified quantitative standards for decision evaluation systems. At the same time, existing systems often cannot effectively adapt to the complex characteristics of multiple accident coupling and multi-stage evolution, resulting in high complexity and limited practicality in the implementation of emergency plan decision support systems.

[0003] Therefore, this invention provides a method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents, thereby improving the aforementioned technical problems. Summary of the Invention

[0004] This invention aims to address the shortcomings of existing technologies by providing a method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents. The invention uses knowledge element theory to construct multi-dimensional accident scenarios, combines fuzzy Bayesian networks to realize dynamic evolution and deduction of scenarios, and uses the entropy weight-TOPSIS method to evaluate and optimize emergency response plans, thereby forming a systematic and quantitative method for constructing accident scenarios and providing decision support.

[0005] The above-mentioned technical objective of this invention is achieved through the following technical solution: a method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents, comprising the following steps: S1. Constructing multi-dimensional scenarios for major chemical safety accidents: Based on the public safety triangle model theory, the SHEM analysis framework is introduced to deconstruct accident scenarios into four core units: scenario state (S), human behavior (H), disaster-bearing environment (E), and emergency management (M); a three-level scenario expression system of "performance-scenario-knowledge" is constructed, and the scenarios are divided into historical, real-time, and predictive scenarios using time slicing technology; the knowledge element theory is used to formally represent scenario elements, forming a hierarchical knowledge element expression system. S2. Construct a scenario inference model based on Bayesian networks and fuzzy mathematics: Analyze the four evolutionary mechanisms of accident propagation, derivation, transformation, and coupling, and clarify the positive and negative paths of accident evolution; transform the qualitative evaluation of experts into quantitative probability values ​​through fuzzy mathematics, and combine the bidirectional reasoning ability of Bayesian networks to construct a multi-scenario, multi-dimensional inference model to quantitatively predict the probability of occurrence of different scenario states. S3. Emergency Decision Evaluation Based on Entropy Weight Method-TOPSIS: Construct a multi-level emergency evaluation index system that includes rescue process, personnel safety, emergency management, and accident site cleanup; calculate the objective weights of the indicators using the entropy weight method, and construct a weighted decision matrix after standardization; use the TOPSIS method to calculate the Euclidean distance and relative closeness between each emergency plan and the positive and negative ideal solutions, and select the optimal emergency decision plan.

[0006] As a preferred technical solution of the present invention, in the SHEM analysis framework, the scenario state unit S characterizes the dynamic characteristics of a specific time section of an accident and provides a spatiotemporal reference; the human behavior unit H is divided into directly triggered and indirectly induced unsafe behaviors; the disaster-bearing environment unit E covers natural environmental factors and safety management issues; and the emergency management unit M examines accident handling measures and effectiveness evaluation.

[0007] As a preferred technical solution of the present invention, in the three-level system of "presentation-scenario-knowledge", the presentation layer describes the macroscopic characteristics of the accident through scenario fragments, the scenario layer covers core elements such as human behavior, disaster-bearing environment and emergency measures, and the knowledge layer uses a knowledge meta-structure to define element categories, attribute states and interaction relationships.

[0008] As a preferred embodiment of the present invention, the knowledge element representation includes: common object knowledge elements, common attribute knowledge elements, and common relation knowledge elements; The common knowledge elements of the objects include a set of concept names, a set of attribute states, and a set of relationships; the common knowledge elements of attributes include descriptive attribute features, measurable dimensions, and time-varying mapping relationships; the common knowledge elements of relationships include relationship features, input and output attribute sets, and mapping relationships.

[0009] As a preferred technical solution of the present invention, in the accident evolution mechanism, the propagation mechanism is that the initiating disaster factor triggers a secondary disaster chain through energy transfer and material migration; the derivative mechanism is that inappropriate emergency response measures lead to more serious events; the transformation mechanism is that the accident is transformed into different types of emergencies through energy conversion and material changes; and the coupling mechanism is that multiple disaster-causing factors interact to aggravate the harm of the accident.

[0010] As a preferred technical solution of the present invention, the fuzzy mathematical processing process includes: using triangular fuzzy numbers or trapezoidal fuzzy numbers to represent the qualitative opinions of experts, defuzzifying them through the integral value method, and converting the aggregated fuzzy numbers into deterministic probability values ​​for assigning probability values ​​to Bayesian network nodes.

[0011] As a preferred technical solution of the present invention, the emergency evaluation index system has a three-dimensional structure. The first-level indicators include the rescue process, personnel safety, emergency management, and accident site cleanup; the second-level indicators include material input, personnel input, affected personnel, and evacuated personnel; and the third-level indicators are specific quantitative evaluation parameters.

[0012] As a preferred technical solution of the present invention, the entropy weight method calculation includes: establishing an evaluation matrix and performing standardization processing, calculating the information entropy value of each indicator, obtaining the coefficient of variation of the indicators through the information entropy value, and determining the objective weight of each indicator based on the coefficient of variation.

[0013] As a preferred technical solution of the present invention, the TOPSIS method evaluation includes: constructing a weighted decision matrix based on the standardized matrix and index weights, determining the positive and negative ideal solutions, calculating the weighted Euclidean distance between each solution and the positive and negative ideal solutions, and ranking and selecting the solutions based on their relative proximity.

[0014] As a preferred technical solution of the present invention, the multiple scenarios include the scenario of illegal storage of hazardous chemicals, the initial stage of a fire, and the scenario of explosion and secondary disasters. Each scenario is dynamically deduced through the causal relationship of key state nodes. The key state nodes of the previous scenario directly affect the evolution direction and probability of occurrence of the next scenario.

[0015] In summary, the present invention has the following beneficial effects: First, based on the SHEM framework and the three-level system of "representation-context-knowledge", it integrates multiple dimensions of situational status, human behavior, disaster-bearing environment, and emergency management, and combines knowledge element representation to achieve a complete description of the evolutionary characteristics of the entire life cycle of an accident, thus solving the problem of incomplete scenario depiction in traditional methods.

[0016] Secondly, by integrating expert experience and quantitative calculations through fuzzy Bayesian networks, the complex mechanisms of accident propagation, derivation, transformation, and coupling are revealed, the probability of occurrence of different evolution paths is clarified, the accuracy of accident development trend prediction is improved, and the limitations of traditional inference models are made up for.

[0017] Third, by using the entropy weight method-TOPSIS combined model, the evaluation indicators are objectively assigned weights, and the comprehensive effectiveness of different emergency plans is quantitatively analyzed, providing a clear basis for decision-making and solving the problem of the lack of unified quantitative standards in traditional decision-making.

[0018] Fourth, through configurable and differentiated scenario units and evaluation factors, it can adapt to the needs of different accident types and evolution stages, without having to build a separate system for each complex accident, thus balancing universality and specificity. Attached Figure Description

[0019] Figure 1 A flowchart illustrating a method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the technical route of a method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents, as provided in an embodiment of the present invention. Figure 3A schematic diagram of the SHEM framework provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the components of a scenario unit for major and serious chemical safety accidents provided in an embodiment of the present invention; Figure 5 A framework diagram of the scenario representation hierarchy model provided in the embodiments of the present invention; Figure 6 This is a knowledge element hierarchy structure diagram provided for embodiments of the present invention; Figure 7 This is a schematic diagram of accident scenario division provided in an embodiment of the present invention; Figure 8 This is a schematic diagram illustrating the evolution of an accident scenario provided in an embodiment of the present invention. Figure 9 This is a schematic diagram of the accident transformation mechanism provided in an embodiment of the present invention; Figure 10 This is a schematic diagram of the accident coupling mechanism provided in an embodiment of the present invention; Figure 11 This is a schematic diagram illustrating the evolution of accident scenarios provided in an embodiment of the present invention. Figure 12 This is a multi-scenario, multi-dimensional scenario analysis framework diagram for major chemical accidents provided in the embodiments of the present invention; Figure 13 This is a schematic diagram illustrating the scenario evolution provided in an embodiment of the present invention; Figure 14 A schematic diagram illustrating the deductive path of a scenario involving the illegal storage of hazardous chemicals, as provided in an embodiment of the present invention; Figure 15 A schematic diagram illustrating the deduction path of the initial fire scenario provided in an embodiment of the present invention; Figure 16 A schematic diagram illustrating the deduction path of an explosion and secondary disaster scenario provided in an embodiment of the present invention; Figure 17 A framework diagram of a multi-scenario reasoning network model for accidents provided in an embodiment of the present invention; Figure 18 A Bayesian network framework diagram for a hazardous chemical storage violation scenario provided in an embodiment of the present invention; Figure 19 A Bayesian network framework diagram for the initial fire scenario provided in an embodiment of the present invention; Figure 20 A Bayesian network framework diagram for an explosion and secondary disaster scenario provided in an embodiment of the present invention; Figure 21 This is a framework diagram of a Bayesian network-based accident multi-dimensional and multi-scenario inference model provided in an embodiment of the present invention. Figure 22 This is a framework diagram of the diagnostic results under the condition of "Ha1" occurrence provided in the embodiments of the present invention; Figure 23 This is a framework diagram of diagnostic results under the condition of "Ha2" occurrence provided in an embodiment of the present invention; Figure 24 This is a framework diagram of the diagnostic results under the condition that "Ha3" does not occur, as provided in the embodiments of the present invention. Detailed Implementation

[0020] The present application will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any way. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application. These all fall within the protection scope of the present application.

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0022] Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items.

[0023] Furthermore, the technical features involved in the various embodiments of this application described below can be combined with each other as long as they do not conflict with each other.

[0024] This disclosure aims to address the problems in emergency management of major chemical safety accidents, such as incomplete scenario characterization, inaccurate evolutionary projections, lack of a unified quantitative system for decision-making and evaluation, and excessively high system complexity due to the inability of the system to adapt to the coupled characteristics of multiple accidents. In view of this, this disclosure proposes a multi-dimensional scenario construction and emergency plan decision-making method for chemical safety accidents to accurately characterize the accident evolution and evaluate the merits of emergency plans. This method employs knowledge element representation, fuzzy Bayesian network inference, and entropy weight-TOPSIS evaluation technology. Through configurable and differentiated scenario units and evaluation factors, it addresses the scenario construction and decision-making needs of different accident types and evolutionary stages, thereby achieving a unified emergency evaluation system for complex accidents and reducing system implementation complexity and cost.

[0025] Please refer to Figure 1-2 , Figure 1-2The flowchart and technical route of a multi-dimensional scenario construction and emergency response decision-making method for chemical safety accidents, as described in this disclosure, are illustrated. The overall process mainly includes the following three steps: S1. Constructing Multi-Dimensional Scenarios for Major Chemical Safety Accidents: First, based on the public safety triangle model theory, a scenario element analysis framework is constructed that includes three dimensions: personnel behavior, disaster-bearing environment, and emergency preparedness. Second, in terms of scenario description methods, a three-level progressive description system of "performance-scenario-knowledge" is proposed, laying a theoretical foundation for subsequent knowledge element modeling. Based on this, a dynamic expression method for accident scenarios based on knowledge elements is proposed, which can fully depict the evolutionary characteristics of the entire life cycle of an event.

[0026] S1.1 Scenario analysis framework based on SHEM framework.

[0027] Combining the public safety triangle theoretical model, the SHEM (Safety-Human-Environment-Management) analytical framework is introduced, such as... Figure 3 As shown, this framework is designed specifically for the characteristics of major chemical accidents and can effectively address the limitations of traditional methods in characterizing such accidents.

[0028] This invention deconstructs the scenario system of major chemical safety accidents based on the SHEM framework into four interconnected core units: scenario state S, human behavior H, disaster-bearing environment E, and emergency management M.

[0029] The scenario state unit S characterizes the dynamic features of an accident at a specific time segment. It reflects the evolution trend of the accident through the state attributes of key elements and provides a spatiotemporal reference for other elements as the main line of system evolution. Human behavior unit H focuses on human error as a key cause. Based on accident case analysis, unsafe human behavior is divided into two categories: direct triggering type and indirect inducing type. The former directly leads to the occurrence of the accident, while the latter indirectly affects the accident process by causing unsafe conditions of equipment defects and other things. Disaster-bearing environmental unit E covers various environmental factors that affect the development of an accident, including natural environmental factors such as severe weather and high temperatures, as well as safety management issues such as untimely equipment maintenance and improper material storage. Emergency management unit M systematically examines the various measures taken to control the accident situation and evaluates their effectiveness.

[0030] The four units, through complex interactions, jointly shape the accident development path. The scenario state serves as the core thread, dynamically integrating the comprehensive impact of various factors on the affected body. Each unit contains several specific elements, and through the coupling effects between these elements, a complete accident scenario representation system is ultimately formed. The specific content of each element unit is as follows: Figure 4 As shown.

[0031] S1.2 Contextual Hierarchy Analysis.

[0032] This invention, based on a comprehensive analysis of the entire accident evolution process, deconstructs the continuous process of a major chemical accident from its initial outbreak to its final handling into multiple time-related scenario segments, each representing a specific state of the accident system. These scenario segments involve multiple object entities with different characteristic attributes within the system, and their dynamic interactions constitute a complex network of accident evolution. Based on a general hierarchical model, a three-level scenario representation system comprising a presentation layer, a scenario layer, and a knowledge layer is constructed. like Figure 5 As shown, the presentation layer, as the outermost framework, provides a general description of the macroscopic characteristics of the accident through scenario fragments. The scenario layer, as the intermediate structure, uses a set of scenario categories to cover core elements such as human behavior, disaster-prone environment, and emergency measures, achieving a systematic decomposition of the presentation layer. The knowledge layer, as the basic expression unit, uses a knowledge meta-structure to formally represent scenario elements, providing a modeling foundation for scenario deduction by defining feature parameters such as element categories, attribute states, and interaction relationships. This hierarchical expression system maintains theoretical consistency with general models while providing specialized extensions for the characteristics of chemical accidents, achieving a systematic expression from macroscopic phenomena to microscopic mechanisms.

[0033] In the dynamic evolution of chemical accidents, the relevant objective entities undergo phased state changes over time, leading to significant differences in their corresponding scenario elements at different time points. This temporal characteristic not only constitutes the sequential dimension of the accident state evolution but also serves as a core entry point for analyzing accident mechanisms and identifying the current situation. Addressing the unique lifecycle characteristics of major chemical accidents, this invention proposes a time-dimensional scenario classification framework: Historical scenarios record the complete evolution trajectory of an accident from its initial outbreak to the present moment; real-time scenarios depict the comprehensive performance of the accident system at the current moment; and predictive scenarios infer the potential development trend of the accident in the future. By introducing time-slicing technology, the continuous accident evolution process is discretized into state units with clear temporal characteristics. This time-series discretization method provides an effective modeling tool for the phased analysis of the development process of chemical accidents, enabling each time slice to accurately characterize the scenario state characteristics under a specific time series.

[0034] S1.3 Contextual knowledge representation based on knowledge elements.

[0035] S1.3.1 Knowledge Meta-Theory.

[0036] Scenario simulations of major chemical safety accidents require an objective description of the accident process and analysis of key elements. Knowledge element theory can be used to systematically integrate the scenario units and elements identified earlier, accurately expressing their inherent logical relationships and providing a structured framework for accident analysis. A temporal scenario chain construction method is employed to systematically decompose the accident development process into several key scenario units. Based on the temporal dimension characteristics of accident evolution, each scenario node is sequentially connected through causal logical relationships, forming a complete accident development chain.

[0037] Based on existing common knowledge element models, the concepts, attributes, and evolutionary laws of objective entities are formally represented. As the smallest independent unit in a knowledge system, a knowledge element can comprehensively represent an objective entity through the systematic integration of multiple elements. In the field of major chemical accident analysis, this theoretical method can achieve a precise description of the constituent elements of an accident scenario at a microscale. For object m, the common knowledge element model is as follows: ; In the formula, For common knowledge elements of objects, For a set of concept names, For attribute state set, For a set of relations; For attribute common knowledge elements, To describe attribute characteristics, The dimensions of measurable attributes, This represents a time-varying mapping relationship for attributes. As a common knowledge element of relationships, To characterize the relationship features (linear, nonlinear, etc.). and These are the input and output attribute sets, respectively. This represents a mapping relationship.

[0038] Major chemical accident scenarios are essentially structured representations of disaster events with spatiotemporal characteristics within the cognitive domain of emergency decision-makers. This representation requires a dual perspective, combining static and dynamic elements. The static dimension focuses on the standardized definition of the scenario's conceptual system and the precise characterization of its attributes, while the dynamic dimension focuses on the formal expression of the uncertainties and time-varying characteristics in the evolution of the emergency. This dual-representation framework ensures both the clarity of scenario elements and fully reflects the dynamic characteristics of accident development, providing a complete knowledge representation foundation for emergency response decision-making.

[0039] S1.3.2 Contextual Representation Based on Knowledge Element Theory.

[0040] In scenario analysis, static feature identification focuses on extracting and describing the state of elements, while dynamic relationship analysis delves into the interaction mechanisms between elements. Decision-makers systematically identify scenario elements and establish their relationships, gradually building a complete cognitive system for emergency scenarios. This cognitive process provides a solid theoretical foundation for accident situation assessment and emergency response decision-making, maintaining the integrity of scenario analysis while fully considering the unique dynamic complexity of chemical accidents. This theoretical framework, by integrating static features and dynamic relationships, achieves a systematic characterization of major chemical accident scenarios.

[0041] The development of major chemical accidents is sequential, and its complete scenario information consists of multiple scenario units across multiple time series. By dividing the entire accident process into n stages, the scenario information can be represented as follows: ; In the formula, This is the initial scenario. This is the final scene of the accident.

[0042] At a single point in time, scenario information reflects the specific state of the accident at that moment, consisting of three components: human behavior (H), the disaster-affected environment (E), and emergency management (M). Based on the analyzed scenario components, a single scenario is represented as: ; Each context element consists of three structured dimensions: concept definition, attribute characteristics, and relationships. These can be systematically represented using the knowledge element standardization method, as follows: ; In the formula, A set of conceptual names for accident scenarios. It is a set of attributes for contextual elements, including two parameter types: qualitative description and quantitative representation. This refers to the coupling relationship between internal elements.

[0043] Feature attribute specifications are described as follows: ; In the formula, For describable characteristics, It is a measurable dimension. It is a time-varying function.

[0044] Based on this, a three-tiered contextual knowledge element representation architecture is established. The top tier represents the overall context, the middle tier comprises three core elements (H, E, M), and the bottom tier provides a detailed description of the attributes, characteristics, and interactions of each element. This hierarchical knowledge element representation system achieves a step-by-step deconstruction from the macro-context to the micro-elements, such as... Figure 6 As shown.

[0045] In summary, the knowledge-based scenario representation model proposed in this application achieves a comprehensive description of major chemical accidents through standardized knowledge representation methods and multi-level subject structure analysis. This model uses knowledge elements to accurately characterize the objective entity states within the accident system, and reveals the interaction mechanisms between various elements through a hierarchical structure, effectively capturing the evolutionary process of major chemical accidents.

[0046] S2. Construct a scenario inference model based on Bayesian networks and fuzzy mathematics.

[0047] S2.1 Analysis of the evolution mechanism of major and serious chemical accidents.

[0048] The emergency scenario evolution of major chemical accidents exhibits typical dynamic and complex characteristics. This process is simultaneously influenced by a triple mechanism: the accident's own developmental laws, human intervention, and surrounding environmental factors, undergoing continuous state transitions from the initial state. The accident situation evolves over time, often accompanied by the continuous accumulation of secondary consequences such as casualties and property damage. This temporal scenario evolution pattern essentially reflects the complex dynamic process of nonlinear interactions between various elements within the emergency system, exhibiting a clear stage-by-stage progression characteristic over time. The entire evolutionary process is governed by the accident's inherent developmental logic and regulated by external intervention measures and environmental conditions, forming a dynamic system behavior driven by multiple coupled factors, such as... Figure 7 As shown.

[0049] like Figure 8 As shown, the scenario evolution of major chemical accidents can be divided into three scenarios. The initial stage is the accident occurrence scenario, which is the basic form of the initial manifestation of the emergency. As time goes by, under the combined effects of the event's own evolutionary laws, human intervention measures, and environmental factors, the accident enters the development and evolution scenario stage, and generates chain effects through the interaction of internal elements and external correlations, specifically manifested in four evolutionary modes: spread, derivation, transformation, and coupling. Finally, with the completion of emergency rescue work, the accident system reaches the disappearance scenario, marking the end of the emergency's life cycle.

[0050] The spread mechanism of major chemical accidents is a nonlinear evolution process that induces secondary disasters. The initiating factors trigger multidimensional secondary disaster chains through energy transfer, material migration, and other pathways, forming a complex situation in which the primary and secondary disaster systems coexist.

[0051] The derivative mechanism of major and serious chemical accidents refers to the occurrence of other types of more serious events due to inappropriate emergency response measures.

[0052] The transformation mechanism of major and catastrophic chemical accidents refers to the qualitative change that occurs in an accident under the influence of internal and external factors, transforming it into different types of emergencies. For example... Figure 9 As shown, when a new type of accident occurs, the original accident scenario is replaced and disappears. This transformation is achieved through energy conversion, material changes, and other means, and is influenced by both the characteristics of the accident itself and the external environment, leading to a fundamental change in the nature and harm of the event.

[0053] The coupling mechanism of major chemical accidents refers to the phenomenon where the interaction of multiple hazard-causing factors exacerbates a sudden event. Specifically, while the initial event A persists, it induces other secondary events. These events intertwine and influence each other, forming a complex coupling effect that leads to a more severe accident. For example... Figure 10 As shown.

[0054] S2.2 Analysis of the evolution patterns and pathways of major and serious chemical accidents.

[0055] An accident scenario analysis framework was constructed from three key dimensions: human behavior, disaster-bearing environment, and emergency management. Within this framework, the evolution of an accident scenario is essentially a process in which various scenario elements interact and jointly drive continuous changes in the event state. Specifically, the initial scenario state S0 transforms under the combined effects of unsafe human behavior, disaster-bearing environmental conditions, and emergency management measures. The previous scenario state, as an input condition, directly influences the evolution direction and probability of subsequent state S1, thus forming a temporally correlated scenario evolution chain.

[0056] To more simply and vividly describe the process of scenario evolution, the accident is divided into n stages, namely: The evolution path of major chemical accidents is as follows: Figure 11 As shown.

[0057] The evolution of major chemical accidents exhibits significant nonlinear characteristics, with its trajectory showing bidirectional bifurcation due to the coupling effects of multiple factors. At any evolutionary node, the accident system may evolve into either a positive or negative path. This evolutionary characteristic dictates that the essence of accident development trend assessment lies in the quantitative evaluation of the probability of different development paths, with the core being the identification of the most probable evolutionary direction. By establishing probabilistic prediction models, emergency decision-makers can be provided with forward-looking situation predictions, thereby enabling targeted intervention measures to effectively prevent the accident from developing in an unfavorable direction.

[0058] The construction process of the scenario simulation system for major chemical accidents comprises four stages: key scenario element extraction, scenario construction, scenario simulation, and analysis and evaluation. In the key scenario element extraction stage, it is necessary to systematically analyze multi-dimensional influencing factors such as environmental characteristic parameters, emergency resource allocation, and accident characteristics, and select key elements with decision-making value. This forms the cornerstone of the entire emergency response decision-making process. The scenario construction stage establishes a complete accident scenario chain model through in-depth research into the dynamic characteristics of accident evolution and the interaction mechanisms between elements. The scenario simulation stage, as the core functional module of the system, uses simulation software to simulate and predict multiple scenarios of accident development trends. The final analysis and evaluation stage analyzes and evaluates the simulation results to ensure the reliability and applicability of the decision support.

[0059] S2.3 Construct a scenario inference model based on Bayesian networks.

[0060] S2.3.1, Bayesian Network.

[0061] In a Bayesian network, when the parent node is x The child node is y In this case, based on the axiom of conditional independence, the following law of total probability can be established: ; As the above analysis shows, the mathematical essence of the law of total probability lies in deriving the posterior probability distribution of child nodes from the prior probability and conditional probability relationship of the parent node. In contrast, Bayes' theorem uses reverse reasoning. When a specific event is certain to occur, it can be used to infer the probability distribution of the potential causes that led to the event. The formula is as follows: ; Bayesian networks, through their unique probabilistic reasoning mechanism, can accurately predict the development trend of accidents and identify potential risks; at the same time, they can also conduct in-depth analysis of accidents that have already occurred, tracing back to key causal factors. Their bidirectional reasoning capability supports both forward prediction based on existing information and reverse diagnosis from result to cause, providing a scientific basis for decision-making in the prevention and emergency response to sudden events.

[0062] S2.3.2 Fuzzy Mathematics Theory.

[0063] By establishing mathematical mapping relationships, fuzzy linguistic variables with qualitative descriptions such as "higher" and "lower" are transformed into calculable quantitative indicators, providing an effective solution for modeling and analyzing uncertainty problems.

[0064] For a triangular fuzzy number A(a, b, c), its membership function is: ; For a trapezoidal fuzzy number A(a, b, c, d), the membership function is expressed as: ; After obtaining the desired fuzzy numerical representation, it needs to be transformed into precise probability values ​​through defuzzification. This application employs the integral value method to achieve this transformation process. This method is based on the cut operation of the membership function, as shown in the following formula: ; In the formula, This represents the fuzzy number. Optimism coefficient It is the integral value of the left and right membership functions of the fuzzy number.

[0065] ; In the formula, Let A be the upper and lower bounds of the fuzzy number A cut set. .

[0066] S2.3.3 Framework for Scenario Simulation Model of Major and Serious Chemical Accidents.

[0067] Major chemical accidents are characterized by complexity and derivative effects, often involving multiple accident scenarios. Traditional "prediction-response" models have proven ineffective in handling complex emergency scenarios. Decision-makers must make appropriate decisions based on different scenarios during the accident's evolution. Therefore, based on a "scenario-response" decision-making model, a multi-scenario, multi-dimensional extrapolation and analysis framework for major chemical accidents has been established, such as... Figure 12 As shown, this framework helps decision-makers assess scenario evolution under different accident scenarios and prevent future trends.

[0068] S3. Emergency Response Plan Decision Evaluation Based on Entropy Weight Method-TOPSIS: First, differentiated response strategies are proposed based on the characteristics of different stages of accident development; second, a comprehensive evaluation system with multi-level indicators is constructed to provide structured standards for decision evaluation; finally, the entropy weight method and TOPSIS method are combined to establish an optimal model for emergency response plan decision schemes.

[0069] S3.1 Comprehensive evaluation index system for emergency response measures based on scenario simulation system.

[0070] Based on the reference to the emergency management evaluation framework of industry benchmark enterprises and the analysis of relevant research literature, this application constructs a comprehensive evaluation system for emergency response decision-making with three dimensions. Following the basic principle of full-cycle management of chemical accidents, the selection of indicators focuses on three core elements: (1) source analysis of accident-causing factors; (2) evaluation of emergency response decision-making efficiency; and (3) evaluation of optimal allocation of rescue resources. Based on this, personnel safety, rescue process, and emergency management are established as primary indicators.

[0071] The specific indicator framework exhibits a hierarchical progression: primary indicators are further subdivided into secondary indicators, such as personnel safety, which can be broken down into two sub-items: disaster victims and evacuees; tertiary indicators establish specific quantitative assessment parameters, forming an operational assessment indicator system. This hierarchical design ensures both the systematic nature of the theoretical framework and the practical guiding value of the assessment process.

[0072] The specific details are shown in Table 3-1 below: Table 3-1 Comprehensive Evaluation System for Emergency Response to Major and Serious Chemical Accidents: ; S3.2 Construct an entropy weight method decision evaluation model.

[0073] S3.2.1 Definition of information entropy.

[0074] In the decision-making process for emergency response plans in major accidents in the chemical industry, key information is often presented in a vague and unstructured language. In this case, information entropy can be used to measure the disorder of decision-making information. Generally speaking, the symbols output by information sources are random, and their uncertainty can be quantified by probability distribution. The higher the probability, the stronger the certainty; the lower the probability, the more significant the uncertainty. The core elements of the entropy weight method include information quantity and information entropy. The former reflects the effective value of decision-making information, while the latter reflects its degree of disorder. The lower the information entropy, the higher the degree of organization of the information.

[0075] Based on this, the entropy weight method optimizes the allocation of indicator weights by calculating the information entropy value, thereby providing an objective quantitative basis for emergency response plan decision-making and evaluation. The formula for calculating information entropy is: ; In the formula, This represents the probability value for decision-making in the corresponding emergency response plan. For the first The value of an emergency response plan decision. ， . This indicates the emergency response decision-making group. Therefore, It contains both the information content of emergency response plan decisions and represents the information entropy value of the decision-making system.

[0076] S3.2.2 Entropy weight analysis.

[0077] In entropy weight analysis, an evaluation matrix must first be established. Where n is the number of candidate indicators and m is the indicator attribute system. To eliminate the influence of dimensions, the original evaluation matrix needs to be standardized to obtain a normalized matrix. The specific steps are as follows: First, for the j-th evaluation index, calculate its information entropy value. : ; In the formula, k is a standardization constant. Let represent the standardized value of the i-th scheme under the j-th indicator.

[0078] Secondly, based on the entropy calculation results, the coefficient of variation of each indicator is calculated. : ; Finally, the weights of each indicator are calculated based on the coefficient of variation.

[0079] ; S3.3 Construct a decision-making and evaluation model for emergency response plans for major and serious chemical safety accidents.

[0080] S3.3.1 Construction of the original matrix of evaluation indicators.

[0081] When applying the entropy weight method in the decision-making and evaluation of emergency response plans for major chemical accidents, the information entropy theory has important guiding significance. Indicators with a larger amount of information often correspond to more effective emergency response plan decision-making schemes, while indicators with a lower information entropy value indicate that the decision-making dimension has higher certainty and reliability.

[0082] Based on this, the present invention constructs an evaluation system containing m candidate emergency response plan decision schemes, whose attribute index set can be formally represented as follows: The evaluation system adopts the three-level indicator structure described above, denoted as... The final original matrix obtained. as follows: ; in, For the corresponding nth indicator, For the first m The decision value of an emergency response strategy for the nth indicator.

[0083] S3.3.2, Normalization of the index matrix.

[0084] In the decision-making and evaluation of emergency response plans for major chemical accidents, the indicator matrix contains multi-dimensional and heterogeneous evaluation indicators. For example, rescue timeliness is measured in time units, while resource adequacy is expressed as a percentage, resulting in significant differences in dimensions and orders of magnitude among the indicators. To achieve comparability and aggregation among the indicators, it is necessary to transform the original indicators into standardized values ​​through dimensionless processing. This normalization process is essentially a key step in eliminating the heterogeneity of indicators through mathematical transformation. The normalization steps are as follows: Positive indicators for dimensional unification: ; Negative indicators of dimensional unification: ; Order of magnitude normalization: ; Will Multiplying by the weights yields the standardized weighted matrix, denoted as . V .

[0085] S3.3.3 Emergency Response Plan Decision-Making Optimization Analysis Based on TOPSIS Method

[0086] In the decision-making process for emergency response plans in major chemical accidents, multiple conflicting decision objectives exist. The TOPSIS method can determine the relative importance of each evaluation indicator through quantitative calculation, thereby selecting the optimal decision scheme that takes into account multiple objectives. The advantage of this method is that it can simultaneously consider the degree to which each scheme is close to the ideal solution and the degree to which it is far from the negative ideal solution, providing a scientific basis for emergency response plan decisions in complex environments.

[0087] For multi-objective decision-making problems such as major chemical accidents, a weighted decision matrix V is first constructed based on a standardized matrix and index weights, and then positive and negative ideal solutions are determined. Subsequently, the weighted Euclidean distance between the index values ​​of each scheme and the reference solution is calculated. Finally, the schemes are ranked and optimized through the relative proximity index, and the larger the index value, the better the overall performance of the scheme.

[0088] Positive Ideal Solution Sure: ; In the formula, .

[0089] Negative ideal solution Sure: ; In the formula, .

[0090] Euclidean distance from solution i to the ideal solution : ; Euclidean distance to the negative ideal solution : ; Attribute closeness calculate: ; The larger the value, the closer the solution i is to the ideal optimal state.

[0091] Example: A company located in a chemical industrial park was established in May 2011. Its business scope included general operations such as import and export trade, container loading and unloading, and logistics information consulting, as well as licensed operations such as port storage and transportation of 74 types of hazardous chemicals, including ammonium nitrate, sodium cyanide, nitrocellulose, calcium carbide, sulfur, and methanol. Its core operating facility was the Yuejin Road Hazardous Materials Warehouse in Tianjin Port, which had separate storage areas for seven types of hazardous chemicals, including flammable solids (nitrocellulose), oxidizers (ammonium nitrate), and corrosives (sodium cyanide), using a mixed open-air and warehouse storage model. At 22:51 on August 12, 2015, a violent explosion of ammonium nitrate and other hazardous chemicals occurred in the company's hazardous materials warehouse due to spontaneous combustion of nitrocellulose. The accident resulted in six large fire points and dozens of smaller fire points, causing 165 deaths, 8 missing persons, and 798 injuries requiring hospitalization.

[0092] According to the accident investigation report released by the National Emergency Management Department, the complete scenario evolution from "precursors-event-response" is described in detail. For example... Figure 13 As shown, the corresponding accident nodes are arranged in chronological order of their occurrence.

[0093] Step 1: Determine the elements of multiple scenes.

[0094] The accident was categorized into three scenarios: illegal storage of hazardous chemicals (risk accumulation phase), initial fire phase (energy accumulation phase), and explosion and secondary disasters (energy release and diffusion phase). Extensive information related to the accident was collected and analyzed, and scientifically sound scenario assumptions and variations were made. Through scenario deduction, the possible scenario elements and their relationships during the accident were analyzed. Detailed descriptions of the proposed solutions and confirmation of their elements are provided below.

[0095] Scenario 1: Illegal storage of hazardous chemicals a Through analysis of the illegal storage of hazardous chemicals, seven scenario states, four types of human behavior, four types of disaster-prone environments, and four types of emergency management were identified. The specific scenario elements are shown in Table 1.

[0096] Table 1. Scenario elements of illegal storage of hazardous chemicals.

[0097] ;

[0098] Scenario 2: Initial Stage of Fire b Similarly, through the analysis of the initial stage of a fire, nine scenario states, five types of human behavior, five types of disaster-bearing environments, and five types of emergency management were identified. The specific scenario elements are shown in Table 2.

[0099] Table 2. Contextual Elements of the Initial Stage of a Fire ; Scenario 3: Explosion and Secondary Disasters c Finally, through the analysis of the explosion and secondary disaster stages, six scenario states, six types of human behavior, six types of disaster-bearing environments, and six types of emergency management were identified. The specific scenario elements are shown in Table 3.

[0100] Table 3 Scenario Elements of Explosion and Secondary Disaster Scenarios ; Step 2: Construction of the scenario inference network based on knowledge element theory.

[0101] After identifying the elements of each accident scenario, and considering the actual circumstances of the emergency, the evolution path of each scenario was constructed using the directed edge method to express the relationships between different elements. During the evolution of an emergency scenario, the development trend is often influenced by the interactions between various scenario elements. The current scenario may evolve into two distinct paths: one is a positive scenario path, where the current scenario develops towards the expected goal with the implementation of rational decision-making measures; the other is a negative scenario path, where the current scenario deteriorates into a negative outcome if measures are not implemented or unreasonable decisions are made, usually exacerbating the losses caused by the accident.

[0102] In the evolution of the scenario of illegal storage of hazardous chemicals, scenario state S a ={S a0 S a2 S a4 S a6} Subject to human behavior H a ={H a0 H a1 H a2 H a3 Disaster-bearing environment E a ={E a0 E a1 E a2 E a3} and emergency management M a ={M a0 M a1 M a2 M a3 The combined effects of these factors lead to a positive or negative evolution of the current situation. The dynamic projection path for scenarios involving the illegal storage of hazardous chemicals is as follows: Figure 14 As shown.

[0103] During the initial fire scenario evolution, scenario state S b Subject to human behavior H b Disaster-bearing environment E b and emergency management M b The combined influence of these factors leads to a positive or negative evolution of the current situation. The dynamic evolution path of the situation at this stage is as follows: Figure 15 As shown.

[0104] During the evolution of the explosion and secondary disaster scenario, scenario state S c Subject to human behavior H c Disaster-bearing environment E c and emergency management M c The combined influence of these factors leads to a positive or negative evolution of the current situation. The dynamic evolution path of the situation at this stage is as follows: Figure 16 As shown.

[0105] Based on a detailed analysis of this major accident, the accident evolution process is divided into three scenarios: illegal storage of hazardous chemicals, initial stage of fire, and explosion and secondary disasters. Importantly, these three scenarios are specific to this case, and different accidents can be divided into different scenarios. Figure 17 A multi-scenario, multi-dimensional reasoning network model was demonstrated. Accident analysis showed that in the scenario of "uncontrolled storage environment and spontaneous combustion" in the case of illegal storage of hazardous chemicals, the S... a6 =0, meaning it does not occur, based on the relationship between the situation stages, S b0 In the initial fire scenario, the hazardous chemicals will remain in a safe state and no fire will occur. Therefore, the scenario state element S a6 For situational state element S b0 The influence allows for dynamic deduction between these two scenarios; similarly, the scenario state element S b8 With situational state element S c0 There is also a process of deduction between them.

[0106] Step 3: Construction and analysis of scenario inference model based on Bayesian network.

[0107] Forward evolution analysis: In order to improve the quantitative analysis capability of the scenario, the BN model needs to determine detailed parameters, including the probability of the root node and the conditional probability table of non-root nodes.

[0108] Expert consultation was conducted using electronic questionnaires and interviews to collect professional opinions from five experts in the field of emergency response. Due to the large number of scenario nodes in the model, this application only considers the "occurrence" state of scenario nodes, simplifying expert evaluation. The probability of a node "not occurring" is calculated by subtracting the probability of "occurrence" from 1. After collecting expert evaluations, it is crucial to summarize these opinions to build consensus. The summarization and calculation of node probabilities utilize the fuzzy mathematics method outlined above.

[0109] Scenario 1: Illegal storage of hazardous chemicals a .

[0110] The following uses the root node "H" a0 The process of expert scoring is illustrated using "humidity testing" as an example.

[0111] The five experts gave "H" respectively a0 The opinion for the "humidity test" is High (H), Relatively High (FH), High (H), Very High (VH), High (H). The process for calculating the aggregated fuzzy number is as follows: ; Because fuzzy numbers possess inherent uncertainty, to address this issue, defuzzification techniques must be employed to transform the aggregated fuzzy values ​​into deterministic probability values ​​usable by Bayesian networks. The process is as follows: (5-2); Similarly, the fuzzy solutions for all root nodes are obtained, and the results are summarized in Table 4. After obtaining the prior probabilities of all root nodes, the conditional probability of scenario state S can also be calculated.

[0112] Table 4. Results of expert evaluation (Scenario 1) ; After assigning probability values ​​to each scenario node, the quantitative analysis method based on Bayesian networks can be implemented. This invention uses GeNle simulation software to construct a probability graph model of accident evolution. First, network nodes are created in the modeling interface, attribute parameters are set for each node, and pre-determined probability distribution data is entered. Second, causally related node elements are connected by directed edges to establish a complete network topology. Finally, the system automatically performs probability propagation calculations, dynamically displaying the occurrence probability of each scenario state in the node attribute panel, thus obtaining... Figure 18 .

[0113] In this application, the obtained parameters are applied to a Bayesian scenario network model of the accident to construct the final Bayesian network model. By utilizing the forward reasoning capability of the Bayesian network, the probability of different scenario states occurring is calculated. Figure 18 As shown, the final result of forward reasoning is displayed. The probabilities of all scenario state nodes are shown in Table 5, where "T" indicates "occurrence".

[0114] Table 5. Probability of occurrence of scenario state nodes (Scenario 1) ; In the BN model for the scenario of illegal storage of hazardous chemicals, the key scenario nodes include "S" a0 "Dry storage of nitrocellulose", "S" a2 "Information concealment", "S" a4 "Mixed storage of hazardous chemicals", "S" a6 "Warehouse environment out of control, spontaneous combustion," these nodes are arranged vertically in the model, such as S a0 -S a2 -S a4 -S a6 This led to an unexpected development path. Simultaneously, horizontally arranged node pairs, such as S... a0 -S a1 S a2 -S a3The arrangement of numbers represents the expected direction of the accident's evolution, revealing different possible development paths during the accident process.

[0115] Analysis of the BN model shows that, "S a0 "Dry storage of nitrocellulose" was the initial scenario, with an occurrence probability of 51%, the highest among all scenarios. This reflects the company's neglect of hazardous chemical management and the dereliction of duty by supervisors. Meanwhile, "S..." a2 "Information concealment", "S" a4 "Mixed storage of hazardous chemicals" and "S" a6 The probabilities of "out-of-control storage environment" are 25%, 27%, and 25%, respectively, indicating that although hazardous chemical storage can develop positively under proper management when nitrocellulose is stored dry, this significantly increases the probability of scenario state S. a3、 S a5、 S a7 The probabilities of occurrence, and the probabilities of death in the accident, are 75%, 73%, and 78%, respectively. Furthermore, in the negative scenario path, from S... a2 The probability of the initial scenario occurring decreased significantly, indicating that "H" a1 Information concealment, E a1 "Label missing" and "M" a1 The scenarios of "rational planning and strengthened supervision" contribute to the positive development of accidents. In the case of illegal storage of hazardous chemicals, reasonable measures and management can effectively and promptly prevent fires. Through case analysis, the rationality and feasibility of the accident deduction model are verified, demonstrating its effectiveness in predicting and understanding such accidents.

[0116] Scenario 2: Initial Stage of Fire b .

[0117] Using the method for determining parameters in Scenario 1, the prior probability table of each root node in Scenario 2 was obtained, as shown in Table 6.

[0118] Table 6. Results of Expert Evaluation (Scenario 2) ; The software inputs probabilities for each node and outputs the results. For example... Figure 19 As shown.

[0119] Analysis of the BN model shows that, "S b0 "Partial spontaneous combustion" is used as the initial scenario, with a probability of occurrence reaching 60%. After H... b0 E b0 M b0 Under the influence of "S" b2The probability of "surrounding nitrocellulose combustion" increased to 47%, indicating that although localized spontaneous combustion of nitrocellulose can reduce the likelihood of surrounding nitrocellulose combustion with proper management measures. Meanwhile, "S" b6 The probability of a "large-scale fire" occurring is 34%, however, its next undesirable scenario "S" b8 The probability of reaching the "explosive limit" is as high as 61%, indicating that when a large-scale fire occurs, it is highly likely to explode due to the influence of the surrounding environment. This is consistent with the actual situation of the accident.

[0120] Scenario 3: Explosion and Secondary Disasters c .

[0121] Similarly, the prior probability table for each root node in scenario 3 was obtained, as shown in Table 7.

[0122] Table 7 Results of Expert Evaluation (Scenario 3) ; The software inputs probabilities for each node and outputs the results. For example... Figure 20 As shown.

[0123] Analysis of the BN model shows that this stage is characterized by "E" c0 There are several containers containing oxidizers and flammable solids in the surrounding area. c0 The "first explosion" is the core high-risk node, with an occurrence probability of 92% and 97% respectively, driving "S". c2 The fire spread (82%) and "S" c2 The second explosion (82%). The critical path is E. c0 →S c0 →S c2 →S c4 Despite "H c0 "Mobilizing multiple professional rescue forces" and "M" c0 The "activation of the emergency response plan" demonstrated a high level of capability, but the risk of explosion remained high, indicating shortcomings in the efficiency of plan implementation or the allocation of fire-fighting resources. Compared to the previous two stages, the probability of an accident increases significantly when hazardous chemicals reach their explosive limits after multiple stages, further highlighting the importance of timely and effective emergency measures and scenario design.

[0124] Through Bayesian network analysis of each scenario path, the relationships and influence levels among the nodes were discovered. Considering three dimensions—human behavior, disaster-prone environment, and emergency management—and the scenarios occurring at different stages of the accident, the evolution paths and probability of accident occurrence at each stage were analyzed in detail. Next, a Bayesian network graph was constructed based on a multi-scenario, multi-dimensional inference network model, and analyzed in conjunction with actual accident scenarios. For example... Figure 21As shown in the figure, a critical scenario state change path S is formed in the hazardous chemical storage scenario. a0 →S a2 →S a4 →S a6 The probability of scenario S is relatively low compared to the latter two scenarios. As the accident evolves, factors such as human error and emergency response measures influence the outcome, and timely and effective measures can prevent escalation. This aligns with the actual situation of the accident, which was caused by the company's illegal storage of ammonium nitrate, coupled with serious overloading and storage, illegal mixed storage, unpacking and handling violations, failure to register and file major hazard sources as required, and a severe lack of safety production education and training. These factors led to the escalation of the accident. Similarly, in the initial fire scenario, S... b0 →S b2 →S b4 →S b6 →S b8 The path shows a trend of gradually increasing scenario probability, eventually leading to large-scale combustion within the area, corresponding to actual accident scenarios; in the explosion and secondary disaster scenario, from S... c0 →S c2 →S c4 The path shows that the initial scene state S c0 The probability of this occurring is relatively high because the node is influenced by its parent node S. b8 The probability value has a significant impact, while the secondary accident scenario S c2 The probability of this happening is relatively high, which is consistent with the actual situation and verifies the rationality of the scenario simulation.

[0125] In scenario simulation analysis, the probability of each node's occurrence changes dynamically over time. By adjusting the probability values ​​of key nodes in each scenario and continuously updating the posterior probability values ​​of the nodes, optimal decision-making and theoretical basis can be provided for different stages of an accident, from its precursors to its occurrence and then to emergency rescue. Given the different development directions of each scenario, several extreme states of the prior probabilities of the nodes are established to observe the changing patterns of the remaining node probabilities. Considering that human behavior is the main factor causing accidents, the impact of the probability changes of key factor human behavior on the accident simulation process is studied in detail. It should be noted that here, "S" is considered... a0 The scenario of "dry storage of nitrocellulose" has already occurred, therefore this analysis only applies to "S". a2 "-"S a6 "Situation and state."

[0126] By analyzing "H" a1 Strict control over "data anomalies" ensures a 100% probability of occurrence, significantly altering the course of incidents, such as... Figure 22 As shown, this control measure has a significant impact on critical probabilities: "S"a2 The probability of "information concealment" decreased from 25% to 8%. a4 The probability of "mixed storage of hazardous chemicals" decreased from 27% to 24%. These changes indicate that the risk of accidents has decreased, and the situation is more likely to develop in a positive direction. Meanwhile, "S" a3 The probability of "accident fatality" increased from 75% to 85%, an increase of 10%, indicating that the accident trajectory is shifting in a more favorable direction. Therefore, the occurrence of data anomalies plays a crucial role in the accident risk at this stage.

[0127] When "H" a2 When the probability of "reporting and handling anomalies" increases to 100%, the analysis results are as follows: Figure 23 As shown. Initially, without being able to determine whether to report and handle the anomaly, S a4 The probability is 27%, but as the anomaly is handled, S a4 The probability dropped to 19%, indicating a significant reduction in risk. a5 The probability of "accidents resulting in death" has risen to 76%, an increase of 3% from the initial state. This change indicates that the accident is developing towards a more favorable outcome. Therefore, timely reporting and handling of anomalies are key measures to prevent the accident from escalating.

[0128] like Figure 24 As shown, "H" a3 The probability of a "fire alarm" has decreased to 0%, resulting in a substantial change in the accident dynamics. This adjustment has led to the "S" a6 The probability of spontaneous combustion due to uncontrolled storage environment increases significantly to 49%, thus significantly increasing the risk of escalation of the accident. Meanwhile, "S" a7 The probability of "fatal accident" decreased to 61%, a 17% reduction from the initial state. Therefore, timely fire alarm is key to guiding an accident in a more favorable direction and improving the overall safety level.

[0129] This application uses the BN scenario model to examine the impact of various human intervention measures on the evolution of accidents. The analysis shows that regular inspections, timely reporting of abnormalities, and timely alarms are key to preventing and controlling the development of major chemical accidents. It is necessary to strengthen safety education for supervisors and eliminate hidden dangers in a timely manner, thereby reducing the probability and severity of accidents.

[0130] Step 4: Emergency plan decision evaluation and analysis.

[0131] Addressing the typical multi-objective decision-making problem of emergency response to major accidents, this application constructs an emergency plan decision-making evaluation framework based on the core hidden dangers exposed by case accidents (including typical problems such as illegal storage of nitrocellulose, overloaded operation, excessive storage of hazardous materials, illegal mixed storage of materials, and failure to register major hazard sources). This framework, based on three dimensions—personnel behavior, characteristics of the disaster-bearing body, and emergency management effectiveness—constitutes 18 specific evaluation indicators, forming a hierarchical evaluation system. To optimize the decision-making process, a rapid expert evaluation mechanism is designed. To improve the decision-making efficiency for complex accidents, five differentiated emergency plan decision-making schemes are developed. Each scheme shows significant differences in its focus, resource allocation, and decision-making sequence. Scheme 1 emphasizes improving emergency effectiveness through the optimized allocation of rescue materials. Scheme 2 emphasizes prioritizing the protection of human life. Scheme 3 considers the safe transfer of equipment and materials after the accident. Scheme 4 focuses on quickly identifying the cause of the accident and controlling its spread. Scheme 5 focuses on post-accident management. A binary evaluation method is used for scheme selection, with experts independently judging (a "1" for qualified and a "0" for unqualified). The sum of the evaluation scores is shown in Table 8 below.

[0132] Table 8 Scoring Table for Decision-Making Schemes ; Calculate the entropy and weights of the indicators: Obtain the initial matrix X based on the table content: ;

[0133] Calculate the normalized matrix Y: ;

[0134] The entropy value of each indicator attribute is calculated using a formula, taking the deployment of rescue equipment as an example. .

[0135] The degree of variability of the index can be obtained from the entropy value. Taking the deployment of rescue equipment as an example, Then, the weights are calculated according to Equation 4-4. Taking the deployment of rescue equipment as an example, Similarly, the weights of other indicators can be obtained. See Table 9.

[0136] Table 9 Scoring Table for Decision-Making Schemes ; Multiply the weights by the standardization matrix to construct the weighted standardization matrix V: ;

[0137] Determine attribute closeness: based on the formula and Find the ideal positive and negative solutions for the index: ;

[0138] According to the formula Japanese style Find the Euclidean distance: ;

[0139] The Euclidean distance is calculated based on the formula. Obtain the attribute similarity of the indicator: ;

[0140] Comparative analysis of five emergency response decision-making schemes revealed that schemes P4 and P1 achieved the most ideal comprehensive evaluation results, with scheme P4 showing the best performance with a proximity score of 0.533. Further analysis showed that the advantages of scheme P4 mainly lie in its focus on rapidly identifying the causes of accidents and controlling their spread, while scheme P1 focuses on improving emergency response effectiveness through the optimized allocation of rescue resources. The successful implementation of these two schemes verifies that the evaluation method proposed in this study can effectively identify the optimal decision-making scheme. Its scientific validity and practicality are mainly reflected in: firstly, the establishment of a systematic scheme comparison mechanism; secondly, the realization of quantitative evaluation of response effectiveness; and thirdly, the accurate identification of key success factors, thus providing a reliable theoretical basis and technical support for emergency response decision-making.

[0141] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents, characterized in that, The method includes the following steps: S1. Constructing multi-dimensional scenarios for major chemical safety accidents: Based on the public safety triangle model theory, the SHEM analysis framework is introduced to deconstruct accident scenarios into four core units: scenario state (S), human behavior (H), disaster-bearing environment (E), and emergency management (M); a three-level scenario expression system of "performance-scenario-knowledge" is constructed, and the scenarios are divided into historical, real-time, and predictive scenarios using time slicing technology; the knowledge element theory is used to formally represent scenario elements, forming a hierarchical knowledge element expression system. S2. Construct a scenario inference model based on Bayesian networks and fuzzy mathematics: Analyze the four evolutionary mechanisms of accident propagation, derivation, transformation, and coupling, and clarify the positive and negative paths of accident evolution; transform the qualitative evaluation of experts into quantitative probability values ​​through fuzzy mathematics, and combine the bidirectional reasoning ability of Bayesian networks to construct a multi-scenario, multi-dimensional inference model to quantitatively predict the probability of occurrence of different scenario states. S3. Emergency Response Plan Decision Evaluation Based on Entropy Weight Method-TOPSIS: Construct a multi-level emergency evaluation index system that includes rescue process, personnel safety, emergency management, and accident site cleanup; calculate the objective weights of the indicators using the entropy weight method, and construct a weighted decision matrix after standardization; use the TOPSIS method to calculate the Euclidean distance and relative closeness between each emergency plan and the positive and negative ideal solutions, and select the optimal emergency response plan decision scheme.

2. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, In the SHEM analysis framework: The scenario state unit S characterizes the dynamic features of an accident at a specific time segment. It reflects the evolution trend of the accident through the state attributes of key elements and provides a spatiotemporal reference for other elements as the main line of system evolution. Human behavior unit H focuses on human error as a key cause. Based on accident case analysis, unsafe human behavior is divided into two categories: direct triggering type and indirect inducing type. The former directly leads to the occurrence of the accident, while the latter indirectly affects the accident process by causing unsafe conditions of equipment defects and other things. Disaster-bearing environmental unit E encompasses various environmental factors affecting the development of an accident, including natural environmental factors such as severe weather and high temperatures, as well as safety management issues such as untimely equipment maintenance and improper material storage; Emergency management unit M systematically examines the various measures taken to control the accident situation and evaluates their effectiveness.

3. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, In the aforementioned three-tiered "representation-scenario-knowledge" system, the representation layer, as the outermost framework, provides a general description of the macroscopic characteristics of an accident through scenario fragments; the scenario layer, as the intermediate structure, covers core elements such as human behavior, disaster-bearing environment, and emergency measures through a set of scenario categories, achieving a systematic decomposition of the representation layer; and the knowledge layer, as the basic expression unit, uses a knowledge metastructure to formally represent scenario elements, providing a modeling foundation for scenario deduction by defining feature parameters such as element categories, attribute states, and interaction relationships.

4. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, The knowledge element representation includes: common knowledge elements of objects, common knowledge elements of attributes, and common knowledge elements of relationships; The knowledge meta-model is as follows: ； In the formula, For common knowledge elements of objects, For a set of concept names, For attribute state set, For a set of relations; For attribute common knowledge elements, To describe attribute characteristics, The dimensions of measurable attributes, This represents a time-varying mapping relationship for attributes. As a common knowledge element of relationships, To characterize relational features, and These are the input and output attribute sets, respectively. This is a mapping relationship; The common knowledge elements of the objects include a set of concept names, a set of attribute states, and a set of relationships; the common knowledge elements of attributes include descriptive attribute features, measurable dimensions, and time-varying mapping relationships; the common knowledge elements of relationships include relationship features, input and output attribute sets, and mapping relationships.

5. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, In the aforementioned accident evolution mechanism, the propagation mechanism is that the initiating disaster factor triggers a secondary disaster chain through energy transfer and material migration; the derivative mechanism is that inappropriate emergency response measures lead to more serious events; the transformation mechanism is that the accident transforms into different types of emergencies through energy conversion and material changes; and the coupling mechanism is that multiple disaster-causing factors interact to exacerbate the harm of the accident.

6. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, The fuzzy mathematical processing includes: using triangular fuzzy numbers or trapezoidal fuzzy numbers to represent expert qualitative opinions, defuzzifying them using the integral value method, and converting aggregated fuzzy numbers into deterministic probability values ​​for assigning probability values ​​to Bayesian network nodes. The membership function of the triangular fuzzy number A(a, b, c) is: ； The trapezoidal fuzzy number A(a, b, c, d) is represented by the membership function as follows: ； After obtaining the desired fuzzy numerical representation, it needs to be transformed into precise probability values ​​through defuzzification. This transformation is achieved using the integral value method, which is based on the cut operation of the membership function, as shown in the following formula: ； In the formula, This represents the fuzzy number. Optimism coefficient , is the integral value of the left and right membership functions of the fuzzy number; ； In the formula, Let A be the upper and lower bounds of the fuzzy number A cut set. .

7. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, The emergency evaluation index system has a three-dimensional structure. The first-level indicators include the rescue process, personnel safety, emergency management, and accident site cleanup; the second-level indicators include material input, personnel input, affected personnel, and evacuated personnel; and the third-level indicators are specific quantitative evaluation parameters.

8. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, The entropy weight method calculation includes: establishing an evaluation matrix and performing standardization processing, calculating the information entropy value of each indicator, obtaining the coefficient of variation of the indicators through the information entropy value, and determining the objective weight of each indicator based on the coefficient of variation. The formula for calculating information entropy is: ； In the formula, This represents the probability value for decision-making in the corresponding emergency response plan. For the first The value of an emergency response plan decision. ， ; This indicates the emergency response decision-making group; The standardization process is as follows: First, for the j-th evaluation index, calculate its information entropy value. : ； In the formula, k is a standardization constant. This represents the standardized value of the i-th scheme under the j-th indicator; Secondly, based on the entropy calculation results, the coefficient of variation of each indicator is calculated. : ； Finally, the weights of each indicator are calculated based on the coefficient of variation. ; 。 9. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, The TOPSIS evaluation method includes: constructing a weighted decision matrix based on the standardized matrix and index weights, determining the positive and negative ideal solutions, calculating the weighted Euclidean distance between each scheme and the positive and negative ideal solutions, and ranking and optimizing the schemes based on their relative proximity. Positive Ideal Solution Sure: ； In the formula, ; Negative ideal solution Sure: ； In the formula, ; Euclidean distance from solution i to the ideal solution : ； Euclidean distance from solution i to the negative ideal solution : ； The closeness calculate: ； The larger the value, the closer the solution i is to the ideal optimal state.

10. The method for constructing multi-dimensional scenarios and making emergency response decisions for chemical safety accidents according to claim 1, characterized in that, The multiple scenarios include: illegal storage of hazardous chemicals, initial stage of fire, and explosion and secondary disaster scenarios. Each scenario is dynamically deduced through the causal relationship of key state nodes. The key state nodes of the previous scenario directly affect the evolution direction and probability of occurrence of the next scenario.

Images