An explosive risk dynamic prediction method and system based on ISM risk coefficient and entropy weight composite empowerment

Abstract

The application discloses an explosion risk dynamic prediction method and system based on ISM risk coefficient and entropy weight compound empowerment, and is applied to the technical field of gas pipeline explosion risk prediction. Firstly, the identification of the hazard source is realized through the fault tree analysis (FTA), the hierarchical transmission relationship among factors is clarified through the interpretative structural modeling (ISM), and the risk coefficient of each factor is quantified based on the reachable matrix, so as to provide a structural basis and quantitative weight correction for system dynamics (SD) model variable aggregation; the entropy theory and the dissipative structure theory are introduced into the SD model, and the entropy weight method is combined with the ISM risk coefficient empowerment; simulation and sensitivity analysis are carried out through the SD simulation software. When the explosion accident prediction model is constructed, the three elements of explosion, i.e., gas leakage, ignition source and limited space, are completely considered, wherein the limited space is introduced into the explosion accident equation as a correction factor, so that the theoretical integrity is ensured, and the simplicity of the model structure is maintained.

Description

Technical Field

[0001] This invention relates to the field of gas pipeline explosion risk prediction technology, and more specifically to a dynamic prediction method and system for explosion risk based on ISM hazard coefficient and entropy weighting. Background Technology

[0002] As a core energy infrastructure in cities, urban gas pipelines are characterized by their massive scale, long service life, and complex operating environments due to accelerated urbanization. Problems such as pipeline corrosion, damage from third-party construction, and negligence in operation and maintenance management are becoming increasingly prominent, leading to frequent fires and explosions caused by gas leaks, which seriously threaten urban public safety. Therefore, achieving accurate identification, dynamic prediction, and quantitative assessment of explosion hazards has become a core issue that the industry urgently needs to address.

[0003] Currently, single or simple combinations of methods such as FTA (Fault Tree Analysis), ISM (Interpretive Structural Model), SD (System Dynamics), and entropy weighting have been preliminarily applied to gas pipeline risk analysis. However, existing technologies still have the following core shortcomings: 1. Hazard identification remains at a qualitative level, lacking quantitative assessment methods. Although existing methods can construct hierarchical relationships between factors through FTA or ISM, they cannot quantify the actual influence and root cause of each factor in the risk transmission network. For example, in ISM applications, only a multi-level hierarchical structure is generated, without calculating the out-degree, in-degree, and hierarchical weights of factors based on the reachability matrix. This makes it impossible to form quantitative indicators such as hazard coefficients, leading to risk classification relying on expert experience and making it difficult to accurately locate the core causes; 2. SD model construction involves arbitrary variable aggregation and highly subjective weight setting: Existing technologies often use equal weights or expert scoring methods when aggregating core SD variables. They neither utilize the ISM hierarchical structure to guide variable classification nor combine factor hazard coefficients for weighted correction. Meanwhile, the lack of entropy theory to explain the system's evolution mechanism makes it impossible to establish a quantitative correlation between positive and negative entropy flows, resulting in insufficient theoretical support for the model and significant simulation bias; 3. Risk assessment is mainly based on static status evaluation and lacks dynamic trend prediction capabilities: Existing technologies are mostly based on historical data for static scoring, which cannot simulate the dynamic evolution of risks. Neither an entropy theory-based SD stock flow map has been constructed, nor has a multi-scenario simulation comparison been conducted. It is impossible to quantify the risk reduction effects of different management intervention measures, making it difficult to support forward-looking decision-making; 4. Lack of a fully integrated technical system: Each analysis link is independent of each other. An integrated system covering data acquisition, ISM quantitative analysis (including automatic calculation of risk coefficients), SD modeling, parameter calibration (including composite weighting), simulation calculation, and visualization early warning has not been constructed. Decision support lacks quantitative basis, and the overall engineering practicality is poor.

[0004] In summary, existing technologies are no longer sufficient to meet the comprehensive requirements of urban gas pipeline safety management for risks that are knowable, assessable, controllable, and predictable. Therefore, how to provide a dynamic prediction method and system for explosion risks based on the ISM hazard coefficient and entropy weighting is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention provides a dynamic prediction method and system for explosion risk based on ISM hazard coefficient and entropy weight, which enables early prediction, accurate assessment and scientific management of explosion risk in urban gas pipelines, and provides core technical support for the safety management of gas pipeline networks throughout their entire life cycle.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A dynamic prediction method for explosion risk based on ISM hazard coefficient and entropy weighting includes the following steps: S1. The fault tree analysis method is used to systematically identify and analyze the hazards of urban gas fire and explosion accidents, and a fault tree diagram with urban gas fire and explosion accidents as the top-level event is established, and the basic events at the bottom level of the fault tree diagram are used as system elements. S2. Construct an explanatory structural model for urban gas fires and explosions and calculate the risk coefficients of factors. Use an iterative algorithm to hierarchically divide the system elements and generate a multi-level hierarchical structural model that reveals the risk transmission path. S3. Based on the multi-level hierarchical structure model and the risk coefficient, the system elements are aggregated into the core variables of system dynamics, and a causal feedback loop describing the positive and negative feedback relationship between the variables is established. S4. Transform the causal feedback loop into a system dynamics stock flow diagram based on entropy theory, and define state variables, rate variables, and auxiliary variables; S5. The objective weights of the sub-factors of the core variables of system dynamics are calculated using the entropy weight method. The objective weights are corrected using the risk coefficient as the weight correction factor. The core mathematical equation of the stock flow diagram of system dynamics is constructed by the multiple regression method. S6. Based on core mathematical equations, simulation prediction is performed. The effectiveness of control measures is quantitatively evaluated through multi-scenario comparison. Sensitivity analysis is combined to identify weak links in risk control and output dynamic prediction results and control decision recommendations.

[0007] Optionally, the fault tree diagram in S1 includes top-level events, intermediate sub-events, and bottom-level basic events; intermediate sub-events are divided into three levels: first-level sub-intermediate events are the core triggering conditions of top-level events, second-level sub-intermediate events are the subdivision of triggering paths, and third-level sub-intermediate events are the cause classification of triggering paths.

[0008] Optionally, the risk coefficients for factors in S2 are calculated as follows: An adjacency matrix is ​​constructed based on the direct influence relationships between system elements in the fault tree diagram. A reachability matrix is ​​then generated through Boolean algebra operations to reveal the direct and indirect transmission relationships between system elements. Based on the reachability matrix, factor hazard coefficients are calculated for each factor. The out-degree is defined as the number of elements with a row value of 1 in the reachability matrix, and the in-degree is defined as the number of elements with a column value of 1 in the reachability matrix, taking into account various factors. Calculate the hazard factor based on the out-degree, in-degree, and position in the hierarchical structure. : ; In the formula, , , These are the weighting coefficients. As factors The degree of exit, For the maximum out-degree of all factors, As factors The level at which it is located The maximum number of levels, As factors in-degree, This represents the maximum in-degree of all factors.

[0009] Optionally, the adjacency matrix A is a 45×45 square matrix, with matrix elements... , Indicator Factors Factors There is a direct impact.

[0010] Optionally, in S3, the system elements are aggregated into the core variables of system dynamics as follows: Based on the hierarchical classification of the multi-level hierarchical structure model, basic events are categorized into material inherent risks, corrosion and environmental risks, anti-corrosion system risks, construction defect risks, management and maintenance risks, external damage risks, equipment failure risks, human operation risks, ignition source risks, and protective facility risks. During the aggregation process, for any aggregation variable, the weights of each sub-factor are determined after normalization based on the corresponding factor hazard coefficient.

[0011] Optionally, entropy theory and dissipative structure theory are introduced in S4 to determine the total entropy change of the system. for: ; In the formula, The entropy generated by irreversible processes within the system. The system is the entropy flow exchanged between itself and the outside world; the sum of the inflow rates of each state variable constitutes the positive entropy source of the system, driving the system to evolve towards a high-risk disordered state; the sum of the outflow rates of each state variable constitutes the negative entropy flow introduced by the system through safety management activities, driving the system to recover to a safe and orderly state; the risk state variable of management and maintenance serves as a bottleneck variable that limits the efficiency of introducing negative entropy flow.

[0012] Optionally, in S5, the entropy weight method is used to calculate the objective weights of the sub-factors of the core variables of system dynamics, and the objective weights are corrected using the hazard coefficient as a weight correction factor. Specifically: by Using a sample of time points, Using factors as indicators, a standardized decision matrix is ​​constructed. Calculate the first Under the first factor The proportion matrix of each sample : ; Calculate the first Entropy of each factor : ; Calculate the coefficient of difference Determine the basic weights of the entropy weight method ISM risk coefficients are introduced for composite weighting:

[0013] In the formula, For composite weights, , As factors , The risk coefficient of the factors.

[0014] Optionally, the core mathematical equations of the system dynamics stock-flow diagram constructed in S5 include variable aggregation equations, management efficiency equations, state variable integral equations, and rate variable equations; among them, the state variable integral equation for an explosion accident is: ; ; In the formula, for The comprehensive index of explosion risk at any given moment. This is the comprehensive index of explosion hazard at the initial moment. To mitigate the risk of equipment failure, Indicates time, For the risk of ignition source, To protect against facility risks, The spatial constraint coefficient characterizes the ventilation conditions and the degree of spatial airtightness of the environment in which the pipeline is located. This represents the leakage-ignition-space coupling function. For leakage-ignition coupling coefficient, To aggravate the failure of protection, To control the intensity of intervention, This is the safety control intervention coefficient.

[0015] Optionally, the sensitivity analysis in S6 specifically includes: Sensitivity analysis was conducted on all input parameters and core variables using the dimensionless sensitivity coefficient method to identify key variables that significantly impact the risk of gas pipeline explosions and pinpoint weak points in risk control. The formula for calculating the dimensionless sensitivity coefficient method is as follows: ; In the formula, For the first Variables Risk of explosion Sensitivity coefficient; Partial derivative; the absolute value of the calculated sensitivity coefficient Variables with a value greater than 1 are identified as highly sensitive variables and are considered as weak links in risk control and core control targets.

[0016] A dynamic explosion risk prediction system based on ISM hazard coefficient and entropy weighting, applying the aforementioned dynamic explosion risk prediction method based on ISM hazard coefficient and entropy weighting, includes: A multi-source data acquisition module is used to collect historical and real-time data from the gas pipeline network; The data preprocessing module is connected to the multi-source data acquisition module to clean, format, and dimensionless process the data. The ISM model building module, connected to the data preprocessing module, is used to automatically build adjacency and reachability matrices, calculate factor risk coefficients, and generate multi-level hierarchical structure models. The SD model building module, connected to the ISM model building module, is based on a multi-level hierarchical structure model and factor risk coefficients. It assists in constructing the core variables of system dynamics, causal feedback loops, and stock flow diagrams according to preset aggregation rules, and defines state variables, rate variables, and auxiliary variables. The parameter calibration and learning module is connected to the data preprocessing module and the SD model construction module. It uses the entropy weight method to calculate objective weights and combines them with factor risk coefficients to perform entropy weight-ISM composite weighting. The model equation parameters are calibrated through multiple regression analysis. The simulation calculation engine module, connected to the SD model building module, is used to load the complete system dynamics model, perform numerical integration calculations, and conduct simulations of baseline scenarios and multi-strategy scenarios. The visualization and early warning decision-making module is connected to the simulation calculation engine module. It is used to receive simulation results, display the explosion hazard trend in the form of charts and GIS heat maps, issue early warnings when the results exceed preset thresholds, and push control decision suggestions based on sensitivity analysis results.

[0017] As can be seen from the above technical solution, compared with the prior art, the present invention provides a dynamic prediction method and system for explosion risk based on ISM hazard coefficient and entropy weighting, which has the following beneficial effects: 1. This invention elevates traditional qualitative structural analysis to quantitative assessment by calculating the risk coefficient of factors based on the ISM reachability matrix, accurately quantifying the influence and root cause of each factor in the risk transmission network, and solving the problem of ambiguous root cause localization in existing technologies; 2. This invention uses the ISM hierarchical structure and hazard coefficients to guide the aggregation of variables in the SD model, and combines the entropy weight method for objective weighting and coefficient correction, forming a scientific composite weighting mechanism that avoids subjective bias and makes the model structure more scientific and the parameters more accurate. 3. This invention introduces entropy theory to explain the system evolution mechanism, constructs a quantitative correlation between positive / negative entropy flow and risk accumulation / decay, realizes dynamic trend prediction of risk and multi-strategy simulation comparison, and promotes the transformation of safety management from experience-based decision-making to vectorized decision-making; 4. By constructing an integrated system, this invention enables fully automated processing from data collection to decision output, thereby improving the ability to implement engineering projects. Attached Figure Description

[0018] To more clearly illustrate the technical solutions 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 embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a flowchart of the dynamic prediction method for explosion risk of the present invention; Figure 2 This is a schematic diagram of the ISM multi-level hierarchical structure model of the present invention; Figure 3 This is a schematic diagram of the causal feedback loop of the present invention; Figure 4 This is the system dynamics stock flow diagram of the present invention; Figure 5This is a fault tree diagram of a gas pipeline explosion accident according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The main body for executing this technical solution is computer equipment (including servers, terminals, simulation workstations, etc.). All data processing, matrix operations, model building, simulation deduction, and result analysis steps are completed by computer equipment through preset programs / professional software (MATLAB, Vensim, etc.), ensuring the accuracy, efficiency, and repeatability of the operation.

[0021] This invention discloses a dynamic prediction method for explosion risk based on a composite weighting of ISM hazard coefficient and entropy weight, such as... Figure 1 As shown, it includes the following steps: S1. The fault tree analysis method is used to systematically identify and analyze the hazards of urban gas fire and explosion accidents, and a fault tree diagram with urban gas fire and explosion accidents as the top-level event is established, and the basic events at the bottom level of the fault tree diagram are used as system elements. S2. Construct an explanatory structural model for urban gas fires and explosions and calculate the risk coefficients of factors. Use an iterative algorithm to hierarchically divide the system elements and generate a multi-level hierarchical structural model that reveals the risk transmission path. S3. Based on the multi-level hierarchical structure model and the risk coefficient, the system elements are aggregated into the core variables of system dynamics, and a causal feedback loop describing the positive and negative feedback relationship between the variables is established. S4. Transform the causal feedback loop into a system dynamics stock flow diagram based on entropy theory, and define state variables, rate variables, and auxiliary variables; S5. The objective weights of the sub-factors of the core variables of system dynamics are calculated using the entropy weight method. The objective weights are corrected using the risk coefficient as the weight correction factor. The core mathematical equation of the stock flow diagram of system dynamics is constructed by the multiple regression method. S6. Based on core mathematical equations, simulation prediction is performed. The effectiveness of control measures is quantitatively evaluated through multi-scenario comparison. Sensitivity analysis is combined to identify weak links in risk control and output dynamic prediction results and control decision recommendations.

[0022] Furthermore, such as Figure 5As shown, the fault tree diagram in S1 includes top-level events, intermediate sub-events, and bottom-level basic events; intermediate sub-events are divided into three levels: first-level sub-intermediate events are the core triggering conditions of top-level events, second-level sub-intermediate events are the subdivision of triggering paths, and third-level sub-intermediate events are the cause classification of triggering paths.

[0023] In this embodiment of the invention, the three primary intermediate events are gas leak (A), ignition source (B), and confined space (C). These three are associated with a top-level event via a door, meaning that an explosion can only occur when the conditions of gas leak, ignition source, and confined space are simultaneously met. A confined space refers to an environment where gas accumulation is difficult to diffuse, including indoor environments, underground pipe trenches, sealed valve wells, and areas with dense obstacles. Secondary intermediate events (path subdivision): For example, gas leak (A) is subdivided into two intermediate events: pipeline leak (A1, covering leaks related to the pipeline body and its ancillary facilities) and user-end leak (A2, covering leaks at the end of gas appliances and indoor pipelines). Tertiary intermediate events (cause classification): The secondary intermediate events are further broken down to form a cause classification layer. For example, under pipeline leak (A1), tertiary sub-events such as corrosion (A11), human damage (A12), natural damage (A13), design and construction defects (A14), and material defects (A15) are set to achieve precise subdivision of the cause path. Based on the three-level sub-intermediate events, and strictly following the constructed fault tree, the underlying basic events are extracted from each of the three-level sub-intermediate events to ensure that each event is an indivisible causal unit. The core elements are as follows: Level 4 Basic Events (Underlying Elements): At the level of pipeline leakage-related (29 items), these include poor material corrosion resistance, defects in internal / external anti-corrosion layers, stress concentration, construction not in accordance with requirements, welding defects, third-party construction damage, poor pipeline management, and use of easily damaged components beyond their service life; At the level of user-end leakage-related (5 items), these include gas appliance leaks, loose ball valve cores, user violations, indoor pipeline damage, and improper fault handling; At the level of ignition source-related (10 items), these include unauthorized hot work in hazardous areas, cable short circuits, static sparks, surge arrester failure, explosion-proof device damage, smoking, non-smoking human-caused ignition sources, poor grounding, non-explosion-proof electrical facilities, and unknown ignition sources.

[0024] Fault tree logical association and graphical construction: Logic gate settings: Based on the causal relationship between events, configure "AND gates" and "OR gates" appropriately: "AND gates" are used between the first-level intermediate event and the top event (the gas leak and ignition source must be satisfied simultaneously); "OR gates" are used between the second-level sub-intermediate event and the first-level intermediate event, and between the third-level sub-event and the second-level sub-intermediate event, according to the causal logic (the occurrence of any sub-event can trigger the previous level event). Graphical drawing: The fault tree is intelligently drawn using the "stem-branch-leaf linking method". First, the "stem" consisting of the top event and the first-level intermediate events is drawn. Then, second-level and third-level sub-events and basic events are added layer by layer. Each event is configured with a unique identifier and pointer (pointing to the parent event, child event, and sibling event) to ensure that the logical relationship is clear and traceable. Finally, a complete graphical file of the fault tree for gas fire and explosion accidents is formed.

[0025] In this embodiment of the invention, 45 disordered factors are transformed into a structural model with a clear hierarchical transmission relationship, and each factor is assigned a quantified hazard index. The determination of the element correlation relationship specifically involves: forming a cross-disciplinary expert team of gas engineering, safety engineering, and operation and maintenance management; combining fault tree logic, accident statistics, and hazard simulation results; and using a "double-blind verification method" to determine the direct influence relationship between any two elements: if... Directly led to If the risk of disaster occurs or is aggravated, it is determined that there is a "direct impact" (denoted as 1); otherwise, it is determined that there is no direct impact (denoted as 0), forming an element correlation matrix table.

[0026] Furthermore, the adjacency matrix A is a 45×45 square matrix, with matrix elements... , Indicator Factors Factors There is a direct impact.

[0027] In this embodiment of the invention, matrix digitization and format standardization are accomplished using MATLAB software. The adjacency matrix A is shown below: A=[ # - Poor corrosion resistance of the material [0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0], # - Stress concentration exists [0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0], # - Internal anti-corrosion layer peeling off [0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0], # -Aging of the internal anti-corrosion layer [0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0], ...,#omit to

[0028] # - Smoking

[0029] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1], # - Static spark [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1], # -Unknown source of fire [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1], # - Explosion [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] ].

[0030] The reachability matrix is ​​calculated as follows: Based on the adjacency matrix A, an identity matrix I (with 1s on the diagonal and 0s elsewhere) is introduced, and matrix exponentiation is performed. (n is a positive integer) until the matrix result no longer changes, thus obtaining the reachable matrix M: Until it is established; Reachability matrix A matrix (elements are either 1 or 0) characterizing the direct or indirect influence relationships between various hazard elements in the system; adjacency matrix. A matrix (elements are either 1 or 0) characterizing the direct influence relationships between various hazard elements in the system; identity matrix. A matrix with 1s on the diagonal and 0s elsewhere is used to represent the influence of elements on itself; n is a positive integer representing the number of times the matrix is ​​exponentiated, and the operation stops when the matrix result no longer changes.

[0031] Elements in reachable matrix M , express It can affect through direct or indirect paths , This indicates that there is no influence relationship, and the calculation process is implemented using MATLAB software to ensure calculation accuracy.

[0032] The reachability matrix M is shown below: M=[ # - Poor corrosion resistance of the material [1,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1], # - Stress concentration exists [1,1,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1], # - Internal anti-corrosion layer peeling off [1,0,1,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1], # -Aging of the internal anti-corrosion layer [1,0,0,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1], ...,#omit to

[0033] # - Smoking

[0034] [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,1], # - Static spark [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,1], # -Unknown source of fire [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1], # - Explosion [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1] ]; The hierarchical division of system elements using an iterative algorithm is specifically as follows: For each element in the reachability matrix M Calculate the three sets: the reachable set R( ): The set of all elements that can be directly / indirectly affected (by M) i The elements corresponding to the columns with a value of 1 in the row constitute the antecedent set A ( ): Can directly / indirectly affect The set of all elements (by M) i The elements corresponding to rows with a value of 1 in the column constitute the intersection R( )∩A( ): Characterization Within the core scope of the system.

[0035] Hierarchical Iterative Division: Following the principle that higher-level elements are only affected and not lower-level elements, the element hierarchy is iteratively divided. L1 (Surface-level direct cause, 16 factors): Satisfies R( )=R( )∩A( The elements that directly trigger gas leaks or ignition sources, such as gas appliance leaks (S... 30 ), unauthorized hot work in dangerous areas (S 38 ), cable short circuit (S) 41 Elements such as these are only affected by mid-level factors and do not affect lower-level factors. L2 (Intermediate indirect causes, 13 factors): After removing L1 factors, the reachability set, antecedent set, and intersection of the remaining factors are recalculated to satisfy the hierarchical determination logic, covering poor quality of the external anti-corrosion layer. Internal stress corrosion () Stealing air ( Indirect influencing factors, such as L3, inherit the fundamental influence of L3 and transmit it to L1; L3 (Deep Root Causes, 15 Factors): The final set of remaining elements, whose reachable set covers multiple elements of L2. The antecedent set only contains its own related fundamental factors, which are the basic and root causes affecting the entire disaster-causing system, including poor material corrosion resistance ( Inadequate anti-corrosion measures Poor pipeline management (e.g., L2 indirectly affects L1 by influencing L2).

[0036] The multi-level hierarchical structure model fully considers the three elements of an explosion. Gas leakage is comprehensively reflected by factors such as equipment failure risk and external damage risk; the ignition source is reflected by the ignition source risk; and confined space conditions are introduced as a correction factor into subsequent explosion accident equations, rather than being set as independent aggregation variables. This approach ensures both the theoretical integrity of the three elements of an explosion and maintains the simplicity and operability of the SD model structure. The hierarchical division results are shown in Table 1. Table 1. Hierarchical Division Results

[0037] like Figure 2 As shown, based on the hierarchical division results, a three-level hierarchical structure model diagram of the ISM hazard source for gas fire and explosion accidents is constructed using a top-down approach and core path labeling. The diagram clearly presents the transmission path of "L3 (poor pipeline management) → L2 (intermediate indirect factors) → L1 (surface direct factors) → top event (gas fire and explosion accident)". The code, name, hazard coefficient and core correlation of each level element are labeled, highlighting the root cause impact of L3 on the entire system.

[0038] Furthermore, the specific calculation method for the risk coefficient of factors in S2 is as follows: An adjacency matrix is ​​constructed based on the direct influence relationships between system elements in the fault tree diagram. A reachability matrix is ​​then generated through Boolean algebra operations to reveal the direct and indirect transmission relationships between system elements. Based on the reachability matrix, factor hazard coefficients are calculated for each factor. The out-degree is defined as the number of elements with a row value of 1 in the reachability matrix, and the in-degree is defined as the number of elements with a column value of 1 in the reachability matrix, taking into account various factors. Calculate the hazard factor based on the out-degree, in-degree, and position in the hierarchical structure. : ; In the formula, , , As the weighting coefficient, this embodiment takes... =0.4, =0.4, =0.2 to highlight the dominant role of influence and root causes in hazard assessment. As factors The degree of exit, For the maximum out-degree of all factors, As factors The level at which it is located The maximum number of levels is 3 in this embodiment. As factors in-degree, This represents the maximum in-degree of all factors.

[0039] In this embodiment of the invention, the out-degree and in-degree calculations of the factors are specifically as follows: Let the reachability matrix be M=[ 45×45, diagonal elements =1, Factors The output (influence) is: (Subtract 1 from the row sum to exclude its own influence); The in-degree (dependency) of a factor is: (The sum of all columns, excluding themselves).

[0040] The calculated 1×45 factor hazard coefficient vector can be used to visually distinguish the hazard level of each factor: Factors with deep-rooted and widespread origins (such as poor pipeline management and corrosion) have a higher risk factor; factors with superficial and small-scale impact (such as smoking) have a lower risk factor. This coefficient vector provides a quantitative basis for subsequent key factor identification, risk ranking, and control strategy formulation. Based on the reachability matrix M, examples of the risk coefficients of each factor are shown in Table 2: Table 2 Examples of Risk Factors

[0041] Furthermore, in S3, the system elements are aggregated into the core variables of system dynamics as follows: Based on the hierarchical classification of the multi-level hierarchical structure model, basic events are categorized into material inherent risks, corrosion and environmental risks, anti-corrosion system risks, construction defect risks, management and maintenance risks, external damage risks, equipment failure risks, human operation risks, ignition source risks, and protective facility risks. During the aggregation process, for any aggregation variable, the weights of each sub-factor are determined after normalization based on the corresponding factor hazard coefficient.

[0042] In this embodiment of the invention, through variable aggregation, feedback relationship analysis, and causal chain construction, a system dynamics causal relationship diagram covering the entire process of risk source transmission, coupling, and inhibition is formed. This clearly characterizes the dynamic interaction mechanism between variables at each level, providing a structured qualitative foundation for subsequent SD mathematical model establishment and simulation. Specifically: 1. Definition of variable aggregation and standardization Based on the ISM hierarchical structure and fault tree original factor coding, and employing the logic of "factor classification - role positioning," 45 basic factors were precisely aggregated into 10 categories of core system dynamics variables. The original factors, hierarchical affiliation, SD roles, and variable definitions were clearly defined for each variable, ensuring standardized variable definitions and clear attributes, aligning with the causal logic and model construction requirements of urban gas pipeline explosion accidents. When aggregating auxiliary variables (such as material inherent risk), their expressions were corrected using the factor hazard coefficients calculated by S2. The aggregated variables are shown in Table 3. Table 3 Aggregate Variables

[0043] like Figure 3 As shown, based on the above variables and combined with the core causal logic of urban gas pipeline explosion accidents, namely "deep root cause → mid-level transmission → surface trigger → risk output → control intervention", a core causal chain is constructed ("+" indicates positive correlation, that is, the enhancement of the previous variable will lead to the enhancement of the next variable; "-" indicates negative correlation, that is, the enhancement of the previous variable will lead to the weakening of the next variable). All causal relationships are consistent with the variable definitions and engineering practice, and strictly follow the ISM three-level hierarchical structure and SD model feedback logic.

[0044] In this embodiment of the invention, the state variable defined in S4 is specifically as follows: Variable type definition, identification, and mathematical relationships between variables: State variables (stock): Construction and design defects become explicit (A), material and operation and maintenance failure risk (P), terminal leakage risk (E), natural environment triggering risk (C), ignition source coupling risk (S), and explosion hazard index (Y). Rate variables (flow rate): growth rate (RA, RC, RP, RE, RS) and decay rate (RB); Auxiliary variables: critical conditions for gas explosion, inherent material risk (M), and construction defect risk (D); Constant: Critical conditions for gas explosion; Exogenous parameters: externally given baseline conditions that can be adjusted in simulation (including all coefficients to be calibrated such as k1 and k2).

[0045] The defined state variables are shown in Table 4: Table 4 State Variables

[0046] The system dynamics stock-flow diagram constructed based on the above parameter definitions is as follows: Figure 4 As shown.

[0047] Furthermore, S4 introduces entropy theory and dissipative structure theory to determine the total entropy change of the system. for: ; In the formula, The entropy generated by irreversible processes within the system. The system is the entropy flow exchanged between itself and the outside world; the sum of the inflow rates of each state variable constitutes the positive entropy source of the system, driving the system to evolve towards a high-risk disordered state; the sum of the outflow rates of each state variable constitutes the negative entropy flow introduced by the system through safety management activities, driving the system to recover to a safe and orderly state; the risk state variable of management and maintenance serves as a bottleneck variable that limits the efficiency of introducing negative entropy flow.

[0048] In this embodiment of the invention, positive entropy includes: variable-driven risk accumulation, such as the accumulation of corrosion and environmental risks (C), corrosion prevention system risks (P), and external damage risks (O); and the increase in explosion risk resulting from the coupling of equipment failure risks (E) and ignition source risks (I). These positive entropies drive the system towards a chaotic, disordered, and high-risk state. Negative entropy flows include: risk elimination driven by safety management interventions, mainly reflected in the management improvement rate (RB) and the outflow rate of various state variables; and risk attenuation brought about by maintenance and repair activities, such as reducing corrosion and environmental risks (C) and equipment failure risks (E) through management. These negative entropy flows offset positive entropy, restoring the system to an orderly and safe state.

[0049] When management and maintenance risks (A) increase, the risk elimination effect produced by the same intervention intensity is significantly reduced.

[0050] The S4 algorithm produces a system dynamics model that can be directly input into computer simulation. Its core outputs include: Visual model file: a stock flow diagram that can be run directly in Vensim; List of variable definitions: Complete definitions of all state variables, rate variables, and auxiliary variables; Mathematical relationships: the quantitative relationships between variables, laying the foundation for the parameterization of S5; System evolution explanation framework: a description of risk evolution mechanism based on entropy theory.

[0051] Furthermore, in S5, the entropy weight method is used to calculate the objective weights of the sub-factors of the core variables of system dynamics, and the objective weights are corrected using the hazard coefficient as a weight correction factor. Specifically: by Using a sample of time points, Using factors as indicators, a standardized decision matrix is ​​constructed. Calculate the first Under the first factor The proportion matrix of each sample : ; Calculate the first Entropy of each factor : ; Calculate the difference coefficient Determine the basic weights of the entropy weight method ISM risk coefficients are introduced for composite weighting:

[0052] In the formula, For composite weights, , As factors , The risk coefficient of the factors.

[0053] In this embodiment of the invention, before determining the index weights based on the entropy weight method, historical data of all factors in layers L1, L2, and L3 are collected and dimensionlessized using the extreme value processing method: ; in For the first i The first sample j The original values ​​of each factor The value is the standardized value, and its range is [0,1]. Taking corrosion and environmental risk (C) as an example, its aggregation formula is: C = × ； in The standardized values ​​of the original factors that constitute C; The entropy weight-ISM composite weight is constructed in a similar way to other aggregate variables (such as corrosion prevention system risk (P), equipment failure risk (E), etc.).

[0054] The general integral form for each state variable is: ; In the formula, For any L1 level state variable at time... t The possible values ​​of ; The initial values ​​for the state variables. State variables At any moment Total inflow rate; State variables At any moment Total outflow rate; integration interval [0, t [] represents the cumulative change from the initial time to the current time.

[0055] Furthermore, the core mathematical equations of the system dynamics stock-flow diagram constructed in S5 include variable aggregation equations, management efficiency equations, state variable integral equations, and rate variable equations; among them, the state variable integral equation for an explosion accident is: ; ; In the formula, for The comprehensive index of explosion risk at any given moment. This is the comprehensive index of explosion hazard at the initial moment. To mitigate the risk of equipment failure, Indicates time, For the risk of ignition source, To protect against facility risks, The spatial constraint coefficient characterizes the ventilation conditions and the degree of spatial airtightness of the environment in which the pipeline is located. This represents the leakage-ignition-space coupling function. For leakage-ignition coupling coefficient, To aggravate the failure of protection, To control the intensity of intervention, is the safety control intervention coefficient.

[0056] The equation of explosion accident Y introduces the space constraint coefficient R to characterize the key impact of the confined space on the explosion risk. The detailed definition, physical meaning, and acquisition method of this coefficient are as follows: Definition and value range of the space constraint coefficient R: R = f (ventilation condition, space volume, obstacle density), R ∈ [0, 1]; Physical meaning of R: R characterizes the difficulty of forming an explosive gas environment after gas leakage. The larger the R value, the more enclosed the space, the easier the gas accumulates, and the higher the explosion risk; R = 1: completely enclosed space (such as indoors, underground pipe trenches, closed valve wells, underground utility tunnels, etc.). In such spaces, the leaked gas is extremely easy to accumulate to reach the explosion limit; R = 0: completely open space (such as an open outdoor area). In such spaces, the leaked gas quickly diffuses and dilutes, and an explosive environment cannot be formed; 0 < R < 1: semi-open or confined space, such as pipe trenches under urban roads, semi-enclosed corridors, industrial factory areas with obstacles, etc. The specific value can be determined by pre-calibration or on-site measurement according to engineering parameters such as regional type, ventilation condition, space volume, and obstacle density.

[0057] Acquisition method of R: In practical applications, the R value can be obtained through one of the following methods: Preset by regional type: Divide the pipe network into different regional types (such as municipal roads, residential courtyards, indoors, valve wells, etc.), and preset a fixed R value for each type; Engineering calculation: Calculate according to parameters such as the space size and ventilation rate of the pipe section through fluid mechanics formulas or CFD simulations; Dynamically adjustable: Set R as an adjustable parameter in the simulation software for strategy analysis in different scenarios.

[0058] By introducing the space constraint coefficient R, this model realizes the differential assessment of explosion risks in different scenarios such as indoors, pipe trenches, and valve wells.

[0059] Furthermore, the sensitivity analysis in S6 is specifically as follows: Use the dimensionless sensitivity coefficient method to conduct sensitivity analysis on all input parameters and core variables, identify the key variables that have a significant impact on the explosion risk of gas pipelines, and locate the weak links in risk control. The calculation formula of the dimensionless sensitivity coefficient method is: ; In the formula, is the th variable 's sensitivity coefficient to the explosion risk ; is the partial derivative; Take the absolute value of the calculated sensitivity coefficient Variables with a value greater than 1 are identified as highly sensitive variables and are considered as weak links in risk control and core control targets.

[0060] In this embodiment of the invention, the system dynamics mathematical equations constructed in S5 are imported into system dynamics simulation software (such as Vensim) to complete the input of model parameters and configuration of variable association. The core simulation parameters are set in combination with the actual needs of urban gas pipeline network operation: simulation duration (such as the next 30 years) and simulation step size (such as 1 year). The current standardized values ​​of each factor after dimensionless processing are used as the initial state values ​​of the model to complete the full-process parameter configuration before simulation, ensuring that the model fits the actual operation scenario. (1) Baseline scenario simulation: The current standardized values ​​of each risk factor and control factor are used as the initial input of the model. The baseline scenario simulation without human intervention is run. Through numerical integration calculation of the simulation software, the trend curve of the dynamic evolution of the explosion accident Y over time is output. The natural development law and change characteristics of the risk of urban gas pipeline explosion under no control measures are clarified, and a benchmark reference is provided for subsequent strategy scenario comparison. (2) Strategy Scenario Simulation: Based on the baseline scenario simulation results, core decision variables in gas pipeline network safety management are selected for gradient adjustment. Multi-dimensional strategy scenario comparison simulation is carried out to quantitatively evaluate the risk reduction effect and implementation synergy of different management intervention measures. Specifically, three types of control strategies are set: Strategy A (fundamental solution strategy): Reduce management and maintenance risks (A) by 50%, simulating fundamental control measures such as improving pipeline management systems, strengthening full life cycle management, and enhancing management standardization. Strategy B (Short-term solution): Reduce external damage risk (O) by 30% by simulating targeted control measures such as strengthening line patrol supervision, adding protective facilities, and enhancing third-party construction management; Strategy C (Combined Approach): Simultaneously implement Strategy A and Strategy B to reduce management and maintenance risks (A) and external damage risks (O) in tandem. Simulate comprehensive intervention measures that combine fundamental control and targeted control, and analyze the synergistic risk reduction effect of the two types of measures.

[0061] By comparing the evolution trend, risk peak, and reduction rate of explosion accident Y under different strategy scenarios, the risk reduction efficiency, implementation cost adaptability, and applicable scenarios of each strategy are quantified, providing quantitative support for actual safety management decisions.

[0062] Based on the sensitivity coefficient calculation results, a variable sensitivity ranking table and a risk control weakness identification report are output, clarifying the priority control objects, core optimization directions and control priorities of gas pipeline network safety management, and providing data support for the accurate implementation of control strategies.

[0063] and Figure 1Corresponding to the method described above, this invention also discloses a dynamic explosion risk prediction system based on a composite weighting of ISM hazard coefficient and entropy weight. The application of the aforementioned dynamic explosion risk prediction method based on a composite weighting of ISM hazard coefficient and entropy weight includes: A multi-source data acquisition module is used to collect historical and real-time data from the gas pipeline network; The data preprocessing module is connected to the multi-source data acquisition module to clean, format, and dimensionless process the data. The ISM model building module, connected to the data preprocessing module, is used to automatically build adjacency and reachability matrices, calculate factor risk coefficients, and generate multi-level hierarchical structure models. The SD model building module, connected to the ISM model building module, is based on a multi-level hierarchical structure model and factor risk coefficients. It assists in constructing the core variables of system dynamics, causal feedback loops, and stock flow diagrams according to preset aggregation rules, and defines state variables, rate variables, and auxiliary variables. The parameter calibration and learning module is connected to the data preprocessing module and the SD model construction module. It uses the entropy weight method to calculate objective weights and combines them with factor risk coefficients to perform entropy weight-ISM composite weighting. The model equation parameters are calibrated through multiple regression analysis. The simulation calculation engine module, connected to the SD model building module, is used to load the complete system dynamics model, perform numerical integration calculations, and conduct simulations of baseline scenarios and multi-strategy scenarios. The visualization and early warning decision-making module is connected to the simulation calculation engine module. It is used to receive simulation results, display the explosion hazard trend in the form of charts and GIS heat maps, issue early warnings when the results exceed preset thresholds, and push control decision suggestions based on sensitivity analysis results.

[0064] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0065] Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An explosive risk dynamic prediction method based on ISM risk coefficient and entropy weight composite weighting, characterized in that, Includes the following steps: S1. The fault tree analysis method is used to systematically identify and analyze the hazards of urban gas fire and explosion accidents, and a fault tree diagram with urban gas fire and explosion accidents as the top-level event is established, and the basic events at the bottom level of the fault tree diagram are used as system elements. S2. Construct an explanatory structural model for urban gas fires and explosions and calculate the risk coefficients of factors. Use an iterative algorithm to hierarchically divide the system elements and generate a multi-level hierarchical structural model that reveals the risk transmission path. S3. Based on the multi-level hierarchical structure model and the risk coefficient, the system elements are aggregated into the core variables of system dynamics, and a causal feedback loop describing the positive and negative feedback relationship between the variables is established. S4. Transform the causal feedback loop into a system dynamics stock flow diagram based on entropy theory, and define state variables, rate variables, and auxiliary variables; S5. The objective weights of the sub-factors of the core variables of system dynamics are calculated using the entropy weight method. The objective weights are corrected using the risk coefficient as the weight correction factor. The core mathematical equation of the stock flow diagram of system dynamics is constructed by the multiple regression method. S6. Based on core mathematical equations, simulation prediction is performed. The effectiveness of control measures is quantitatively evaluated through multi-scenario comparison. Sensitivity analysis is combined to identify weak links in risk control and output dynamic prediction results and control decision recommendations.

2. The dynamic prediction method of explosion risk based on ISM risk coefficient and entropy weight composite weighting according to claim 1, characterized in that, The fault tree diagram in S1 includes top-level events, intermediate sub-events, and bottom-level basic events. Intermediate sub-events are divided into three levels: first-level sub-intermediate events are the core triggering conditions of top-level events, second-level sub-intermediate events are the subdivision of triggering paths, and third-level sub-intermediate events are the cause classification of triggering paths.

3. The method for dynamic prediction of explosion risk based on ISM hazard coefficient and entropy weighting as described in claim 1, characterized in that, The specific calculation method for the risk coefficient of factors in S2 is as follows: An adjacency matrix is ​​constructed based on the direct influence relationships between system elements in the fault tree diagram. A reachability matrix is ​​generated through Boolean algebra operations to reveal the direct and indirect transmission relationships between system elements. Factor hazard coefficients are calculated based on the reachability matrix for each factor. The out-degree is defined as the number of elements with a row value of 1 in the reachability matrix, and the in-degree is defined as the number of elements with a column value of 1 in the reachability matrix, taking into account various factors. Calculate the hazard factor based on the out-degree, in-degree, and position in the hierarchical structure. : ； In the formula, , , These are the weighting coefficients. As factors The degree of exit, For the maximum out-degree of all factors, As factors The level at which it is located The maximum number of levels, As factors in-degree, This represents the maximum in-degree of all factors.

4. The method for dynamic prediction of explosion risk based on ISM hazard coefficient and entropy weighting as described in claim 3, characterized in that, The adjacency matrix A is a 45×45 square matrix with matrix elements... , Indicator Factors Factors There is a direct impact.

5. The method for dynamic prediction of explosion risk based on ISM hazard coefficient and entropy weighting as described in claim 1, characterized in that, In S3, the system elements are aggregated into the core variables of system dynamics as follows: Based on the hierarchical classification of the multi-level hierarchical structure model, the basic events are classified into material inherent risks, corrosion and environmental risks, anti-corrosion system risks, construction defect risks, management and maintenance risks, external damage risks, equipment failure risks, human operation risks, ignition source risks, and protective facility risks. During the aggregation process, for any aggregation variable, the weights of each sub-factor are determined after normalization based on the corresponding factor hazard coefficient.

6. The method for dynamic prediction of explosion risk based on ISM hazard coefficient and entropy weighting as described in claim 1, characterized in that, S4 introduces entropy theory and dissipative structure theory to determine the total entropy change of the system. for: ； In the formula, The entropy generated by irreversible processes within the system. The system is the entropy flow exchanged between itself and the outside world; the sum of the inflow rates of each state variable constitutes the positive entropy source of the system, driving the system to evolve towards a high-risk disordered state; the sum of the outflow rates of each state variable constitutes the negative entropy flow introduced by the system through safety management activities, driving the system to recover to a safe and orderly state; the risk state variable of management and maintenance serves as a bottleneck variable that limits the efficiency of introducing negative entropy flow.

7. The method for dynamic prediction of explosion risk based on ISM hazard coefficient and entropy weighting as described in claim 1, characterized in that, In S5, the entropy weight method is used to calculate the objective weights of the sub-factors of the core variables of system dynamics. The objective weights are then corrected using the hazard coefficient as a weight correction factor. Specifically: by Using a sample of time points, Using factors as indicators, a standardized decision matrix is ​​constructed. Calculate the first Under the first factor The proportion matrix of each sample : ； Calculate the first Entropy of each factor : ； Calculate the difference coefficient Determine the basic weights of the entropy weight method ISM risk coefficients are introduced for composite weighting: In the formula, For composite weights, , As factors , The risk coefficient of the factors.

8. The method for dynamic prediction of explosion risk based on ISM hazard coefficient and entropy weighting as described in claim 1, characterized in that, The core mathematical equations of the system dynamics stock-flow diagram constructed in S5 include variable aggregation equations, management efficiency equations, state variable integral equations, and rate variable equations; among them, the state variable integral equation for an explosion accident is: ； ； In the formula, for The comprehensive index of explosion risk at any given moment. This is the comprehensive index of explosion hazard at the initial moment. To mitigate the risk of equipment failure, Indicates time, For the risk of ignition source, To protect against facility risks, The spatial constraint coefficient characterizes the ventilation conditions and the degree of spatial airtightness of the environment in which the pipeline is located. This represents the leakage-ignition-space coupling function. For leakage-ignition coupling coefficient, To aggravate the failure of protection, To control the intensity of intervention, This is the safety control intervention coefficient.

9. The method for dynamic prediction of explosion risk based on ISM hazard coefficient and entropy weighting as described in claim 1, characterized in that, The sensitivity analysis in S6 specifically includes: Sensitivity analysis was conducted on all input parameters and core variables using the dimensionless sensitivity coefficient method to identify key variables that significantly impact the risk of gas pipeline explosions and pinpoint weak points in risk control. The formula for calculating the dimensionless sensitivity coefficient method is as follows: ； In the formula, For the first Variables Risk of explosion Sensitivity coefficient; Partial derivative; the absolute value of the calculated sensitivity coefficient Variables with a value greater than 1 are identified as highly sensitive variables and are considered as weak links in risk control and core control targets.

10. A dynamic prediction system for explosion risk based on ISM hazard coefficient and entropy weighting, characterized in that, The method for dynamic prediction of explosion risk based on ISM hazard coefficient and entropy weighting as described in any one of claims 1-9 includes: A multi-source data acquisition module is used to collect historical and real-time data from the gas pipeline network; The data preprocessing module is connected to the multi-source data acquisition module to clean, format, and dimensionless process the data. The ISM model building module, connected to the data preprocessing module, is used to automatically build adjacency and reachability matrices, calculate factor risk coefficients, and generate multi-level hierarchical structure models. The SD model building module, connected to the ISM model building module, is based on a multi-level hierarchical structure model and factor risk coefficients. It assists in constructing the core variables of system dynamics, causal feedback loops, and stock flow diagrams according to preset aggregation rules, and defines state variables, rate variables, and auxiliary variables. The parameter calibration and learning module is connected to the data preprocessing module and the SD model construction module. It uses the entropy weight method to calculate objective weights and combines them with factor risk coefficients to perform entropy weight-ISM composite weighting. The model equation parameters are calibrated through multiple regression analysis. The simulation calculation engine module, connected to the SD model building module, is used to load the complete system dynamics model, perform numerical integration calculations, and conduct simulations of baseline scenarios and multi-strategy scenarios. The visualization and early warning decision-making module is connected to the simulation calculation engine module. It is used to receive simulation results, display the explosion hazard trend in the form of charts and GIS heat maps, issue early warnings when the results exceed preset thresholds, and push control decision suggestions based on sensitivity analysis results.

Images