Chemical safety risk management and control method and device based on knowledge graph, equipment and medium

Abstract

The invention relates to the technical field of chemical safety risk management and control. The chemical safety risk management and control method and device based on the knowledge graph, the equipment and the medium are provided, and the method comprises the following steps: constructing a dynamic risk propagation graph based on a spatial topology of a preset process unit; constructing a chemical system knowledge graph; associating the dynamic risk propagation map with the chemical engineering system knowledge map through the real-time sensor data flow to generate a chemical engineering system risk knowledge network; carrying out coupling effect analysis through a graph neural network algorithm, and generating risk conduction path weight distribution; performing three-dimensional risk field rendering processing to generate a visual risk perspective interface; through a visual risk perspective interface, interactive risk traceability verification is carried out, and a protection measure optimization scheme is output, so that the technical effects of improving the dynamic risk response real-time performance, reducing the cross-unit linkage risk misjudgment rate and optimizing the protection decision economy are achieved.

Description

Technical Field

[0001] This invention relates to the field of chemical safety risk management technology, and in particular to chemical safety risk management methods, devices, equipment and media based on knowledge graphs. Background Technology

[0002] With the continuous expansion of the chemical industry and the increasing complexity of its processes, safety risk management in chemical production systems has become a core issue in ensuring the safety of personnel and property and maintaining production continuity. Efficient and precise risk management technologies are not only crucial for the long-term stable operation of chemical plants, but also directly impact the safety and resilience of the industrial and supply chains.

[0003] However, traditional risk identification methods based on static rule bases and offline fault tree models are difficult to integrate with real-time sensor data streams, which can easily lead to delayed responses to chain failures in equipment. When faced with multi-dimensional coupled analysis of spatial topology connections, process logic dependencies, and fluid dynamics behavior, traditional technologies often adopt a fragmented modeling approach. Due to the lack of a collaborative analysis mechanism, this leads to quantitative deviations in the transmission of risks from multiple factors. In addition, traditional technologies rely on manual experience to backtrack system topology and formulate protective measures, which is inefficient in large-scale node analysis. Furthermore, the lack of a cost-effectiveness quantification model can easily lead to improper allocation of operation and maintenance resources. Summary of the Invention

[0004] Therefore, it is necessary to provide chemical safety risk management methods, devices, equipment and media based on knowledge graphs to address the above-mentioned technical problems, so as to improve the real-time performance of dynamic risk response, reduce the misjudgment rate of cross-unit chain risks, and optimize the economic efficiency of protection decisions.

[0005] Firstly, this application provides a chemical safety risk management method based on knowledge graphs, which includes:

[0006] Based on the spatial topology of the preset process units, a dynamic risk propagation graph is constructed; based on the interaction relationships of the preset process units, a chemical system knowledge graph is constructed; through real-time sensor data streams, the dynamic risk propagation graph and the chemical system knowledge graph are linked to generate a chemical system risk knowledge network.

[0007] Based on the risk knowledge network of chemical system, a graph neural network algorithm is used to analyze the coupling effect and generate the risk transmission path weight distribution.

[0008] Based on the risk transmission path weight distribution, a three-dimensional risk field rendering process is performed to generate a visual risk perspective interface.

[0009] Through a visual risk perspective interface, interactive risk tracing and verification can be conducted, and optimized protective measures can be output.

[0010] In one embodiment, based on the risk transmission path weight distribution, a three-dimensional risk field rendering process is performed to generate a visualized risk perspective interface, including:

[0011] Spatial field intensity mapping is performed on the weight distribution of risk transmission paths to generate a three-dimensional risk intensity field.

[0012] Based on the three-dimensional risk intensity field, dynamic particle trajectory simulation is performed to generate a dynamic model of risk diffusion.

[0013] The dynamic risk diffusion model is processed by multi-view fusion rendering to generate a visual risk perspective interface.

[0014] In one embodiment, a multi-view fusion rendering process is performed on the risk diffusion dynamic model to generate a visualized risk perspective interface, including:

[0015] Multi-dimensional data segmentation processing is performed on the dynamic model of risk diffusion to generate spatial distribution view, temporal evolution view and intensity hierarchy view;

[0016] The spatial distribution view is reconstructed using three-dimensional isosurfaces to generate a spatial risk distribution layer for equipment.

[0017] Dynamic particle flow rendering is applied to the time evolution view to generate a risk transmission trajectory layer;

[0018] The intensity hierarchy view is subjected to thermal gradient coloring to generate a risk thermal field layer.

[0019] By using a perspective-adaptive fusion algorithm, the equipment space risk distribution layer, risk transmission trajectory layer, and risk thermal field layer are simultaneously overlaid to generate a visual risk perspective interface.

[0020] In one embodiment, an interactive risk tracing and verification process is performed through a visual risk perspective interface, outputting an optimized protective measure plan, including:

[0021] Based on the risk point selection command triggered by the visual risk perspective interface, generate target risk point identifiers;

[0022] Multi-level causal chain backtracking analysis is performed on the identified target risk points to generate a set of critical failure paths;

[0023] Virtual protection units are inserted at the nodes of the critical failure path set to simulate the propagation of the blocking effect and generate a protection effectiveness evaluation matrix.

[0024] Based on the protection effectiveness assessment matrix and preset economic parameters, multi-objective optimization decisions are made to generate optimized protection measures.

[0025] In one embodiment, a multi-level causal chain backtracking analysis is performed on the target risk point identification to generate a set of critical failure paths, including:

[0026] Based on the dynamic risk propagation map, the spatial adjacency relationship of the target risk point is traced back to generate a set of physical propagation paths;

[0027] Based on the knowledge graph of chemical system, the process logic dependency relationship of the target risk point is traced back to generate a set of functional failure chains.

[0028] By using a cross-level association matching algorithm, a spatiotemporal coupling analysis is performed on the set of physical propagation paths and the set of functional failure chains to generate a set of critical failure paths.

[0029] In one embodiment, based on the protection effectiveness evaluation matrix and preset economic parameters, multi-objective optimization decision-making is performed to generate an optimized protection measure scheme, including:

[0030] The safety performance dimension of the protection effectiveness assessment matrix is ​​decoupled to generate a risk blocking rate vector and an accident consequence suppression rate vector.

[0031] The preset economic parameters are processed through full life cycle cost modeling to generate an economic impact vector that includes equipment purchase cost information, installation and commissioning cost information, and maintenance and operation cost information;

[0032] The Pareto front analysis algorithm is used to perform multi-objective collaborative optimization on the risk blocking rate vector, accident consequence inhibition rate vector and economic impact vector to generate a non-dominated solution set.

[0033] Based on the preset security integrity level constraints, the non-dominated solution set is subjected to feasibility screening to generate an optimized protection measure scheme.

[0034] In one embodiment, the three-dimensional risk intensity field is obtained using the following formula, including:

[0035]

[0036] in, Represents the three-dimensional risk intensity field. Representing a path The source point weight, Representing a path Spatial length, Indicates the distance attenuation coefficient. Representing a path The initial risk value, and Both represent nonlinear mapping parameters. This represents the total number of risk transmission paths.

[0037] Secondly, this application also provides a chemical safety risk management device based on a knowledge graph, the device comprising:

[0038] The risk knowledge network construction module is used to construct a dynamic risk propagation graph based on the spatial topology of preset process units; construct a chemical system knowledge graph based on the interaction relationships of preset process units; and generate a chemical system risk knowledge network by associating the dynamic risk propagation graph and the chemical system knowledge graph through real-time sensor data streams.

[0039] The risk coupling analysis module is used to analyze coupling effects based on the risk knowledge network of chemical systems and generate risk transmission path weight distribution by using graph neural network algorithm.

[0040] The risk field visualization module is used to perform three-dimensional risk field rendering based on the weight distribution of risk transmission paths, and generate a visualized risk perspective interface.

[0041] The risk tracing and optimization module is used to conduct interactive risk tracing verification through a visual risk perspective interface and output optimized protection measures.

[0042] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods in the first aspect of this application.

[0043] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the methods in the first aspect of this application.

[0044] This application provides a knowledge graph-based method, device, equipment, and medium for chemical safety risk management: By constructing a dynamic risk propagation graph and a chemical system knowledge graph, and using real-time sensor data streams to link the two to generate a chemical system risk knowledge network, risk management can access dynamic data in real time, breaking through the limitations of traditional static analysis and thus improving the real-time performance of dynamic risk response. A graph neural network algorithm is used to analyze the coupling effect of the chemical system risk knowledge network, generating a risk transmission path weight distribution. This captures the correlation and influence between multiple factors, avoiding the bias caused by traditional fragmented modeling, and thus reducing the misjudgment rate of cross-unit cascading risks. A three-dimensional risk field is rendered based on the risk transmission path weight distribution to obtain a visual interface, supporting interactive risk tracing and verification. Combined with multi-objective optimization decision-making, protective measures are output, which can ensure safety while considering economy, achieving optimized protective decision-making. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a flowchart of a knowledge graph-based chemical safety risk management method according to one embodiment of the present invention;

[0047] Figure 2 This is a flowchart of an optimized protection measure scheme, which is generated through an interactive risk tracing and verification process using a visual risk perspective interface in one embodiment of the present invention.

[0048] Figure 3 This is a structural diagram of a knowledge graph-based chemical safety risk management device according to one embodiment of the present invention. Detailed Implementation

[0049] To make the above-mentioned objects, features, and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0050] First, the application scenarios of the embodiments of this application are described. In the embodiments of this application, a knowledge graph-based chemical safety risk management method, device, equipment, and medium are provided, applicable to scenarios such as production control of large-scale petrochemical complexes, safety monitoring of continuous reaction processes in fine chemicals, and risk early warning of complex coal chemical processes.

[0051] For example, in large-scale petrochemical complexes, the equipment includes multiple interconnected process units such as cracking, separation, and refining. There is close interaction between materials and energy between these units, and multiple operating parameters such as pressure, temperature, and flow rate need to be monitored in real time. The embodiments of this application can achieve dynamic control of the overall safety risks of the equipment in such scenarios. In continuous reaction processes in fine chemicals, the reaction process involves multiple hazardous chemicals, and the process conditions are sensitive and prone to cascading risks due to parameter fluctuations. The embodiments of this application can be adapted to the risk identification and protection decision-making needs in such scenarios. In complex coal chemical processes, the system includes multiple complex processes such as gasification, conversion, and synthesis, with a large number of equipment, complex spatial layout, and diverse risk transmission paths. The embodiments of this application can also be applied to safety risk management in such scenarios.

[0052] In illustrative purposes, the chemical safety risk management method, device, equipment and medium based on knowledge graph provided in this application embodiment can also be applied to other application scenarios, such as safety risk management of batch reaction processes in pharmaceutical and chemical industries, and collaborative risk prevention and control of related processes among multiple enterprises in chemical industrial parks. These are only examples and do not limit the specific application scenarios.

[0053] like Figure 1 As shown, this application provides a chemical safety risk management method based on knowledge graphs, which includes:

[0054] S101: Based on the spatial topology of the preset process units, construct a dynamic risk propagation graph; based on the interaction relationship of the preset process units, construct a chemical system knowledge graph; through real-time sensor data streams, link the dynamic risk propagation graph and the chemical system knowledge graph to generate a chemical system risk knowledge network.

[0055] For example, the physical spatial locations, connection methods, and spatial association paths of pre-designed process units in chemical production are analyzed. Based on the above spatial topology information, the propagation rules of risks between pre-designed process units through spatial contact or proximity are clarified, thereby constructing a dynamic risk propagation map. The material transportation, energy exchange, control signal interaction, and process logic dependencies of pre-designed process units in the production process are analyzed, and the functional attributes and constraints corresponding to the above interactions are integrated to construct a chemical system knowledge graph.

[0056] Real-time sensor data streams, including the operating status, environment, and key operating parameters of preset process units, are collected. A mapping relationship is established between this data stream and the spatial nodes of preset process units in the dynamic risk propagation graph and the interactive nodes of preset process units in the chemical system knowledge graph. Dynamic information is integrated into the above two types of graphs to achieve correlation and fusion, generating a chemical system risk knowledge network.

[0057] S102: Based on the risk knowledge network of chemical system, the coupling effect is analyzed by graph neural network algorithm to generate the risk transmission path weight distribution.

[0058] For example, features are extracted from nodes and edges in a chemical system risk knowledge network. Node features include the operating status, function, and spatial location attributes of preset process units, while edge features include the spatial correlation strength, interaction tightness, and dynamic correlation features reflected in real-time sensor data streams between preset process units. These features are then transformed into an input format recognizable by a graph neural network algorithm. The formatted features are input into the graph neural network algorithm, which uses its inter-layer propagation mechanism to analyze the spatial coupling effect, process logic coupling effect, and dynamic data coupling effect between preset process units, clarifying the degree of influence of various coupling effects on risk transmission.

[0059] Based on the analytical results, the correlation stability, impact range, and transmission efficiency of each risk transmission path are quantitatively assessed to determine the importance coefficient of each path. These importance coefficients are then integrated to generate the risk transmission path weight distribution.

[0060] S103: Based on the risk transmission path weight distribution, perform three-dimensional risk field rendering processing to generate a visual risk perspective interface.

[0061] For example, the weight values ​​of each path in the risk transmission path weight distribution are spatially mapped and associated with the actual spatial coordinates of a preset process unit to determine the risk intensity benchmark at different spatial locations, generating initial risk intensity distribution data in three-dimensional space. Combining the geometric structural features and spatial topological relationships of the preset process unit, the risk spatial attenuation law and the risk superposition rule between paths are incorporated into the basic spatial model of the three-dimensional risk field to complete the numerical construction of the three-dimensional risk field.

[0062] The system dynamically renders a 3D risk field, presenting the distribution of preset process units through spatial rendering, showcasing differences in risk intensity through intensity rendering, and simulating risk diffusion trends through dynamic rendering. It integrates various rendering results and adds interactive control functions to generate a visual risk perspective interface.

[0063] S104: Through a visual risk perspective interface, conduct interactive risk tracing and verification, and output optimized protection measures.

[0064] For example, a user-initiated risk point selection command for a specific risk area or risk signal is received through a visual risk perspective interface. Based on this command, the corresponding target risk point is located, and its associated spatial distribution data, temporal evolution data, and intensity level data are retrieved. According to the associated data of the target risk point, its spatial adjacency path in the dynamic risk propagation graph is traced back, and its process logic dependency chain in the chemical system knowledge graph is also traced. By cross-matching the spatial adjacency path and the process logic dependency chain, the key failure sources and intermediate transmission nodes of the target risk point are determined, completing interactive risk tracing verification.

[0065] Different types of virtual protection units are pre-set at key failure sources and intermediate transmission nodes. The blocking effect of various virtual protection units on the risk transmission path is simulated, generating protection effectiveness evaluation data including risk blocking rate and accident consequence suppression rate. Combined with pre-set equipment purchase cost parameters, installation and commissioning cost parameters, and maintenance and operation cost parameters, the protection effectiveness and economy of different virtual protection units are analyzed collaboratively. The protection combination that meets safety requirements and has the best cost is selected, and the optimized protection measure scheme is output.

[0066] This application provides a knowledge graph-based chemical safety risk management method: by constructing a dynamic risk propagation graph and a chemical system knowledge graph, and using real-time sensor data streams to link the two to generate a chemical system risk knowledge network, risk management can access dynamic data in real time, breaking the limitations of traditional static analysis and thus improving the real-time performance of dynamic risk response. A graph neural network algorithm is used to analyze the coupling effect of the chemical system risk knowledge network, generating a risk transmission path weight distribution. This captures the correlation and influence between multiple factors, avoiding the bias caused by traditional fragmented modeling, and thus reducing the misjudgment rate of cross-unit chain risks. A three-dimensional risk field is rendered based on the risk transmission path weight distribution to obtain a visual interface, supporting interactive risk tracing and verification. Combined with multi-objective optimization decision-making, protective measures are output, which can ensure safety while taking into account economy, achieving optimization of protective decisions.

[0067] In one embodiment, based on the risk transmission path weight distribution, a three-dimensional risk field rendering process is performed to generate a visualized risk perspective interface, including:

[0068] (1) Perform spatial field strength mapping on the risk transmission path weight distribution to generate a three-dimensional risk intensity field.

[0069] For example, the weight values ​​of each path in the risk transmission path weight distribution and the spatial coordinate information of the corresponding path are extracted. According to the preset spatial field strength mapping rules, the weight value of each path is converted into the risk intensity value of the corresponding spatial coordinate point. At the same time, considering the superposition effect of risk intensity between different paths, the risk intensity value of the overlapping spatial area is comprehensively calculated to generate a risk intensity data matrix covering the entire spatial range of the chemical production scenario. Based on this data matrix, a three-dimensional risk intensity field that can reflect the difference in risk intensity at various points in space is constructed.

[0070] Among them, the preset spatial field strength mapping rules include the conversion ratio between weight values ​​and risk intensity values, the setting of the attenuation coefficient of risk intensity for different spatial distances, and the superposition effect of risk intensity including the arithmetic superposition or weighted superposition calculation method of risk intensity of multiple paths in the same region.

[0071] (2) Based on the three-dimensional risk intensity field, dynamic particle trajectory simulation is performed to generate a dynamic model of risk diffusion.

[0072] For example, based on the risk intensity values ​​of each spatial point in the three-dimensional risk intensity field, the initial source spatial location and initial risk intensity level of risk diffusion are determined. According to the preset risk diffusion rate parameters and diffusion direction rules, the trajectory of risk moving from the initial source to the surrounding area in the form of particles is simulated in the three-dimensional space. The spatial coordinates and corresponding risk intensity values ​​of each particle at different time nodes are updated in real time. The path changes and intensity decay during the particle movement process are recorded. The dynamic trajectory data of all particles are integrated to construct a dynamic risk diffusion model that can intuitively show the process of risk diffusion in space over time.

[0073] Among them, the preset risk diffusion rate parameters include risk diffusion speed thresholds under different media environments, diffusion direction rules include the setting of preferred diffusion directions based on spatial topological relationships, and particle morphology risk simulation includes the setting of the risk unit magnitude represented by a single particle and the time interval of particle movement step.

[0074] (3) Perform multi-view fusion rendering on the risk diffusion dynamic model to generate a visual risk perspective interface.

[0075] For example, the risk diffusion dynamic model is processed by data segmentation, and the spatial distribution data, temporal evolution data and intensity level data are extracted from the model respectively. Based on the spatial distribution data, a spatial distribution view showing the location distribution of the risk in three-dimensional space is generated. Based on the temporal evolution data, a temporal evolution view showing the process of risk change over time is generated. Based on the intensity level data, an intensity level view showing the difference in risk intensity in different regions is generated.

[0076] The spatial distribution view is reconstructed using 3D isosurfaces to highlight the spatial relationship between pre-defined process units and risk areas. The temporal evolution view is rendered with dynamic particle flow to enhance the dynamic visual effect of risk diffusion. The intensity level view is colored with thermal gradients to differentiate risk intensity through color differences. Then, an adaptive viewpoint fusion algorithm is used to synchronously overlay the processed spatial distribution view, temporal evolution view, and intensity level view, ensuring that the spatial coordinates and time nodes of each view match. Interactive functions such as view switching, zooming, and risk information query are added to generate a visualized risk perspective interface.

[0077] The data segmentation process includes classifying and extracting risk diffusion dynamic model data according to spatial, temporal, and intensity dimensions; the three-dimensional isosurface reconstruction includes a spatial closed surface generation method based on risk intensity thresholds; the thermal gradient coloring includes setting color gradient intervals from low risk to high risk; and the viewpoint adaptive fusion algorithm includes rules for adjusting the viewpoint angle and adapting image transparency when multiple views are overlaid.

[0078] In one embodiment, a multi-view fusion rendering process is performed on the risk diffusion dynamic model to generate a visualized risk perspective interface, including:

[0079] (1) Perform multi-dimensional data segmentation on the risk diffusion dynamic model to generate spatial distribution view, time evolution view and intensity level view.

[0080] For example, spatial coordinate data, time series data, and risk intensity data are extracted from the risk diffusion dynamic model. The data are classified, filtered, and structured according to the three dimensions of space, time, and intensity. Redundant data is removed and data association identifiers are added. A dataset reflecting the location distribution of risk in three-dimensional space is constructed for the spatial dimension data, a dataset reflecting the process of risk change over time is constructed for the time dimension data, and a dataset reflecting the differences in risk intensity in different regions is constructed for the intensity dimension data. Based on the datasets of the three dimensions, a spatial distribution view, a time evolution view, and an intensity hierarchy view are generated respectively.

[0081] (2) Perform three-dimensional isosurface reconstruction on the spatial distribution view to generate a spatial risk distribution layer for equipment.

[0082] For example, the geometric structure data of the preset process units and the spatial coordinate data of the risk areas are extracted from the spatial distribution view. The threshold for generating isosurfaces of risk intensity is set. Based on the three-dimensional isosurface reconstruction algorithm, the spatial coordinate points corresponding to the same risk intensity threshold are connected to generate closed three-dimensional isosurfaces. The contour features and spatial positional relationships of the preset process units are integrated into the isosurface model. The distance parameters between each preset process unit and the surrounding risk isosurfaces are marked. The spatial hierarchy of the isosurfaces is enhanced by lighting rendering technology to generate a spatial risk distribution layer for the equipment.

[0083] Among them, the thresholds for generating isosurfaces of risk intensity include low-risk threshold, medium-risk threshold and high-risk threshold.

[0084] (3) Perform dynamic particle flow rendering on the time evolution view to generate a risk transmission trajectory layer.

[0085] For example, extract the time node data of risk diffusion and the spatial trajectory data of the corresponding nodes in the time evolution view, set the number of dynamic particles generated and particle attribute parameters, and generate a corresponding number of dynamic particles at each time node based on the particle flow rendering algorithm. Bind the initial position of the particles to the spatial coordinates of the risk source at that time node, keep the movement direction of the particles consistent with the risk diffusion direction, match the movement speed of the particles with the risk diffusion rate, and present the transmission process of risk over time through the continuous movement trajectory of the particles. At the same time, mark the correlation information between the particle distribution density and the risk intensity at each time node to generate a risk transmission trajectory layer.

[0086] Among them, particle attribute parameters include particle size, transparency and color characteristics, particle distribution density includes the number of particles in a unit space area, and risk intensity correlation information includes the correspondence between particle color and risk intensity level.

[0087] (4) Perform thermal gradient coloring on the intensity level view to generate the risk thermal field layer.

[0088] For example, risk intensity numerical data of each spatial region in the intensity hierarchy view is extracted, the thermal gradient interval of risk intensity and the corresponding color system are set, the risk intensity values ​​are classified according to the gradient interval, and a unique color value is matched for each gradient interval. Based on the heat map rendering algorithm, the risk intensity values ​​of each spatial region are converted into the corresponding color presentation. During the rendering process, the spatial outline of the preset process unit is retained as a background reference, the risk intensity range corresponding to different thermal gradient intervals is marked, and the spatial distribution difference of risk intensity is intuitively displayed through the color gradient transition, thereby generating a risk thermal field layer.

[0089] The thermal gradient range includes a continuous gradient division from low risk to high risk, the color system includes a corresponding gradient from cool to warm colors, and the risk intensity range includes the maximum and minimum risk intensity values ​​for each gradient range.

[0090] (5) Through the perspective adaptive fusion algorithm, the equipment space risk distribution layer, risk transmission trajectory layer and risk thermal field layer are synchronously superimposed to generate a visual risk perspective interface.

[0091] For example, the spatial coordinate system and time synchronization parameters of the equipment space risk distribution layer, risk transmission trajectory layer, and risk thermal field layer are obtained. Based on the perspective adaptive fusion algorithm, the spatial scaling ratio and perspective angle of the three layers are adjusted to ensure that the spatial coordinates of each layer are aligned. The overlay transparency parameters of each layer are set so that the risk transmission trajectory layer and the risk thermal field layer can be clearly presented on the background of the equipment space risk distribution layer. At the same time, a time synchronization mechanism is established to ensure that the dynamic changes of the risk transmission trajectory and the intensity updates of the risk thermal field are consistent at the time nodes. Layer on / off control and risk information query interaction functions are added to generate a visual risk perspective interface.

[0092] Among them, the overlay transparency parameter includes the visibility ratio of each layer when overlaying, and the time synchronization mechanism includes a time-stamp-based layer data synchronization call method.

[0093] like Figure 2 As shown, an interactive risk tracing and verification process is conducted through a visual risk perspective interface, outputting optimized protective measures, including:

[0094] S201: Based on the risk point selection command triggered by the visual risk perspective interface, generate target risk point identifiers.

[0095] For example, the visual risk perspective interface is pre-set with a risk point selection interaction function. This function allows users to trigger risk point selection instructions through clicks, box selections, and other operations. After receiving the risk point selection instructions initiated by the user, the interface backend automatically locates the spatial coordinates and risk intensity information corresponding to the instructions, retrieves the preset process unit identifier, real-time operating parameters, and risk level data associated with the location, integrates the above information in a structured manner, generates a unique identifier code, and generates a target risk point identifier.

[0096] The risk point selection interaction function includes operation modes such as mouse interaction and touch screen interaction, and the identification code includes a combination of spatial coordinate code, preset process unit number and risk level code.

[0097] S202: Perform multi-level causal chain backtracking analysis on the target risk point identification to generate a set of critical failure paths.

[0098] For example, dynamic risk propagation graphs associated with target risk point identifiers and chemical system knowledge graph data are retrieved. Based on the dynamic risk propagation graph, spatial adjacency nodes corresponding to the target risk point identifiers are traced, and the physical propagation links from spatial adjacency nodes to the target risk point identifiers are identified. Simultaneously, based on the chemical system knowledge graph, process logic dependency nodes corresponding to the target risk point identifiers are traced, and functional failure links from process logic dependency nodes to the target risk point identifiers are identified. A cross-graph association algorithm is used to perform spatiotemporal matching of the physical propagation links and functional failure links, filtering out failure links that simultaneously satisfy both spatial propagation conditions and process logic conditions. These failure links are then sorted by impact level and integrated to generate a set of critical failure paths.

[0099] Among them, spatial adjacency nodes include preset process unit nodes that are directly connected or adjacent to the target risk point identifier in space, process logic dependent nodes include nodes that provide materials, energy or control signals to the preset process unit corresponding to the target risk point identifier, and cross-graph association algorithms include link matching rules based on timestamps and node attributes.

[0100] S203: Insert virtual protection units at the nodes of the critical failure path set, simulate the propagation of the blocking effect, and generate a protection effectiveness evaluation matrix.

[0101] For example, the functional types and risk transmission roles of each node in the set of critical failure paths are analyzed to determine the appropriate type of virtual protection unit (VPU) for each node. VPU types include physical isolation, parameter control, and early warning / interception. Different types of VPUs are inserted at each node in the set of critical failure paths to construct a simulated scenario containing VPUs. Using a risk propagation simulation algorithm, the changes in risk transmission intensity, transmission speed, and impact range of the critical failure paths before and after inserting VPUs are calculated. These changes are quantified into indicators such as risk blocking rate and consequence suppression rate. Based on the combination relationship between VPU type and node location, the indicator data are organized into a matrix to generate a protection effectiveness evaluation matrix.

[0102] The functional types include material conveying nodes, reaction control nodes, and energy supply nodes; the risk transmission roles include risk source nodes, intermediate transmission nodes, and terminal impact nodes; and the risk propagation simulation algorithm includes a risk transmission calculation model based on path weights.

[0103] S204: Based on the protection effectiveness assessment matrix and preset economic parameters, perform multi-objective optimization decisions to generate optimized protection measures.

[0104] For example, the risk blocking rate vector and accident consequence suppression rate vector are obtained from the protection effectiveness evaluation matrix, while preset economic parameters are retrieved. These preset economic parameters include the equipment purchase cost, installation and commissioning cost, maintenance and operation cost, and total life cycle cost of the virtual protection unit. A multi-objective optimization function is constructed by using the risk blocking rate and accident consequence suppression rate as safety objectives and various costs as economic objectives. Solving the multi-objective optimization function yields a non-dominated solution set, which contains combinations of protection schemes that satisfy different safety and economic balance relationships. Based on preset security integrity level constraints, the non-dominated solution set is screened, eliminating schemes that fail to meet safety performance standards or exceed budgetary economic costs, retaining the optimal balance scheme, and compiling it into a specification document to generate an optimized protection measure scheme.

[0105] Among them, the multi-objective optimization algorithm includes the Pareto front analysis algorithm, the safety integrity level constraints include the minimum threshold for risk blocking rate, the minimum threshold for accident consequence suppression rate, and the maximum threshold for cost, and the specification documents include the protection unit type, installation location, expected performance, and cost details.

[0106] In one embodiment, a multi-level causal chain backtracking analysis is performed on the target risk point identification to generate a set of critical failure paths, including:

[0107] (1) Based on the dynamic risk propagation map, the spatial adjacency relationship of the target risk point is traced back to generate a set of physical propagation paths.

[0108] For example, the system retrieves the spatial topology data of the preset process unit stored in the dynamic risk propagation map, locates the spatial node of the preset process unit corresponding to the target risk point identifier, queries the spatial adjacency relationship data recorded in the dynamic risk propagation map based on the spatial node, determines other preset process unit spatial nodes that are directly connected to the spatial node or within the preset spatial distance range, sorts out the spatial connection paths from these adjacent spatial nodes to the spatial node corresponding to the target risk point identifier, marks the spatial distance, connection method and risk propagation direction of each path, and organizes and archives all spatial connection paths that meet the conditions to generate a set of physical propagation paths.

[0109] The preset spatial distance range is set according to the safety protection level of the chemical production scenario. The connection methods include pipeline connection, equipment proximity, and regional association. The risk propagation direction includes attributes such as one-way propagation and two-way propagation.

[0110] (2) Based on the knowledge graph of chemical system, the process logic dependency relationship of the target risk point is traced back to generate a set of functional failure chains.

[0111] For example, the system retrieves the preset process unit interaction relationship data stored in the chemical system knowledge graph, locates the preset process unit functional node corresponding to the target risk point identifier, queries the process logic dependency relationship data recorded in the chemical system knowledge graph based on the functional node, determines the upstream preset process unit functional node that provides material input, energy supply, control signals or process support for the functional node, sorts out the process logic links from the upstream functional node to the functional node corresponding to the target risk point identifier, marks the dependency type, interaction parameter requirements and functional impact of each link, organizes and archives all the process logic links that meet the conditions, and generates a set of functional failure chains.

[0112] The dependency types include material dependency, energy dependency, control dependency, and process sequence dependency. The interaction parameter requirements include material flow range, energy supply stability threshold, and control signal response time. The functional impact level is classified into critical impact, important impact, and general impact.

[0113] (3) Through cross-level association matching algorithm, spatiotemporal coupling analysis is performed on the physical propagation path set and the functional failure chain set to generate a critical failure path set.

[0114] For example, the spatial node sequence, risk propagation timestamp, and spatial correlation strength data of each path in the physical propagation path set are extracted. Simultaneously, the functional node sequence, logical dependency time series, and functional impact coefficient data of each link in the functional failure chain set are extracted. The spatial node sequences of the physical propagation paths and the functional node sequences of the functional failure chains are matched to ensure a one-to-one correspondence between spatial nodes and functional nodes within the same predefined process unit. Based on the risk propagation timestamp and logical dependency time series, time-dimensional coupling is performed to filter out path and link combinations that are time-synchronized or have temporal correlations. Weight calculations are performed by combining the spatial correlation strength data and the functional impact coefficient data, retaining combinations with weight values ​​higher than a preset threshold. The paths and links corresponding to these combinations are integrated into complete failure paths, generating a set of critical failure paths.

[0115] Among them, node matching includes precise matching based on the unique identifier of the preset process unit, time synchronization means that the risk propagation time and the logical dependency time are in the same time interval, time sequence association means that the risk propagation time and the logical dependency time have a sequential triggering relationship, and the preset threshold is set according to the safety risk assessment standard of chemical system.

[0116] In one embodiment, based on the protection effectiveness evaluation matrix and preset economic parameters, multi-objective optimization decision-making is performed to generate an optimized protection measure scheme, including:

[0117] (1) Decouple the safety performance dimension of the protection effectiveness assessment matrix to generate the risk blocking rate vector and the accident consequence suppression rate vector.

[0118] For example, the effectiveness data of various protection units recorded in the protection effectiveness evaluation matrix at different nodes are extracted, and the core dimensions of safety effectiveness are determined to be risk blocking and accident consequence suppression. According to the combination dimension of protection units and nodes, the corresponding data in the protection effectiveness evaluation matrix are classified and organized respectively, data redundancy items and outliers are removed, and consistency verification is performed on the same type of data to ensure that the risk blocking data and accident consequence suppression data of each protection unit at each node are accurately matched. Then, the organized risk blocking data is arranged in a preset order to generate a risk blocking rate vector, and the accident consequence suppression data is arranged in the same order to generate an accident consequence suppression rate vector.

[0119] Among them, the combination dimension of protection units and nodes refers to the combination of different nodes in the set of critical failure paths corresponding to different types of virtual protection units. Data redundancy items include duplicate recorded performance data, outliers refer to data that exceeds the reasonable performance range, and preset order refers to the fixed sorting rules according to the type of protection unit and the location of the nodes.

[0120] (2) Perform full life cycle cost modeling on the preset economic parameters to generate an economic impact vector that includes equipment purchase cost information, installation and commissioning cost information and maintenance and operation cost information.

[0121] For example, the cost categories included in the preset economic parameters are sorted out, and the core cost elements are identified as equipment purchase cost, installation and commissioning cost, and maintenance and operation cost. According to the requirements of full life cycle cost modeling, each cost category is broken down into cost sub-items related to the protection unit type and installation node. The calculation basis and value source of each cost sub-item are determined, and the association mapping relationship between cost sub-items and protection units and nodes is established. The equipment purchase cost sub-items, installation and commissioning cost sub-items, and maintenance and operation cost sub-items of the same protection unit at the same node are summarized and integrated. The summarized cost data is arranged according to the combination order of protection units and nodes, and an economic impact vector including various cost information is generated.

[0122] The cost sub-items include equipment purchase costs such as unit price and transportation costs; installation and commissioning costs such as labor costs and consumable costs; and maintenance and operation costs such as regular inspection costs and consumable replacement costs. The calculation basis includes market quotations and industry quota standards, and the value sources include price lists provided by suppliers and historical operation and maintenance cost databases.

[0123] (3) Using the Pareto front analysis algorithm, the risk blocking rate vector, the accident consequence inhibition rate vector and the economic impact vector are processed by multi-objective collaborative optimization to generate a non-dominated solution set.

[0124] For example, the safety objective corresponding to the risk blocking rate vector is to maximize the risk blocking effect, the safety objective corresponding to the accident consequence suppression rate vector is to maximize the accident consequence suppression effect, and the economic objective corresponding to the economic impact vector is to minimize the total life cycle cost. The values ​​of the three vectors are standardized to eliminate the difference in dimensions between different indicators. The standardized vectors are input into the Pareto front analysis algorithm, and the optimization iteration conditions and convergence criteria of the algorithm are set. The algorithm calculates the performance of different protection unit and node combination schemes in the three objective dimensions, and selects the schemes that do not have other schemes that are better than itself in all three objective dimensions at the same time. These schemes are sorted according to their objective performance to generate a non-dominated solution set.

[0125] (4) Based on the preset security integrity level constraints, the non-dominated solution set is subjected to feasibility screening to generate an optimized protection measure scheme.

[0126] For example, the specific indicators included in the preset safety integrity level constraints are clearly defined, including the minimum threshold for risk blocking rate, the minimum threshold for accident consequence suppression rate, and the maximum threshold for total life cycle cost. The risk blocking rate value, accident consequence suppression rate value, and total life cycle cost value corresponding to each scheme are extracted from the non-dominated solution set. The value of each scheme is compared with the corresponding constraint indicators one by one. Schemes with risk blocking rate values ​​not lower than the minimum threshold for risk blocking rate, accident consequence suppression rate values ​​not lower than the minimum threshold for accident consequence suppression rate, and total life cycle cost values ​​not higher than the maximum threshold for total life cycle cost are selected. The selected schemes are comprehensively evaluated, and the schemes with better safety or economic performance on the basis of meeting the constraints are given priority. The finally selected schemes are compiled into a document containing the protection unit type, installation node location, expected safety performance, and cost details, i.e., the protection measure optimization scheme.

[0127] In one embodiment, the three-dimensional risk intensity field is obtained using the following formula, including:

[0128]

[0129] in, Represents the three-dimensional risk intensity field. Representing a path The source point weight, Representing a path Spatial length, Indicates the distance attenuation coefficient. Representing a path The initial risk value, and Both represent nonlinear mapping parameters. This represents the total number of risk transmission paths.

[0130] For example, the total number of risk transmission paths and other relevant parameters are defined, the distance attenuation term and the initial risk nonlinear mapping term for each path are calculated, the risk intensity contribution value of a single path to a spatial point is then calculated, the contribution values ​​of all paths are summed, and the risk intensity values ​​of each point in three-dimensional space are integrated to generate a three-dimensional risk intensity field.

[0131] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0132] In one embodiment, this application also provides a chemical safety risk management and control device 300 based on a knowledge graph, the device 300 comprising:

[0133] The risk knowledge network construction module 301 is used to construct a dynamic risk propagation graph based on the spatial topology of a preset process unit; construct a chemical system knowledge graph based on the interaction relationship of the preset process unit; and generate a chemical system risk knowledge network by associating the dynamic risk propagation graph and the chemical system knowledge graph through real-time sensor data streams.

[0134] The risk coupling analysis module 302 is used to analyze the coupling effect based on the chemical system risk knowledge network and generate the risk transmission path weight distribution by using a graph neural network algorithm.

[0135] The risk field visualization module 303 is used to perform three-dimensional risk field rendering based on the risk transmission path weight distribution and generate a visualized risk perspective interface.

[0136] The risk tracing and optimization module 304 is used to conduct interactive risk tracing verification through a visual risk perspective interface and output optimized protection measures.

[0137] Specifically, the risk knowledge network construction module 301 acquires spatial topology information such as the spatial location and connection relationships of preset process units, and constructs a dynamic risk propagation graph that reflects the propagation pattern of risks in space based on the above information. It acquires interaction relationship information such as material interaction, energy interaction, and control interaction between preset process units, and constructs a chemical system knowledge graph including the process logic of the chemical system based on the above information. It receives real-time sensor data streams, extracts risk-related real-time parameters from them, and associates and matches these parameters with nodes and edges in the dynamic risk propagation graph and the chemical system knowledge graph to establish a mapping relationship between real-time data and the graph, generating a chemical system risk knowledge network.

[0138] The risk coupling analysis module 302 extracts data such as node attributes and edge weights from the risk knowledge network of the chemical system, and inputs this data into a graph neural network algorithm. The graph neural network algorithm performs feature learning and propagation on the nodes and edges in the network, analyzes the coupling mechanism of risks between different preset process units, and calculates the degree of mutual influence of risks on each transmission path. Based on the calculation results, each risk transmission path is assigned a corresponding weight value, and the weight values ​​of all paths are organized into an ordered set to generate a risk transmission path weight distribution.

[0139] The risk field visualization module 303 spatially locates the weight values ​​in the risk transmission path weight distribution. Combined with the spatial topology of a preset process unit, it maps the weight value of each path to its corresponding position in three-dimensional space, generating a three-dimensional risk intensity field reflecting the risk intensity at each point in space. Based on this three-dimensional risk intensity field, it simulates the trajectory of risk spreading in space in particle form, considering factors such as the direction and speed of risk propagation, generating a dynamic risk diffusion model that dynamically displays the risk diffusion process. The dynamic risk diffusion model is rendered from different perspectives, integrating spatial distribution, temporal evolution, intensity levels, and other view information to generate a visualized risk perspective interface.

[0140] The risk tracing and optimization module 304 receives user input on the visualized risk perspective interface, selects risk points, determines target risk points, and generates unique target risk point identifiers. Based on these identifiers, it traces multi-level causal relationships within the chemical system's risk knowledge network, analyzes the spatial adjacency relationships of equipment leading to risks and the process logic dependencies, identifies critical failure paths, and generates a set of critical failure paths. At the nodes of this set, different types of virtual protection units are simulated and inserted, and the effectiveness of these units in blocking risk propagation is calculated, generating a protection effectiveness evaluation matrix containing performance data for each protection unit. Combining preset economic parameters, the protection effectiveness evaluation matrix undergoes multi-objective optimization analysis, balancing safety effectiveness with economic costs, selecting the optimal protection scheme, and outputting an optimized protection measure scheme.

[0141] The risk field visualization module 303 is also used for:

[0142] Spatial field intensity mapping is performed on the weight distribution of risk transmission paths to generate a three-dimensional risk intensity field.

[0143] Based on the three-dimensional risk intensity field, dynamic particle trajectory simulation is performed to generate a dynamic model of risk diffusion.

[0144] The dynamic risk diffusion model is processed by multi-view fusion rendering to generate a visual risk perspective interface.

[0145] The risk field visualization module 303 is also used for:

[0146] Multi-dimensional data segmentation processing is performed on the dynamic model of risk diffusion to generate spatial distribution view, temporal evolution view and intensity hierarchy view;

[0147] The spatial distribution view is reconstructed using three-dimensional isosurfaces to generate a spatial risk distribution layer for equipment.

[0148] Dynamic particle flow rendering is applied to the time evolution view to generate a risk transmission trajectory layer;

[0149] The intensity hierarchy view is subjected to thermal gradient coloring to generate a risk thermal field layer.

[0150] By using a perspective-adaptive fusion algorithm, the equipment space risk distribution layer, risk transmission trajectory layer, and risk thermal field layer are simultaneously overlaid to generate a visual risk perspective interface.

[0151] The risk tracing optimization module 304 is also used for:

[0152] Based on the risk point selection command triggered by the visual risk perspective interface, generate target risk point identifiers;

[0153] Multi-level causal chain backtracking analysis is performed on the identified target risk points to generate a set of critical failure paths;

[0154] Virtual protection units are inserted at the nodes of the critical failure path set to simulate the propagation of the blocking effect and generate a protection effectiveness evaluation matrix.

[0155] Based on the protection effectiveness assessment matrix and preset economic parameters, multi-objective optimization decisions are made to generate optimized protection measures.

[0156] The risk tracing optimization module 304 is also used for:

[0157] Based on the dynamic risk propagation map, the spatial adjacency relationship of the target risk point is traced back to generate a set of physical propagation paths;

[0158] Based on the knowledge graph of chemical system, the process logic dependency relationship of the target risk point is traced back to generate a set of functional failure chains.

[0159] By using a cross-level association matching algorithm, a spatiotemporal coupling analysis is performed on the set of physical propagation paths and the set of functional failure chains to generate a set of critical failure paths.

[0160] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0161] In one embodiment, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0162] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0163] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A chemical safety risk management method based on knowledge graphs, characterized in that, The method includes: Based on the spatial topology of preset process units, a dynamic risk propagation graph is constructed; based on the interaction relationships of preset process units, a chemical system knowledge graph is constructed; through real-time sensor data streams, the dynamic risk propagation graph and the chemical system knowledge graph are linked to generate a chemical system risk knowledge network. Based on the aforementioned chemical system risk knowledge network, a graph neural network algorithm is used to analyze the coupling effect and generate the risk transmission path weight distribution. Based on the weight distribution of the risk transmission path, a three-dimensional risk field rendering process is performed to generate a visual risk perspective interface. The visualized risk perspective interface allows for interactive risk tracing and verification, and outputs optimized protection measures.

2. The chemical safety risk management method based on knowledge graphs according to claim 1, characterized in that, The step of performing three-dimensional risk field rendering based on the risk transmission path weight distribution to generate a visualized risk perspective interface includes: The risk transmission path weight distribution is subjected to spatial field intensity mapping processing to generate a three-dimensional risk intensity field; Based on the aforementioned three-dimensional risk intensity field, dynamic particle trajectory simulation is performed to generate a dynamic risk diffusion model. The risk diffusion dynamic model is subjected to multi-view fusion rendering to generate the visualized risk perspective interface.

3. The chemical safety risk management method based on knowledge graphs according to claim 2, characterized in that, The process of performing multi-view fusion rendering on the dynamic risk diffusion model to generate the visualized risk perspective interface includes: The risk diffusion dynamic model is subjected to multi-dimensional data segmentation processing to generate a spatial distribution view, a temporal evolution view, and an intensity hierarchy view; The spatial distribution view is reconstructed using three-dimensional isosurfaces to generate a spatial risk distribution layer for the equipment. The time evolution view is subjected to dynamic particle flow rendering to generate a risk transmission trajectory layer; The intensity hierarchy view is subjected to thermal gradient coloring to generate a risk thermal field layer; The device space risk distribution layer, the risk transmission trajectory layer, and the risk thermal field layer are synchronously superimposed using a perspective adaptive fusion algorithm to generate the visualized risk perspective interface.

4. The chemical safety risk management method based on knowledge graphs according to claim 1, characterized in that, The interactive risk tracing and verification process, conducted through the visualized risk perspective interface, and the output of optimized protective measures include: Based on the risk point selection command triggered by the visual risk perspective interface, a target risk point identifier is generated. A multi-level causal chain backtracking analysis is performed on the identified target risk points to generate a set of critical failure paths; Virtual protection units are inserted at the nodes of the set of critical failure paths to simulate the propagation of the blocking effect and generate a protection effectiveness evaluation matrix. Based on the protection effectiveness evaluation matrix and preset economic parameters, multi-objective optimization decisions are made to generate the optimized protection measures scheme.

5. The chemical safety risk management method based on knowledge graphs according to claim 4, characterized in that, The step of performing multi-level causal chain backtracking analysis on the target risk point identifiers to generate a set of critical failure paths includes: Based on the dynamic risk propagation map, the spatial adjacency relationship of the target risk point is traced back to generate a set of physical propagation paths; Based on the chemical system knowledge graph, the process logic dependency relationship of the target risk point is traced back to generate a set of functional failure chains. The set of critical failure paths is generated by performing spatiotemporal coupling analysis on the set of physical propagation paths and the set of functional failure chains through a cross-level association matching algorithm.

6. The chemical safety risk management method based on knowledge graphs according to claim 4, characterized in that, The process of generating an optimized protection measure scheme based on the protection effectiveness evaluation matrix and preset economic parameters through multi-objective optimization decision-making includes: The protection effectiveness evaluation matrix is ​​decoupled according to the safety effectiveness dimension to generate a risk blocking rate vector and an accident consequence suppression rate vector. The preset economic parameters are subjected to full life cycle cost modeling to generate an economic impact vector that includes equipment purchase cost information, installation and commissioning cost information, and maintenance and operation cost information; The Pareto front analysis algorithm is used to perform multi-objective collaborative optimization on the risk blocking rate vector, the accident consequence inhibition rate vector, and the economic impact vector to generate a non-dominated solution set. Based on the preset security integrity level constraints, the non-dominated solution set is subjected to feasibility screening to generate the optimized protection measures.

7. The chemical safety risk management method based on knowledge graphs according to claim 2, characterized in that, The three-dimensional risk intensity field is obtained using the following formula, including: in, Represents the three-dimensional risk intensity field. Representing a path The source point weight, Representing a path Spatial length, Indicates the distance attenuation coefficient. Representing a path The initial risk value, and Both represent nonlinear mapping parameters. This represents the total number of risk transmission paths.

8. A chemical safety risk management and control device based on knowledge graphs, characterized in that, The device includes: The risk knowledge network construction module is used to construct a dynamic risk propagation graph based on the spatial topology of a preset process unit; construct a chemical system knowledge graph based on the interaction relationship of the preset process unit; and generate a chemical system risk knowledge network by associating the dynamic risk propagation graph with the chemical system knowledge graph through real-time sensor data streams. The risk coupling analysis module is used to analyze the coupling effect based on the risk knowledge network of the chemical system and generate the risk transmission path weight distribution by using a graph neural network algorithm. The risk field visualization module is used to perform three-dimensional risk field rendering based on the weight distribution of the risk transmission path and generate a visualized risk perspective interface. The risk tracing and optimization module is used to perform interactive risk tracing verification through the visual risk perspective interface and output optimized protection measures.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the chemical safety risk management method based on knowledge graphs as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the chemical safety risk management method based on knowledge graphs as described in any one of claims 1 to 7.

Images