Risk propagation analysis method and system based on power grid fault chain evolution simulation

Abstract

This invention provides a risk propagation analysis method and system based on power grid fault chain evolution simulation, belonging to the field of power grid fault analysis technology. The method includes: constructing a power grid digital twin model; injecting initial fault events into the power grid digital twin model; establishing an event propagation chain node set; establishing cross-layer fault triggering relationships; performing risk coupling and transmission analysis of the event propagation chain node set based on the cross-layer fault triggering relationships; establishing an event propagation chain; performing evolution simulation of the event propagation chain; generating a time-series fault propagation graph; calculating the propagation risk degree based on node consequence weights; establishing a fault chain risk degree identifier; and generating a risk propagation early warning. This solves the technical problem that existing technologies struggle to comprehensively and accurately reflect the dynamic propagation characteristics of power grid faults. It achieves the technical effect of comprehensively and accurately reflecting the dynamic propagation characteristics of power grid faults, improving the timeliness of fault diagnosis and the accuracy of risk assessment.

Description

Technical Field

[0001] This invention relates to the field of power grid fault analysis technology, and specifically to a risk propagation analysis method and system based on power grid fault chain evolution simulation. Background Technology

[0002] As the scale of power grid systems continues to expand, the safe and stable operation of power grids faces increasing uncertainties. The trend of power grid faults evolving from single-point sudden events to multi-node, multi-level, chain-like propagation is becoming increasingly apparent. When local equipment or lines malfunction, electrical parameter disturbances may propagate across different levels through the interaction of electrical topology, communication networks, and control systems, forming a fault chain. Existing power grid fault analysis methods are mostly based on single-level or static models, making it difficult to comprehensively and accurately reflect the dynamic propagation characteristics of power grid faults. This leads to incomplete fault diagnosis, inaccurate risk assessment, increased power grid operational risks, and reduced power grid reliability and stability.

[0003] Existing technologies have limitations in fully and accurately reflecting the dynamic propagation characteristics of power grid faults, leading to untimely fault diagnosis and inaccurate risk assessment. Summary of the Invention

[0004] The purpose of this application is to provide a risk propagation analysis method and system based on power grid fault chain evolution simulation, in order to solve the technical problem that existing technologies cannot fully and accurately reflect the dynamic propagation characteristics of power grid faults, resulting in untimely power grid fault diagnosis and inaccurate risk assessment.

[0005] In view of the above problems, this application provides a risk propagation analysis method and system based on power grid fault chain evolution simulation.

[0006] The first aspect of this application provides a risk propagation analysis method based on power grid fault chain evolution simulation. The method includes: constructing a power grid digital twin model, the power grid digital twin model including a physical layer twin, a communication layer twin, and a control layer twin synchronized with the actual power grid topology, operating state parameters, and control logic; injecting an initial fault event into the power grid digital twin model, establishing an event propagation chain node set, establishing cross-layer fault triggering relationships through a multi-layer propagation mapping matrix, performing risk coupling and transmission analysis of the event propagation chain node set based on the cross-layer fault triggering relationships, and establishing an event propagation chain; performing evolution simulation of the event propagation chain, calculating the dynamic triggering strength of nodes through a propagation kernel function, and generating a time-series fault propagation graph; calculating the propagation risk degree using the time-series fault propagation graph and node consequence weights, and establishing a fault chain risk degree identifier; and generating a risk propagation early warning based on the risk degree identifier and the time-series fault propagation graph.

[0007] Optionally, based on the multi-source fault feature vector set, a propagation correlation matrix of the multi-source fault vector set is established through spatiotemporal similarity calculation between events, power flow sensitivity analysis, and protection action logic reasoning; historical fault chain samples are obtained, and the propagation probability iterative analysis between nodes is performed using the historical fault chain samples and the propagation correlation matrix to establish probability weights; using the probability weights and the propagation delay distribution function, triggerable multi-level event node identification is performed within a preset time window to construct an event propagation chain node set.

[0008] Optionally, the propagation delay distribution function is used to dynamically model the temporal correlation of event nodes, and a time decay function for node triggering intensity is established to perform time-series alignment and interpolation completion of node state data in the physical layer twin, communication layer twin, and control layer twin, thereby establishing a node state sequence. Within the preset time window, the probability integral of the event node triggering intensity under the time decay function is performed based on the node state sequence to calculate a multi-level triggering threshold. The multi-level triggering threshold is used to perform triggering analysis of node triggering intensity to complete multi-level event node identification.

[0009] Optionally, node state evolution analysis is performed based on the event propagation chain. The node state change rate in the node state evolution analysis is constructed using a propagation kernel function. The propagation kernel function includes a time decay term, a topological coupling term, and a feedback correction term, which are used to characterize the temporal dependence of risk triggering between nodes and the multi-layer topological coupling effect. During the simulation period, the dynamic triggering intensity sequence of each node is calculated using the propagation kernel function, and a temporal risk influence matrix between nodes is constructed using clustering operations. Dynamic clustering analysis is performed using the temporal risk influence matrix to generate a temporal fault propagation map.

[0010] Optionally, the fitting residual of the propagation kernel function is calculated by obtaining the node state deviation during the simulation period; the attenuation coefficient of the time decay term and the coupling weight of the topology coupling term of the propagation kernel function are dynamically adjusted using the fitting residual to complete the dynamic update.

[0011] Optionally, based on the temporal fault propagation map, the temporal trigger intensity sequence of each node is extracted to establish a dynamic risk transmission matrix between nodes; the dynamic risk transmission matrix is ​​weighted and corrected using the node consequence weights to obtain a comprehensive risk transmission matrix that integrates node importance and consequence sensitivity; the cumulative risk contribution value of each node in the propagation chain is calculated based on the comprehensive risk transmission matrix, and a risk decay function is established according to the propagation path length and trigger frequency; the cumulative risk contribution value is weighted and integrated using the risk decay function to generate a node-level propagation risk degree, and a risk degree identifier for the fault chain is established based on the node-level propagation risk degree.

[0012] Optionally, the node-level propagation risk degree is clustered and reduced, and divided into multiple risk clustering intervals based on the correlation of risk transmission between nodes and the temporal evolution characteristics; for each risk clustering interval, the risk center value and risk diffusion coefficient are calculated, and a risk degree identifier for the fault chain is established.

[0013] Optionally, the risk level identifier and the time-series fault propagation map are used to identify the warning type and warning intensity, and a warning matching feature set is extracted; the warning signal matching feature set is used to match and identify the warning signal, and a matching and identification result is generated; risk propagation warning management is carried out based on the matching and identification result.

[0014] Optionally, the warning signal status is determined based on the risk level identifier and the matching identification result. The warning signal status includes a retained warning status and a reported warning status. Risk propagation warning management is performed based on the warning signal status determination result and the matching identification result.

[0015] A second aspect of this application provides a risk propagation analysis system based on power grid fault chain evolution simulation. The system includes: a digital twin model construction module for constructing a power grid digital twin model, the power grid digital twin model including a physical layer twin, a communication layer twin, and a control layer twin synchronized with the actual power grid topology, operating state parameters, and control logic; an event propagation chain establishment module for injecting initial fault events into the power grid digital twin model, establishing an event propagation chain node set, establishing cross-layer fault triggering relationships through a multi-layer propagation mapping matrix, performing risk coupling and transmission analysis of the event propagation chain node set based on the cross-layer fault triggering relationships, and establishing an event propagation chain; a fault propagation graph generation module for performing evolution simulation of the event propagation chain, calculating the dynamic triggering strength of nodes through a propagation kernel function, and generating a time-series fault propagation graph; a risk degree identification establishment module for calculating the propagation risk degree using the time-series fault propagation graph and node consequence weights, and establishing a risk degree identification for the fault chain; and a risk propagation early warning generation module for generating a risk propagation early warning based on the risk degree identification and the time-series fault propagation graph.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] The method provided in this application constructs a digital twin model of the power grid, which includes a physical layer twin, a communication layer twin, and a control layer twin synchronized with the actual power grid topology, operating state parameters, and control logic. An initial fault event is injected into the power grid digital twin model to establish an event propagation chain node set. Cross-layer fault triggering relationships are established through a multi-layer propagation mapping matrix. Risk coupling and transmission analysis of the event propagation chain node set is performed based on the cross-layer fault triggering relationships to establish the event propagation chain. Evolution simulation of the event propagation chain is executed, and the dynamic triggering strength of nodes is calculated using a propagation kernel function to generate a time-series fault propagation graph. The propagation risk degree is calculated using the time-series fault propagation graph and node consequence weights to establish a risk degree identifier for the fault chain. A risk propagation early warning is generated based on the risk degree identifier and the time-series fault propagation graph. This achieves the technical effect of comprehensively and accurately reflecting the dynamic propagation characteristics of power grid faults, improving the timeliness of fault diagnosis and the accuracy of risk assessment, thereby effectively reducing the technical risks of power grid operation.

[0018] The above description is merely an overview of the technical solution of this application. To enable a clearer understanding of the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application 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 merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 A flowchart illustrating the risk propagation analysis method based on power grid fault chain evolution simulation provided in this application.

[0021] Figure 2 A schematic diagram of the risk propagation analysis system based on power grid fault chain evolution simulation provided in this application.

[0022] Figure labeling: Digital twin model construction module 11, event propagation chain establishment module 12, fault propagation map generation module 13, risk level identification establishment module 14, risk propagation early warning generation module 15. Detailed Implementation

[0023] This application provides a risk propagation analysis method and system based on power grid fault chain evolution simulation. It addresses the technical problem that existing technologies struggle to comprehensively and accurately reflect the dynamic propagation characteristics of power grid faults, leading to untimely fault diagnosis and inaccurate risk assessment. The method achieves the technical effect of comprehensively and accurately reflecting the dynamic propagation characteristics of power grid faults, improving the timeliness of fault diagnosis and the accuracy of risk assessment, thereby effectively reducing power grid operational risks.

[0024] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0025] Example 1, as Figure 1 As shown, this application provides a risk propagation analysis method based on power grid fault chain evolution simulation. The risk propagation analysis method based on power grid fault chain evolution simulation includes:

[0026] A digital twin model of a power grid is constructed, which includes a physical layer twin, a communication layer twin, and a control layer twin that are synchronized with the actual power grid topology, operating status parameters, and control logic.

[0027] Specifically, a power grid digital twin model refers to a virtual model established in a computer simulation environment that is consistent with the physical entity structure, operating mechanism, and dynamic behavior of the power grid, achieving real-time mapping, dynamic interaction, and feedback optimization between the virtual space and the physical space of the power grid. The power grid digital twin model includes a physical layer twin, a communication layer twin, and a control layer twin. The physical layer twin synchronously corresponds to the actual power grid topology, operating status parameters, and control logic, reflecting electrical topology relationships and electrical states. Power system simulation software, such as PowerFactor, is used for modeling and simulation. Power grid topology information, including the connection relationships of power grid nodes, lines, transformers, buses, and circuit breakers, is acquired in real time through SCADA systems and EMS. Graph theory methods are used to represent the nodes and lines of the power grid as vertices and edges of a graph, or a topology data model based on the CIM standard is used to map the physical structure of the power grid into a graph data structure, constructing a dynamically updatable power grid topology twin structure.

[0028] Real-time measurement data such as voltage phase angle, frequency, current, voltage, and power factor are acquired through vector measurement units. Combined with monitoring data from the SCADA system, power grid operating status parameters are obtained. Using data synchronization technology, these parameters are used as input parameters to the power grid digital twin model, and transmitted in real-time to the model to construct the operating status model of the physical layer twin. For example, message queues and data buses are employed to ensure the real-time performance and accuracy of data transmission. Furthermore, the control logic of the physical layer twin is consistent with the protection status and specific control logic in the actual power grid, enabling real-time reflection and simulation of the operating status and behavior of the control logic in the power grid.

[0029] The communication layer twin reflects the topology, link characteristics, and information transmission of the power grid monitoring and control communication network. Topology information, including routing nodes, link delays, and bandwidth utilization, is obtained from the Network Management System (NMS) and the communication monitoring platform. A network simulation framework, such as OPNET or NS-3, is used to construct the communication layer twin, which is then node-level bound to the physical layer topology via a device mapping table. Within the communication layer twin, the transmission delay function and packet loss rate model of the communication links are defined to reflect the impact of power grid congestion, faults, and delays on control information transmission. Specifically, Poisson or exponential distribution communication delays can be used for timing simulation. The control layer twin simulates the logical behavior of power grid dispatching, protection, and control strategies. Protection settings, action logic, and control parameters are extracted from the Dispatch Automation System (DAS) and protection status configuration library. These extracted settings, action logic, and control parameters are used as inputs, and an event-driven simulation framework, such as DEVS modeling or SimEvents, is used to construct the control layer twin to describe the execution flow of dispatching commands and protection actions. By associating the control layer twin y with the physical layer twin in real time, and synchronizing the actual control actions and protection operation states to the power grid digital twin model through the control simulation engine, the real-time mapping of the control layer twin actions and the physical layer twin operation states is realized, enabling the power grid digital twin model to simultaneously reflect the electrical state of the power grid and the dynamic behavior affected by the control logic strategy.

[0030] Based on the CIM and IEC 61850 communication protocols, data interaction and collaborative work are ensured between the physical layer twin, communication layer twin, and control layer twin. The twins at each layer are merged to form a complete power grid digital twin. MQTT or Kafka message middleware is used as a real-time data stream channel to ensure data consistency among the twins at each layer.

[0031] The initial fault event is injected into the power grid digital twin model to establish an event propagation chain node set. Cross-layer fault triggering relationships are established through a multi-layer propagation mapping matrix. Based on the cross-layer fault triggering relationships, risk coupling transmission analysis of the event propagation chain node set is performed to establish the event propagation chain.

[0032] Furthermore, establishing an event propagation chain node set includes: using the initial fault event as the original matching feature, extracting associated node electrical parameters, control action signals, and communication topology data from the physical layer twin, communication layer twin, and control layer twin to establish a multi-source fault feature vector set; based on the multi-source fault feature vector set, establishing a propagation correlation matrix of the multi-source fault vector set through spatiotemporal similarity calculation between events, power flow sensitivity analysis, and protection action logic reasoning; acquiring historical fault chain samples, using the historical fault chain samples and the propagation correlation matrix to perform iterative analysis of propagation probability between nodes, and establishing probability weights; using the probability weights and the propagation delay distribution function, performing triggerable multi-level event node identification within a preset time window to construct the event propagation chain node set.

[0033] Specifically, the process involves obtaining initial fault events that actually occur in the power grid or are based on simulations. These initial fault events refer to the first abnormal event occurring in the power grid that can trigger subsequent fault propagation or cascading events. The initial fault events are used as initial matching features to identify and track fault propagation in the power grid digital twin model. The initial fault events include fault location, fault type, occurrence time, and related protection actions. Fault location refers to the specific location of the power grid component or node where the fault occurs, used to locate the fault element. Fault type refers to the fault category, such as single-phase ground fault, three-phase short circuit, equipment overload, etc. Related protection actions refer to the actions or control commands of protection equipment triggered after the fault occurs, such as circuit breaker tripping actions, bus protection relay actions, etc. The initial fault events are injected into the power grid digital twin model to simulate the operating state at the time of the initial fault event. Electrical parameters of the nodes associated with the fault are extracted from the physical layer twin, including but not limited to voltage, current, and power. Communication topology data, including but not limited to bandwidth and delay, is extracted from the communication layer twin. Control action information, including protection action signals and control commands, is extracted from the control layer twin. The extracted multi-source data are integrated to form a multi-source fault feature vector set. Each feature vector represents the state of a node, including electrical parameters, communication status, and control actions.

[0034] Based on the multi-source fault feature vector set, Euclidean distance, cosine similarity, or dynamic time warping algorithms are used to calculate the temporal and spatial similarity between different nodes, assessing the probability of fault propagation between different nodes. For example, using the Euclidean distance algorithm, the Euclidean distance between the feature vectors of node i and node j is calculated; the smaller the distance, the more similar the node features, and the higher the probability of fault propagation. Simultaneously, a linear power flow sensitivity matrix or Jacobian matrix is ​​used to perform power flow sensitivity analysis on the nodes, assessing the degree of response of the node's electrical state to faults in neighboring nodes. For example, the power flow sensitivity matrix is ​​used to characterize the sensitivity of node i's state to power changes in node j; the higher the sensitivity, the greater the impact of node i on the fault in node j, and the higher the propagation probability. Furthermore, based on a rule-based reasoning engine, such as Drools, the protection action logic is analyzed to infer the fault triggering path and possible multi-level actions. For example, if a short circuit occurs at node i, the protection relay will trigger the circuit breaker at node j to trip, forming a secondary event. The tripping of node j will then affect node k, triggering load shedding, forming a tertiary event. Based on spatiotemporal similarity, power flow sensitivity, and protection action logic, a propagation correlation matrix R=[r] is formed for the multi-source fault vector set between nodes. ij ], matrix element r ij It integrates spatiotemporal similarity, current sensitivity, and protection action logic reasoning to reflect the triggering probability or correlation strength between nodes. For example, a weighted calculation method is used to integrate spatiotemporal similarity, current sensitivity, and protection action logic reasoning, and the weight coefficients can be adjusted based on experience or historical samples.

[0035] Historical fault chain samples are collected from historical power grid operation records. These samples include data such as the node triggering sequence, triggering time, and triggering intensity at the time of a power grid fault. The historical fault chain samples are combined with the propagation correlation matrix, and propagation probability iterative analysis is performed using iterative algorithms, such as Markov chain iteration or Bayesian update iteration, to establish probability weights between nodes. These probability weights reflect the likelihood and strength of fault propagation between nodes. For example, based on the propagation correlation matrix, the propagation probability matrix between nodes is initialized. Based on the historical fault chain samples, the triggering probability of each node in the historical data is calculated. Based on the historical fault chain samples and the propagation correlation matrix, the conditional probability that node j will be triggered after node i is triggered is calculated. The propagation probability between nodes is updated according to the Bayesian update rule. This process is repeated until the change in the propagation probability matrix is ​​less than a preset threshold, at which point the iteration is considered converged. If the convergence condition is not met, iteration continues until the convergence condition is met or the maximum number of iterations is reached. After iterative convergence, the resulting propagation probability matrix is ​​used as the probability weights.

[0036] The node propagation delay function describes the time required for a fault to propagate from one node to another in a power grid. This propagation time is influenced by various factors, including communication delay, control action delay, and physical state response time. Delay data for communication links, including bandwidth, latency, and packet loss rate, is collected from the communication layer twin. Action delay data for protection devices and automated control systems, including protection action time and control command execution time, is collected from the control layer twin. Physical state response times of equipment, including switching action time and relay action time, are collected from the physical layer twin. Using exponential, Poisson, or normal distributions, and based on the characteristics of the communication link, the node propagation delay function is constructed by combining communication delay, control action delay, and physical state response time. Within a preset time window, node triggering analysis is performed based on probability weights and the delay function to identify multi-level event nodes. Based on these multi-level event nodes, an event propagation chain node set is constructed. This event propagation chain node set refers to the set of nodes that may be triggered in chronological order within the fault propagation chain, reflecting the hierarchical structure of power grid fault propagation.

[0037] Based on probability weights, the node triggering relationships of the physical layer twin, communication layer twin, and control layer twin are integrated to form a multi-layer propagation mapping matrix. Matrix elements in the multi-layer propagation mapping matrix represent the probability or intensity of cross-layer fault triggering between nodes. Probability thresholds are set based on expert experience or historical data, and the triggering relationships between node pairs in the matrix are determined according to these thresholds, establishing cross-layer fault triggering relationships. Then, risk coupling and transmission analysis is performed on the event propagation chain node set based on the cross-layer fault triggering relationships. Specifically, risk coupling and transmission analysis calculates the multi-level risk accumulation between nodes in the event propagation chain node set, quantifying the risk contribution of each node in each propagation path. By iteratively accumulating the risk of each node in the propagation chain, a complete event propagation chain is generated. The event propagation chain refers to the node sequence formed according to the fault triggering order and cross-layer relationships. By establishing the event propagation chain node set and the event propagation chain, multi-level event nodes that can be triggered by faults in the power grid can be comprehensively identified, thereby improving the accuracy of assessing the propagation of power grid risk faults.

[0038] Furthermore, using the probability weights and propagation delay distribution function, triggerable multi-level event node identification is performed within a preset time window. This includes: dynamically modeling the temporal correlation of event nodes using the propagation delay distribution function, establishing a time decay function for node triggering intensity, performing time-series alignment and interpolation completion of node state data in the physical layer twin, communication layer twin, and control layer twin, and establishing a node state sequence; within the preset time window, calculating the probability integral of the event node triggering intensity under the time decay function based on the node state sequence, and calculating a multi-level triggering threshold; and using the multi-level triggering threshold to perform triggering analysis of node triggering intensity to complete multi-level event node identification.

[0039] Specifically, the propagation delay distribution function is used to dynamically model the temporal correlation of event nodes. That is, based on the propagation delay distribution function, an exponential decay function or a Gaussian decay function is used to model the decay of node triggering intensity over time, establishing a time decay function for node triggering intensity. This time decay function refers to a function that gradually weakens the fault triggering intensity over time. For example, based on the propagation delay, the node triggering intensity is modeled to decay over time, establishing a time decay function as follows: , among which, I i (t) represents the trigger strength of node i at time t, I0 is the initial trigger strength, and λ is the time decay coefficient, set according to the power grid topology and equipment response characteristics. Based on the established time decay function, the node state data of the physical layer, communication layer, and control layer are aligned in chronological order to ensure temporal consistency of data across different layers. For missing or discontinuous data points, interpolation methods such as linear interpolation and spline interpolation are used to complete the data, ensuring the continuity and accuracy of the node state sequence. The processed node state data are then aligned in chronological order to form a node state sequence, including electrical parameters, communication status, and control signals.

[0040] A preset time window is set, which refers to the time interval used to evaluate the triggering intensity of nodes in fault propagation analysis. This preset time window is set according to the characteristics of fault propagation and actual needs, for example, 5 seconds or 10 seconds. Within the preset time window, the probability integral of the triggering intensity of event nodes in the node state sequence under the time decay function is performed. The integral result reflects the cumulative triggering probability of the node within the preset time window. Based on the probability integral result, the cumulative triggering intensity of the node within the preset time window is obtained. The triggering intensity is divided into multiple level thresholds, where nodes below the minimum threshold are determined to be untriggered, nodes within the threshold range are determined to be triggered at the corresponding level, and nodes above the maximum threshold are determined to be triggered at a higher level. This is used to determine whether a node has been triggered. Using these multi-level triggering thresholds, the triggering intensity of each node is analyzed, that is, multiple triggering thresholds are compared with the triggering intensity of each node to determine whether the node has been triggered and its triggering level. This completes the multi-level event node identification, and the nodes determined to be triggered are added to the event propagation chain node set.

[0041] By utilizing time decay functions and propagation delay distribution functions, the propagation process of power grid faults between nodes can be dynamically analyzed, improving the accuracy and real-time performance of the analysis. Through the calculation and trigger analysis of multi-level trigger thresholds, multiple event nodes in the fault propagation process can be identified. Time sequence alignment and interpolation completion ensure the temporal consistency and integrity of node state data at different levels. The identification of multi-level event nodes enhances the comprehensiveness and accuracy of power grid fault risk assessment, thereby improving the reliability and safety of power grid operation.

[0042] The evolution simulation of the event propagation chain is performed, and the dynamic triggering strength of the nodes is calculated through the propagation kernel function to generate a time-series fault propagation map.

[0043] Furthermore, the evolution simulation of the event propagation chain is performed, and the dynamic triggering intensity of nodes is calculated through the propagation kernel function to generate a time-series fault propagation map. This includes: performing node state evolution analysis based on the event propagation chain, wherein the node state change rate in the node state evolution analysis is constructed through the propagation kernel function; the propagation kernel function includes a time decay term, a topological coupling term, and a feedback correction term, which are used to characterize the time-series dependence of risk triggering between nodes and the multi-layer topological coupling effect; within the simulation period, the dynamic triggering intensity sequence of each node is calculated through the propagation kernel function, and a time-series risk influence matrix between nodes is constructed using clustering operations; dynamic clustering analysis is performed using the time-series risk influence matrix to generate a time-series fault propagation map.

[0044] Specifically, based on actual needs and historical data, a reasonable simulation period is set. The simulation period refers to the time interval used to evaluate node state changes in the evolutionary simulation. For example, it is set to 5 minutes. Within the simulation period, node state evolution simulation analysis is performed based on the event propagation chain. Node states include physical layer electrical parameters, communication link states, and control layer protection action states, and a continuous node state sequence is formed through time alignment and interpolation. The node state change rate in the node state evolution analysis is constructed using a propagation kernel function. The propagation kernel function includes a time decay term, a topology coupling term, and a feedback correction term, used to characterize the time-series dependence of risk triggering between nodes and the multi-layer topology coupling effect. The time decay term is a time decay function, characterizing that the node triggering intensity gradually weakens with the event. The topology coupling term is β∑w. ij I j (t) characterizes the influence of adjacent node states on node triggering strength, reflecting the multi-layer topological coupling effect, and the feedback correction term is γF. i (t) is used to correct the node trigger strength based on real-time feedback data from the digital twin model. The dynamic trigger strength sequence of each node is calculated using a propagation kernel function to achieve dynamic simulation of node state changes. Euclidean distance, dynamic time warping, or cosine similarity are used to evaluate the similarity of trigger strengths among different nodes, analyzing the similarity between nodes and the risk transmission relationship, and generating a time-series risk impact matrix. Then, dynamic clustering algorithms, such as K-means clustering and hierarchical clustering, are used to perform dynamic clustering analysis on the time-series risk impact matrix, identifying node groups with similar behaviors. Based on the dynamic clustering analysis results, the dynamic process of fault propagation is displayed using time series plots, heatmaps, etc., generating a time-series fault propagation map. This fault propagation map is a visual representation of node state changes within the simulation period.

[0045] By calculating the dynamic trigger strength sequence using a propagation kernel function, node state changes can be dynamically simulated, improving the accuracy and real-time performance of fault propagation analysis. Through time-series risk impact matrices and dynamic clustering analysis, groups of nodes with similar behaviors can be identified, generating fault propagation maps. This provides a clear visual representation of the dynamic process of power fault propagation, enabling rapid identification of potential grid risks and improving grid operational stability and security.

[0046] Furthermore, the parameters of the propagation kernel function are dynamically updated based on the feedback data of the digital twin model, including: obtaining the node state deviation during the simulation period to calculate the fitting residual of the propagation kernel function; and using the fitting residual to dynamically adjust the decay coefficient of the time decay term and the coupling weight of the topology coupling term of the propagation kernel function to complete the dynamic update.

[0047] Specifically, the parameters of the propagation kernel function are dynamically updated based on feedback data from the digital twin model. The actual states of each node within the simulation period are obtained in real-time from the digital twin model, including physical layer electrical parameters, communication link states, and control layer action states. Correspondingly, a dynamic trigger intensity sequence, i.e., the simulation prediction state, is obtained based on the propagation kernel function. The node state deviation between the actual state and the simulated prediction state of each node is calculated; this deviation reflects the accuracy of the simulation prediction. Statistical analysis is performed on all node state deviations within the simulation period, and the fitting residual of the propagation kernel function is calculated using mean squared error or mean absolute error. The fitting residual is used to quantify the magnitude of the fitting error of the propagation kernel function. Based on the fitting residual, an optimization algorithm is used to dynamically adjust the decay coefficient of the time decay term and the coupling weight of the topology coupling term. For example, a gradient descent optimization algorithm is used to dynamically optimize the parameters of the propagation kernel function based on the magnitude of the fitting residual. The sensitivity statement of the fitting residual to each parameter is calculated, i.e., the direction and magnitude of the influence of parameter changes on the residual. The parameters are gradually adjusted in this direction to gradually reduce the residual. In each iteration, the adjustment magnitude of the parameters is controlled by the learning rate to ensure stable adjustment and avoid over-correction. Through continuous iterative updates, the decay coefficient of the time decay term and the coupling weight of the topology coupling term are gradually optimized until convergence or a preset number of iterations are reached. This yields optimized decay coefficients for the time decay term and coupling weights for the topology coupling term, making the node dynamic triggering intensity output by the propagation kernel function closer to the feedback data from the digital twin model. The adjusted parameters are then used to update the propagation kernel function, and the node state change rate is reconstructed using the updated kernel function. The dynamic triggering intensity sequence for each node is then calculated, improving the accuracy of the time-series fault propagation map and further enhancing the accuracy and reliability of power grid fault propagation analysis.

[0048] The propagation risk degree is calculated using the time-series fault propagation graph and node consequence weights, and a risk degree identifier for the fault chain is established.

[0049] Furthermore, the propagation risk degree is calculated using the time-series fault propagation graph and node consequence weights, and a risk degree identifier for the fault chain is established. This includes: extracting the time-series trigger intensity sequence of each node based on the time-series fault propagation graph, and establishing a dynamic risk transmission matrix between nodes; using the node consequence weights to perform weighted correction on the dynamic risk transmission matrix to obtain a comprehensive risk transmission matrix that integrates node importance and consequence sensitivity; calculating the cumulative risk contribution value of each node in the propagation chain based on the comprehensive risk transmission matrix, and establishing a risk attenuation function according to the length of the propagation path and the trigger frequency; using the risk attenuation function to perform weighted integration on the cumulative risk contribution value to generate a node-level propagation risk degree, and establishing a risk degree identifier for the fault chain based on the node-level propagation risk degree.

[0050] Specifically, the temporal trigger intensity sequence of each node is extracted from the temporal fault propagation map. This sequence reflects the trigger intensity of the node at different times. Based on the extracted temporal trigger intensity sequences of each node, the dynamic risk transmission relationship between nodes is calculated. Using matrix operations or graph network tools such as NetworkX, the temporal risk correlation between nodes is quantified, and a dynamic risk transmission matrix between nodes is established. The matrix element r... ij(t) represents the degree of risk impact of node i on node j at time t. The consequence weights of nodes are determined based on their electrical parameters, equipment type, and historical fault data. Specifically, weight coefficients are assigned to each feature type according to expert rules or data-driven approaches, such as 0.3 for equipment type, 0.25 for electrical parameters, 0.25 for critical loads, and 0.2 for historical losses. Then, a weighted sum is calculated for each node to obtain its consequence weight, characterizing the node's importance and consequence sensitivity. The more important the node, the higher its consequence sensitivity, and the greater its consequence weight. The dynamic risk transmission matrix is ​​weighted according to the node consequence weights to generate a comprehensive risk transmission matrix that integrates node importance and consequence sensitivity. The matrix elements in the comprehensive risk transmission matrix comprehensively reflect the contribution of the triggering strength between nodes and the node's importance to risk propagation. Based on the comprehensive risk transmission matrix, the cumulative risk contribution value of each node in the propagation chain is calculated. According to the propagation path length and trigger frequency, the path contribution is reasonably attenuated / amplified based on the path length and trigger frequency to avoid unlimited propagation and reflect the reality that remote or sporadic triggers have lower contributions. A risk attenuation function is established, which adjusts the node contribution weight based on the propagation path length or trigger frequency, making the risk assessment more consistent with actual propagation characteristics. The cumulative risk contribution value is weighted using the risk attenuation function to generate a node-level propagation risk degree. This node-level propagation risk degree is then normalized or classified to generate a risk degree identifier for each node, used to quantify the node's risk importance in the entire fault chain. For example, K-means or hierarchical clustering can be used to classify node-level propagation risk degrees, identifying high, medium, and low-risk nodes and assigning risk degree identifiers. By establishing node-level risk degree identifiers, the risk level of each node in the fault chain can be visually displayed, providing decision support for power grid operation and maintenance, and optimizing resource allocation and fault handling strategies.

[0051] Furthermore, establishing a risk level identifier for the fault chain based on the node-level propagation risk level includes: clustering and reducing the node-level propagation risk level, dividing it into multiple risk clustering intervals based on the correlation of risk transmission between nodes and the temporal evolution characteristics; calculating the risk center value and risk diffusion coefficient for each risk clustering interval, and establishing a risk level identifier for the fault chain.

[0052] Specifically, based on the correlation and temporal evolution characteristics of risk transmission among nodes, cluster feature vectors are determined, including the cumulative risk contribution value, trigger frequency, and path length of each node. Clustering algorithms, such as K-means and DBSCAN, are used to perform structured reduction and cluster analysis on the node-level propagation risk, grouping nodes with similar risk characteristics into the same risk cluster interval, resulting in multiple risk cluster intervals. After obtaining multiple risk cluster intervals, the risk center value of each risk cluster interval is calculated by averaging the values ​​of each interval. The risk center value reflects the overall risk level of each cluster interval. Simultaneously, the standard deviation of each risk cluster interval is calculated to obtain the risk diffusion coefficient, which represents the fluctuation or dispersion of node risk within that interval. By integrating the risk center values ​​and risk diffusion coefficients of multiple risk clustering intervals, a risk level identifier for power grid fault chains is formed. This can quantify the risk characteristics of each risk clustering interval, comprehensively and accurately assess the risk of power grid fault chains, provide specific numerical values ​​for fault chain risk assessment, and thus provide decision support for power grid operation and maintenance, optimize resource allocation and fault handling strategies, and improve the stability and security of power grid operation.

[0053] A risk propagation warning is generated based on the risk level identifier and the time-series fault propagation map.

[0054] Furthermore, generating a risk propagation warning based on the risk level identifier and the time-series fault propagation map includes: using the risk level identifier and the time-series fault propagation map to identify the warning type and warning intensity, and extracting a warning matching feature set; using the warning matching feature set to match and identify warning signals, and generating a matching identification result; and managing the risk propagation warning based on the matching identification result.

[0055] Specifically, based on the risk center value and risk diffusion coefficient in the risk level identifier, the early warning intensity is calculated, and risks are divided into different levels, such as low risk, medium risk, and high risk. Based on the propagation path and trigger intensity in the time-series fault propagation map, the specific power grid early warning type is determined, such as localized transmission risk and cascading trigger risk. After identifying the risk propagation early warning type and intensity, an early warning matching feature set is extracted. This feature set includes multi-dimensional parameters such as node risk intensity, propagation path length, node trigger frequency, risk propagation direction, propagation delay, topology information, and node importance, used to comprehensively characterize the risk propagation status. The current power grid operating status parameters, such as voltage, current, power, communication delay, and control signal status, are compared and analyzed with the early warning matching feature set to generate a matching identification result. This result includes information such as early warning type and early warning intensity. Matching analysis can be implemented using various methods, such as Euclidean distance, cosine similarity, or dynamic time warping algorithms, to calculate the similarity between the current node state sequence and the early warning matching feature set. Alternatively, risk triggering rules can be defined using the early warning matching feature set to determine whether an early warning threshold has been reached, thus achieving risk propagation early warning matching. After obtaining the matching identification results, risk propagation early warning management is carried out based on the type and intensity of the early warning signal in the matching identification results. This ensures that early warning signals are issued in a timely and accurate manner, helping operators take measures to reduce the risk of cascading failures or large-scale power outages.

[0056] Furthermore, risk propagation early warning management based on the matching and identification results also includes: determining the early warning signal status based on the risk level identifier, wherein the early warning signal status includes a retained early warning status and an issued early warning status; and performing risk propagation early warning management based on the early warning signal status determination results and the matching and identification results.

[0057] Specifically, based on the risk center value and risk diffusion coefficient in the risk level identifier, a threshold is set, and the status of the early warning signal matching the identification results is analyzed to determine whether the risk level of the current node meets the preset early warning conditions. When the node's risk level is near the threshold but does not exceed it, it is determined to be in a reserved early warning state, indicating that the node has potential risks but has not yet met the criteria for triggering an external early warning, and it is continuously monitored and dynamically tracked. When the node's risk level exceeds the threshold of the corresponding risk clustering interval, and the risk propagation trend is accelerating or spreading, it is determined to be in an early warning state. Then, risk propagation early warning management is carried out based on the early warning signal status judgment results and matching identification results, including continuous monitoring and updating of the reserved early warning state, and graded processing and response control of the reported early warning signals. For the reported early warning state, corresponding prevention and control measures are triggered according to the fault chain propagation path and risk level, such as communication link isolation, protection device verification, or emergency dispatch strategy execution. This enables the power grid to accurately and comprehensively identify high-risk propagation trends and take proactive defense measures in the early stages of fault chain evolution, improve the accuracy of power grid risk prevention and control and real-time response capabilities, thereby effectively reducing power grid operation risks and improving power grid operation stability and security.

[0058] Example 2, based on the same inventive concept as the risk propagation analysis method based on power grid fault chain evolution simulation in the previous examples, such as... Figure 2 As shown, this application provides a risk propagation analysis system based on power grid fault chain evolution simulation, wherein the risk propagation analysis system based on power grid fault chain evolution simulation includes:

[0059] The digital twin model construction module 11 is used to construct a power grid digital twin model, which includes a physical layer twin, a communication layer twin, and a control layer twin synchronized with the actual power grid topology, operating status parameters, and control logic. The event propagation chain establishment module 12 is used to inject initial fault events into the power grid digital twin model, establish an event propagation chain node set, establish cross-layer fault triggering relationships through a multi-layer propagation mapping matrix, perform risk coupling and transmission analysis of the event propagation chain node set based on the cross-layer fault triggering relationships, and establish the event propagation chain. The fault propagation graph generation module 13 is used to perform evolution simulation of the event propagation chain, calculate the dynamic triggering strength of nodes through a propagation kernel function, and generate a time-series fault propagation graph. The risk degree identification establishment module 14 is used to calculate the propagation risk degree using the time-series fault propagation graph and node consequence weights, and establish a risk degree identification for the fault chain. The risk propagation early warning generation module 15 is used to generate a risk propagation early warning based on the risk degree identification and the time-series fault propagation graph.

[0060] Furthermore, the event propagation chain establishment module 12 is also used to: use the initial fault event as the original matching feature to extract associated node electrical parameters, control action signals, and communication topology data from the physical layer twin, communication layer twin, and control layer twin to establish a multi-source fault feature vector set; based on the multi-source fault feature vector set, establish a propagation correlation matrix of the multi-source fault vector set through spatiotemporal similarity calculation between events, power flow sensitivity analysis, and protection action logic reasoning; obtain historical fault chain samples, use the historical fault chain samples and the propagation correlation matrix to perform iterative analysis of propagation probability between nodes, and establish probability weights; use the probability weights and propagation delay distribution function to perform triggerable multi-level event node identification within a preset time window, and construct an event propagation chain node set.

[0061] Furthermore, the event propagation chain establishment module 12 is also used to: dynamically model the time correlation of event nodes using the propagation delay distribution function, establish a time decay function for node triggering intensity, perform time alignment and interpolation completion of node state data of physical layer twin, communication layer twin, and control layer twin, and establish a node state sequence; within the preset time window, perform probability integral of event node triggering intensity under the time decay function based on the node state sequence, and calculate multi-level triggering thresholds; use the multi-level triggering thresholds to perform triggering analysis of node triggering intensity, so as to complete multi-level event node identification.

[0062] Furthermore, the fault propagation map generation module 13 is also used for: performing node state evolution analysis based on the event propagation chain, wherein the node state change rate in the node state evolution analysis is constructed through a propagation kernel function; the propagation kernel function includes a time decay term, a topological coupling term, and a feedback correction term, which are used to characterize the temporal dependence and multi-layer topological coupling effect of risk triggering between nodes; within the simulation period, the dynamic trigger intensity sequence of each node is calculated through the propagation kernel function, and a temporal risk influence matrix between nodes is constructed using clustering operations; dynamic clustering analysis is performed using the temporal risk influence matrix to generate a temporal fault propagation map.

[0063] Furthermore, the fault propagation map generation module 13 is also used to: obtain the node state deviation of the simulation cycle to calculate the fitting residual of the propagation kernel function; and use the fitting residual to dynamically adjust the decay coefficient of the time decay term and the coupling weight of the topology coupling term of the propagation kernel function to complete the dynamic update.

[0064] Furthermore, the risk level identification module 14 is also used to: extract the temporal trigger intensity sequence of each node based on the temporal fault propagation map, and establish a dynamic risk transmission matrix between nodes; use the node consequence weight to perform weighted correction of the dynamic risk transmission matrix to obtain a comprehensive risk transmission matrix that integrates node importance and consequence sensitivity; calculate the cumulative risk contribution value of each node in the propagation chain based on the comprehensive risk transmission matrix, and establish a risk decay function according to the length of the propagation path and the trigger frequency; use the risk decay function to perform weighted integration on the cumulative risk contribution value to generate a node-level propagation risk level, and establish a risk level identification of the fault chain based on the node-level propagation risk level.

[0065] Furthermore, the risk identification module 14 is also used to: cluster and reduce the node-level propagation risk, and divide it into multiple risk clustering intervals based on the correlation of risk transmission between nodes and the temporal evolution characteristics; calculate the risk center value and risk diffusion coefficient for each risk clustering interval, and establish the risk identification of the fault chain.

[0066] Furthermore, the risk propagation early warning generation module 15 is also used to: identify the early warning type and early warning intensity using the risk level identifier and the time-series fault propagation map, and extract the early warning matching feature set; use the early warning matching feature set to perform early warning signal matching and identification, and generate matching and identification results; and perform risk propagation early warning management based on the matching and identification results.

[0067] Furthermore, the risk propagation early warning generation module 15 is also used to: determine the early warning signal status based on the risk level identifier and the matching identification result, wherein the early warning signal status includes a retained early warning status and a reported early warning status; and perform risk propagation early warning management based on the early warning signal status determination result and the matching identification result.

[0068] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. 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 this application. Therefore, this application 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.

[0069] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A risk propagation analysis method based on power grid fault chain evolution simulation, characterized in that, The method includes: A digital twin model of a power grid is constructed, comprising a physical layer twin, a communication layer twin, and a control layer twin that are synchronized with the actual power grid topology, operating status parameters, and control logic. The initial fault event is injected into the power grid digital twin model to establish an event propagation chain node set. Cross-layer fault triggering relationships are established through a multi-layer propagation mapping matrix. Based on the cross-layer fault triggering relationships, risk coupling transmission analysis of the event propagation chain node set is performed to establish the event propagation chain. The evolution simulation of the event propagation chain is performed, and the dynamic triggering strength of the nodes is calculated through the propagation kernel function to generate a time-series fault propagation map; Using the aforementioned temporal fault propagation graph and node consequence weights, the propagation risk degree is calculated, and a risk degree identifier for the fault chain is established, including: Based on the aforementioned time-series fault propagation map, the time-series trigger intensity sequence of each node is extracted, and a dynamic risk transmission matrix between nodes is established. The dynamic risk transmission matrix is ​​dynamically weighted and corrected using the node consequence weights to obtain a comprehensive risk transmission matrix that integrates node importance and consequence sensitivity. The cumulative risk contribution value of each node in the propagation chain is calculated based on the comprehensive risk transmission matrix, and a risk attenuation function is established according to the length of the propagation path and the triggering frequency. The cumulative risk contribution value is weighted and integrated using the risk decay function to generate a node-level propagation risk degree, and a risk degree identifier for the fault chain is established based on the node-level propagation risk degree. A risk propagation warning is generated based on the risk level identifier and the time-series fault propagation map. The risk level identifier for establishing the fault chain based on the node-level propagation risk level includes: The node-level propagation risk is clustered and reduced, and divided into multiple risk clustering intervals based on the correlation of risk transmission between nodes and the temporal evolution characteristics; For each risk cluster interval, calculate the risk center value and risk diffusion coefficient, and establish the risk level identifier of the failure chain.

2. The risk propagation analysis method based on power grid fault chain evolution simulation as described in claim 1, characterized in that, Establish a set of event propagation chain nodes, including: Using the initial fault event as the original matching feature, related node electrical parameters, control action signals and communication topology data are extracted from the physical layer twin, communication layer twin and control layer twin to establish a multi-source fault feature vector set; Based on the multi-source fault feature vector set, a propagation correlation matrix of the multi-source fault vector set is established through spatiotemporal similarity calculation between events, power flow sensitivity analysis, and protection action logic reasoning. Obtain historical failure chain samples, and use the historical failure chain samples and propagation correlation matrix to perform iterative analysis of propagation probability between nodes, and establish probability weights; Using the probability weights and propagation delay distribution function, triggerable multi-level event node identification is performed within a preset time window to construct an event propagation chain node set.

3. The risk propagation analysis method based on power grid fault chain evolution simulation as described in claim 2, characterized in that, Using the aforementioned probability weights and propagation delay distribution function, triggerable multi-level event node identification is performed within a preset time window, including: The propagation delay distribution function is used to dynamically model the time correlation of event nodes, and a time decay function of node triggering intensity is established to perform time sequence alignment and interpolation of node state data of physical layer twin, communication layer twin and control layer twin, and establish node state sequence; Within the preset time window, the event node trigger intensity under the time decay function is integrally calculated based on the node state sequence to determine the multi-level trigger threshold. The multi-level trigger thresholds are used to perform trigger strength analysis on nodes in order to complete the identification of multi-level event nodes.

4. The risk propagation analysis method based on power grid fault chain evolution simulation as described in claim 1, characterized in that, Perform an evolution simulation of the event propagation chain, calculate the dynamic triggering strength of nodes using a propagation kernel function, and generate a time-series fault propagation graph, including: The node state evolution analysis is performed based on the event propagation chain, and the node state change rate in the node state evolution analysis is constructed by the propagation kernel function; The propagation kernel function includes a time decay term, a topology coupling term, and a feedback correction term, which are used to characterize the temporal dependence of risk triggering between nodes and the multi-layer topology coupling effect. During the simulation period, the dynamic triggering intensity sequence of each node is calculated through the propagation kernel function, and the temporal risk influence matrix between nodes is constructed using clustering operations. Dynamic clustering analysis is performed using the aforementioned time-series risk impact matrix to generate a time-series fault propagation map.

5. The risk propagation analysis method based on power grid fault chain evolution simulation as described in claim 4, characterized in that, The parameters of the propagation kernel function are dynamically updated based on feedback data from the digital twin model, including: Obtain the node state deviation during the simulation cycle and calculate the fitting residual of the propagation kernel function; The decay coefficient of the time decay term and the coupling weight of the topological coupling term of the propagation kernel function are dynamically adjusted using the fitted residuals to complete the dynamic update.

6. The risk propagation analysis method based on power grid fault chain evolution simulation as described in claim 1, characterized in that, Based on the risk level identifier and the time-series fault propagation map, a risk propagation warning is generated, including: The risk level identifier and the time-series fault propagation map are used to identify the warning type and warning intensity, and to extract the warning matching feature set. The warning signal matching feature set is used to perform warning signal matching and identification, and a matching and identification result is generated. Risk propagation early warning management is carried out based on the matching and identification results.

7. The risk propagation analysis method based on power grid fault chain evolution simulation as described in claim 6, characterized in that, Risk propagation early warning management based on the matching and identification results also includes: The warning signal status is determined based on the risk level identifier and the matching identification result. The warning signal status includes a retained warning status and a reported warning status. Risk propagation early warning management is carried out based on the status judgment result of the early warning signal and the matching identification result.

8. A risk propagation analysis system based on power grid fault chain evolution simulation, characterized in that, The steps for implementing the risk propagation analysis method based on power grid fault chain evolution simulation as described in any one of claims 1 to 7 include: A digital twin model construction module is used to construct a power grid digital twin model, which includes a physical layer twin, a communication layer twin, and a control layer twin that are synchronized with the actual power grid topology, operating status parameters, and control logic. The event propagation chain establishment module is used to inject the initial fault event into the power grid digital twin model, establish an event propagation chain node set, establish cross-layer fault triggering relationships through a multi-layer propagation mapping matrix, perform risk coupling transmission analysis of the event propagation chain node set based on the cross-layer fault triggering relationships, and establish the event propagation chain. The fault propagation graph generation module is used to perform evolution simulation of the event propagation chain, calculate the dynamic triggering strength of nodes through the propagation kernel function, and generate a time-series fault propagation graph. The risk identification module is used to calculate the propagation risk using the time-series fault propagation graph and node consequence weights, and to establish the risk identification of the fault chain. The risk propagation warning generation module is used to generate risk propagation warnings based on the risk level identifier and the time-series fault propagation map.

Images