A command and control system safety risk assessment method based on linear effects

Abstract

The application discloses a kind of command and control system safety risk assessment methods based on linear influence.First, the physical connection diagram and information flow diagram of command and control system are established, and the bottom map is aggregated to form the bottom map of causal diagram in layers;Non-causal relationship in the bottom map is eliminated from the opposite direction of the form of spanning tree to the cause transmission;An index abstraction system is established, and the resource nodes in the causal diagram are decomposed into index nodes;Input system operation raw data is normalized and data interpolation processed;The linear relationship of abnormal node to each path on the subnode is calculated by least square estimation;After traversing all paths of abnormal node, the influence degree of abnormal node on each path is calculated;The influence degree of abnormal node on subsequent nodes, influence domain range and alarm level are output.The method can effectively calculate the influence result of abnormal node under different risk scenarios, improve the reliability and safety of command and control system.

Description

Technical Field

[0001] This invention relates to the field of command and control system security risk assessment technology, specifically, a command and control system security risk assessment method based on linear effects. Background Technology

[0002] With the rapid development of robotics technology, robot swarms are increasingly widely used in various fields, such as industrial production, logistics and distribution, and environmental monitoring. However, the complexity of robot swarm collaboration and management is also increasing, especially when the number of robots is large and the task environment is complex. The stability and security of the robot swarm command and control system become key factors determining the effective operation of the system. Existing robot swarm command and control systems generally adopt a multi-level, multi-node structure design, encompassing different functional modules and robot units. These modules and units collaborate through complex communication networks. In this process, due to the openness, distribution, and interactivity of the system, it is susceptible to various external or internal risk factors, such as communication failures, hardware damage, software vulnerabilities, environmental interference, and human error.

[0003] To mitigate potential risks, reduce system defects, and enhance system robustness, safety risk assessment has become a crucial method, significantly impacting the reliability of robot swarm command and control systems in future automated tasks. Current technologies for safety risk assessment of robot swarm command and control systems largely rely on qualitative analysis and empirical judgment, lacking precise means to quantify risks. Furthermore, existing assessment systems often neglect the dynamic interactions between internal system components and real-time response capabilities under abnormal conditions. This makes it difficult to accurately predict and assess the potential impact of system interference or malfunctions in practical applications, thereby affecting the scientific nature of command decisions and the targeted nature of system optimization. Summary of the Invention

[0004] To address the shortcomings of the aforementioned technologies, this invention provides a safety risk assessment method for command and control systems based on linear influence. By constructing a dynamic task influence estimation model, it can perform real-time safety risk assessments of robot swarm command and control systems under different hazardous scenarios, providing a more accurate and real-time risk assessment tool for robot swarm command and control systems.

[0005] The technical solution of this invention to solve the technical problem is: a method for safety risk assessment of command and control systems based on linear effects, comprising the following steps:

[0006] S1: Establish the physical connection diagram and information flow diagram of the command and control system, and perform cross-layer aggregation of the base diagram to form a cause-effect graph base diagram;

[0007] S2: Eliminate non-causal relationships in the base graph by using a spanning tree to obtain a causal graph from the opposite direction of the motivation transmission;

[0008] S3: Establish an indicator extraction system to decompose resource nodes in the cause-effect graph into indicator nodes;

[0009] S4: Perform input normalization processing on the raw data of the command and control system.

[0010] S5: Perform data interpolation on the data processed in step S4;

[0011] S6: Calculate the linear relationship from the outlier node to the child nodes on each path using least squares estimation;

[0012] S7: Traverse all paths following the abnormal node and calculate the degree of impact of the abnormal node on each path;

[0013] S8: Based on the calculation results of step S7, output the degree of influence of the abnormal node on subsequent nodes and the range of influence domain to obtain the alarm level of the command and control system.

[0014] Further, in step S1, the step of establishing the physical connection diagram includes: identifying each subsystem in the command and control system and treating it as a node in the diagram; identifying the physical connection medium between subsystems, determining the physical connection relationship between each subsystem, and representing it in the diagram through the connection of edges between nodes; and determining the direction of the edges in the diagram according to the communication direction between each subsystem.

[0015] The steps for creating an information flow diagram include: identifying each subsystem in the command and control system and treating it as a node in the diagram; analyzing the data flow between subsystems, determining the propagation path of data from one subsystem to another, and representing it in the diagram through the connection of edges between nodes; and determining the direction of the edges in the diagram based on the data propagation flow between subsystems.

[0016] The implementation steps for cross-layer aggregation of the base graph include: mapping all subsystem nodes in the physical connection graph and the information flow graph; merging the physical connection graph and the information flow graph, retaining the connection relationships and directions between nodes in both graphs to obtain the causal graph base graph.

[0017] Further, in step S2, in the causal graph base graph generated in step S1, nodes that will not be affected are eliminated from the opposite direction of the cause propagation in the form of a spanning tree. First, an abnormal node set is generated, which includes abnormal nodes marked due to faults, errors, or abnormal behavior. Nodes in the set are selected sequentially as the current processing node. If the node is not in the taboo list, a set of neighboring resource nodes directly connected to it is generated, and a neighboring resource node is selected from it. It is checked whether it is the parent node of the current node. If it is, it is added as a neighboring node and a connection is established. If not, other neighboring resource nodes are selected until all neighboring nodes have been traversed. This process is repeated until all abnormal nodes have been processed. Finally, the processed causal graph is output.

[0018] Further, in step S3, index extraction involves decomposing the resource nodes in the causal graph into quantitatively describable index nodes. Based on the specific resource node, state parameters characterizing the node's performance are selected as the extracted index nodes, establishing an index extraction system. This system consists of two parts: intra-resource index extraction and inter-resource index extraction. Intra-resource index extraction includes packet output density, memory utilization, processing latency, and other indicators that characterize the internal state parameters of the resource. Inter-resource index extraction includes access packet density, number of output targets, number of access targets, and other indicators that characterize the inter-resource state parameters. Based on the established index extraction system, the extracted indexes corresponding to each resource node in the system are determined, and each resource node is decomposed into an index node.

[0019] Further, in step S4, based on the index nodes obtained from step S3, the state data corresponding to each index node during normal operation of the command and control system are collected as raw data. The raw data is then processed using a linear normalization method, the formula of which is:

[0020]

[0021] In the formula, x is the original data value, x max and x min These are the maximum and minimum values ​​in the original data, x. norm These are the normalized data values.

[0022] Furthermore, in step S5, data interpolation is used to process the data obtained in step S4, as shown in the following formula:

[0023]

[0024] In the formula, t i t represents the time point of interpolation. start and t end These represent the start and end time points of the interpolation interval, respectively. This indicates the time after interpolation is t. i Data values, and They represent time t respectively start and t end The data value.

[0025] Furthermore, in step S6, the method for calculating the linear relationship from the abnormal node to the child nodes on each path using least squares estimation is as follows:

[0026] When the relationship between two indicator nodes is linear, y = ax + b, the measurement data can be expressed in the following form:

[0027] Y = Hx + e

[0028] in:

[0029]

[0030] In the formula {x0,x1,…,x n} and {y0,y1,…,y n} represents the sampled data for two indicator nodes, {e0,e1,…,e n} represents process noise;

[0031] The residual ε of the i-th measurement i for:

[0032]

[0033] Among them, Y i For the sampled data of the i-th indicator node, H i This represents the i-th row of the linear relationship coefficient matrix;

[0034] The sum of all measurement errors, J, is:

[0035]

[0036] Let J be The first-order partial derivative is equal to 0, which allows us to calculate the optimal estimate. The following formula is used for calculation:

[0037]

[0038] Furthermore, in step S7, all paths following the abnormal node are traversed using a depth-first search algorithm to calculate the degree of influence of the abnormal node on each path. The method for calculating the degree of influence on a single path is as follows:

[0039] There exists a fault propagation path A→B→C→…→N. The linear relationship from A to B is y=a1x+b1, the linear relationship from A to C is y=a2x+b2, and the linear relationship from A to the nth child node is y=a n x+b n ;

[0040] The steps for calculating the impact of abnormal nodes on this path are as follows:

[0041] The absolute values ​​of the linear coefficients from the outlier node to all child nodes on the path are summed using the following formula:

[0042] a sum =|a1|+|a2|+…+|a n |

[0043] Where n is the total number of child nodes, a i This represents the weight of the impact of the abnormal node on the i-th child node;

[0044] The activation function is used to activate a as follows: sum Mapping to the range of 0 to 1 to obtain the corresponding degree of influence Inf val :

[0045]

[0046] Further, in step S8, by comparing the impact of the abnormal node obtained in step S7 on all subsequent paths, the propagation path with the greatest impact is obtained, and the resource set to which the node on this path belongs is the influence domain of the abnormal node; the maximum impact value obtained in step S7 is compared with a preset threshold, and the current alarm level Inf of the command and control system is determined according to the following formula. lv It is divided into three levels: mild, moderate, and severe.

[0047]

[0048] Compared with the prior art, the significant advantages of this invention are:

[0049] Based on the task requirements of robot swarm control in the information age, this invention constructs a network topology and index extraction system for robot swarm command and control systems, simplifies the risk assessment process of complex robot swarm command and control systems, and establishes a dynamic task impact estimation model for robot swarm command and control systems based on linear least squares. This model can simply and effectively calculate the impact results under different risk scenarios, improves the shortcomings of traditional risk assessment methods in qualitative analysis and real-time performance, and has significant implications for macro-control and system optimization in robot swarm control. Attached Figure Description

[0050] Figure 1This is a flowchart illustrating the overall process framework of the command and control system safety risk assessment method based on linear influence of the present invention.

[0051] Figure 2 This is a schematic diagram of cross-layer polymerization of the base layer in this invention.

[0052] Figure 3 This is a schematic diagram illustrating the reverse elimination of hybridization according to the present invention.

[0053] Figure 4 This is a schematic diagram illustrating the extraction of indicators in this invention. Detailed Implementation

[0054] To better understand the present invention, the embodiments of the present invention will be explained in detail below with reference to the accompanying drawings and examples.

[0055] The present invention provides a safety risk assessment method for command and control systems based on linear effects, the overall process framework of which is as follows: Figure 1 As shown, it includes the following steps:

[0056] S1: Establish the physical connection diagram and information flow diagram of the command and control system, and perform cross-layer aggregation of the base diagram to form a cause-effect graph base diagram;

[0057] S2: Eliminate non-causal relationships in the base graph by using a spanning tree to obtain a causal graph from the opposite direction of the motivation transmission;

[0058] S3: Establish an indicator extraction system to decompose resource nodes in the cause-effect graph into indicator nodes;

[0059] S4: Perform input normalization on the raw data of the command and control system.

[0060] S5: Perform data interpolation on the data processed in step S4;

[0061] S6: Calculate the linear relationship from the outlier node to the child nodes on each path using least squares estimation;

[0062] S7: Traverse all paths following the abnormal node and calculate the degree of impact of the abnormal node on each path;

[0063] S8: Based on the calculation results of step S7, output the degree of influence of the abnormal node on subsequent nodes and the range of influence domain to obtain the alarm level of the command and control system.

[0064] Furthermore, in step S1, since the linear influence algorithm can only be calculated on a single graph, cross-layer aggregation of the base graph is required. The steps for establishing the physical connection graph include: identifying each subsystem in the command and control system and treating it as a node in the graph; identifying the physical connection medium between subsystems, determining the physical connection relationships between each subsystem, and representing them in the graph through the connections between nodes; and determining the direction of the edges in the graph based on the communication direction between each subsystem. The steps for establishing the information flow graph include: identifying each subsystem in the command and control system and treating it as a node in the graph; analyzing the data flow direction between each subsystem, determining the propagation path of data from one subsystem to another, and representing it in the graph through the connections between nodes; and determining the direction of the edges in the graph based on the data propagation direction between subsystems. The steps for implementing cross-layer aggregation of the base graph include: mapping all subsystem nodes in the physical connection graph and the information flow graph; merging the physical connection graph and the information flow graph, retaining the connection relationships and directions between the nodes in both.

[0065] Furthermore, in step S2, since the influence degree and influence domain of an abnormal node are identified from the child node direction as the initial point in the causal graph, and the parent node and ancestor node of the abnormal node are not affected, the nodes that are not affected are eliminated from the opposite direction of the cause propagation by the form of a spanning tree, thereby achieving the purpose of eliminating clutter, reducing the computational scale, and improving the identification efficiency. First, an abnormal node set is generated, which includes abnormal nodes marked due to faults, errors, or abnormal behavior. Nodes in the set are selected as the current processing node in turn. If the node is not in the taboo list, a set of its neighboring resource nodes is generated, and a neighboring resource node is selected from it. It is checked whether it is the parent node of the current node. If it is, it is added as a neighboring node and a connection is established. If not, other neighboring resource nodes are selected until all neighboring nodes have been traversed. This process is repeated until all abnormal nodes have been processed, and finally the generated tree structure is output.

[0066] Further, in step S3, indicator extraction involves decomposing the resource nodes in the causal graph into quantitatively describable indicator nodes. Based on the specific resource node, state parameters characterizing the node's performance are selected as the extracted indicator nodes, establishing an indicator extraction system. The established indicator extraction system consists of two parts: intra-resource indicator extraction and inter-resource indicator extraction. Intra-resource indicator extraction includes packet output density, memory utilization, processing latency, and other indicators that characterize the internal state parameters of the resource. Inter-resource indicator extraction includes access packet density, number of output targets, number of access targets, and other indicators that characterize the inter-resource state parameters. Based on the established indicator extraction system, the extraction indicators corresponding to each resource node in the system are determined, and each resource node is decomposed into indicator nodes.

[0067] Furthermore, in step S4, based on the index nodes obtained from step S3, the state data corresponding to each index node during normal operation of the command and control system are collected as raw data. Since the input data types differ, there is a significant difference in magnitude. Input normalization can avoid prediction errors caused by large differences in magnitude. The processing method used is linear normalization, and its formula is:

[0068]

[0069] In the formula, x is the original data value, x max and x min These are the maximum and minimum values ​​in the original data, x. norm These are the normalized data values.

[0070] Furthermore, in step S5, since the sampling frequency and the number of samples collected by different indicator nodes are different within a certain sampling time T, data interpolation is used to synchronize the data of different indicator nodes in time. The formula is as follows:

[0071]

[0072] In the formula, t i t represents the time point of interpolation. start and t end These represent the start and end time points of the interpolation interval, respectively. This indicates the time after interpolation is t. i Data values, and They represent time t respectively start and t end The data value.

[0073] For clarity, assume that the sampling frequencies for indicator nodes A, B, and C are 2s, 3s, and 5s, respectively, and the sampling time is 10s. Then the sampled data for indicator node A is {d}. A0 ,d A2 ,d A4 ,d A6 ,d A8 ,d A10 The sampled data for indicator node B is {d}. B0 ,d B3 ,d B6 ,d B9 The sampled data for indicator node C is {d}. C0 ,d C5 ,d C10 The sampling data index indicates the data sampled at a certain node at time}. Now, the data of indicator nodes A and C are synchronized to the time axis of B. After time synchronization, the data of indicator node A will be {d}.A0 ,d A3 ,d A6 ,d A9}, the data for indicator node C is {d C0 ,d C3 ,d C6 ,d C9}

[0074] Furthermore, in step S6, the method for calculating the linear relationship from the abnormal node to the child nodes on each path using least squares estimation is as follows:

[0075] When the relationship between two indicator nodes is linear, y = ax + b, the measurement data can be expressed in the following form:

[0076] Y = Hx + e

[0077] in:

[0078]

[0079] In the formula {x0,x1,…,x n} and {y0,y1,…,y n} represents the sampled data for the two indicator nodes,

[0080] {e0,e1,…,e n} represents process noise;

[0081] The residual ε of the i-th measurement i for:

[0082]

[0083] Among them, Y i For the i-th sampled data of the indicator node, H i This represents the i-th row of the linear relationship coefficient matrix;

[0084] The sum of all measurement errors, J, is:

[0085]

[0086] Let J be The first-order partial derivative is equal to 0, which allows us to calculate the optimal estimate. The following formula is used for calculation:

[0087]

[0088] Least squares estimation can be used to calculate the linear relationship between the abnormal node and the child nodes on each path, laying the groundwork for subsequent influence search.

[0089] Furthermore, in step S7, all paths following the abnormal node are traversed using a depth-first search algorithm to calculate the degree of influence of the abnormal node on each path. The method for calculating the degree of influence on a single path is as follows:

[0090] There exists a fault propagation path A→B→C→…→N. The linear relationship from A to B is y=a1x+b1, the linear relationship from A to C is y=a2x+b2, and the linear relationship from A to the nth child node is y=a n x+b n .

[0091] The steps for calculating the impact of abnormal nodes on this path are as follows:

[0092] The absolute values ​​of the linear coefficients from the outlier node to all child nodes on the path are summed using the following formula:

[0093] a sum =|a1|+|a2|+…+|a n |

[0094] Where n is the total number of child nodes, a i The weight represents the influence of the abnormal node on the i-th child node. The summation of the influence weights is to calculate the overall influence of the abnormal node on the propagation path rather than the influence on a single node.

[0095] Because of a sum The output is a linear combination of the influencing weights, which has low data representation power. The activation function is used as follows to activate a... sum Mapping to the range of 0 to 1 to obtain the corresponding degree of influence Inf val This enhances its expressive power and decision boundaries, and resolves the problem of indivisible results.

[0096]

[0097] Furthermore, in step S8, the influence domain range (Inf) a The set of resource nodes on the path that has the greatest impact on all paths within a certain time window is defined as the range of influence of the abnormal node. By comparing the impact of the abnormal node obtained in step S7 on all subsequent paths, the propagation path with the greatest impact is obtained, and the set of resource nodes on this path is the range of influence of the abnormal node.

[0098] Alarm Level (Inf) lv This is used to assess the severity of the impact of a resource node on other resource nodes in the command and control system within a specified time window. The maximum impact level of the abnormal node obtained in step S7 is compared with a preset threshold. Based on the following formula, the current alarm level of the command and control system can be divided into three levels: minor, moderate, and severe.

[0099]

[0100] To facilitate understanding of the present invention, the security risk assessment method for the command and control system of the present invention will be described below with reference to specific embodiments:

[0101] Example 1

[0102] like Figure 2 As shown, the constructed robot swarm command and control system includes 5 visual perception systems, 3 sensor perception systems, 3 data processing systems, and 5 decision analysis systems. The abnormal node is the 012 visual perception system; its failure causes excessively high message output density. A safety risk assessment method for the command and control system based on linear effects is proposed, and the method includes the following steps:

[0103] S1: Establish the physical connection diagram and information flow diagram of the robot cluster command and control system, and perform cross-layer aggregation of the base diagram to form a causal graph base diagram, as shown below. Figure 2 As shown;

[0104] S2: As Figure 2 As shown, if the 012 visual perception system malfunctions, the 200 integrated processing system, 103 command and control system, 403 command and control system, 301 situation analysis system, 302 command and guidance system, and 303 mission planning system will be affected. The remaining nodes will not be affected. Eliminating the unaffected nodes and processing the resulting cause-effect graph is shown below. Figure 3 As shown;

[0105] S3: As Figure 4 As shown, an indicator extraction system is established, and the resource nodes in the influence transmission diagram obtained in step S2 are decomposed into indicator nodes. For the 012 visual perception system, it is decomposed into the 012 output message density. For the 200 integrated processing system, it is decomposed into the 200 access message density, 200 memory utilization, 200 processing latency, and 200 distribution target. For the 103 command and control system and the 403 command and control system, it is decomposed into the 103 target acquisition quantity and the 403 target acquisition quantity. Since there is no causal relationship between the abnormal indicators of the 200 integrated processing system and the indicators of the 301 situation analysis system after decomposition, the 301 situation analysis system and its subsequent sub-nodes are eliminated.

[0106] S4: Perform input normalization on the raw data of the command and control system.

[0107] S5: Perform data interpolation on the data processed in step S4;

[0108] S6: The linear relationship between the abnormal node and the child node on each path is calculated by least squares estimation. The resulting 012 output message density and the absolute value of the linear coefficient between each child node are shown in Table 1.

[0109] Table 1 Relationship of linear model coefficients among nodes

[0110]

[0111]

[0112] S7: Traverse all subsequent paths of the abnormal node. There are two propagation paths. Calculate the impact of the abnormal node on each path. The propagation path with the greatest impact is: 012 Output packet density → 200 Access packet density → 200 Memory utilization → 200 Processing latency → 200 Distribute target → 103 Obtain target quantity. Sum all child nodes of this path and map them through the activation function to obtain the impact of this propagation path as 0.87.

[0113] S8: Based on the calculation results of step S7, combined with... Figure 3 and Figure 4 The results obtained are as follows: the degree of impact of the abnormal node on subsequent nodes, the range of the impact domain, and the current alarm level of the command and control system.

[0114] Affected area: 200 integrated processing system, 103 command and control system;

[0115] Impact level: 0.87;

[0116] Current alert level: Critical.

[0117] Although the specific implementation of the invention has been described above in conjunction with the accompanying drawings and embodiments, it is not intended to limit the scope of protection of the invention. Based on the technical solution of the invention, various modifications or variations that can be made by those skilled in the art without creative effort are still within the scope of protection of the invention.

Claims

1. A method for safety risk assessment of command and control systems based on linear effects, characterized in that, Includes the following steps: S1: Establish the physical connection diagram and information flow diagram of the command and control system, and perform cross-layer aggregation of the base diagram to form a cause-effect graph base diagram; S2: Eliminate non-causal relationships in the base graph by using a spanning tree to move the motivation in the opposite direction to obtain a causal graph; S3: Establish an indicator extraction system to decompose resource nodes in the cause-effect graph into indicator nodes; S4: Perform input normalization processing on the raw data of the command and control system. S5: Perform data interpolation on the data processed in step S4; S6: Calculate the linear relationship from the outlier node to each child node on each path using least squares estimation; the implementation method is as follows: When the relationship between two indicator nodes is linear, y = ax + b, the measurement data can be expressed in the following form: Y = Hx + e in: In the formula {x0,x1,…,x n } and {y0,y1,…,y n } represents the sampled data for two indicator nodes, {e0,e1,…,e n } represents process noise; The residual ε1 of the i-th measurement is: Among them, Y i For the sampled data of the i-th indicator node, H i This represents the i-th row of the linear relationship coefficient matrix; The sum of all measurement errors, J, is: Let J be The first-order partial derivative is equal to 0, which allows us to calculate the optimal estimate. The following formula is used for calculation: S7: Traverse all paths following the abnormal node and calculate the impact of the abnormal node on each path; the calculation method is as follows: There exists a fault propagation path A→B→C→…→N. The linear relationship from A to B is y=a1x+b1, the linear relationship from A to C is y=a2x+b2, and the linear relationship from A to the nth child node is y=a n x+b n ; The steps for calculating the impact of abnormal nodes on this path are as follows: The absolute values ​​of the linear coefficients from the outlier node to all child nodes on the path are summed using the following formula: a sum =|a1|+|a2|+…+|a n | Where n is the total number of child nodes, a i This represents the weight of the impact of the abnormal node on the i-th child node; The activation function is used to activate a as follows: sum Mapping to the range of 0 to 1 to obtain the corresponding degree of influence Inf val : S8: Based on the calculation results of step S7, output the degree of influence of the abnormal node on subsequent nodes and the range of influence domain to obtain the current alarm level of the command and control system.

2. The method for safety risk assessment of command and control systems based on linear effects according to claim 1, characterized in that: In step S1, the steps for establishing the physical connection diagram include: identifying each subsystem in the command and control system and treating it as a node in the diagram; identifying the physical connection medium between subsystems, determining the physical connection relationship between each subsystem, and representing it in the diagram through the connection of edges between nodes; and determining the direction of the edges in the diagram based on the communication direction between each subsystem. The steps for creating an information flow diagram include: identifying each subsystem in the command and control system and treating it as a node in the diagram; analyzing the data flow between subsystems, determining the propagation path of data from one subsystem to another, and representing it in the diagram through the connection of edges between nodes; and determining the direction of the edges in the diagram based on the data propagation flow between subsystems. The implementation steps for cross-layer aggregation of the base graph include: mapping all subsystem nodes in the physical connection graph and the information flow graph; merging the physical connection graph and the information flow graph, retaining the connection relationships and directions between nodes in both graphs to obtain the causal graph base graph.

3. The method for safety risk assessment of command and control systems based on linear effects according to claim 1, characterized in that: In step S2, in the causal graph base graph generated in step S1, nodes that will not be affected are eliminated from the opposite direction of the cause propagation in the form of a spanning tree. First, an abnormal node set is generated, which includes abnormal nodes marked due to faults, errors, or abnormal behavior. Nodes in the set are selected as the current processing node in turn. If the node is not in the taboo list, a set of neighboring resource nodes directly connected to it is generated, and a neighboring resource node is selected from it. It is checked whether it is the parent node of the current node. If it is, it is added as a neighboring node and a connection is established. If not, other neighboring resource nodes are selected until all neighboring nodes have been traversed. This process is repeated until all abnormal nodes have been processed. Finally, the processed causal graph is output.

4. The method for safety risk assessment of command and control systems based on linear effects according to claim 1, characterized in that: In step S3, indicator extraction involves decomposing resource nodes in the causal graph into quantitatively describable indicator nodes. Based on the specific resource node, state parameters characterizing the node's performance are selected as extracted indicator nodes, establishing an indicator extraction system. This system consists of two parts: intra-resource indicator extraction and inter-resource indicator extraction. Intra-resource indicator extraction includes packet output density, memory utilization, processing latency, and other indicators characterizing the internal state parameters of the resource. Inter-resource indicator extraction includes access packet density, number of output targets, number of access targets, and other indicators characterizing the inter-resource state parameters. Based on the established indicator extraction system, the extraction indicators corresponding to each resource node in the system are determined, and each resource node is decomposed into indicator nodes.

5. The method for safety risk assessment of command and control systems based on linear effects according to claim 1, characterized in that: In step S4, based on the index nodes obtained from step S3, the status data corresponding to each index node during normal operation of the command and control system are collected as raw data. The raw data is then processed using a linear normalization method, the formula of which is: In the formula, x is the original data value, x max and x min These are the maximum and minimum values ​​in the original data, x. norm These are the normalized data values.

6. The method for safety risk assessment of command and control systems based on linear effects according to claim 1, characterized in that: In step S5, data interpolation is used to process the data obtained in step S4, as shown in the following formula: In the formula, t i t represents the time point of interpolation. start and t end These represent the start and end time points of the interpolation interval, respectively. This indicates the time after interpolation is t. i Data values, and They represent time t respectively start and t end The data value.

7. The method for safety risk assessment of command and control systems based on linear effects according to claim 1, characterized in that: In step S8, by comparing the impact of the abnormal node obtained in step S7 on all subsequent paths, the propagation path with the greatest impact is obtained, and the resource set to which the node on this path belongs is the influence domain of the abnormal node; the maximum impact value obtained in step S7 is compared with a preset threshold, and the current alarm level Inf of the command and control system is determined according to the following formula. lv It is divided into three levels: mild, moderate, and severe.

Images