Coupling coordination driven enterprise security risk dynamic assessment method and system

Abstract

The embodiment of the invention provides a coupling coordination driven enterprise safety risk dynamic assessment method and system, and relates to the technical field of enterprise safety assessment, and the method comprises the steps: determining a safety risk candidate index pool of a target coal mine enterprise, carrying out the staged clustering iteration screening of the safety risk candidate index pool, and constructing a risk assessment index set; determining a weight system structure, an index layer weight set and a criterion layer weight set; performing fuzzy risk quantitative evaluation, and determining an initial risk evaluation score; in a preset window, performing dynamic memory iteration coupling coordination analysis on the target coal mine enterprise based on the weight system structure, and determining a coupling coordination degree; and performing dynamic correction, and determining a target enterprise security risk dynamic assessment result. The problems that in the prior art, a closed-loop dynamic optimization mechanism is lacked, and continuous iterative sensing and active prevention and control of the enterprise security risk state cannot be achieved are solved. The effects of improving the risk dynamic response rate and improving the evaluation reliability are achieved.

Description

Technical Field

[0001] This invention relates to the field of enterprise security assessment technology, and in particular to a method and system for dynamic assessment of enterprise security risks driven by coupling and coordination. Background Technology

[0002] Coal mines are a vital energy industry in my country, and their safety plays a crucial role in national economic development and social stability. Traditional risk management methods in coal mines are relatively rudimentary, with insufficient application of information technology. However, the application of big data, the Internet of Things, and artificial intelligence in coal mines provides the technological means to identify and assess coal mine safety risks and transform the "Internet+" coal mine supervision and inspection model.

[0003] Traditional risk assessment methods rely on static indicators and human experience, making it difficult to dynamically capture real-time risk changes, resulting in serious delays in early warning. At the same time, existing technologies for analyzing sensor data are limited to single threshold judgments and lack multi-dimensional correlation mining, making it impossible to identify systemic risks caused by minor local failures.

[0004] Furthermore, the assessment model is insensitive to sudden changes in key factors, making it difficult to quantify the transmission effect of local anomalies on the overall safety status. In addition, the low utilization rate of massive monitoring data, redundant indicators, and inability to be simplified in real time result in low assessment efficiency.

[0005] In summary, the existing technology system has not formed a closed-loop dynamic optimization mechanism, and cannot achieve continuous iterative perception and proactive prevention and control of risk status.

[0006] It should be noted that the information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0007] To address the aforementioned deficiencies or improvement needs of existing technologies, this invention provides a coupled coordination-driven dynamic assessment method and system for enterprise safety risks. It solves the problem of existing technologies lacking a closed-loop dynamic optimization mechanism, thus failing to achieve continuous iterative perception and proactive prevention of enterprise safety risk status. By clustering and iteratively screening and performing hierarchical analysis on the indicators of the target coal mining enterprise, a weighting system and fuzzy risk quantification are determined. Dynamic memory iterative coupled coordination analysis is then performed to determine the degree of coupling coordination, followed by dynamic correction to determine the dynamic assessment result of the target enterprise's safety risks. This achieves the technical effects of improving the dynamic response rate of risks and enhancing the reliability of the assessment. The specific technical solution is as follows:

[0008] According to a first aspect of the present invention, a method for dynamic assessment of enterprise security risks driven by coupling and coordination is provided, the method comprising:

[0009] A pool of candidate safety risk indicators for the target coal mining enterprise is determined. This pool is then subjected to phased clustering and iterative screening to construct a risk assessment indicator set. The risk assessment indicator set is then processed using the Analytic Hierarchy Process (AHP) to establish a progressive hierarchical structure, determining the weighting system, indicator layer weight set, and criterion layer weight set. Fuzzy risk quantification is performed using these weighting systems to determine an initial risk assessment score. Within a preset window, dynamic memory iterative coupling and coordination analysis is conducted on the target coal mining enterprise based on the weighting system to determine the coupling coordination degree. Based on the coupling coordination degree, the initial risk assessment score is dynamically corrected to determine the dynamic safety risk assessment result for the target enterprise.

[0010] In one implementation, a pool of candidate safety risk indicators for the target coal mine enterprise is determined. The pool is then subjected to phased clustering and iterative screening to construct a risk assessment indicator set. The following further processing is performed: First-stage base class indicator identification is conducted on the candidate safety risk indicators to determine K base class indicators. These base class indicators are the K indicators in the candidate safety risk indicators whose mean Euclidean distance from other indicators is ranked from largest to smallest, where K is a positive integer greater than or equal to 3. Second-stage nearest neighbor ablation processing is performed on the K base class indicators to obtain M ablation base class indicators, where M is a positive integer less than or equal to K. Third-stage clustering analysis and intra-cluster iterative screening are then performed on the candidate safety risk indicators based on the M ablation base indicators to determine the risk assessment indicator set.

[0011] In one implementation, a second-stage nearest-neighbor ablation processing is performed on the K base class indices to obtain M ablation base class indices, where M is a positive integer less than or equal to K. The following processing is also performed: pairwise nearest-neighbor ablation coefficient identification is performed on each of the K base class indices to obtain K sets of nearest-neighbor ablation coefficients; the mean of the K sets of nearest-neighbor ablation coefficients is calculated to determine the mean of the K nearest-neighbor ablation coefficients; the base class index corresponding to the maximum value among the K nearest-neighbor ablation coefficient means is used as the ablation guidance base class index; using the ablation guidance base class index as an index, nearest-neighbor ablation processing is performed on the base class indices corresponding to the nearest-neighbor ablation coefficients in the corresponding sets of the K sets of nearest-neighbor ablation coefficients that are greater than or equal to a preset nearest-neighbor ablation coefficient threshold, to obtain M ablation base class indices.

[0012] In one implementation, based on the M ablation base class indicators, a third-stage cluster analysis and intra-cluster iterative screening are performed on the safety risk candidate indicator pool to determine the risk assessment indicator set. The following processing is also performed: Indicators from the safety risk candidate indicator pool are added to the set corresponding to the ablation base class indicator with the closest Euclidean distance among the M ablation base class indicators, obtaining a set of M ablation base class indicator clusters; using the M ablation base class indicators as the starting point for the M intra-cluster iterative screening, the M ablation base class indicator clusters are iteratively screened until a preset iterative screening stop condition is met, obtaining M representative indicators; the M representative indicators are summarized to obtain the risk assessment indicator set; wherein, the preset iterative screening stop condition is that the indicator neighborhood size in the current iteration is less than the indicator neighborhood size in the previous iteration and / or the number of iterations is greater than the maximum number of iterations.

[0013] In one implementation, the M ablation base class indicators are used as the starting points for M intra-cluster iterative screening. The clusters of the M ablation base class indicators are iteratively screened until a preset iterative screening stop condition is met, resulting in M ​​representative indicators. The following processing is also performed: According to a preset intra-cluster iterative bandwidth, the neighborhoods of the M intra-cluster iterative screening starting points are constructed, and initial iterations are performed on the M intra-cluster iterative screening starting points to obtain M intra-cluster iterative screening stage points; the neighborhoods of the M intra-cluster iterative screening stage points are traversed based on the preset intra-cluster iterative bandwidth to obtain M intra-cluster iterative screening stage point neighborhoods; the neighborhoods of the M intra-cluster iterative screening stage points are compared with the neighborhoods of the M intra-cluster iterative screening starting points to determine whether the preset iterative screening stop condition is met. If yes, the M intra-cluster iterative screening starting points are used as the M representative indicators; otherwise, iterations are continued on the M intra-cluster iterative screening stage points according to the preset intra-cluster iterative bandwidth, and the number of iterations is counted. This process is repeated until the preset iterative screening stop condition is no longer met, resulting in the M representative indicators.

[0014] In one implementation, the risk assessment indicator set is processed using a progressive hierarchical structure through the analytic hierarchy process (AHP) to determine the weight structure, indicator layer weight set, and criterion layer weight set. The following processing is also performed: According to the risk assessment objective, the risk assessment indicator set is divided into a three-layer progressive structure: objective layer, criterion layer, and indicator layer, to obtain the weight structure; Each indicator within each criterion in the weight structure is traversed, and pairwise importance comparisons are performed to determine the indicator comparison judgment matrix, and weight vectors are calculated to obtain the indicator layer weight set; Multiple criteria in the weight structure are traversed, and pairwise importance comparisons are performed to construct an N×N criterion comparison judgment matrix, and weight vectors are calculated to obtain the criterion layer weight set, where N is a positive integer.

[0015] In one implementation, fuzzy risk quantification assessment is performed by combining the weight system architecture, the indicator layer weight set, and the criterion layer weight set to determine the initial risk assessment score. The following processes are also performed: pre-constructing a safety risk level evaluation set and corresponding risk scale vectors; traversing the weight system architecture to perform fuzzy membership degree analysis of the indicator layer of the safety risk level evaluation set to obtain an indicator membership degree vector set; constructing a fuzzy judgment matrix based on the indicator membership degree vector set, and performing fuzzy synthesis of the criterion layer using the indicator layer weight set to obtain a criterion layer membership degree vector set; performing matrix synthesis based on the criterion layer membership degree vector set, and performing fuzzy synthesis of the target layer using the criterion layer weight set to obtain a target layer membership degree vector; and defuzzifying the target layer membership vector using the risk scale vector to obtain the initial risk assessment score.

[0016] In one implementation, within a preset window, dynamic memory iterative coupling coordination analysis is performed on the target coal mining enterprise based on the weighting system architecture to determine the coupling coordination degree. The following processes are also performed: Within the preset window, indicator monitoring is conducted on the target coal mining enterprise in conjunction with the weighting system architecture to obtain a monitoring weighting system architecture sequence; criterion-level dynamic memory iteration is performed on the first and second monitoring weighting systems in the monitoring weighting system architecture sequence to determine a first dynamic memory iteration matrix; criterion-level dynamic memory iteration is performed on the second and third monitoring weighting systems in the monitoring weighting system architecture sequence to determine a second dynamic memory iteration matrix; the second dynamic memory iteration matrix is ​​enhanced using the first dynamic memory iteration matrix through iterative matrix interaction to obtain a second enhanced dynamic memory iteration matrix; and so on, after multiple criterion-level dynamic memory iterations and iterative matrix interaction enhancements, an Nth enhanced dynamic memory iteration matrix is ​​determined; coupling coordination analysis is performed based on the Nth enhanced dynamic memory iteration matrix to determine the coupling coordination degree.

[0017] In one implementation, a criterion-level dynamic memory iteration is performed on the first and second monitoring weight architectures in the monitoring weight architecture sequence to determine a first dynamic memory iteration matrix. The following processing is also performed: criterion-level index deviation analysis is performed on the first and second monitoring weight architectures to obtain a first criterion-level index deviation coefficient set; the first criterion-level index deviation coefficient set is normalized within the set to obtain a first criterion-level index deviation coefficient normalized value set; the first criterion-level index deviation coefficient normalized value set is filled into an initially empty matrix to obtain the first dynamic memory iteration matrix.

[0018] According to a second aspect of the present invention, a coupled and coordinated driven dynamic assessment system for enterprise security risks is provided, the system comprising:

[0019] The system comprises the following modules: a clustering iterative screening module for determining a pool of candidate safety risk indicators for the target coal mining enterprise, and a risk assessment indicator set for constructing a phased clustering iterative screening of the candidate safety risk indicator pool; a structure processing module for performing progressive hierarchical structure processing on the risk assessment indicator set using the analytic hierarchy process (AHP) to determine the weight system structure, indicator layer weight set, and criterion layer weight set; a risk quantification assessment module for performing fuzzy risk quantification assessment by combining the weight system structure, indicator layer weight set, and criterion layer weight set to determine the initial risk assessment score; a coupling coordination degree determination module for performing dynamic memory iterative coupling coordination analysis on the target coal mining enterprise within a preset window based on the weight system structure to determine the coupling coordination degree; and a dynamic correction module for dynamically correcting the initial risk assessment score based on the coupling coordination degree to determine the dynamic safety risk assessment result of the target enterprise.

[0020] Beneficial effects of the embodiments of the present invention:

[0021] The solution provided in this invention involves determining a pool of candidate safety risk indicators for a target coal mine enterprise, performing phased clustering and iterative screening of this pool to construct a risk assessment indicator set, and then using the analytic hierarchy process (AHP) to process the risk assessment indicator set into a progressive hierarchical structure to determine the weighting system structure, indicator layer weight set, and criterion layer weight set. Fuzzy risk quantification assessment is then performed using the weighting system structure, indicator layer weight set, and criterion layer weight set to determine the initial risk assessment score. Subsequently, within a preset window, dynamic memory iterative coupling and coordination analysis is conducted on the target coal mine enterprise based on the weighting system structure to determine the coupling coordination degree. The initial risk assessment score is then dynamically corrected based on the coupling coordination degree to determine the dynamic safety risk assessment result for the target enterprise. This achieves intelligent enhancement of risk weights, significantly improves the early warning sensitivity of minor hidden dangers triggering major accidents, overcomes the limitations of traditional models in responding to sudden anomalies, optimizes the dynamic safety risk assessment mechanism, and improves the reliability of the assessment. Of course, implementing any product or method of this invention does not necessarily require achieving all of the above advantages simultaneously. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This invention illustrates a flowchart of the coupled coordination-driven dynamic assessment method for enterprise security risks provided by the present invention.

[0024] Figure 2A schematic diagram of the structure of the enterprise security risk dynamic assessment system driven by coupling and coordination provided by the present invention is shown.

[0025] Figure labeling: Clustering Iterative Screening Module 11, Structure Processing Module 12, Risk Quantification Assessment Module 13, Coupling Coordination Determination Module 14, Dynamic Correction Module 15. Detailed Implementation

[0026] To facilitate understanding of the present invention, a more complete description of the invention will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein; rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the invention.

[0027] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0028] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0029] Unless otherwise expressly stated, throughout the specification and claims, the term "comprising" or its variations such as "including" or "comprises" shall be understood to include the stated elements or components without excluding other elements or other components.

[0030] The present invention provides a coupled coordination-driven dynamic assessment method and system for enterprise security risks, which is used to solve the problem of lacking a closed-loop dynamic optimization mechanism and being unable to achieve continuous iterative perception and proactive prevention and control of enterprise security risk status.

[0031] Example 1: See Figure 1 The flowchart of the coupled coordination-driven dynamic assessment method for enterprise security risks provided in this embodiment of the invention includes:

[0032] S1: Determine the safety risk candidate indicator pool of the target coal mining enterprise, and perform phased clustering and iterative screening on the safety risk candidate indicator pool to construct a risk assessment indicator set;

[0033] In one implementation, a pool of candidate safety risk indicators for the target coal mining enterprise is determined, and the pool of candidate safety risk indicators is subjected to phased clustering and iterative screening to construct a risk assessment indicator set. Step S1 may further include:

[0034] The first stage of base class indicator identification is performed on the security risk candidate indicator pool to determine K base class indicators. The base class indicators are the K indicators in the security risk candidate indicator pool whose mean Euclidean distance from other indicators is ranked from largest to smallest, and K is a positive integer greater than or equal to 3.

[0035] The K base class indices are subjected to a second-stage nearest neighbor ablation processing to obtain M ablation base class indices, where M is a positive integer less than or equal to K;

[0036] Based on the M ablation base class indicators, a third-stage cluster analysis and intra-cluster iterative screening are performed on the pool of candidate safety risk indicators to determine the risk assessment indicator set.

[0037] In one embodiment, the safety risk candidate indicator pool refers to a complete set of risk indicators that can reflect the status of personnel, equipment, environment, management, and disaster behavior in coal mining enterprises. These indicators originate from safety monitoring systems, personnel positioning systems, video AI systems, geological and production data, management ledgers, etc., and have high dimensionality and redundancy. Therefore, it is necessary to perform phased clustering and iterative screening to reduce the amount of data analyzed. The mean Euclidean distance is calculated by averaging the Euclidean distances between the vector describing a certain indicator and the vectors describing all other indicators, reflecting the independence or isolation degree of that indicator.

[0038] Specifically, a candidate indicator pool for safety risks is constructed from the massive heterogeneous monitoring indicators of the target coal mining enterprises. Based on accident causation theory, the causes of safety risks and accidents mainly include four aspects: unsafe human behavior, unsafe production environments, unsafe conditions of equipment, and lack of management. Following the classification principles of risk factors in classic accident causation theory, and based on the indicator system in the coal mine safety risk assessment system and inspection standard research report, indicators are selected from a range including historical factors, natural environmental factors, personnel factors, mine equipment and facility information, safety management factors, and monitoring information. Furthermore, dynamic identification results are introduced from video AI analysis monitoring and early warning systems and big data analysis monitoring and early warning systems to initially construct the candidate indicator pool required for coal enterprise safety risk assessment.

[0039] The process then proceeds to a phased clustering and iterative screening. In the first phase, the mean Euclidean distance of each indicator's description to other indicators is calculated for all indicators in the candidate pool of security risk indicators. This determines the correlation between each indicator and other indicators, and the K indicators with the highest mean are selected as base class indicators, serving as the most discriminative feature cores in the risk space. In the second phase, nearest neighbor ablation is performed on these K base class indicators. The nearest neighbor ablation coefficient is calculated by considering the pairwise distances between base class indicators. This coefficient reflects the similarity between each base class indicator and other base class indicators. If the coefficient is too large, it indicates a high overlap between the corresponding base class indicators, requiring nearest neighbor ablation. Base class indicators that are too close or have highly overlapping information are deleted, ultimately retaining M non-redundant ablation base class indicators. In the third stage, with M ablation base class indicators as the initial cluster centers, targeted clustering analysis is performed on the safety risk candidate indicator pool. All indicators are automatically assigned to their most similar base class clusters, and the intra-cluster iterative screening bandwidth is used to iterate the most representative indicators step by step, finally forming a unique set of representative indicators for each cluster, thereby constructing a lightweight and non-redundant risk assessment indicator set.

[0040] In one implementation, the K base class indices undergo a second-stage nearest-neighbor ablation processing to obtain M ablation base class indices, where M is a positive integer less than or equal to K. Step S1 may further include:

[0041] Each of the K base class indices is subjected to pairwise nearest neighbor ablation coefficient identification to obtain a set of K nearest neighbor ablation coefficients.

[0042] Calculate the mean of the set of K nearest neighbor ablation coefficients, determine the mean of the K nearest neighbor ablation coefficients, and take the base class index corresponding to the maximum value of the K nearest neighbor ablation coefficients as the ablation guidance base class index;

[0043] Using the ablation guidance base class index as an index, the base class index corresponding to the nearest neighbor ablation coefficient that is greater than or equal to the preset nearest neighbor ablation coefficient threshold in the corresponding nearest neighbor ablation coefficient set of the K nearest neighbor ablation coefficient sets is subjected to nearest neighbor ablation processing to obtain M ablation base class indices.

[0044] In one embodiment, the ablation guide base class indicator is the base class indicator with the largest mean, which is selected as the center of this round of ablation. It is responsible for identifying the set of redundant indicators most similar to itself and performing ablation actions. The preset nearest neighbor ablation coefficient threshold is a similarity threshold predefined by those skilled in the art. When the nearest neighbor ablation coefficient of a base class indicator and the ablation guide indicator is greater than or equal to this threshold, it is considered a redundant indicator and needs to be ablated.

[0045] For the K base class indicators, pairwise nearest neighbor ablation coefficient identification is performed. By calculating the similarity between each pair of base class indicators, such as the inverse Euclidean distance, a set of K nearest neighbor ablation coefficients is formed. Subsequently, the mean of each set is calculated to quantify the overall redundancy of each base class indicator. The base class indicator with the largest mean is automatically identified as the ablation guide base class indicator. This avoids the false diversity caused by high overlap among the K initial base class indicators. By dynamically selecting the guide indicator and eliminating its highly similar nearest neighbors, redundant base classes can be significantly reduced, enhancing the independence and distribution balance of the final M ablation base class indicators. This provides more stable and representative initial cluster centers for the subsequent third-stage cluster analysis, improving the overall clustering result quality and the reliability of the final indicator set construction.

[0046] In one implementation, a third-stage clustering analysis and intra-cluster iterative screening are performed on the safety risk candidate indicator pool based on the M ablation base class indicators to determine the risk assessment indicator set. Step S1 may further include:

[0047] The indicators in the safety risk candidate indicator pool are added to the set corresponding to the ablation base class indicator with the closest Euclidean distance among the M ablation base class indicators, to obtain the M ablation base class indicator cluster set;

[0048] Using the M ablation base class indicators as the starting point for iterative screening within the M clusters, iterative screening is performed on the cluster sets of the M ablation base class indicators until the preset iterative screening stop condition is met, thereby obtaining M representative indicators.

[0049] By summarizing the M representative indicators, the risk assessment indicator set is obtained.

[0050] The preset iterative screening stopping condition is that the index neighborhood size in the current iteration is less than the index neighborhood size in the previous iteration and / or the number of iterations is greater than the maximum number of iterations.

[0051] In one implementation, the M ablation base class indicators are used as the starting point for iterative screening within the M clusters. The clusters of the M ablation base class indicators are iteratively screened until a preset iterative screening stopping condition is met, resulting in M ​​representative indicators. Step S1 may further include:

[0052] According to the preset intra-cluster iteration bandwidth, construct the neighborhood of the M intra-cluster iteration screening starting point and perform initial iteration on the M intra-cluster iteration screening starting point to obtain the M intra-cluster iteration screening stage points.

[0053] Traverse the M cluster-in-cluster iterative screening stage points and construct neighborhoods based on the preset cluster-in-cluster iterative bandwidth to obtain the neighborhoods of the M cluster-in-cluster iterative screening stage points.

[0054] Compare the neighborhood size of the M cluster-in-cluster iterative screening stage point neighborhood with the neighborhood size of the M cluster-in-cluster iterative screening start point neighborhood to determine whether the preset iterative screening stop condition is met. If so, the M cluster-in-cluster iterative screening start points are taken as M representative indicators.

[0055] If not, continue iterating over the M intra-cluster iterative screening stage points according to the preset intra-cluster iterative bandwidth, and count the number of iterations. Continue in this manner until the preset iterative screening stop condition is no longer met, and obtain the M representative indicators.

[0056] In one embodiment, the intra-cluster iteration bandwidth describes the discrete width of the neighborhood range of the current centroid. The index neighborhood quantity is the number of indices falling within the neighborhood range of a given iteration centroid, used to determine whether the cluster is converging. The preset iteration screening stopping condition is that the index neighborhood quantity of the current iteration is less than that of the previous iteration and / or the number of iterations is greater than the maximum number of iterations. When the neighborhood quantity decreases, it indicates that moving the centroid or further shrinking is meaningless. When the number of iterations reaches the upper limit, it also needs to be stopped to prevent infinite loops, used to determine the intra-cluster iteration stability.

[0057] Using M ablation base class indicators as cluster centers, all indicators in the safety risk candidate indicator pool are assigned to corresponding clusters according to the Euclidean distance principle, forming a set of M ablation base class indicator clusters. Then, using the ablation base class indicator of each cluster as the starting point for intra-cluster iterative screening, a neighborhood is constructed based on a preset intra-cluster iterative bandwidth, and the first iteration is performed to generate intra-cluster iterative screening stage points. For each stage point, a neighborhood is constructed again based on the iterative bandwidth, and the number of indicators within the neighborhood is calculated. If the neighborhood size in this round is less than that in the previous round, it indicates that the cluster is shrinking or the number of iterations has exceeded the maximum number, and the iteration is considered to have converged, with the stage starting point considered the final representative indicator within the cluster. If the stopping condition is not met, iteration continues with the stage point as the center, repeating the neighborhood construction and neighborhood size detection process until the stopping condition is met. Repeating the above process yields M core indicators with the strongest representativeness within the cluster. Finally, the M representative indicators are summarized to form a risk assessment indicator set.

[0058] By clustering a large pool of indicators and iteratively shrinking it, the representative indicators of each cluster are purified and screened. Compared with the traditional hard clustering method of direct K-Means, this method enables the representative indicators to converge adaptively from the data structure itself, thus enhancing the adaptability of the indicator system to dynamic data structures.

[0059] S2: The risk assessment indicator set is processed into a progressive hierarchical structure using the analytic hierarchy process to determine the weight system structure, indicator layer weight set, and criterion layer weight set.

[0060] In one implementation, the risk assessment index set is processed using the analytic hierarchy process (AHP) to determine the weight system structure, the index layer weight set, and the criterion layer weight set. Step S2 may further include:

[0061] According to the risk assessment objectives, the risk assessment indicator set is divided into a three-layer progressive structure of target layer, criterion layer and indicator layer to obtain a weighted system structure.

[0062] By traversing each indicator within each criterion in the weight system architecture, pairwise importance comparisons are performed to determine the indicator comparison judgment matrix, and weight vectors are calculated to obtain the indicator layer weight set.

[0063] The system iterates through multiple criteria in the weighting architecture, performs pairwise importance comparisons, constructs an N×N criterion comparison judgment matrix, and calculates weight vectors to obtain the criterion layer weight set, where N is a positive integer.

[0064] In one embodiment, the target layer is used for comprehensive enterprise security risk assessment, the criteria layer includes major risk categories such as video monitoring, equipment status, personnel behavior, and management level, and the indicator layer consists of specific risk indicators under each criterion, which are representative indicators selected in step S1. The weighting structure is used to clarify the risk impact path and the hierarchical relationship of the weights.

[0065] The indicator comparison judgment matrix is ​​used to analyze the relative importance of multiple indicators within a criterion. The matrix consists of a 1-9 rating scale, with the importance of each pairwise comparison determined by experts or models. For example: 1: equally important, 3: relatively important, 5: significantly important, 7: strongly important, 9: extremely important. This matrix is ​​used to determine the weights at the indicator level. The criterion comparison judgment matrix is ​​used to assess the degree of importance of different criteria to the overall risk assessment objective. The matrix dimension is N×N, where N is the number of criteria, such as 12 categories.

[0066] Based on the overall objectives of safety risk assessment for coal mining enterprises, the selected risk assessment indicator set is divided into a three-tiered progressive structure: the top layer is the comprehensive safety risk assessment target layer, below which is the criterion layer divided according to indicator categories (e.g., video AI early warning, gas monitoring, personnel behavior, etc.), and the bottom layer consists of multiple specific risk indicators under each criterion. After this structure is completed, the system performs pairwise importance comparisons of indicators within each criterion, constructs an indicator comparison judgment matrix, and uses the AHP maximum eigenvalue method to calculate the indicator layer weights, thereby obtaining the importance distribution of each indicator within each criterion. Subsequently, an N×N comparison matrix between criteria is constructed in the same way, and the criterion layer weight set is obtained by solving the eigenvectors, clarifying the contribution of each major risk category to the overall assessment.

[0067] S3: Combine the weight system structure, indicator layer weight set and criterion layer weight set to perform fuzzy risk quantification assessment and determine the initial risk assessment score;

[0068] In one implementation, fuzzy risk quantification assessment is performed by combining the weight system architecture, the indicator layer weight set, and the criterion layer weight set to determine the initial risk assessment score. Step S3 may further include:

[0069] Pre-construct a safety risk level evaluation set and the corresponding risk scale vector;

[0070] By traversing the weight system structure, fuzzy membership analysis of the indicator layer of the security risk level evaluation set is performed to obtain the indicator membership vector set.

[0071] Based on the index membership vector set, a fuzzy judgment matrix is ​​constructed, and combined with the index layer weight set, a criterion layer fuzzy synthesis is performed to obtain the criterion layer membership vector set.

[0072] Matrix synthesis is performed based on the criterion layer membership vector set, and target layer fuzzy synthesis is performed by combining the criterion layer weight set to obtain the target layer membership vector.

[0073] The risk scale vector is used to defuzzify the membership vector of the target layer to obtain the initial risk assessment score.

[0074] Preferably, the safety risk level evaluation set refers to a set of linguistic levels used to describe the enterprise's risk status, such as: {negligible, minor, moderate, severe, critical}. It is a linguistic variable in fuzzy comprehensive evaluation, used to map discrete risk performance to level intervals. The risk scale vector is a vector that maps each risk level to a quantitative score, such as 20, 40, 60, 80, 100, used for final defuzzification. This converts fuzzy linguistic information into concrete numerical values ​​that can be used for automated decision-making.

[0075] First, based on the classification standards for risk events such as gas concentration, equipment failure rate, and personnel violations in the Coal Mine Safety Regulations, a safety risk level evaluation set containing five levels—negligible, minor, moderate, severe, and critical—is pre-constructed. Each level is then mapped to a risk scale vector, such as 20, 40, 60, 80, and 100, and stored in a parameter library for later use. Subsequently, the system traverses the weight structure and, for each indicator's real-time acquired monitoring value, calls the corresponding membership function according to its indicator type. Numerical indicators use triangular or trapezoidal membership functions, while categorical indicators use expert experience table mapping functions. The system calculates the membership degree value at each of the five risk levels level by level. For example, when processing methane concentration, the system calculates the membership degree of each level according to the interval it falls into using a piecewise function, thereby generating the membership degree vector of that indicator. Then, the system assembles the membership degree vectors of all indicators under the same criterion into an N×5 fuzzy judgment matrix by row, and performs matrix multiplication with the indicator layer weight vector corresponding to that criterion to obtain the criterion layer membership degree vector reflecting the overall contribution of the criterion to the five risk levels.

[0076] Furthermore, when there are 12 criteria, all criterion-level membership vectors are combined column-wise to form a 12×5 criterion-level composite matrix, and then matrix-weighted composite is performed with the criterion-level weight set to finally obtain the five-dimensional membership vector of the target layer. Finally, the risk scale vector is called to perform defuzzification calculation on the membership of the target layer using the weighted average method, that is, multiplying the membership of each risk level with its corresponding scale and summing them to generate an initial quantitative risk assessment score that can be used for real-time early warning, trend analysis and subsequent coupling and coordination correction.

[0077] S4: Within a preset window, perform dynamic memory iterative coupling coordination analysis on the target coal mining enterprise based on the weight system architecture to determine the coupling coordination degree;

[0078] S5: Based on the coupling coordination degree, dynamically correct the initial risk assessment score to determine the dynamic assessment result of the target enterprise's security risk.

[0079] In one implementation, within a preset window, dynamic memory iterative coupling coordination analysis is performed on the target coal mining enterprise based on the weighted system architecture to determine the coupling coordination degree. Step S4 may further include:

[0080] Within a preset window, the target coal mining enterprise is monitored in accordance with the weight system architecture to obtain a monitoring weight system architecture sequence.

[0081] Perform criterion-level dynamic memory iteration on the first and second monitoring weight architectures in the monitoring weight architecture sequence to determine the first dynamic memory iteration matrix;

[0082] Perform criterion-level dynamic memory iteration on the second and third monitoring weight architectures in the monitoring weight architecture sequence to determine the second dynamic memory iteration matrix;

[0083] The second dynamic memory iteration matrix is ​​enhanced by iterative matrix interaction using the first dynamic memory iteration matrix to obtain the second enhanced dynamic memory iteration matrix.

[0084] By analogy, after multiple iterations of criterion-level dynamic memory and interactive enhancement of the iteration matrix, the Nth enhanced dynamic memory iteration matrix is ​​determined.

[0085] Based on the Nth enhanced dynamic memory iteration matrix, a coupling coordination analysis is performed to determine the coupling coordination degree.

[0086] In one implementation, a criterion-level dynamic memory iteration is performed on the first and second monitoring weight architectures in the monitoring weight architecture sequence to determine the first dynamic memory iteration matrix. Step S4 may further include:

[0087] A criterion-level index deviation analysis is performed on the first and second monitoring weight structures to obtain a set of first criterion-level index deviation coefficients.

[0088] The set of deviation coefficients of the first criterion-level indicators is normalized within the set to obtain the set of normalized values ​​of the deviation coefficients of the first criterion-level indicators.

[0089] The first set of deviation coefficient normalization values ​​of the first criterion-level index is filled into the initially empty matrix to obtain the first dynamic memory iteration matrix.

[0090] In one embodiment, the preset window is a pre-defined time period for a single analysis, used by those skilled in the art to dynamically collect data on enterprise safety risks, such as 10 minutes or 30 minutes. Within this window, real-time monitoring data of the coal mine enterprise is read according to a sampling cycle, such as 1 minute, and a monitoring weight structure is generated for each sampling moment based on the pre-constructed weight structure, i.e., a weight snapshot of each criterion at that moment. All snapshots constitute a monitoring weight structure sequence in chronological order.

[0091] Subsequently, Euclidean distance is calculated for the weight snapshots of two adjacent time points in the sequence to determine the magnitude of change for each criterion between these two time points. These changes are then organized into a set according to the criterion order. To avoid inconsistencies in scale between different criteria, this set is internally normalized to reflect the relative intensity of change for each criterion within the time period. After processing, this set is filled into an empty matrix, which becomes the dynamic memory iteration matrix for the current time point, used to record the overall shape of this weight perturbation.

[0092] For each pair of adjacent time points within the window, the above steps are repeated, generating a new dynamic memory iteration matrix each time. After generating the second matrix, the system also interacts and enhances the previous matrix with the current matrix. That is, through element-wise fusion between matrices, the perturbation information of adjacent periods can be mutually reinforced, thus preserving short-term sudden changes without being amplified by accidental noise.

[0093] Preferably, the network layer constructed based on the feedforward neural network is trained by using multiple sample first dynamic memory iteration matrices, multiple sample second dynamic memory iteration matrices, and multiple sample augmentation dynamic memory iteration matrices as training data, with the sample first dynamic memory iteration matrices and sample second dynamic memory iteration matrices as inputs and the sample augmentation dynamic memory iteration matrix as outputs.

[0094] Specifically, the input layer takes two matrices as input, expanded into vector form (e.g., expanded row-wise or according to criteria). It receives and organizes feature information from two time points, enabling subsequent network layers to perceive it simultaneously. It unifies the format of the first and second dynamic memory iteration matrices, allowing the network to identify structural differences and trends between the two matrices. The feature fusion layer, also known as the first hidden layer, contains several neurons and fuses features from the two input matrices. It extracts key change features based on differences, similarities, and directional changes, fusing the criterion change patterns of the first and second matrices to provide preliminary trend and perturbation information for subsequent analysis, such as which criteria continuously strengthen and which suddenly decrease. The deep feature extraction layer, also known as the second hidden layer, typically contains more neurons than the first hidden layer. It extracts deeper-level change patterns, such as stable changes, anomalous jumps, trend consistency, and noise patterns. By learning the temporal structural logic of the dynamic memory iteration matrix, such as continuously strengthening trends, sudden anomalous oscillations, and inter-criterion linkages, the network can determine which information should be retained or amplified, and which should be filtered out as noise. The enhancement pattern modeling layer, the third hidden layer, simulates the behavior of matrix interaction enhancement, such as changes in consistent enhancement directions and changes in conflicting suppression directions. The output layer has the same number of neurons as the enhancement dynamic memory iteration matrix. The output is the expanded form of the enhancement matrix generated by the network, which is then reassembled into a matrix structure in post-processing.

[0095] The trained network layers are used to perform iterative matrix interaction enhancement on the first dynamic memory iteration matrix and the second dynamic memory iteration matrix to obtain the second enhanced dynamic memory iteration matrix. All matrices in the entire window are subjected to dynamic memory iteration and interaction enhancement processing in sequence, and finally an enhanced dynamic memory iteration matrix is ​​obtained at the end of the window. This matrix intuitively presents the fluctuation trend and perturbation strength of each criterion in the time dimension within the window.

[0096] By analyzing the intensity and consistency of changes in each criterion term in the Nth enhanced dynamic memory iteration matrix, the overall coordination status is determined. If the changes in each criterion are consistent in direction and similar in magnitude, the overall coordination is considered good. If some criteria deviate significantly from other criteria, it indicates a significant imbalance and decreased coordination. Based on the above results, the coupling coordination degree is finally output to reflect the overall coordination level of coal mine safety status within the window and its sensitivity to risk.

[0097] The degree of coupling coordination reflects the operational status of the risk system. For example, a high degree of coordination indicates that the changing trends among the criteria are consistent and there are no obvious conflicts. In this case, the system is operating stably, and the initial risk assessment score should be appropriately lowered to reflect a good risk status. Conversely, a low degree of coordination indicates that some key criteria have significant deviations within the window, or that disturbances have been amplified by the enhancement matrix, resulting in a systemic transmission phenomenon. In this case, the system is unstable, and the initial risk assessment score should be increased to fully reflect the potential risk amplification effect. During the correction process, not only the overall level of coordination should be considered, but the duration of sudden anomalies within the window, the number of anomalous criteria, and the intensity of conflicts between criteria can also be used to adjust the correction magnitude. This ensures that the corrected score reflects the impact of local anomalies on the overall risk while avoiding exaggeration of risk due to short-term noise.

[0098] Example 2: Based on the same inventive concept as the coupled coordination-driven dynamic assessment method for enterprise security risks in the foregoing examples, this invention provides a coupled coordination-driven dynamic assessment system for enterprise security risks. See [link to example]. Figure 2 As shown, the system includes:

[0099] Clustering iterative screening module 11 is used to determine the safety risk candidate indicator pool of the target coal mining enterprise, perform phased clustering iterative screening on the safety risk candidate indicator pool, and construct a risk assessment indicator set.

[0100] The structure processing module 12 is used to perform progressive hierarchical structure processing on the risk assessment index set using the analytic hierarchy process, and to determine the weight system structure, the index layer weight set, and the criterion layer weight set.

[0101] Risk quantification assessment module 13 is used to perform fuzzy risk quantification assessment by combining the weight system structure, indicator layer weight set and criterion layer weight set to determine the initial risk assessment score;

[0102] The coupling coordination degree determination module 14 is used to perform dynamic memory iterative coupling coordination analysis on the target coal mining enterprise based on the weight system architecture within a preset window to determine the coupling coordination degree.

[0103] The dynamic correction module 15 is used to dynamically correct the initial risk assessment score based on the coupling coordination degree, and determine the dynamic assessment result of the target enterprise's security risk.

[0104] In one implementation, the cluster iterative screening module 11 is used to perform the following steps:

[0105] The first stage of base class indicator identification is performed on the security risk candidate indicator pool to determine K base class indicators. The base class indicators are the K indicators in the security risk candidate indicator pool whose mean Euclidean distance from other indicators is ranked from largest to smallest, and K is a positive integer greater than or equal to 3.

[0106] The K base class indices are subjected to a second-stage nearest neighbor ablation processing to obtain M ablation base class indices, where M is a positive integer less than or equal to K;

[0107] Based on the M ablation base class indicators, a third-stage cluster analysis and intra-cluster iterative screening are performed on the pool of candidate safety risk indicators to determine the risk assessment indicator set.

[0108] In one implementation, the cluster iterative screening module 11 is used to perform the following steps:

[0109] Each of the K base class indices is subjected to pairwise nearest neighbor ablation coefficient identification to obtain a set of K nearest neighbor ablation coefficients.

[0110] Calculate the mean of the set of K nearest neighbor ablation coefficients, determine the mean of the K nearest neighbor ablation coefficients, and take the base class index corresponding to the maximum value of the K nearest neighbor ablation coefficients as the ablation guidance base class index;

[0111] Using the ablation guidance base class index as an index, the base class index corresponding to the nearest neighbor ablation coefficient that is greater than or equal to the preset nearest neighbor ablation coefficient threshold in the corresponding nearest neighbor ablation coefficient set of the K nearest neighbor ablation coefficient sets is subjected to nearest neighbor ablation processing to obtain M ablation base class indices.

[0112] In one implementation, the cluster iterative screening module 11 is used to perform the following steps:

[0113] The indicators in the safety risk candidate indicator pool are added to the set corresponding to the ablation base class indicator with the closest Euclidean distance among the M ablation base class indicators, to obtain the M ablation base class indicator cluster set;

[0114] Using the M ablation base class indicators as the starting point for iterative screening within the M clusters, iterative screening is performed on the cluster sets of the M ablation base class indicators until the preset iterative screening stop condition is met, thereby obtaining M representative indicators.

[0115] By summarizing the M representative indicators, the risk assessment indicator set is obtained.

[0116] The preset iterative screening stopping condition is that the index neighborhood size in the current iteration is less than the index neighborhood size in the previous iteration and / or the number of iterations is greater than the maximum number of iterations.

[0117] In one implementation, the cluster iterative screening module 11 is used to perform the following steps:

[0118] According to the preset intra-cluster iteration bandwidth, construct the neighborhood of the M intra-cluster iteration screening starting point and perform initial iteration on the M intra-cluster iteration screening starting point to obtain the M intra-cluster iteration screening stage points.

[0119] Traverse the M cluster-in-cluster iterative screening stage points and construct neighborhoods based on the preset cluster-in-cluster iterative bandwidth to obtain the neighborhoods of the M cluster-in-cluster iterative screening stage points.

[0120] Compare the neighborhood size of the M cluster-in-cluster iterative screening stage point neighborhood with the neighborhood size of the M cluster-in-cluster iterative screening start point neighborhood to determine whether the preset iterative screening stop condition is met. If so, the M cluster-in-cluster iterative screening start points are taken as M representative indicators.

[0121] If not, continue iterating over the M intra-cluster iterative screening stage points according to the preset intra-cluster iterative bandwidth, and count the number of iterations. Continue in this manner until the preset iterative screening stop condition is no longer met, and obtain the M representative indicators.

[0122] In one implementation, the structure processing module 12 is used to perform the following steps:

[0123] According to the risk assessment objectives, the risk assessment indicator set is divided into a three-layer progressive structure of target layer, criterion layer and indicator layer to obtain a weighted system structure.

[0124] By traversing each indicator within each criterion in the weight system architecture, pairwise importance comparisons are performed to determine the indicator comparison judgment matrix, and weight vectors are calculated to obtain the indicator layer weight set.

[0125] The system iterates through multiple criteria in the weighting architecture, performs pairwise importance comparisons, constructs an N×N criterion comparison judgment matrix, and calculates weight vectors to obtain the criterion layer weight set, where N is a positive integer.

[0126] In one implementation, the risk quantification assessment module 13 is used to perform the following steps:

[0127] Pre-construct a safety risk level evaluation set and the corresponding risk scale vector;

[0128] By traversing the weight system structure, fuzzy membership analysis of the indicator layer of the security risk level evaluation set is performed to obtain the indicator membership vector set.

[0129] Based on the index membership vector set, a fuzzy judgment matrix is ​​constructed, and combined with the index layer weight set, a criterion layer fuzzy synthesis is performed to obtain the criterion layer membership vector set.

[0130] Matrix synthesis is performed based on the criterion layer membership vector set, and target layer fuzzy synthesis is performed by combining the criterion layer weight set to obtain the target layer membership vector.

[0131] The risk scale vector is used to defuzzify the membership vector of the target layer to obtain the initial risk assessment score.

[0132] In one implementation, the coupling coordination degree determination module 14 is used to perform the following steps:

[0133] Within a preset window, the target coal mining enterprise is monitored in accordance with the weight system architecture to obtain a monitoring weight system architecture sequence.

[0134] Perform criterion-level dynamic memory iteration on the first and second monitoring weight architectures in the monitoring weight architecture sequence to determine the first dynamic memory iteration matrix;

[0135] Perform criterion-level dynamic memory iteration on the second and third monitoring weight architectures in the monitoring weight architecture sequence to determine the second dynamic memory iteration matrix;

[0136] The second dynamic memory iteration matrix is ​​enhanced by iterative matrix interaction using the first dynamic memory iteration matrix to obtain the second enhanced dynamic memory iteration matrix.

[0137] By analogy, after multiple iterations of criterion-level dynamic memory and interactive enhancement of the iteration matrix, the Nth enhanced dynamic memory iteration matrix is ​​determined.

[0138] Based on the Nth enhanced dynamic memory iteration matrix, a coupling coordination analysis is performed to determine the coupling coordination degree.

[0139] In one implementation, the coupling coordination degree determination module 14 is used to perform the following steps:

[0140] A criterion-level index deviation analysis is performed on the first and second monitoring weight structures to obtain a set of first criterion-level index deviation coefficients.

[0141] The set of deviation coefficients of the first criterion-level indicators is normalized within the set to obtain the set of normalized values ​​of the deviation coefficients of the first criterion-level indicators.

[0142] The first set of deviation coefficient normalization values ​​of the first criterion-level index is filled into the initially empty matrix to obtain the first dynamic memory iteration matrix.

[0143] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0144] The foregoing description of specific exemplary embodiments of the invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it will be apparent that many changes and variations can be made in accordance with the foregoing teachings. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling those skilled in the art to implement and utilize various different exemplary embodiments of the invention, as well as various different choices and variations. The scope of the invention is intended to be defined by the claims and their equivalents.

Claims

1. A dynamic assessment method for enterprise security risks driven by coupling and coordination, characterized in that, The method includes: A pool of candidate safety risk indicators for the target coal mining enterprise is determined, and the pool of candidate safety risk indicators is subjected to phased clustering and iterative screening to construct a risk assessment indicator set. The risk assessment indicator set is processed into a progressive hierarchical structure using the analytic hierarchy process (AHP) to determine the weight system structure, indicator layer weight set, and criterion layer weight set. Fuzzy risk quantification assessment is performed by combining the weight system structure, indicator layer weight set, and criterion layer weight set to determine the initial risk assessment score; Within a preset window, dynamic memory iterative coupling coordination analysis is performed on the target coal mining enterprise based on the weight system architecture to determine the coupling coordination degree. The initial risk assessment score is dynamically corrected based on the coupling coordination degree to determine the dynamic assessment result of the target enterprise's security risk.

2. The enterprise security risk dynamic assessment method driven by coupling and coordination as described in claim 1, characterized in that, A pool of candidate safety risk indicators for the target coal mining enterprise is determined, and the pool is subjected to phased clustering and iterative screening to construct a risk assessment indicator set, including: The first stage of base class indicator identification is performed on the security risk candidate indicator pool to determine K base class indicators. The base class indicators are the K indicators in the security risk candidate indicator pool whose mean Euclidean distance from other indicators is ranked from largest to smallest, and K is a positive integer greater than or equal to 3. The K base class indices are subjected to a second-stage nearest neighbor ablation processing to obtain M ablation base class indices, where M is a positive integer less than or equal to K; Based on the M ablation base class indicators, a third-stage cluster analysis and intra-cluster iterative screening are performed on the pool of candidate safety risk indicators to determine the risk assessment indicator set.

3. The enterprise security risk dynamic assessment method driven by coupling and coordination as described in claim 2, characterized in that, The K base class indices are subjected to a second-stage nearest neighbor ablation processing to obtain M ablation base class indices, where M is a positive integer less than or equal to K, including: Each of the K base class indices is subjected to pairwise nearest neighbor ablation coefficient identification to obtain a set of K nearest neighbor ablation coefficients. Calculate the mean of the set of K nearest neighbor ablation coefficients, determine the mean of the K nearest neighbor ablation coefficients, and take the base class index corresponding to the maximum value of the K nearest neighbor ablation coefficients as the ablation guidance base class index; Using the ablation guidance base class index as an index, the base class index corresponding to the nearest neighbor ablation coefficient that is greater than or equal to the preset nearest neighbor ablation coefficient threshold in the corresponding nearest neighbor ablation coefficient set of the K nearest neighbor ablation coefficient sets is subjected to nearest neighbor ablation processing to obtain M ablation base class indices.

4. The enterprise security risk dynamic assessment method driven by coupling and coordination as described in claim 3, characterized in that, Based on the M ablation base class indicators, a third-stage cluster analysis and intra-cluster iterative screening are performed on the safety risk candidate indicator pool to determine the risk assessment indicator set, including: The indicators in the safety risk candidate indicator pool are added to the set corresponding to the ablation base class indicator with the closest Euclidean distance among the M ablation base class indicators, to obtain the M ablation base class indicator cluster set; Using the M ablation base class indicators as the starting point for iterative screening within the M clusters, iterative screening is performed on the cluster sets of the M ablation base class indicators until the preset iterative screening stop condition is met, thereby obtaining M representative indicators. By summarizing the M representative indicators, the risk assessment indicator set is obtained. The preset iterative screening stopping condition is that the index neighborhood size in the current iteration is less than the index neighborhood size in the previous iteration and / or the number of iterations is greater than the maximum number of iterations.

5. The enterprise security risk dynamic assessment method driven by coupling and coordination as described in claim 4, characterized in that, Using the M ablation base class indicators as the starting point for iterative screening within the M clusters, the clusters of the M ablation base class indicators are iteratively screened until a preset iterative screening stopping condition is met, resulting in M ​​representative indicators, including: According to the preset intra-cluster iteration bandwidth, construct the neighborhood of the M intra-cluster iteration screening starting point and perform initial iteration on the M intra-cluster iteration screening starting point to obtain the M intra-cluster iteration screening stage points. Traverse the M cluster-in-cluster iterative screening stage points and construct neighborhoods based on the preset cluster-in-cluster iterative bandwidth to obtain the neighborhoods of the M cluster-in-cluster iterative screening stage points. Compare the neighborhood size of the M cluster-in-cluster iterative screening stage point neighborhood with the neighborhood size of the M cluster-in-cluster iterative screening start point neighborhood to determine whether the preset iterative screening stop condition is met. If so, the M cluster-in-cluster iterative screening start points are taken as M representative indicators. If not, continue iterating over the M intra-cluster iterative screening stage points according to the preset intra-cluster iterative bandwidth, and count the number of iterations. Continue in this manner until the preset iterative screening stop condition is no longer met, and obtain the M representative indicators.

6. The enterprise security risk dynamic assessment method driven by coupling and coordination as described in claim 1, characterized in that, The risk assessment indicator set is processed using the analytic hierarchy process (AHP) to create a progressive hierarchical structure, determining the weight system structure, indicator layer weight set, and criterion layer weight set, including: According to the risk assessment objectives, the risk assessment indicator set is divided into a three-layer progressive structure of target layer, criterion layer and indicator layer to obtain a weighted system structure. By traversing each indicator within each criterion in the weight system architecture, pairwise importance comparisons are performed to determine the indicator comparison judgment matrix, and weight vectors are calculated to obtain the indicator layer weight set. The system iterates through multiple criteria in the weighting architecture, performs pairwise importance comparisons, constructs an N×N criterion comparison judgment matrix, and calculates weight vectors to obtain the criterion layer weight set, where N is a positive integer.

7. The enterprise security risk dynamic assessment method driven by coupling and coordination as described in claim 1, characterized in that, By combining the aforementioned weighting system architecture, indicator layer weight set, and criterion layer weight set, a fuzzy risk quantification assessment is performed to determine the initial risk assessment score, including: Pre-construct a safety risk level evaluation set and the corresponding risk scale vector; By traversing the weight system structure, fuzzy membership analysis of the indicator layer of the security risk level evaluation set is performed to obtain the indicator membership vector set. Based on the index membership vector set, a fuzzy judgment matrix is ​​constructed, and combined with the index layer weight set, a criterion layer fuzzy synthesis is performed to obtain the criterion layer membership vector set. Matrix synthesis is performed based on the criterion layer membership vector set, and target layer fuzzy synthesis is performed by combining the criterion layer weight set to obtain the target layer membership vector. The risk scale vector is used to defuzzify the membership vector of the target layer to obtain the initial risk assessment score.

8. The enterprise security risk dynamic assessment method driven by coupling and coordination as described in claim 1, characterized in that, Within a preset window, dynamic memory iterative coupling coordination analysis is performed on the target coal mining enterprise based on the weighted system architecture to determine the coupling coordination degree, including: Within a preset window, the target coal mining enterprise is monitored in accordance with the weight system architecture to obtain a monitoring weight system architecture sequence. Perform criterion-level dynamic memory iteration on the first and second monitoring weight architectures in the monitoring weight architecture sequence to determine the first dynamic memory iteration matrix; Perform criterion-level dynamic memory iteration on the second and third monitoring weight architectures in the monitoring weight architecture sequence to determine the second dynamic memory iteration matrix; The second dynamic memory iteration matrix is ​​enhanced by iterative matrix interaction using the first dynamic memory iteration matrix to obtain the second enhanced dynamic memory iteration matrix. By analogy, after multiple iterations of criterion-level dynamic memory and interactive enhancement of the iteration matrix, the Nth enhanced dynamic memory iteration matrix is ​​determined. Based on the Nth enhanced dynamic memory iteration matrix, a coupling coordination analysis is performed to determine the coupling coordination degree.

9. The enterprise security risk dynamic assessment method driven by coupling and coordination as described in claim 8, characterized in that, Perform criterion-level dynamic memory iteration on the first and second monitoring weight architectures in the monitoring weight architecture sequence to determine the first dynamic memory iteration matrix, including: A criterion-level index deviation analysis is performed on the first and second monitoring weight structures to obtain a set of first criterion-level index deviation coefficients. The set of deviation coefficients of the first criterion-level indicators is normalized within the set to obtain the set of normalized values ​​of the deviation coefficients of the first criterion-level indicators. The first set of deviation coefficient normalization values ​​of the first criterion-level index is filled into the initially empty matrix to obtain the first dynamic memory iteration matrix.

10. A coupled and coordinated enterprise security risk dynamic assessment system, characterized in that, For implementing the method steps of any one of claims 1 to 9, including: The clustering iterative screening module is used to determine the safety risk candidate indicator pool of the target coal mining enterprise, and to perform phased clustering iterative screening on the safety risk candidate indicator pool to construct a risk assessment indicator set. The structure processing module is used to perform progressive hierarchical structure processing on the risk assessment index set using the analytic hierarchy process, and to determine the weight system structure, the index layer weight set, and the criterion layer weight set. The risk quantification assessment module is used to perform fuzzy risk quantification assessment by combining the weight system structure, indicator layer weight set and criterion layer weight set to determine the initial risk assessment score. The coupling coordination degree determination module is used to perform dynamic memory iterative coupling coordination analysis on the target coal mining enterprise based on the weight system architecture within a preset window to determine the coupling coordination degree. The dynamic correction module is used to dynamically correct the initial risk assessment score based on the coupling coordination degree, and determine the dynamic assessment result of the target enterprise's security risk.

Images