A method and system for constructing a multi-dimensional risk management maturity model

Abstract

The application discloses a kind of multidimensional risk management maturity model construction method and system, including the monitoring data and risk management data of the acquisition preset power system, the monitoring data and the risk management data are preprocessed;Risk identification is carried out to the monitoring data to obtain risk data, risk type is obtained by priori algorithm through the risk data, control data is obtained by power prediction through the risk management data;Dynamic causal matching is carried out to the control data and the risk type to obtain matching degree, management data and fuzzy data are obtained, according to the historical data and the fuzzy data, longitudinal comparison and evaluation are carried out to obtain optimization degree, the fuzzy data of the optimization degree greater than optimization threshold is added to the management data;According to the management data, horizontal longitudinal maturity comprehensive comparison is constructed to form maturity evaluation function, and the power risk management maturity model is constructed according to the maturity evaluation function.

Description

Technical Field

[0001] This invention relates to the field of risk management, and in particular to a method and system for constructing a multi-dimensional risk management maturity model. Background Technology

[0002] With the transformation of the energy structure and the rapid development of new power systems, the risks faced by power systems are becoming increasingly complex and dynamic. Traditional power system risk management mainly relies on static analysis of a single risk dimension, which is insufficient to address the coupling effects of multi-source heterogeneous risks. Existing risk management models are mostly based on linear regression of historical data or constructed using a single indicator system, lacking the ability to quantitatively analyze the dynamic correlations and causal relationships of multi-dimensional risks, resulting in delayed risk identification and insufficient adaptability of control measures.

[0003] At the technical level, the existing risk management framework has the following bottlenecks: First, the data fusion mechanism is imperfect. Monitoring data and risk management data have not yet formed a unified multi-dimensional feature extraction and standardized processing flow, which restricts the efficiency of risk collaborative analysis. Second, risk type identification relies heavily on expert experience or single algorithm models, which have limited ability to distinguish between "single risk" and "multiple risk coupling" scenarios and lack a dynamic classification mechanism based on prior knowledge base. Third, the matching of control strategies and risk types relies on a static rule base, which cannot adapt to the dynamic changes in the real-time operating status of the power system and lacks self-optimization capabilities for fuzzy data.

[0004] Existing maturity models often employ fixed weight allocation and horizontal benchmarking, neglecting the evolutionary patterns of historical data and the dynamic optimization paths of risk management measures. Furthermore, traditional causal reasoning methods struggle to quantify the nonlinear characteristics of risk transmission paths when dealing with risk events across multiple time scales, resulting in insufficient accuracy and adaptability in the construction of maturity assessment functions. Therefore, there is an urgent need to construct a multi-dimensional risk management maturity model that integrates dynamic analysis of multi-source data, intelligent matching of risk types, and self-optimizing iteration of fuzzy data. This model would address the shortcomings of existing methods in data fusion, dynamic causal correlation, and cross-cycle assessment, providing systematic and forward-looking decision support for risk prevention and control in new power systems. Summary of the Invention

[0005] The purpose of this invention is to provide a method for constructing a multi-dimensional risk management maturity model.

[0006] To achieve the above objectives, the present invention is implemented according to the following technical solution:

[0007] This invention includes the following steps:

[0008] The system collects monitoring data and risk management data from a pre-defined power system, and preprocesses the monitoring data and risk management data. The risk management data includes equipment failure risk management, fuel supply risk management, power grid operation safety risk management, distributed resource aggregation risk management, and power supply shortage or surplus risk management. The monitoring data includes data to be analyzed and historical data.

[0009] Risk data is obtained by performing risk identification on the monitoring data, and risk types are obtained by using a priori algorithms through the risk data. Power forecasting is then performed using the risk management data to obtain control data. The risk types include single risk types and multiple risk types.

[0010] Dynamic causal matching is performed on the control data and the risk type to obtain the matching degree. Control data with a matching degree greater than the matching threshold is used as management data, and otherwise as fuzzy data. The degree of optimization is obtained by longitudinal comparison and evaluation based on the historical data and the fuzzy data. The fuzzy data with an optimization degree greater than the optimization threshold is added to the management data.

[0011] A maturity assessment function is constructed based on a comprehensive horizontal and vertical maturity comparison of the management data. A power risk management maturity model is then built based on the maturity assessment function, and the target model is output.

[0012] Furthermore, a method for obtaining risk data by risk identification of the monitoring data includes:

[0013] A decision tree algorithm is used to construct a decision tree based on equipment operating parameters and maintenance records. Equipment failure risks are identified through the score rules of the decision tree. A time series analysis algorithm is used to model fuel consumption data, inventory data, and supply cycles. By analyzing the trend, seasonality, and periodic characteristics of historical data, fuel demand and supply are predicted. When supply is lower than demand, a fuel supply risk is identified. An association rule mining algorithm is used to identify grid operation safety risks based on the association rules between parameters in grid operation data, identifying those that do not meet the association rules. A fuzzy comprehensive evaluation algorithm is used to quantify stability and environmental factors, establishing a fuzzy evaluation matrix combined with objective weights to evaluate distributed resource aggregation risks. When the evaluation value is greater than a threshold, a distributed resource aggregation risk is identified. A Monte Carlo simulation algorithm is used to simulate power supply and demand. Through random sampling and simulation calculations, the probability distribution of power supply and demand is obtained. When the power distribution is higher than a threshold, a power supply shortage or surplus risk is identified.

[0014] Furthermore, a method for obtaining a matching degree by dynamically matching the control data and the risk type includes:

[0015] The control data is sorted by time, and the sorted control data is divided into time windows. Control variables are obtained from the control data, and risk variables and risk factors are obtained according to the risk type.

[0016] Construct a causal graph, using the control variables and risk variables that are added or eliminated as the time window slides as nodes of the causal tree, and the causal relationships between nodes as edges of the causal graph;

[0017] We employ weighted auto-resampled sampling for control data of sparse control variables, conduct prior conditional independence tests on continuous risk factors, and perform chi-square tests on discrete multi-risk factors to obtain the cumulative statistic for conditional independence test optimization, the regression coefficient matrix of the structure vector autoregressive model, and the residual covariance.

[0018] Non-instantaneous causal relationships can be directly determined through temporal rules, while instantaneous causal relationships can be obtained by examining the relationship between control variables within the same time window through a structural vector autoregression model.

[0019] Calculate the single risk matching degree:

[0020] ,

[0021] The exponential decay of the lag w is: The maximum lag period is The single risk matching degree is The standard deviation of the causal influence coefficient is The single risk variable at time t is The risk factors at time tw are Risk factors For a single risk variable The causal influence coefficient is The dynamically adjusted weight at time t is The forgetting factor is ;

[0022] Principal component analysis (PCA) is performed to reduce the dimensionality of multiple risk types, obtain the principal components, and calculate the multi-risk matching degree.

[0023] ,

[0024] The a-th principal component is The number of principal components is The variance contribution rate of the a-th principal component is The predictor variable is Predictor variables For principal components The Granger causality test results are Multiple risk matching degree is ;

[0025] Calculate the rolling mean and standard deviation based on the matching degree, and set an adaptive threshold. The expression is:

[0026] ,

[0027] Where the convergence rate coefficient of the control threshold is... The rolling mean is The standard deviation is The current time is t, and the adaptive threshold is... ;

[0028] When the matching degree is lower than the adaptive threshold, nodes are dynamically added or removed, and the conditional independence test is updated until all control data and risk types are traversed, and the matching degree is output.

[0029] Furthermore, a method for obtaining the degree of optimization by performing a longitudinal comparative evaluation based on the historical data and the fuzzy data includes:

[0030] The fuzzy data is classified according to the monitoring type. The fuzzy data far from the classification center is separated by a Gaussian mixture model to obtain classified data. The monitoring types include environmental parameters, operating parameters, economic factors and equipment status.

[0031] Align historical data and categorized data according to timestamps, and calculate the overall optimization level based on objective weighting:

[0032] ,

[0033] The overall optimization level of the b-th risk management is: The upper limit of the monitoring time is The monitoring started at the time of The monitoring deadline is The response time for the bth risk management event is The data for the x-th category in the z-th category is Categorized data The value at time t is The value of the x-th historical data at time t is The number of categories is The xth weight coefficient is .

[0034] Furthermore, the method for constructing a maturity assessment function based on a comprehensive horizontal and vertical maturity comparison of the management data includes:

[0035] Obtain risk management data from different regions of other power systems as horizontal comparison data, and use management data from the same power learning platform at different points in time as vertical comparison data, and normalize the horizontal comparison data;

[0036] The trend of change in longitudinal comparative data was calculated using the exponential smoothing method:

[0037] ;

[0038] Where the smoothness coefficient is The trend value is h, and the trend value at time t-1 is h. The longitudinal comparison data at time t is ;

[0039] A cross-sectional maturity assessment was conducted based on cross-sectional comparison data to obtain the cross-sectional maturity score, expressed as follows:

[0040] ,

[0041] Among them, horizontal maturity is The number of cross-sectional comparison data is The i-th horizontal comparison data is The reference value for the i-th horizontal comparison data is The response time of the i-th horizontal comparison data is ;

[0042] Based on longitudinal comparison data and trends, a longitudinal maturity assessment is conducted to obtain the longitudinal maturity level, expressed as:

[0043] ,

[0044] Among them, vertical maturity is The number of longitudinal comparison data is The upper limit of longitudinal observation time is The longitudinal comparison data at time t is The reference value for the longitudinal comparison data at time t is The trend of the vth longitudinal comparison data at time t is as follows: The response time of the vth longitudinal comparison data is The reference response time for the vth longitudinal comparison data is ;

[0045] A maturity assessment function is constructed based on horizontal and vertical maturity levels, and its expression is as follows:

[0046] ,

[0047] The overall optimization level of the b-th risk management is: The effectiveness of the risk management strategy in the bth instance is: The number of risk management items is M, and the optimization term is... The horizontal weight is The vertical weight is Optimize weights as The mapping function is The maturity assessment function for the b-th risk management is: The vertical maturity level of the bth risk management is The horizontal maturity level of the bth risk management is .

[0048] Furthermore, the method for constructing a power risk management maturity model based on the maturity assessment function includes:

[0049] By managing data, multi-dimensional evaluation indicators are selected, and the weights of the evaluation indicators are obtained using the entropy weight method. The evaluation indicators include equipment performance indicators, power grid operation indicators, economic indicators, environmental protection indicators, and user satisfaction indicators.

[0050] The objective function for constructing the power risk management maturity model is expressed as follows:

[0051] ,

[0052] The objective function for the b-th risk management is: The maturity assessment function for the j-th assessment indicator in the b-th risk management is: The j-th evaluation indicator for the b-th risk management is The loss function is The number of evaluation indicators is D;

[0053] Maturity levels are categorized based on the objective function: a score less than 0.219 indicates risk-free power system management; a score greater than 0.219 but less than 0.417 indicates level four power system risk management capability; a score greater than 0.417 but less than 0.609 indicates level three; a score greater than 0.609 but less than 0.794 indicates level two; and a score greater than 0.794 indicates level one.

[0054] Secondly, a system for constructing a multi-dimensional risk management maturity model includes:

[0055] Data acquisition module: used to collect monitoring data and risk management data of a preset power system, and to preprocess the monitoring data and risk management data; the risk management data includes equipment failure risk management, fuel supply risk management, power grid operation safety risk management, distributed resource aggregation risk management, and power supply shortage or surplus risk management; the monitoring data includes data to be analyzed and historical data;

[0056] Risk type identification module: performs risk identification on the monitoring data to obtain risk data, uses a priori algorithm to obtain risk type through the risk data, and performs power forecasting through the risk management data to obtain control data; the risk type includes single risk type and multiple risk type;

[0057] Prediction and matching module: used to obtain control data by power forecasting through the risk management data, perform dynamic causal matching between the control data and the risk type to obtain the matching degree, and use control data with a matching degree greater than the matching threshold as management data, and otherwise as fuzzy data;

[0058] The comparison and evaluation module performs dynamic causal matching on the control data and the risk type to obtain the matching degree. Control data with a matching degree greater than the matching threshold is used as management data, and otherwise as fuzzy data. The module performs longitudinal comparison and evaluation on the historical data and the fuzzy data to obtain the degree of optimization. The fuzzy data with an optimization degree greater than the optimization threshold is added to the management data.

[0059] Modeling and optimization module: Construct a maturity assessment function based on a comprehensive horizontal and vertical maturity comparison of the management data, build a power risk management maturity model based on the maturity assessment function, and output the target model.

[0060] The beneficial effects of this invention are:

[0061] This invention provides a method and system for constructing a multi-dimensional risk management maturity model. Compared with existing technologies, this invention has the following technical advantages:

[0062] This invention improves the accuracy of risk management maturity model construction through preprocessing, risk identification, risk type acquisition, power forecasting, dynamic causal matching, longitudinal comparative assessment, maturity assessment function construction, and model building steps. This enhances the precision of risk management maturity model construction, significantly saves resources, improves work efficiency, and enables the scientific construction of risk management maturity models. Real-time dynamic causal matching and maturity assessment function construction are crucial for risk management maturity model development. This invention is adaptable to the construction of risk management maturity models with different standards and to meet diverse needs, demonstrating a degree of universality. Attached Figure Description

[0063] Figure 1 This is a flowchart illustrating the steps involved in constructing a multi-dimensional risk management maturity model according to the present invention. Detailed Implementation

[0064] The present invention will be further described below through specific embodiments. The illustrative embodiments and descriptions herein are used to explain the present invention, but are not intended to limit the present invention.

[0065] The present invention provides a method and system for constructing a multi-dimensional risk management maturity model, comprising the following steps:

[0066] like Figure 1 As shown, this embodiment includes the following steps:

[0067] The system collects monitoring data and risk management data from a pre-defined power system, and preprocesses the monitoring data and risk management data. The risk management data includes equipment failure risk management, fuel supply risk management, power grid operation safety risk management, distributed resource aggregation risk management, and power supply shortage or surplus risk management. The monitoring data includes data to be analyzed and historical data.

[0068] In actual assessments, the data to be analyzed includes transformer operating temperature, line power transmission value, generator speed, raw material consumption data, raw material inventory data, raw material supply cycle information, voltage, current, power factor, voltage stability, irradiance, wind speed, power generation, and electricity consumption; historical data includes historical operating status, maintenance records, historical consumption of raw materials, raw material inventory, raw material supply, historical operating data, and historical data on distributed resources and power supply and demand.

[0069] This study takes a power system as the research object, uses the period from May 2021 to May 2024 as historical data, and uses the period from June 2024 as data to be analyzed; the risk management mainly involves generator speed, solar irradiance, and wind speed.

[0070] Risk data is obtained by performing risk identification on the monitoring data, and risk types are obtained by using a priori algorithms through the risk data. Power forecasting is then performed using the risk management data to obtain control data. The risk types include single risk types and multiple risk types.

[0071] In actual assessments, the control data include transformer operating temperature, line power transmission value, generator speed, voltage stability, irradiance, and wind speed.

[0072] Dynamic causal matching is performed on the control data and the risk type to obtain the matching degree. Control data with a matching degree greater than the matching threshold is used as management data, and otherwise as fuzzy data. The degree of optimization is obtained by longitudinal comparison and evaluation based on the historical data and the fuzzy data. The fuzzy data with an optimization degree greater than the optimization threshold is added to the management data.

[0073] A maturity assessment function is constructed based on a comprehensive horizontal and vertical maturity comparison of the management data. A power risk management maturity model is then built based on the maturity assessment function, and the target model is output.

[0074] In this embodiment, the method for obtaining risk data by risk identification of the monitoring data includes:

[0075] A decision tree algorithm is used to construct a decision tree based on equipment operating parameters and maintenance records. Equipment failure risks are identified through the score rules of the decision tree. A time series analysis algorithm is used to model fuel consumption data, inventory data, and supply cycles. By analyzing the trend, seasonality, and periodic characteristics of historical data, fuel demand and supply are predicted. When supply is lower than demand, a fuel supply risk is identified. An association rule mining algorithm is used to identify grid operation safety risks based on the association rules between parameters in grid operation data, identifying those that do not meet the association rules. A fuzzy comprehensive evaluation algorithm is used to quantify stability and environmental factors, establishing a fuzzy evaluation matrix combined with objective weights to evaluate distributed resource aggregation risks. When the evaluation value is greater than a threshold, a distributed resource aggregation risk is identified. A Monte Carlo simulation algorithm is used to simulate power supply and demand. Through random sampling and simulation calculations, the probability distribution of power supply and demand is obtained. When the power distribution is higher than a threshold, a power supply shortage or surplus risk is identified.

[0076] In this embodiment, the method for obtaining a matching degree by dynamically matching the control data and the risk type includes:

[0077] The control data is sorted by time, and the sorted control data is divided into time windows. Control variables are obtained from the control data, and risk variables and risk factors are obtained according to the risk type.

[0078] Construct a causal graph, using the control variables and risk variables that are added or eliminated as the time window slides as nodes of the causal tree, and the causal relationships between nodes as edges of the causal graph;

[0079] We employ weighted auto-resampled sampling for control data of sparse control variables, conduct prior conditional independence tests on continuous risk factors, and perform chi-square tests on discrete multi-risk factors to obtain the cumulative statistic for conditional independence test optimization, the regression coefficient matrix of the structure vector autoregressive model, and the residual covariance.

[0080] Non-instantaneous causal relationships can be directly determined through temporal rules, while instantaneous causal relationships can be obtained by examining the relationship between control variables within the same time window through a structural vector autoregression model.

[0081] Calculate the single risk matching degree:

[0082] ,

[0083] The exponential decay of the lag w is: The maximum lag period is The single risk matching degree is The standard deviation of the causal influence coefficient is The single risk variable at time t is The risk factors at time tw are Risk factors For a single risk variable The causal influence coefficient is The dynamically adjusted weight at time t is The forgetting factor is ;

[0084] Principal component analysis (PCA) is performed to reduce the dimensionality of multiple risk types, obtain the principal components, and calculate the multi-risk matching degree.

[0085] ,

[0086] The a-th principal component is The number of principal components is The variance contribution rate of the a-th principal component is The predictor variable is Predictor variables For principal components The Granger causality test results are Multiple risk matching degree is ;

[0087] Calculate the rolling mean and standard deviation based on the matching degree, and set an adaptive threshold. The expression is:

[0088] ,

[0089] Where the convergence rate coefficient of the control threshold is... The rolling mean is The standard deviation is The current time is t, and the adaptive threshold is... ;

[0090] When the matching degree is lower than the adaptive threshold, nodes are dynamically added or removed, and the conditional independence test is updated until all control data and risk types are traversed, and the matching degree is output.

[0091] In the actual evaluation, the forgetting factor was 0.297, the maximum lag period was 5, and the matching degree of transformer operating temperature, line power transmission value, generator speed, voltage stability, light intensity, and wind speed were 0.671, 0.594, 0.307, 0.761, 0.293, and 0.458, respectively.

[0092] In this embodiment, the method for obtaining the degree of optimization by performing a longitudinal comparison evaluation based on the historical data and the fuzzy data includes:

[0093] The fuzzy data is classified according to the monitoring type. The fuzzy data far from the classification center is separated by a Gaussian mixture model to obtain classified data. The monitoring types include environmental parameters, operating parameters, economic factors and equipment status.

[0094] Align historical data and categorized data according to timestamps, and calculate the overall optimization level based on objective weighting:

[0095] ,

[0096] The overall optimization level of the b-th risk management is: The upper limit of the monitoring time is The monitoring started at the time of The monitoring deadline is The response time for the bth risk management event is The data for the x-th category in the z-th category is Categorized data The value at time t is The value of the x-th historical data at time t is The number of categories is The xth weight coefficient is ;

[0097] In the actual evaluation, the optimization level was 0.247.

[0098] In this embodiment, the method for constructing a maturity assessment function based on a comprehensive horizontal and vertical maturity comparison of the management data includes:

[0099] Obtain risk management data from different regions of other power systems as horizontal comparison data, and use management data from the same power learning platform at different points in time as vertical comparison data, and normalize the horizontal comparison data;

[0100] The trend of change in longitudinal comparative data was calculated using the exponential smoothing method:

[0101] ;

[0102] Where the smoothness coefficient is The trend value is h, and the trend value at time t-1 is h. The longitudinal comparison data at time t is ;

[0103] A cross-sectional maturity assessment was conducted based on cross-sectional comparison data to obtain the cross-sectional maturity score, expressed as follows:

[0104] ,

[0105] Among them, horizontal maturity is The number of cross-sectional comparison data is The i-th horizontal comparison data is The reference value for the i-th horizontal comparison data is The response time of the i-th horizontal comparison data is ;

[0106] Based on longitudinal comparison data and trends, a longitudinal maturity assessment is conducted to obtain the longitudinal maturity level, expressed as:

[0107] ,

[0108] Among them, vertical maturity is The number of longitudinal comparison data is The upper limit of longitudinal observation time is The longitudinal comparison data at time t is The reference value for the longitudinal comparison data at time t is The trend of the vth longitudinal comparison data at time t is as follows: The response time of the vth longitudinal comparison data is The reference response time for the vth longitudinal comparison data is ;

[0109] A maturity assessment function is constructed based on horizontal and vertical maturity levels, and its expression is as follows:

[0110] ,

[0111] The overall optimization level of the b-th risk management is: The effectiveness of the risk management strategy in the bth instance is: The number of risk management items is M, and the optimization term is... The horizontal weight is The vertical weight is Optimize weights as The mapping function is The maturity assessment function for the b-th risk management is: The vertical maturity level of the bth risk management is The horizontal maturity level of the bth risk management is .

[0112] In this embodiment, the method for constructing a power risk management maturity model based on the maturity assessment function includes:

[0113] By managing data, multi-dimensional evaluation indicators are selected, and the weights of the evaluation indicators are obtained using the entropy weight method. The evaluation indicators include equipment performance indicators, power grid operation indicators, economic indicators, environmental protection indicators, and user satisfaction indicators.

[0114] The objective function for constructing the power risk management maturity model is expressed as follows:

[0115] ,

[0116] The objective function for the b-th risk management is: The maturity assessment function for the j-th assessment indicator in the b-th risk management is: The j-th evaluation indicator for the b-th risk management is The loss function is The number of evaluation indicators is D;

[0117] Maturity levels are categorized based on the objective function: a score less than 0.219 indicates risk-free power system management; a score greater than 0.219 but less than 0.417 indicates level four power system risk management capability; a score greater than 0.417 but less than 0.609 indicates level three; a score greater than 0.609 but less than 0.794 indicates level two; and a score greater than 0.794 indicates level one.

[0118] In actual evaluation, equipment performance indicators include equipment failure rate, maintenance cycle, and reliability; power grid operation indicators include power supply stability, load fluctuation, and response time; economic indicators include production cost, operating cost, and power loss; environmental indicators include carbon emission level and renewable energy ratio; and user satisfaction indicators include customer complaint rate and service response time.

[0119] Secondly, a system for constructing a multi-dimensional risk management maturity model includes:

[0120] Data acquisition module: used to collect monitoring data and risk management data of a preset power system, and to preprocess the monitoring data and risk management data; the risk management data includes equipment failure risk management, fuel supply risk management, power grid operation safety risk management, distributed resource aggregation risk management, and power supply shortage or surplus risk management; the monitoring data includes data to be analyzed and historical data;

[0121] Risk type identification module: performs risk identification on the monitoring data to obtain risk data, uses a priori algorithm to obtain risk type through the risk data, and performs power forecasting through the risk management data to obtain control data; the risk type includes single risk type and multiple risk type;

[0122] Prediction and matching module: used to obtain control data by power forecasting through the risk management data, perform dynamic causal matching between the control data and the risk type to obtain the matching degree, and use control data with a matching degree greater than the matching threshold as management data, and otherwise as fuzzy data;

[0123] The comparison and evaluation module performs dynamic causal matching on the control data and the risk type to obtain the matching degree. Control data with a matching degree greater than the matching threshold is used as management data, and otherwise as fuzzy data. The module performs longitudinal comparison and evaluation on the historical data and the fuzzy data to obtain the degree of optimization. The fuzzy data with an optimization degree greater than the optimization threshold is added to the management data.

[0124] Modeling and optimization module: Construct a maturity assessment function based on a comprehensive horizontal and vertical maturity comparison of the management data, build a power risk management maturity model based on the maturity assessment function, and output the target model.

[0125] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a multi-dimensional risk management maturity model, characterized in that, The method comprises the following steps: Collecting monitoring data and risk management data of a preset power system, and preprocessing the monitoring data and the risk management data; The risk management data comprises equipment failure risk management, fuel supply risk management, power grid operation safety risk management, distributed resource aggregation risk management, power supply shortage or excess risk management; the monitoring data comprises to-be-analyzed data and historical data; Risk identification is performed on the monitoring data to obtain risk data, a priori algorithm is used to obtain a risk type through the risk data, and power prediction is performed through the risk management data to obtain management data; The risk type comprises a single risk type and a multiple risk type; Dynamic causal matching is performed on the management data and the risk type to obtain a matching degree, management data with a matching degree greater than a matching threshold is taken as management data, otherwise, it is taken as fuzzy data, longitudinal comparison and evaluation are performed on the historical data and the fuzzy data to obtain an optimization degree, and the fuzzy data with an optimization degree greater than an optimization threshold is added to the management data; A maturity evaluation function is constructed according to comprehensive comparison of the management data in the horizontal and vertical directions, a power risk management maturity model is constructed according to the maturity evaluation function, and a target model is output; The method for obtaining an optimization degree according to longitudinal comparison and evaluation of the historical data and the fuzzy data comprises: Classifying the fuzzy data according to monitoring types, dividing fuzzy data far away from a classification center through a Gaussian mixture model, and obtaining classified data; the monitoring types comprise environmental parameters, operation parameters, economic factors and equipment states; Aligning the historical data and the classified data according to time stamps, and calculating a comprehensive optimization degree according to objective weighting: ； The comprehensive optimization degree of the bth risk management is , the upper limit of the monitoring time is , the starting time of the monitoring is , the end time of the monitoring is , the response time of the bth risk management is , the xth classification data of the zth classification is , the value of the classification data at the tth time is , the value of the xth historical data at the tth time is , the number of the classifications is , and the xth weight coefficient is ; The method for constructing a maturity evaluation function according to comprehensive comparison of the management data in the horizontal and vertical directions comprises: Obtaining risk management data of different regions of other power systems as horizontal comparison data, obtaining management data of the same power system at different time points as longitudinal comparison data, and normalizing the horizontal comparison data; Calculating a change trend of the longitudinal comparison data by using an exponential smoothing method: ； wherein the smoothing coefficient is , the trend value is h, the trend value at the t-1 time is , and the longitudinal control data at the t time is ; Performing horizontal maturity evaluation according to the horizontal comparison data to obtain horizontal maturity, and the expression is: ； wherein the lateral maturity is , the number of lateral control data is , the ith lateral control data is , the reference value of the ith lateral control data is , and the response time of the ith lateral control data is ; Performing longitudinal maturity evaluation according to the longitudinal comparison data and the change trend to obtain longitudinal maturity, and the expression is: ； wherein the longitudinal maturity is , the number of longitudinal control data is , the upper limit of longitudinal observation time is , the longitudinal control data at the t time is , the reference value of the longitudinal control data at the t time is , the change trend of the vth longitudinal control data at the t time is , the response time of the vth longitudinal control data is , and the reference response time of the vth longitudinal control data is ; Constructing a maturity evaluation function according to the horizontal maturity and the longitudinal maturity, and the expression is: ； wherein the comprehensive optimization degree of the bth risk management is , the strategy effectiveness of the bth risk management is , the number of risk management is M, the optimization item is , the horizontal weight is , the vertical weight is , the optimization weight is , the mapping function is , the maturity assessment function of the bth risk management is , the vertical maturity of the bth risk management is , the horizontal maturity of the bth risk management is .

2. The method of claim 1, wherein the multi-dimensional risk management maturity model is constructed by, The method for obtaining risk data by performing risk identification on the monitoring data comprises: The decision tree algorithm is used to construct a decision tree according to the operation parameters and maintenance records of the equipment, and the equipment failure risk is identified through the score rules of the decision tree; the time series analysis algorithm is used to perform time series modeling on the fuel consumption data, inventory data and supply cycle, the trend, seasonality and periodicity characteristics of the historical data are analyzed, the fuel demand and supply situation are predicted, and the fuel supply risk is determined when the supply situation is lower than the fuel demand; the association rule mining algorithm is used to determine the grid operation safety risk when the parameters in the grid operation data do not meet the association rules; the fuzzy comprehensive evaluation algorithm is used to quantize the stability and environmental factors, a fuzzy evaluation matrix is established, and the distributed resource aggregation risk is evaluated by combining the objective weight, and the distributed resource aggregation risk is determined when the evaluation value is greater than the threshold; the Monte Carlo simulation algorithm is used to simulate the power supply and demand, the probability distribution of the power supply and demand is obtained through random sampling and simulation calculation, and the power supply shortage or excess risk is determined when the power distribution is higher than the threshold.

3. The method of claim 1, wherein the multi-dimensional risk management maturity model is constructed by: The method for dynamically matching the control data and the risk types to obtain the matching degree comprises the following steps: The control data is sorted according to time, and the sorted control data is divided into time windows; control variables are obtained according to the control data, and risk variables and risk factors are obtained according to the risk types; A causal diagram is constructed, and the control variables and risk variables that are newly added or eliminated with the time window sliding are taken as nodes of a causal tree, and the causal relationship between the nodes is taken as edges of the causal diagram; The control data of sparse control variables is resampled by weighted bootstrap, the continuous risk factors are subjected to prior condition independence test, and the discrete multiple risk factors are subjected to chi-square test, to obtain the cumulative statistics, the regression coefficient matrix of the structural vector autoregressive model and the residual covariance of the condition independence test optimization; The non-instantaneous causal relationship is directly determined by the time sequence rule, and the instantaneous causal relationship is obtained by testing the relationship between the control variables in the same time window by the structural vector autoregressive model; The single risk matching degree is calculated: ； The exponential decay of the lag w is: The maximum lag period is The single risk matching degree is The standard deviation of the causal influence coefficient is The single risk variable at time t is The risk factors at time tw are Risk factors For a single risk variable The causal influence coefficient is The dynamically adjusted weight at time t is The forgetting factor is ; The multiple risk types are subjected to principal component analysis for dimension reduction, to obtain principal components, and the multiple risk matching degree is calculated: ； Wherein the a th principal component is , the number of principal components is , the a th principal component variance contribution rate is , the prediction variable is , the prediction variable The Granger causality test result of the principal component Is , the multi-risk matching degree is ; The rolling mean and standard deviation are calculated according to the matching degree, and the adaptive threshold is set, and the expression is: ； Wherein the control threshold convergence speed coefficient is , the rolling mean is , the standard deviation is , the current time is t, and the adaptive threshold is ; When the matching degree is lower than the adaptive threshold, the nodes are dynamically added or deleted, the condition independence test is updated, and all the control data and risk types are traversed until the matching degree is output.

4. The method of claim 1, wherein the multi-dimensional risk management maturity model is constructed by: The method for constructing a power risk management maturity model according to the maturity evaluation function comprises the following steps: Multi-dimensional evaluation indexes are selected through management data, and the weights of the evaluation indexes are obtained by using an entropy weight method; wherein the evaluation indexes include equipment performance indexes, grid operation indexes, economic indexes, environmental protection indexes and user satisfaction indexes; A target function of the power risk management maturity model is constructed, and the expression is: ； Wherein the objective function of the bth risk management is , the maturity assessment function of the jth assessment index of the bth risk management is , the jth assessment index of the bth risk management is , the loss function is , and the number of assessment indexes is D. According to the target function, when the target function score is less than 0.219, the power system is risk-free management; when the target function score is greater than 0.219 and less than 0.417, the power system risk management ability is level four; when the target function score is greater than 0.417 and less than 0.609, the power system risk management ability is level three; when the target function score is greater than 0.609 and less than 0.794, the power system risk management ability is level two; when the target function score is greater than 0.794, the power system risk management ability is level one.

5. A system for constructing a multi-dimensional risk management maturity model for performing the method of any one of claims 1-4, characterized in that, It comprises: A data acquisition module is used to collect monitoring data and risk management data of a preset power system, and the monitoring data and the risk management data are preprocessed; The risk management data includes equipment failure risk management, fuel supply risk management, power grid operation safety risk management, distributed resource aggregation risk management, and power supply shortage or excess risk management; the monitoring data includes to-be-analyzed data and historical data; A risk type identification module is used to identify risks in the monitoring data to obtain risk data, use a priori algorithm to obtain risk types from the risk data, and obtain management and control data by power prediction based on the risk management data; The risk types include single risk type and multiple risk types; A prediction matching module is used to obtain management and control data by power prediction based on the risk management data, dynamically and causally match the management and control data and the risk types to obtain a matching degree, and take the management and control data with a matching degree greater than a matching threshold as management data, and vice versa as fuzzy data; A comparison and evaluation module is used to dynamically and causally match the management and control data and the risk types to obtain a matching degree, take the management and control data with a matching degree greater than a matching threshold as management data, and vice versa as fuzzy data, longitudinally compare and evaluate the historical data and the fuzzy data to obtain an optimization degree, and add the fuzzy data with an optimization degree greater than an optimization threshold to the management data; A modeling and optimization module is used to construct a maturity evaluation function according to a horizontal and vertical maturity comprehensive comparison of the management data, construct a power risk management maturity model according to the maturity evaluation function, and output a target model.

Images