Multi-dimensional risk factor index identification method based on artificial intelligence technology

Abstract

The invention provides a multi-dimensional risk factor index identification method based on an artificial intelligence technology, and belongs to the technical field of safety production. The method comprises the following steps: firstly, determining the influence of risk factors on the overall security risk, and forming a multi-dimensional risk matrix and a multi-dimensional risk factor dynamic time sequence model; secondly, in combination with the multi-dimensional risk matrix and the multi-dimensional risk factor dynamic time sequence model, constructing a multi-dimensional risk factor index group and an operation risk prediction model; according to the method, automation and informatization of risk factor index identification can be further improved, digitization and intelligentization are realized, the production efficiency is greatly improved, and the safety risk is reduced.

Description

Technical Field

[0001] The present invention relates to the field of safe production technology, and in particular to a multi-dimensional risk factor indicator identification method based on artificial intelligence technology. Background Art

[0002] The digital and intelligent transformation of power systems has been an important development trend in the power industry in recent years. From the initial manual operation to automation and informatization, to the current digitalization and intelligence, each technological upgrade of the power system has greatly improved production efficiency and reduced safety risks.

[0003] In the existing power production process, production safety is becoming more and more important. On the one hand, more and more laws and regulations emphasize the deployment of law-based public security and implement more stringent and standardized safety production standards. In the face of new situations and new requirements, it is necessary to promote law-based public security and ensure that safety production management and behavior are in compliance with the law. On the other hand, the existing production site operation risk control capabilities are not strong, and the on-site operation safety level is not high. The production site operation risk control system is not sound, and it is impossible to strengthen source prevention and hierarchical control, and it is difficult to effectively improve the level of production site operation risk control. At the same time, the supervision of risk control at the operation site involves all levels and professions. The existing methods require a large amount of manpower for on-site surveys, data collection, communication and research, offline compilation, decision-making analysis and other work, which consumes too much manpower and material resources, wastes costs, and has complex operations and low work efficiency. Summary of the invention

[0004] The purpose of the present invention is to provide a multi-dimensional risk factor indicator identification method based on artificial intelligence technology, which can further enhance the automation and informatization of risk factor indicator identification, realize digitization and intelligence, greatly improve production efficiency and reduce safety risks.

[0005] In order to achieve the above object, the present invention provides a multi-dimensional risk factor indicator identification method based on artificial intelligence technology, the method comprising:

[0006] Determine the impact of risk factors on overall safety risk, conduct work hazard analysis (JHA), identify the hazard sources in the production process system and determine the operability problems of the process system through hazard and operability analysis (HAZOP). The power transmission specialty adopts the work hazard analysis (JHA) method, the transformer inspection specialty adopts the work hazard analysis (JHA) method, and the distribution specialty adopts the work hazard analysis (JHA) method and hazard and operability analysis (HAZOP) method. Obtain the unilateral influencing factor data of the risk factors of the production operations of transformer maintenance, transmission line maintenance, and distribution maintenance, and construct a multi-dimensional risk matrix and a multi-dimensional risk factor dynamic time series model;

[0007] Based on the impact of the coupling of multidimensional risk factors on the overall safety of production operations, a multidimensional risk matrix for substation maintenance, transmission line maintenance, and distribution maintenance production operations is constructed. Combined with the multidimensional risk factor dynamic time series model, a multidimensional risk factor indicator group and operation risk prediction model are constructed.

[0008] Preferably, determining the impact of risk factors on the overall security risk includes:

[0009] Subdivide the work activities into specific work tasks (HTA) step by step according to the principle of relative independence, identify the safety risks of each work task and the specific hazard sources that lead to safety risks, conduct risk assessment based on the identified specific hazard sources that lead to safety risks, determine the acceptability of risks and add measures to further control safety risks;

[0010] Based on the analysis of the possible "deviations" of system elements and their characteristics and intentions under the "guide words", further analyze the "consequences" that may be caused by the "deviations", and analyze the safety risks that may be caused by the "deviations"; further analyze the factors that may cause "deviations" in the system to realize the process of identifying the hazard sources for system elements; and conduct risk assessment based on the specific hazard sources that cause safety risks identified, determine the acceptability of the risks and add measures to further control safety risks;

[0011] By sorting out and summarizing hazard source information and control measures, a multidimensional risk factor indicator group was constructed during the digital transformation period, and the impact of the coupling of multidimensional risk factors during the digital transformation period on the overall safety of production operations was obtained.

[0012] Preferably, risk assessment includes: combining typical risk database and violation database information, based on the safety supervision big model, based on the PDCA theory, starting from the five dimensions of people, machines, materials, methods and environment, analyzing the influencing factors of risk assessment level determination in various operation scenarios; using the multi-dimensional risk factor coupling weights and the multi-dimensional risk factor dynamic time series model to construct a safety assurance risk factor prediction model for substation maintenance, transmission line maintenance, and distribution maintenance production operations, and mining the data value through machine learning and big data models to obtain multiple risk evolution mechanisms during the risk occurrence process.

[0013] Preferably, constructing a multidimensional risk factor indicator group under the background of digital transformation includes: based on the cumulative effect of risk transmission, obtaining the unilateral influencing factor data of risk factors and constructing a risk factor indicator group;

[0014] Constructing a multidimensional risk matrix includes: analyzing the possibility of accidents in the risk process based on historical data and real-time data of production safety risks, constructing a multidimensional risk matrix based on the severity of actual accidents, and obtaining a dynamic time series model of multidimensional risk factors.

[0015] Preferably, establishing a risk factor prediction model based on resilience protection includes: based on the analysis of the mechanism of multi-dimensional risk factors and their interactions, mining the data value through machine learning, big data models, etc., combining the resilience characteristics of the power grid in responding to risks, obtaining multiple risk evolution mechanisms during the risk occurrence process, quickly perceiving and predicting the coupling relationship between risk factors, and forming a spatial risk prediction model for multiple uncertain elements.

[0016] Preferably, the subjective fuzzy weighting method is used to obtain the subjective weight range of the indicator in view of the weight that should be given to each indicator in the multidimensional risk factor indicator group; at the same time, based on the maximum entropy criterion, the information entropy of the factor weight eigenvector is used to quantify the elasticity of the risk identification result, and the obtained indicator weight range is converted into a definite numerical value; and, an adaptive weighting method of dynamic fusion of subjective and objective factors is adopted, combined with the constraint conditions of each indicator under different types of operation scenarios, to construct a multidimensional operation risk adaptive quantitative assessment model, and the factor weight mechanism dynamically adjusts the weight distribution based on the adaptive criterion.

[0017] Preferably, the construction of the multi-dimensional risk factor indicator group and the operation risk prediction model includes real-time monitoring and querying of the risk verification status of the operation link, construction of a multi-dimensional operation risk assessment model and data input model assessment; wherein,

[0018] Real-time monitoring and query of risk verification of operation links includes obtaining multi-dimensional monitoring data on equipment status and operating environment changes at power sites through the data center, business center, and enterprise-level meteorological center;

[0019] The multi-dimensional operation risk assessment model includes: a data receiving module, which is used to receive data through the data interface of the data center and the business center, and obtain field operation data, accident records, maintenance logs and operator information; a data preprocessing module, which is used to filter the noise in the received data through Kalman filtering, fill in the missing values ​​through KNN, and use the Z-Score standardization method to convert the data to a unified scale; an equipment status assessment module, which is used to diagnose the operating status of the equipment based on historical data and real-time data through the equipment status assessment model built by the support vector machine; an environmental risk assessment module, which is used to diagnose the operating status of the equipment based on the environmental risk assessment model built by the support vector machine according to the enterprise-level meteorological center Data evaluation of operating environment risks; personnel risk assessment module, which is used to identify potential personnel risks based on the personnel risk assessment model constructed by support vector machine and the personnel-related data of the risk management platform; multi-dimensional risk association module, which is used to integrate the risk data of equipment, environment and personnel through DS evidence theory, and to generate a risk association matrix by analyzing the association relationship between different risks through a priori knowledge algorithm; comprehensive risk assessment module, which is used to comprehensively evaluate the risk information of each dimension based on the fused data and the risk association matrix based on the fuzzy comprehensive evaluation method to generate the final risk assessment result; risk information output module, which is used to output the final risk assessment result through a compatible RESTful API interface;

[0020] The data input model evaluation includes inputting the collected real-time data and historical data into the multi-dimensional operation risk assessment model to obtain the risks and risk information existing in the current power field operations. It is divided into a risk data fusion unit and a risk relationship analysis unit, and the output end of the risk data fusion unit is connected to the input end of the risk relationship analysis unit.

[0021] Preferably, a risk data fusion unit is used to fuse the risk data of equipment, environment and personnel through DS evidence theory to generate a comprehensive risk data set, including:

[0022] Obtain risk assessment data of the equipment status assessment module, the environmental risk assessment module, and the personnel risk assessment module, which are recorded as E1, E2, and E3 respectively;

[0023] The basic probability distribution of each risk assessment data is calculated by formula (1):

[0024]

[0025] Among them, m i (θ) is the basic probability distribution function of the ith risk dimension, E i is the risk assessment data of the ith risk dimension, θ is the risk state set, k is the slope adjustment parameter, t is the threshold, α is the amplitude of the sine function, and β is the frequency of the sine function;

[0026] The basic probability distribution of each risk assessment data is synthesized through the synthesis rule of DS evidence theory, and the comprehensive basic probability distribution is obtained according to formula (2):

[0027]

[0028] Among them, m(θ) is the comprehensive basic probability distribution function, θ is the risk state set, K is the standardization coefficient, A, B and C are the subsets corresponding to different risk dimensions, and are the basic probability distributions of subsets corresponding to different risk dimensions, γ is the amplitude of the cosine function, and δ is the frequency of the cosine function;

[0029] According to formula (3), the comprehensive basic probability distribution is normalized to obtain the comprehensive risk assessment value:

[0030]

[0031] Among them, R is the comprehensive risk assessment value, Θ is the set of all risk states, m(θ) is the comprehensive basic probability distribution function, θ is the set of risk states, λ is the amplitude of the tangent function, and μ is the frequency of the tangent function.

[0032] Preferably, the risk relationship analysis unit is used to mine frequent item sets and association rules from the comprehensive risk data set through an association rule mining method based on a priori knowledge algorithm, analyze the association relationship between risks of different dimensions, and generate a risk association matrix, including:

[0033] The comprehensive risk data set is preprocessed through Kalman filtering and the data cleaning tool OpenRefine to remove noise and outliers from the comprehensive risk data set;

[0034] Mining frequent itemsets in the comprehensive risk dataset through prior knowledge algorithms;

[0035] Extract association rules from frequent item sets through prior knowledge algorithms to obtain the support and confidence of association rules;

[0036] Filter high-credibility association rules based on support and confidence thresholds, and obtain the association strength of association rules through the Lift indicator;

[0037] The association strength of the association rule is filled into the risk association matrix, and the elements of the risk association matrix represent the association strength between different risk dimensions.

[0038] Preferably, a comprehensive risk assessment module is used to conduct a comprehensive assessment of the risk information of each dimension based on the fused data and the risk association matrix based on a fuzzy comprehensive evaluation method to generate a final risk assessment result, including:

[0039] Obtaining a comprehensive risk data set and a risk association matrix generated by the multi-dimensional risk association module;

[0040] The membership degree of each risk item is calculated by the fuzzy comprehensive evaluation method, and the fuzzy evaluation matrix is ​​constructed to map each risk item in the comprehensive risk data set to the fuzzy set. The membership degree calculation formula (4) of each risk item is:

[0041]

[0042] Among them, μ i (x) is the membership degree of risk item i, x is the risk assessment value, c i is the central value of risk term i, σ i is the width of the fuzzy set, α i is the adjustment parameter;

[0043] According to the risk association matrix and the membership of each risk item, the comprehensive membership of risk items of different dimensions is calculated by Sugeno fuzzy reasoning. The calculation formula (5) is:

[0044]

[0045] Among them, μ all (x) is the comprehensive membership, w i is the weight of risk term i, μ i (x) is the membership degree of risk item i, Lift i is the correlation strength of risk term i, β i is the correlation strength influence coefficient;

[0046] The final risk assessment result is generated through fuzzy decision making based on the comprehensive membership of risk items in different dimensions;

[0047] The severity of each risk item is ranked by fuzzy analytic hierarchy process to identify high-risk areas.

[0048] Through the above technical solution, we first determine the impact of risk factors on the overall safety risk, and form a multidimensional risk matrix and a multidimensional risk factor dynamic time series model. Secondly, we combine the multidimensional risk matrix and the multidimensional risk factor dynamic time series model to construct a multidimensional risk factor indicator group and an operational risk prediction model, which improves the automation and informatization of risk factor indicator identification, realizes digitization and intelligence, greatly improves production efficiency, and reduces safety risks.

[0049] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0051] Figure 1 It is a schematic diagram of the hazard identification and risk assessment process;

[0052] Figure 2 It is a comparison chart of hazard identification and risk assessment methods;

[0053] Figure 3 It is the JHA operation procedure division diagram;

[0054] Figure 4 This is a graph of the JHA summary analysis results;

[0055] Figure 5 It is the HAZOP operation procedure division diagram;

[0056] Figure 6 It is the HAZOP summary analysis result diagram;

[0057] Figure 7 It is a flowchart of time series analysis;

[0058] Figure 8 It is a schematic diagram of the numerical change law of the time series;

[0059] Fig. 9 It is a multi-dimensional operation risk assessment system diagram;

[0060] Fig.10 It is a detailed multi-dimensional operational risk assessment;

[0061] Fig.11 It is a schematic diagram of risk factor weighting;

[0062] Fig.12 It is a data collection detail sheet;

[0063] Fig.13 It is the operational risk assessment indicator system table;

[0064] Fig.14 It is a structural diagram of the multi-dimensional operation risk assessment model;

[0065] Fig.15 It is a flow chart of the work of the risk data fusion unit;

[0066] Fig.16It is a flow chart of the work of the risk relationship analysis unit;

[0067] Fig.17 It is a flow chart of the work of the integrated risk assessment module;

[0068] Fig.18 It is a structure diagram of risk assessment indicators for power operations;

[0069] Fig.19 It is a schematic diagram of the dynamic time series model of risk factors. DETAILED DESCRIPTION

[0070] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0071] See also Figure 1 In the present invention, the concepts and characteristics involved in hazard identification and risk assessment theory are first clarified, including:

[0072] Hazard source: a source that may cause injury and damage to health;

[0073] Hazard identification: The process of identifying hazards and determining their characteristics;

[0074] Risk: The impact of uncertainty; in order to better solve the problem of uncertainty, the science of risk management has emerged. The basic goal of enterprise safety management is to control accidents;

[0075] Accidents: Accidents are random events with uncertainty; enterprise safety management belongs to the category of risk management science and should be based on the scientific principles of risk management;

[0076] Risk identification: The output information of the hazard identification process must include possible injuries and health damage. Generally, the safety risks in the workplace can be identified from two aspects: empirical analysis and energy analysis.

[0077] According to the scientific principles of risk management, in order to control safety risks, enterprises are required to identify safety risks (possible accidents) and possible hazards that may cause safety risks through the hazard identification process; evaluate the degree of risk through the risk assessment process, determine whether the risk is acceptable, and determine the effectiveness of existing control measures; based on the hazard identification and risk assessment results, determine the control measures to be taken for the hazard sources to achieve control of accidents or safety risks.

[0078] When carrying out hazard identification and risk assessment work and choosing the hazard identification and risk assessment methods, the following factors should be considered:

[0079] (1) The purpose and scope of hazard identification and risk assessment;

[0080] (2) Research and development, planning, detailed design, and operation of activities during the life cycle of the workplace;

[0081] (3) The components of the workplace (people, machines, materials, methods, and environment);

[0082] (4) Resources available for hazard identification and risk assessment;

[0083] Combination Figure 2 Based on the comparison results in , the present invention selects Job Hazard Analysis (JHA) and Hazard and Operability Analysis (HAZOP) to carry out the analysis. Specifically, Job Hazard Analysis (JHA) is a method that can be used to identify hazards for work activities, and its identification steps include:

[0084] (1) First, the work activities are subdivided into specific work tasks according to the principle of relative independence (hierarchical task analysis, HTA);

[0085] (2) Secondly, identify the safety risks (possible accidents) of each work task;

[0086] (3) Next, identify the specific sources of danger that lead to safety risks (unsafe conditions of objects, unsafe behaviors of people, including causal chain events that lead to accidents);

[0087] (4) Finally, based on the identified specific hazard sources that cause safety risks (unsafe conditions of objects, unsafe behaviors of people), use applicable risk assessment methods to conduct risk assessment, determine the acceptability of the risks, and add measures to further control safety risks.

[0088] The premise of using JHA is the safety risks to the operators when an accident occurs at the work site. The safety risks identified in the work activities include:

[0089] (1) Safety risks caused by tools or machinery used by operators during operations, such as wrenches, lifts, cranes, etc.;

[0090] (2) Safety risks caused by the working environment during operation, such as strong winds, high temperatures, proximity to live objects, etc.;

[0091] (3) Safety risks arising from special operations, such as fire and earthwork;

[0092] (4) Safety risks unique to the work activities being carried out, that is, other safety risks other than those caused by tools or machinery, working environment and special operations, such as slips, falls, injuries, bumps, falls from heights, etc.

[0093] Hazard and operability analysis HAZOP is a systematic and structured analysis method used to identify hazards in production process systems and determine operability problems of process systems. Its analysis steps include:

[0094] (1) Hazard and operability analysis is an analysis of the system elements and their characteristics and possible "deviations" from their intent based on the "guide words";

[0095] (2) Further analyze the “consequences” that may result from the deviation, that is, analyze the safety risks (possible accidents) that may result from the deviation;

[0096] (3) Further analyze the factors that may cause deviations in the system and implement the hazard source identification process for system elements;

[0097] (4) Finally, based on the identified specific hazard sources that cause safety risks (unsafe conditions of objects, unsafe behaviors of people), use applicable risk assessment methods to conduct risk assessment, determine the acceptability of the risks, and add measures to further control safety risks.

[0098] The above-mentioned Hazard and Operability Analysis (HAZOP) is mainly used for systematic safety analysis of systems including but not limited to industrial and power systems. Its use premise is that personnel operations may cause safety risks to the system or remote locations (i.e. not at the workplace at the time), including:

[0099] (1) Common safety risks of primary and secondary equipment during switching operations;

[0100] (2) Safety risks arising from special operations, such as working at heights;

[0101] (3) Safety risks unique to the operational activities being carried out, that is, safety risks other than those arising from general and special operations, such as power outages, equipment damage, etc.

[0102] To sum up, when each profession selects a risk assessment method, the power transmission profession adopts the job hazard analysis JHA method, the transformer inspection profession adopts the job hazard analysis JHA method, and the distribution profession adopts the job hazard analysis JHA and hazard and operability analysis HAZOP method.

[0103] In the study of the coupling effect of multi-dimensional risk factors in the present invention, when applying the job hazard analysis JHA, first determine the work activities and collect data on the operating procedures and other similar operating procedures, and then gradually divide the operating procedures (HTA) until they are divided into specific operating behaviors:

[0104] like Figure 3 As shown, the operation activities are divided into specific work tasks step by step according to the principle of relative independence (taking transformer inspection as an example): the table header is divided according to the various equipment under maintenance, such as transformer maintenance operation activities, disconnector maintenance operation activities, etc.; the first column of the information table is to divide each maintenance equipment into different components; the second column of the information table is the maintenance activities performed on the components in the first column; the third column of the information table is the specific operation steps when performing the maintenance activities in the second column;

[0105] Then, risk identification is carried out to analyze the safety risks (possible accidents) in each specific operation step. The third column mentioned above is the operation step when performing the operation. According to the cause and effect chain of accidents, the fourth column is the safety risk generated by each operation step.

[0106] Then, we will identify the hazard sources, study the coupling of multi-dimensional risk factors, and further analyze the possible causes of accidents (hazard sources / hazard factors). The fifth column (factors that may cause accidents) analyzes what caused the previous safety risks.

[0107] Conduct risk assessment and formulate control measures, use applicable risk assessment methods to conduct risk assessment, determine the acceptability of risks, and consider adding measures to further control safety risks. The sixth column obtains effective measures to control safety risks through engineering calculations and empirical analysis, that is, what measures can be taken to control factors that may cause accidents. The control requirements of this column must correspond one-to-one with the factors that may cause accidents in the previous fifth column, and the factors that may cause accidents must be controlled one by one.

[0108] Finally, the analysis results are summarized, usually in the form of a table (see Figure 4 ).

[0109] In the multi-dimensional risk factor coupling study of the present invention, when applying the hazard and operability analysis HAZOP, the operation activities are first determined and the data of the operation procedures and other similar operation procedures are collected. Then the operation procedures are gradually divided (HTA) until they are divided into specific operation behaviors. Since there are few operation procedures in the power industry, the safety risk, hazard source and control requirement information table of the substation operation procedure is clearly defined; Figure 5As shown in the figure, the switching operation is divided into line operation, busbar operation, main transformer operation, bypass operation, operation, low-resistance operation, main transformer neutral point reactance and station transformer operation; the operating activities are specifically divided into work tasks, and the work tasks are further specifically divided into operating behaviors.

[0110] Then conduct a guide word check deviation and further analyze the deviation. For example, if the operation behavior is to transfer the load, then the guide word can be selected more, and the deviation is excessive transfer of load.

[0111] Then, risk identification is carried out. The sixth column is safety risk. This column analyzes the previous deviations and analyzes what safety risks these deviations may cause. For example, the safety risk that the deviation "excessive load transfer" may bring to the production system is "it may cause the relevant components to exceed the limit and cause damage to the primary equipment (knife switch, grounding knife switch)".

[0112] After that, the hazard sources are identified. The seventh column is to analyze the factors that may cause deviations. This column analyzes the previous deviations and analyzes what caused these deviations.

[0113] Carry out risk assessment and formulate control measures. In the eighth column, control requirements should be formulated to effectively control the factors (hazard sources / hazardous factors) that may cause deviations.

[0114] Finally, the analysis results are summarized. Generally, the HAZOP analysis results are summarized in the form of a table (see Figure 6 ).

[0115] Based on the above method, the unilateral influencing factor data of the risk factors of substation maintenance, transmission line maintenance, and distribution maintenance production operations are obtained, the hazard source information and control measures are sorted and summarized, and a multidimensional risk factor indicator group during the digital transformation period of the power grid company is constructed. The impact of the coupling of multidimensional risk factors during the digital transformation period on the overall safety of production operations is studied.

[0116] In this implementation, based on the impact of the coupling of multidimensional risk factors on the overall safety of production operations, a multidimensional risk matrix for substation maintenance, transmission line maintenance, and distribution maintenance production operations is constructed, and a dynamic time series model of multidimensional risk factors is studied (see Figure 7 ).

[0117] The dynamic time series model refers to a numerical sequence formed by arranging the indicator values ​​of a certain phenomenon in chronological order. Time series analysis can be divided into three parts: describing the past, analyzing the rules, and predicting the future. The core is to predict the data for a period of time in the future, that is, to predict the risk level of multi-dimensional risk factors.

[0118] Time series consists of two components, namely the time element and the value element.

[0119] Time series are divided into period time series and point time series according to the different time and numerical properties. In period time series, numerical elements reflect the results of the development of phenomena in a certain period; in point time series, numerical elements reflect the instantaneous level of phenomena at a certain point in time.

[0120] Since time series is the numerical expression of the long-term changes in the value of a certain indicator, the changes in the value of the time series must contain the regularity of the numerical transformation, and these regularities are the entry points for time series analysis. Figure 8 As shown in the figure, in general, there are four types of numerical change rules for time series, and a time series is often a superposition of these four types of changes:

[0121] Secular trend (T) refers to a statistical indicator that shows a continuous upward or downward trend over a fairly long period of time, which is usually represented by the letter T.

[0122] Seasonal Variation (S) refers to the periodic change of indicator values ​​due to the change of seasons. The season here is broad, usually in months, quarters, weeks, and not in years.

[0123] Cyclical Variation (C) is different from seasonal variation in that it usually takes several years as a cycle and is represented as a wave-like cyclical variation on a graph. This cyclical variation is characterized by alternating increases and decreases, but does not have a strict regular cyclical continuous variation.

[0124] Irregular variation (I) is a numerical change caused by certain random factors. The effects of these factors are unpredictable and irregular, and can be regarded as the impact of many accidental factors on the time series (also called disturbance terms in regression).

[0125] The relationship between the four types of changes and the changes in indicators includes superposition and product relationships. As t is deduced, the greater the change in various relationships, the more it is a product relationship, and if it remains constant, it is a superposition relationship:

[0126] (1) If the four changes are independent of each other, the superposition model can be expressed as:

[0127] Y = T + S + C + I;

[0128] (2) If there is a mutual influence between the four changes, then use the product model:

[0129] Y=T×S×C×I; where

[0130] Y is the final change in the indicator value, T is the long-term trend, S is the seasonal change, C is the cyclical change, and I is the irregular change.

[0131] like Fig. 9 and Fig.10 As shown in the figure, combined with the information of typical risk database and violation database, based on the safety supervision big model and based on the PDCA theory, starting from the five dimensions of personnel risk factors, equipment risk factors, tool risk factors, method risk factors and environmental risk factors, the influencing factors of risk assessment level judgment in various operation scenarios are analyzed; using the multi-dimensional risk factor coupling weights and the multi-dimensional risk factor dynamic time series model, a prediction model for safety assurance risk factors of substation maintenance, transmission line maintenance, and distribution maintenance production operations is constructed, and the data value is mined through machine learning and big data models, and the multiple risk evolution mechanisms during the risk occurrence process are studied.

[0132] In this embodiment, constructing a multidimensional risk factor indicator group under the background of digital transformation includes: based on the cumulative effect of risk transmission, obtaining unilateral influencing factor data of risk factors and constructing a risk factor indicator group.

[0133] In this embodiment, constructing a multidimensional risk matrix includes: analyzing the possibility of accidents in the risk process based on historical data and real-time data of production safety risks, constructing a multidimensional risk matrix based on the severity of actual accidents, and obtaining a dynamic time series model of multidimensional risk factors.

[0134] In this embodiment, establishing a risk factor prediction model based on resilience protection includes: based on the analysis of the mechanism of multi-dimensional risk factors and their interactions, mining the data value through machine learning, big data models, etc., combining the resilience characteristics of the company's power grid in responding to risks, and studying the multiple risk evolution mechanisms during the risk occurrence process, so as to quickly perceive and predict the coupling relationship between risk factors and form a spatial risk prediction model for multiple uncertain factors.

[0135] like Fig.11As shown in the figure, for the weight that should be given to each indicator in the multidimensional risk factor indicator group, a subjective fuzzy weighting method is adopted. A number of risk control experts in the industry independently compare and evaluate the weights of each indicator under different operation types and different risk level scenarios to obtain the subjective weight range of the indicator; at the same time, based on the maximum entropy criterion, the information entropy of the factor weight eigenvector is used to quantify the elasticity of the risk identification results, so as to reduce the unreasonable subjective assumptions caused by the lack of objective parameters, and convert the obtained indicator weight range into a certain value; in addition, an adaptive weighting method of dynamic fusion of subjective and objective factors is adopted, combined with the constraints of each indicator under different types of operation scenarios, to construct a multi-dimensional operation risk adaptive quantitative assessment model, and the factor weight mechanism dynamically adjusts the weight distribution based on the adaptive criterion to meet the accurate assessment of risk levels under different operation scenarios.

[0136] In this embodiment, the construction of the multi-dimensional risk factor indicator group and the operation risk prediction model includes real-time monitoring and querying of the operation link risk verification status, construction of a multi-dimensional operation risk assessment model and data input model evaluation; wherein,

[0137] like Fig.12 and Fig.13 As shown, real-time monitoring and querying of risk verification of operation links includes obtaining multi-dimensional monitoring data of equipment status and changes in the operating environment of the power site through the data middle platform, business middle platform and enterprise-level meteorological center, including obtaining data such as team members' years of service, violations, operation frequency, equipment type, tool configuration and operation duration from the risk management platform, obtaining data such as strong wind, precipitation, lightning and high temperature from the enterprise-level meteorological center, obtaining data such as power grid topology, power grid resource location information and geographical location information from a power grid map and power grid GIS platform, matching risk factor-related data from the data middle platform, monitoring changes in equipment status and operating environment of the power site from multiple dimensions, and real-time monitoring and querying of risk verification of operation links;

[0138] Construct a multi-dimensional operational risk assessment model (see Fig.14 )include:

[0139] The data receiving module is used to receive data through the data interface of the data center and business center, and obtain on-site operation data, accident records, maintenance logs and operator information;

[0140] The data preprocessing module is used to filter the noise in the received data through Kalman filtering, fill the missing values ​​through KNN, and convert the data to a uniform scale using the Z-Score standardization method;

[0141] The equipment status assessment module is used to diagnose the equipment's operating status based on historical data and real-time data through the equipment status assessment model built by the support vector machine;

[0142] Environmental risk assessment module, which is used to assess the operating environment risk based on the enterprise-level meteorological center data through the environmental risk assessment model built by support vector machine;

[0143] The personnel risk assessment module is used to identify potential personnel risks based on the personnel-related data of the risk management and control platform through the personnel risk assessment model built by the support vector machine;

[0144] The multi-dimensional risk association module is used to integrate the risk data of equipment, environment and personnel through the DS evidence theory, and to generate a risk association matrix by analyzing the association relationship between different risks through a priori knowledge algorithm;

[0145] The comprehensive risk assessment module is used to conduct a comprehensive assessment of the risk information of each dimension based on the fused data and the risk association matrix based on the fuzzy comprehensive evaluation method to generate the final risk assessment result;

[0146] The risk information output module is used to output the final risk assessment results through a compatible RESTful API interface;

[0147] The data input model evaluation includes inputting the collected real-time data and historical data into the multi-dimensional operation risk assessment model to obtain the risks and risk information existing in the current power field operation, wherein the risk information includes the risk type, risk assessment value and risk level; it is mainly divided into a risk data fusion unit and a risk relationship analysis unit. The risk data fusion unit fuses the risk data of equipment, environment and personnel through the DS evidence theory to generate a comprehensive risk data set; the risk relationship analysis unit mines frequent item sets and association rules from the comprehensive risk data set through an association rule mining method based on a priori knowledge algorithm, analyzes the association relationship between risks of different dimensions and generates a risk association matrix; and the output end of the risk data fusion unit is connected to the input end of the risk relationship analysis unit; wherein,

[0148] See also Fig.15 ,The working steps of the risk data fusion unit are:

[0149] Step A, obtaining risk assessment data of the equipment status assessment module, the environmental risk assessment module and the personnel risk assessment module, which are recorded as E1, E2 and E3 respectively;

[0150] Step B: Calculate the basic probability distribution of each risk assessment data through the basic probability distribution function, the formula is:

[0151]

[0152] Among them, m i (θ) is the basic probability distribution function of the ith risk dimension, E iis the risk assessment data of the ith risk dimension, θ is the risk state set, k is the slope adjustment parameter, t is the threshold, α is the amplitude of the sine function, and β is the frequency of the sine function;

[0153] Step C: synthesize the basic probability distribution of each risk assessment data through the synthesis rule of DS evidence theory to obtain the comprehensive basic probability distribution, the formula is:

[0154]

[0155] Among them, m(θ) is the comprehensive basic probability distribution function, θ is the risk state set, K is the standardization coefficient, A, B and C are the subsets corresponding to different risk dimensions, and are the basic probability distributions of subsets corresponding to different risk dimensions, γ is the amplitude of the cosine function, and δ is the frequency of the cosine function;

[0156] Step D: normalize the comprehensive basic probability distribution to obtain a comprehensive risk assessment value. The normalization formula is:

[0157]

[0158] Where R is the comprehensive risk assessment value, Θ is the set of all risk states, m(θ) is the comprehensive basic probability distribution function, θ is the set of risk states, λ is the amplitude of the tangent function, and μ is the frequency of the tangent function;

[0159] See also Fig.16 , the working steps of the risk relationship analysis unit are:

[0160] Step a: Preprocess the comprehensive risk data set by using Kalman filtering and data cleaning tool OpenRefine to remove noise and outliers from the comprehensive risk data set;

[0161] Step b: mining frequent item sets in the comprehensive risk data set through a priori knowledge algorithm;

[0162] Step c: extracting association rules from frequent item sets through a priori knowledge algorithm to obtain support and confidence of the association rules;

[0163] Step d: Filter high-credibility association rules according to support and confidence thresholds, and obtain the association strength of the association rules through the Lift indicator;

[0164] Step e: Fill the association strength of the association rule into the risk association matrix, where the elements of the risk association matrix represent the association strength between different risk dimensions;

[0165] See also Fig.17,The working steps of the comprehensive risk assessment module are:

[0166] Step 1: Obtain a comprehensive risk data set and a risk association matrix generated by the multi-dimensional risk association module;

[0167] Step 2: Calculate the membership of each risk item through the fuzzy comprehensive evaluation method, and construct a fuzzy evaluation matrix to map each risk item in the comprehensive risk data set to the fuzzy set. The membership calculation formula of each risk item is:

[0168]

[0169] Among them, μ i (x) is the membership degree of risk item i, x is the risk assessment value, c i is the central value of risk term i, σ i is the width of the fuzzy set, α i is the adjustment parameter;

[0170] Step 3: According to the risk association matrix and the membership of each risk item, the comprehensive membership of risk items of different dimensions is calculated by Sugeno fuzzy reasoning. The calculation formula is:

[0171]

[0172] Among them, μ all (x) is the comprehensive membership, w i is the weight of risk term i, μ i (x) is the membership degree of risk item i, Lift i is the correlation strength of risk term i, β i is the correlation strength influence coefficient;

[0173] Step 4: Generate the final risk assessment result through fuzzy decision-making according to the comprehensive membership of risk items in different dimensions;

[0174] Step 5: Use the fuzzy analytic hierarchy process to rank the severity of each risk item and identify high-risk areas.

[0175] Through the above technical scheme, the present invention first determines the impact of risk factors on the overall safety risk, forms a multidimensional risk matrix and a multidimensional risk factor dynamic time series model; secondly, combines the multidimensional risk matrix and the multidimensional risk factor dynamic time series model to construct a multidimensional risk factor indicator group and an operation risk prediction model, which improves the automation and informatization of risk factor indicator identification, realizes digitization and intelligence, greatly improves production efficiency, and reduces safety risks.

[0176] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0177] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0178] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0179] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0180] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0181] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0182] Computer readable media include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0183] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0184] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A multi-dimensional risk factor indicator identification method based on artificial intelligence technology, characterized in that: The method comprises: Determine the impact of risk factors on overall safety risk, conduct work hazard analysis (JHA), identify the hazard sources in the production process system and determine the operability problems of the process system through hazard and operability analysis (HAZOP). The power transmission specialty adopts the work hazard analysis (JHA) method, the transformer inspection specialty adopts the work hazard analysis (JHA) method, and the distribution specialty adopts the work hazard analysis (JHA) method and hazard and operability analysis (HAZOP) method. Obtain the unilateral influencing factor data of the risk factors of the production operations of transformer maintenance, transmission line maintenance, and distribution maintenance, and construct a multi-dimensional risk matrix and a multi-dimensional risk factor dynamic time series model; Based on the impact of the coupling of multidimensional risk factors on the overall safety of production operations, a multidimensional risk matrix for substation maintenance, transmission line maintenance, and distribution maintenance production operations is constructed. Combined with the multidimensional risk factor dynamic time series model, a multidimensional risk factor indicator group and operation risk prediction model are constructed.

2. The multidimensional risk factor indicator identification method based on artificial intelligence technology according to claim 1 is characterized in that: Determine the impact of risk factors on overall security risk including: Subdivide the work activities into specific work tasks (HTA) step by step according to the principle of relative independence, identify the safety risks of each work task and the specific hazard sources that lead to safety risks, conduct risk assessment based on the identified specific hazard sources that lead to safety risks, determine the acceptability of risks and add measures to further control safety risks; Based on the analysis of the possible "deviations" of the system elements and their characteristics and intentions under the "guide words", further analyze the "consequences" that may be caused by the "deviations" and analyze the safety risks that may be caused by the "deviations"; further analyze the factors that may cause the "deviations" of the system to realize the process of identifying the hazard sources for the system elements; and conduct risk assessment based on the specific hazard sources that cause safety risks identified, determine the acceptability of the risks and add measures to further control safety risks; By sorting out and summarizing hazard source information and control measures, a multidimensional risk factor indicator group was constructed during the digital transformation period, and the impact of the coupling of multidimensional risk factors during the digital transformation period on the overall safety of production operations was obtained.

3. The multidimensional risk factor indicator identification method based on artificial intelligence technology according to claim 2 is characterized in that: Risk assessment includes: combining typical risk database and violation database information, based on the safety supervision big model, based on the PDCA theory, starting from the five dimensions of people, machines, materials, methods, and environment, analyzing the influencing factors of risk assessment level determination in various operation scenarios; using the multi-dimensional risk factor coupling weights and the multi-dimensional risk factor dynamic time series model to construct a safety assurance risk factor prediction model for substation maintenance, transmission line maintenance, and distribution maintenance production operations, and through machine learning and big data models to mine data value, obtain multiple risk evolution mechanisms during the risk occurrence process.

4. The multidimensional risk factor indicator identification method based on artificial intelligence technology according to claim 1 is characterized in that: Constructing a multidimensional risk factor indicator group under the background of digital transformation includes: based on the cumulative effect of risk transmission, obtaining the data of unilateral influencing factors of risk factors and constructing a risk factor indicator group; Constructing a multidimensional risk matrix includes: analyzing the possibility of accidents in the risk process based on historical data and real-time data of production safety risks, constructing a multidimensional risk matrix based on the severity of actual accidents, and obtaining a dynamic time series model of multidimensional risk factors.

5. The multidimensional risk factor indicator identification method based on artificial intelligence technology according to claim 1 is characterized in that: Establishing a risk factor prediction model based on resilience protection includes: based on the analysis of the mechanism of multi-dimensional risk factors and their interactions, mining the data value through machine learning, big data models, etc., combining the resilience characteristics of the power grid in responding to risks, obtaining multiple risk evolution mechanisms during the risk occurrence process, quickly perceiving and predicting the coupling relationship between risk factors, and forming a spatial risk prediction model for multiple uncertain factors.

6. The multidimensional risk factor indicator identification method based on artificial intelligence technology according to claim 1 is characterized in that: In view of the weights that should be given to each indicator in the multidimensional risk factor indicator group, the subjective fuzzy weighting method is used to obtain the subjective weight interval of the indicator; at the same time, based on the maximum entropy criterion, the information entropy of the factor weight eigenvector is used to quantify the elasticity of the risk identification result, and the obtained indicator weight interval is converted into a definite value; In addition, an adaptive weighting method that dynamically integrates subjective and objective factors is adopted, and the constraints of various indicators under different types of operation scenarios are combined to construct a multi-dimensional adaptive quantitative assessment model for operation risks. The factor weight mechanism dynamically adjusts the weight distribution based on the adaptive criterion.

7. The multidimensional risk factor indicator identification method based on artificial intelligence technology according to claim 5 or 6 is characterized in that: The construction of multi-dimensional risk factor indicator group and operation risk prediction model includes real-time monitoring and query of operation link risk verification status, construction of multi-dimensional operation risk assessment model and data input model assessment; among them, Real-time monitoring and query of risk verification of operation links includes obtaining multi-dimensional monitoring data on equipment status and operating environment changes at power sites through the data center, business center, and enterprise-level meteorological center; The multi-dimensional operation risk assessment model includes: a data receiving module, which is used to receive data through the data interface of the data center and the business center, and obtain field operation data, accident records, maintenance logs and operator information; a data preprocessing module, which is used to filter the noise in the received data through Kalman filtering, fill in the missing values ​​through KNN, and use the Z-Score standardization method to convert the data to a unified scale; an equipment status assessment module, which is used to diagnose the operating status of the equipment based on historical data and real-time data through the equipment status assessment model built by the support vector machine; an environmental risk assessment module, which is used to diagnose the operating status of the equipment based on the environmental risk assessment model built by the support vector machine according to the enterprise-level meteorological center Data evaluation of operating environment risks; personnel risk assessment module, which is used to identify potential personnel risks based on the personnel risk assessment model constructed by support vector machine and the personnel-related data of the risk management platform; multi-dimensional risk association module, which is used to integrate the risk data of equipment, environment and personnel through DS evidence theory, and to generate a risk association matrix by analyzing the association relationship between different risks through a priori knowledge algorithm; comprehensive risk assessment module, which is used to comprehensively evaluate the risk information of each dimension based on the fused data and the risk association matrix based on the fuzzy comprehensive evaluation method to generate the final risk assessment result; risk information output module, which is used to output the final risk assessment result through a compatible RESTful API interface; The data input model evaluation includes inputting the collected real-time data and historical data into the multi-dimensional operation risk assessment model to obtain the risks and risk information existing in the current power field operations. It is divided into a risk data fusion unit and a risk relationship analysis unit, and the output end of the risk data fusion unit is connected to the input end of the risk relationship analysis unit.

8. The multi-dimensional risk factor indicator identification method based on artificial intelligence technology according to claim 7 is characterized in that: The risk data fusion unit is used to fuse the risk data of equipment, environment and personnel through DS evidence theory to generate a comprehensive risk data set, including: Obtain risk assessment data of the equipment status assessment module, the environmental risk assessment module, and the personnel risk assessment module, which are recorded as E1, E2, and E3 respectively; The basic probability distribution of each risk assessment data is calculated by formula (1): Among them, m i (θ) is the basic probability distribution function of the ith risk dimension, E i is the risk assessment data of the ith risk dimension, θ is the risk state set, k is the slope adjustment parameter, t is the threshold, α is the amplitude of the sine function, and β is the frequency of the sine function; The basic probability distribution of each risk assessment data is synthesized through the synthesis rule of DS evidence theory, and the comprehensive basic probability distribution is obtained according to formula (2): Among them, m(θ) is the comprehensive basic probability distribution function, θ is the risk state set, K is the standardization coefficient, A, B and C are the subsets corresponding to different risk dimensions, and are the basic probability distributions of subsets corresponding to different risk dimensions, γ is the amplitude of the cosine function, and δ is the frequency of the cosine function; According to formula (3), the comprehensive basic probability distribution is normalized to obtain the comprehensive risk assessment value: Among them, R is the comprehensive risk assessment value, Θ is the set of all risk states, m(θ) is the comprehensive basic probability distribution function, θ is the set of risk states, λ is the amplitude of the tangent function, and μ is the frequency of the tangent function.

9. The multidimensional risk factor indicator identification method based on artificial intelligence technology according to claim 7 is characterized in that: The risk relationship analysis unit uses the association rule mining method based on prior knowledge algorithm to mine frequent item sets and association rules from the comprehensive risk data set, analyze the association relationship between risks of different dimensions, and generate a risk association matrix, including: The comprehensive risk data set is preprocessed through Kalman filtering and the data cleaning tool OpenRefine to remove noise and outliers from the comprehensive risk data set; Mining frequent itemsets in the comprehensive risk dataset through prior knowledge algorithms; Extract association rules from frequent item sets through prior knowledge algorithms to obtain the support and confidence of association rules; Filter high-credibility association rules based on support and confidence thresholds, and obtain the association strength of association rules through the Lift indicator; The association strength of the association rule is filled into the risk association matrix, and the elements of the risk association matrix represent the association strength between different risk dimensions.

10. The multidimensional risk factor indicator identification method based on artificial intelligence technology according to claim 7 is characterized in that: The comprehensive risk assessment module uses the fuzzy comprehensive evaluation method to conduct a comprehensive assessment of the risk information of each dimension based on the fused data and the risk association matrix to generate the final risk assessment results, including: Obtaining a comprehensive risk data set and a risk association matrix generated by the multi-dimensional risk association module; The membership degree of each risk item is calculated by the fuzzy comprehensive evaluation method, and the fuzzy evaluation matrix is ​​constructed to map each risk item in the comprehensive risk data set to the fuzzy set. The membership degree calculation formula (4) of each risk item is: Among them, μ i (x) is the membership degree of risk item i, x is the risk assessment value, c i is the central value of risk term i, σ i is the width of the fuzzy set, α i is the adjustment parameter; According to the risk association matrix and the membership of each risk item, the comprehensive membership of risk items of different dimensions is calculated by Sugeno fuzzy reasoning. The calculation formula (5) is: Among them, μ all (x) is the comprehensive membership, w i is the weight of risk term i, μ i (x) is the membership degree of risk item i, Lift i is the correlation strength of risk term i, β i is the correlation strength influence coefficient; The final risk assessment result is generated through fuzzy decision making based on the comprehensive membership of risk items in different dimensions; The severity of each risk item is ranked by fuzzy analytic hierarchy process to identify high-risk areas.

Images