Electric energy meter fault operation and maintenance method based on electricity utilization acquisition data analysis

Abstract

The invention provides an electric energy meter fault operation and maintenance method based on power utilization acquisition data analysis, and relates to the technical field of electric energy meter fault operation and maintenance. Taking an ammeter assembly node as an initial node, mining a survival function and an electric energy meter survival label under a covariable, and constructing electric energy meter life chain information; introducing a triangular fuzzy number, and constructing an index triple for fault evaluation; constructing a survival analysis model by combining an analytic hierarchy process principle and introducing a hierarchical comparison matrix; interacting the power data flow, performing data flow classification and decision logic endowing, performing fault decision positioning, and determining an operation and maintenance fault list; and performing electric energy meter operation and maintenance management under the power system. According to the method and the device, the technical problem of insufficient fault processing accuracy caused by inflexible complex fault mode recognition of fault diagnosis in the prior art can be solved, and the fault processing accuracy is improved through real-time monitoring and positioning of the faults of the electric energy meter.

Description

Technical Field

[0001] The present application relates to the technical field of electric energy meter fault operation and maintenance, and in particular to an electric energy meter fault operation and maintenance method based on analysis of electricity consumption collection data. Background Art

[0002] As the complexity and scale of power systems continue to increase, the stability and accuracy of the operation of electric energy meters, as key components in the power system, are crucial to the reliability of the entire system. Electric energy meters are gradually shifting towards intelligence and digitization. Traditional mechanical electric energy meters are gradually being replaced by smart electric energy meters. Smart electric energy meters have functions such as automatic meter reading, two-way communication and remote monitoring, which greatly improve the operation and maintenance management level of electric energy meters. Some advanced smart electric energy meters have self-healing capabilities. For example, when a minor fault is detected (such as network interruption or short-term voltage fluctuation), the electric energy meter can automatically perform recovery operations to avoid the expansion of the fault or affect the power supply quality. This self-healing function reduces the dependence on manual maintenance. Although the electric energy meter has self-diagnosis capabilities, the diagnostic flexibility of existing technologies is still insufficient when faced with complex failure modes intertwined with multiple factors (such as power fluctuations, equipment aging and environmental factors).

[0003] In summary, the prior art has a technical problem in that the existing fault diagnosis method lacks flexibility in processing complex fault modes, resulting in insufficient accuracy in fault processing. Summary of the invention

[0004] The purpose of this application is to provide an electric energy meter fault operation and maintenance method based on electricity consumption collection data analysis, so as to solve the technical problem in the prior art that the existing fault diagnosis method lacks flexibility in handling complex fault modes, resulting in insufficient accuracy of fault handling.

[0005] In view of the above problems, the present application provides an electric energy meter fault operation and maintenance method based on power consumption collection data analysis, wherein the electric energy meter fault operation and maintenance method based on power consumption collection data analysis includes: retrieving and retrieving the ammeter fault record, wherein the ammeter fault record includes a self-test fault record and a data flow record; taking the ammeter assembly node as the starting node, based on the ammeter fault record, mining the survival function and the electric energy meter survival label under the covariate, and constructing the electric energy meter life chain information, wherein the covariate is a strong correlation factor; introducing triangular fuzzy numbers, and constructing an indicator triple for fault evaluation, wherein the indicator triple is composed of a minimum value, a possible value, and a maximum value; based on the electric energy meter life chain information and the indicator triple, combining the hierarchical analysis principle and introducing a hierarchical comparison matrix, constructing a survival analysis model, wherein the analysis hierarchy includes a target layer, a rule layer, and an indicator layer, and the hierarchical comparison matrix is ​​used for indicator weighting; interacting with the power data flow, classifying the data flow and assigning decision logic, locating the fault decision in the survival analysis model, and determining the operation and maintenance fault sheet; based on the operation and maintenance fault sheet, performing operation and maintenance management of the electric energy meter under the power system.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: By retrieving and retrieving the ammeter fault record, the ammeter fault record includes the self-test fault record and the data flow record; taking the ammeter assembly node as the starting node, based on the ammeter fault record, mining the survival function and the electric energy meter survival label under the covariate, and constructing the electric energy meter life chain information, the covariate is a strong correlation factor; introducing triangular fuzzy numbers, constructing the indicator triple of fault evaluation, the indicator triple consists of the minimum value, the possible value and the maximum value; based on the electric energy meter life chain information and the indicator triple, combining the hierarchical analysis principle and introducing the hierarchical comparison matrix, constructing the survival analysis model, wherein the analysis hierarchy includes the target layer, the rule layer and the indicator layer, the hierarchical comparison matrix is ​​used for indicator weighting; interactive power data flow, data flow classification and decision logic assignment, fault decision location in the survival analysis model, determine the operation and maintenance fault ticket; based on the operation and maintenance fault ticket, the electric energy meter operation and maintenance management under the power system is carried out. In other words, through comprehensive fault record retrieval and analysis, the introduction of triangular fuzzy numbers and hierarchical analysis method, data classification and decision logic assignment, flexible processing of complex fault modes, accurate identification of faults, and dynamic decision-making based on real-time data, the accuracy of fault handling is improved.

[0007] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0009] Figure 1 This is a flow chart of the electric energy meter fault operation and maintenance method based on the analysis of electricity consumption collection data in this application.

[0010] Figure 2 This is a flow chart of determining an operation and maintenance fault ticket in the electric energy meter fault operation and maintenance method based on analysis of electricity consumption collection data in this application. DETAILED DESCRIPTION

[0011] This application solves the technical problem in the prior art that the existing fault diagnosis method lacks flexibility in handling complex fault modes, resulting in insufficient accuracy of fault handling, by providing a fault maintenance method for electric energy meters based on power consumption data analysis. Through comprehensive fault record retrieval and analysis, the introduction of triangular fuzzy numbers and hierarchical analysis method, data classification and decision logic assignment, flexible handling of complex fault modes, accurate fault identification, and dynamic decision making based on real-time data, the accuracy of fault handling is improved.

[0012] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0013] For examples, please see the attached Figure 1The present application provides an electric energy meter fault operation and maintenance method based on power consumption collection data analysis, wherein the electric energy meter fault operation and maintenance method based on power consumption collection data analysis specifically includes the following steps: Step 1: Retrieve and retrieve the ammeter fault record, which includes a self-test fault record and a data flow record.

[0014] Specifically, the fault-related data of the ammeter is obtained from the monitoring and data acquisition system of the power system, including the fault record generated during the self-test of the ammeter and the data flow record before and after the fault occurs. The self-test fault record refers to any abnormal or fault information detected by the ammeter during the self-test process, which is usually the self-test performed by the built-in diagnostic program of the ammeter during startup or operation, while the data flow record contains the operation data of the ammeter over a period of time, covering the entire cycle of the ammeter from normal operation to failure and then back to normal, such as the real-time measurement values ​​of parameters such as current and voltage. The data flow record usually refers to the current data continuously recorded by the ammeter under normal working conditions. When the ammeter is working normally, the recorded data flow is used to monitor the performance of the ammeter to ensure that it operates accurately within the specified working range. When the data flow shows abnormal patterns or exceeds the normal range, the data flow can help detect potential faults, including data before, during and after the fault occurs. After the fault occurs, analyzing the data flow records before and after the fault can help technicians understand the cause and development process of the fault. By retrieving and retrieving the fault records of the ammeter, ensuring that the fault diagnosis and analysis are based on comprehensive and accurate data, it helps to improve the accuracy of fault diagnosis and reduce the possibility of misdiagnosis and missed diagnosis.

[0015] Step 2: Taking the ammeter assembly node as the starting node, based on the ammeter fault record, mining the survival function and the electric energy meter survival label under the covariate, and constructing the electric energy meter life chain information, wherein the covariate is a strong correlation factor.

[0016] Specifically, the assembly node of the ammeter is determined as the starting point of the analysis, that is, the time point when the electric energy meter completes assembly and starts its life cycle. Using fault record data, factors highly correlated with ammeter failure are mined as covariates, including environmental factors, operating conditions, electric energy meter characteristics, etc. Based on fault records and covariates, the probability of the ammeter not failing at each time point is calculated, and a label is assigned to each ammeter to indicate whether it fails at a specific time point. The fault phase node of each ammeter is associated with its corresponding survival function and survival label, and how the covariate affects the survival time and failure risk is analyzed. A time series model is constructed to show the entire life cycle of the ammeter from assembly to possible failure, and the life chain information of the electric energy meter is obtained. The survival function is a statistical method used to estimate the probability that a device will not fail before a specific time point. The survival label is a label assigned to the electric energy meter based on the results of the survival function analysis, which is used to represent its failure risk and health status. By taking the ammeter assembly node as the starting node and mining the survival function and the energy meter survival label under the covariate based on the ammeter fault record, a detailed energy meter life chain information model is created, which can not only describe the survival status of the ammeter at different time points, but also reveal the key factors affecting its reliability.

[0017] Step 3: Introduce triangular fuzzy numbers and construct an indicator triplet for fault evaluation, which consists of a minimum value, a possible value, and a maximum value.

[0018] Specifically, triangular fuzzy numbers are a mathematical tool for dealing with uncertainty, which can describe the uncertainty distribution of a variable within a certain interval. Triangular fuzzy numbers can be expressed as (a, m, b), where: a is the minimum value (lower bound) of the fuzzy number, m is the possible value (most likely or central value) of the fuzzy number, and b is the maximum value (upper bound) of the fuzzy number. For each evaluation index, a triple is constructed, which is composed of triangular fuzzy numbers, including the probability of failure, mean time between failures, fault repair time, degree of fault impact, maintenance cost, etc. Based on this value, an evaluation is performed within the interval, and the index is evaluated within the range between the minimum and maximum values, while the possible value gives the most likely result, which improves the accuracy of the analysis. Considering all values ​​within this range, rather than just a single point estimate, helps avoid missed evaluations and misjudgments. By introducing triangular fuzzy numbers and constructing a triplet of indicators for fault evaluation, the uncertainty of the fault indicator is accurately described, and missed evaluations and misjudgments are avoided, which helps improve the accuracy of fault diagnosis and analysis, thereby more effectively identifying and predicting faults.

[0019] Step 4: Based on the life chain information of the energy meter and the indicator triplet, combined with the hierarchical analysis principle and introducing the hierarchical comparison matrix, a survival analysis model is constructed, wherein the analysis hierarchy includes the target layer, the rule layer and the indicator layer, and the hierarchical comparison matrix is ​​used for indicator weighting.

[0020] Specifically, the life chain information of the electric energy meter is combined with the indicator triples, and then the analytic hierarchy process (AHP) and hierarchical comparison matrix are used to determine the weight of each indicator, and finally a model for survival analysis is constructed. The analytic hierarchy process is a decision analysis method that combines qualitative and quantitative methods. When constructing a survival analysis model, the problem is decomposed into a target layer, a rule layer, and an indicator layer. The target layer is the highest level of decision-making, which usually represents the goal or final result that the decision-making hopes to achieve. The rule layer is the middle layer, which contains the various criteria or rules that need to be considered to achieve the goal. The indicator layer is the lowest level, which contains specific indicators or variables for evaluating and comparing different solutions. The hierarchical comparison matrix is ​​used to compare the relative importance of different indicators and determine the element values ​​in the matrix through expert scoring or other methods. The weight of each indicator is calculated according to the hierarchical comparison matrix for indicator weighting. After completing the hierarchical comparison matrix, the weight of each indicator is calculated, which is usually completed by calculating the maximum eigenvalue of the matrix and the corresponding eigenvector. The elements in the eigenvector are the weight of each indicator, reflecting the relative importance of each indicator in the overall decision. The indicators are weighted by the hierarchical analysis method to accurately reflect the impact of different factors on the survival time of the electricity meter, which helps to accurately assess the failure risk of the electricity meter.

[0021] Step 5: Interact with the power data stream, classify the data stream and assign decision logic, locate the fault decision in the survival analysis model, and determine the operation and maintenance fault ticket.

[0022] Specifically, interacting with the power data stream, that is, receiving and processing the real-time collected electric energy meter data, reflects the instantaneous state of the system. Traversing the ammeter fault records, clustering processing is performed based on the decision dimension, that is, the data stream is divided into different categories according to the characteristics of the data stream (such as current size, voltage fluctuation, power factor, etc.), and the distinctive features and fault decision logic are mined to build a logical classifier. The data stream is classified using a logical classifier, and the corresponding decision logic is assigned, and a specific decision logic is defined for each classification. The survival analysis model is used to locate the fault decision of these classified and logically assigned data streams to locate possible fault points. Based on the results of the fault decision location, an operation and maintenance fault ticket is generated, including the fault type, fault level, recommended maintenance or inspection steps, expected maintenance time window, required resources and tools, etc. By interacting with the power data stream, classifying the data stream and assigning decision logic, and locating the fault decision in the survival analysis model, the state of the power system is monitored in real time, and the fault is accurately diagnosed, which improves the accuracy and efficiency of fault diagnosis, helps to reduce fault downtime, and improves the reliability and operation efficiency of the power system.

[0023] Step 6: Based on the operation and maintenance fault sheet, perform operation and maintenance management of the electric energy meter in the power system.

[0024] Specifically, the generated operation and maintenance trouble tickets are reviewed to ensure that the fault description is accurate and the recommended maintenance steps are reasonable. The priority of fault handling is determined according to the severity and urgency of the fault. For example, emergency faults need to be handled immediately, while routine faults can be arranged during off-peak hours. According to the needs of the trouble ticket, the necessary resources and tools, such as test equipment, spare parts, etc., are dispatched, and the operation and maintenance team with corresponding skills and experience is assigned to handle the fault. According to the inspection results, appropriate maintenance measures are taken, such as replacing faulty parts, adjusting settings, cleaning equipment, etc. After the maintenance is completed, functional tests are performed to ensure that the electric energy meter resumes normal operation. The process of fault handling, the resources used, and the maintenance results are recorded in detail, and the process of fault handling, the resources used, and the maintenance results are recorded in detail. By performing operation and maintenance management of electric energy meters under the power system based on operation and maintenance trouble tickets, timely respond to the results of fault diagnosis and prediction, and take appropriate maintenance and replacement measures.

[0025] Furthermore, step 2 of this application includes: Determine a preset sampling method, wherein the preset sampling method is identified with a relative proportion and is a hybrid sampling method based on oversampling and undersampling, and is used to eliminate class imbalance in samples; based on the preset sampling method, traverse the ammeter fault records to perform information mining.

[0026] Specifically, a suitable sampling strategy is selected to solve the class imbalance problem in the data set, and the selected sampling method is used to process the ammeter fault records to mine valuable information. Since fault data often has class imbalance problems, that is, normal data is far more than fault data, it will lead to insufficient prediction ability of machine learning models for fault data during training. Class imbalance refers to the large difference in the number of samples of different categories in the data set, which makes the machine learning model overly sensitive to the majority class and insensitive to the minority class during training, thereby affecting the performance of the model. In data processing, the sampling method refers to the method of selecting a subset from the original data set. The preset sampling method is usually a pre-set rule or algorithm based on a specific data analysis goal. It combines oversampling and undersampling hybrid sampling methods to eliminate or reduce the impact of class imbalance by adjusting the number of samples. Oversampling is to balance the data set by increasing the number of minority class samples, which is achieved by copying existing samples, generating new synthetic samples (such as using the SMOTE algorithm), etc., while undersampling is to reduce the number of majority class samples and randomly delete majority class samples. Combining oversampling and undersampling, the number of minority class samples is increased and the number of majority class samples is reduced to achieve a better class balance. In sampling, relative proportion refers to the proportional relationship between samples of different categories.

[0027] In a specific example, suppose there are 100 fault samples and 1000 normal samples in a dataset about ammeter faults. In order to deal with this imbalance, a hybrid sampling method based on oversampling and undersampling is determined, and the relative ratio is set to 1:1. First, the fault samples are oversampled, and new synthetic fault samples are generated using the SMOTE algorithm, so that the number of fault samples increases to 200. Then the normal samples are undersampled, and some normal samples are randomly deleted, so that the number of normal samples is reduced to 200. A balanced data set is obtained, with 200 samples in each category.

[0028] After determining the sampling strategy, the ammeter fault records are deeply analyzed to extract valuable information from a large amount of fault data. The sampled data set is analyzed to extract useful information about the ammeter fault, and data mining techniques (such as association rule mining, cluster analysis, etc.) are used to discover fault patterns and related features. By determining a preset sampling method and traversing the ammeter fault records based on this method to conduct information mining, the class imbalance problem in the data set is effectively solved, the performance of the fault diagnosis model is improved, and the model is ensured not to be biased towards the majority class, thereby reducing misdiagnosis and missed diagnosis of minority classes (such as fault classes).

[0029] Furthermore, the present application also includes the following steps: Based on the preset sampling method, a fault sample set is determined; the fault sample set is traversed to construct a life chain with potential fault phase nodes, wherein the fault phase nodes are constrained by a preset frequency; the covariates under the link nodes are mined for each life chain to determine a survival function and a survival label of an electric energy meter; the life chain, the survival function and the survival label of the electric energy meter are associated to determine the life chain information of the electric energy meter.

[0030] Specifically, according to the determined sampling strategy, a fault sample set is selected from the ammeter fault record, which contains various parameters when the fault occurs, such as current, voltage, temperature, etc. The fault sample set is traversed, and each fault sample is analyzed to identify the phase node where the fault occurs, that is, the exact time point when the fault occurs. The fault phase node represents the key moments from the normal operation of the ammeter to the failure, that is, the specific time point before and after the failure occurs. Potential fault phase nodes refer to the moments or states that may indicate an impending failure in the life cycle of the ammeter. For example, an abnormal increase in the current of the ammeter may be a potential fault phase node. When constructing the life chain, the preset frequency (for example, daily, weekly) is used as a constraint to ensure that the time interval considered in the analysis process is consistent. By connecting the fault phase nodes, a link representing the entire life cycle of the ammeter from assembly to failure is constructed. The preset frequency is a constraint for screening the fault phase nodes when constructing the life chain, and the number of occurrences of the fault phase nodes is limited to avoid misjudging accidental events as fault signs. The life chain is a tracking and description of the entire process of the ammeter from normal operation to failure, which usually includes a series of key nodes, such as the assembly of the ammeter, normal operation, signs of failure, and final failure.

[0031] Along the constructed life chain, each link node is analyzed to identify and record the covariates that may affect the survival time of the ammeter. Covariates are other variables in the statistical model other than the main research variables, such as ambient temperature, humidity, current load, etc. The survival function is calculated for each link node using statistical methods (such as Kaplan-Meier estimation), that is, the probability that the ammeter is still working properly at this node. According to the analysis results of the survival function and covariates, a label is assigned to each ammeter to represent its failure risk and health status, indicating whether the electric energy meter is not faulty or has failed at a specific time point. The information of each node in the life chain is combined with the corresponding survival function and survival label to form a complete life chain information of the electric energy meter, which contains the entire history of the ammeter from assembly to failure, as well as the survival probability and status at each key time node. By determining the fault sample set and constructing the life chain information of the electric energy meter, a deeper understanding of the failure mode and survival characteristics of the electric energy meter is achieved, which helps to predict the remaining service life of the electric energy meter and provide a basis for maintenance and replacement decisions of the power system.

[0032] Further, step five of this application includes: The ammeter fault records are traversed, clustering is performed based on the decision dimension, and N cluster clusters are determined; for the N cluster clusters, classification features and fault decision logic are mined, and the classification features are distinguishing features; a logic classifier is constructed based on the classification features and the fault decision logic; based on the logic classifier, data stream classification and decision logic assignment are performed.

[0033] Specifically, the ammeter fault records are checked one by one to ensure that all relevant data are taken into account, and the fault records are divided into different clusters according to decision dimensions (such as fault type, fault severity, etc.). Decision dimensions refer to the characteristics or attributes used to judge the similarity of data points in the clustering process. For example, in the clustering of ammeter fault records, the decision dimensions may include fault type, fault occurrence time, fault severity, etc. The fault records are grouped using a clustering algorithm (such as K-means, hierarchical clustering, or density-based clustering methods) to determine N clusters, where the value of N is usually determined based on the distribution of the data and the purpose of clustering. Clustering is an unsupervised learning method that divides data into several categories so that the data within the same category has a high similarity, while the data between different categories has a low similarity.

[0034] Analyze the characteristics of each cluster, identify the distinctive features that distinguish different clusters, and establish fault decision logic based on these features. Fault decision logic refers to the logical rules or models used to determine which cluster a new fault record belongs to. Traverse each item of the classification feature, complete the construction of N classification layers, traverse the fault decision logic, match and mark the records in each subset of the fault record, and determine the fault decision logic corresponding to each subset. At each classification layer, mark the data according to the current classification features and fault decision logic. Associate the first classification layer with the subsequent classification layers according to the importance of the features or the order of decision logic to form a hierarchical logical classifier, and each classification layer is part of the logical classifier. Use the constructed logical classifier to classify the data stream and assign the corresponding decision logic according to the classification results. Data stream classification refers to classifying the real-time collected ammeter data stream according to the rules of the logical classifier to identify the fault type and fault level. Decision logic assignment refers to assigning corresponding processing suggestions or operation instructions to each fault record according to the classification results. For example, if the fault record is classified as a serious fault, it is assigned an operation instruction to immediately shut down and check.

[0035] For example, assume that there is a logic classifier containing the following fault decision logic: if the current value exceeds threshold A, the fault record belongs to the severe fault category; if the current value is between threshold A and threshold B, and the temperature value exceeds threshold C, the fault record belongs to the medium fault category; if the current value is between threshold A and threshold B, and the temperature value is between threshold C and threshold D, and the humidity value exceeds threshold E, the fault record belongs to the minor fault category. By traversing the ammeter fault records and performing clustering processing, understanding the fault mode, and providing a basis for fault diagnosis and prediction, mining classification features and fault decision logic, and constructing a logic classifier, the accuracy and efficiency of fault classification are improved, data stream classification and decision logic are assigned based on the logic classifier, and real-time monitoring and intelligent decision-making of electric energy meter faults are realized.

[0036] Furthermore, the present application also includes the following steps: Based on the classification feature, one item is randomly extracted to construct the first classification layer, and the fault decision logic is traversed to perform matching marking; the classification feature is traversed to complete the construction of the Nth classification layer and the fault decision logic marking; the first classification layer is hierarchically associated to the Nth classification layer to generate the logical classifier.

[0037] Specifically, according to the classification feature, one item is randomly selected as the basis for the first classification layer, and the fault records are divided into several subsets based on the selected feature, each of which represents a different range or category of the feature. For example, assuming that the randomly selected classification feature is the current value, the constructed first classification layer will divide the fault records into several groups based on the current value, such as normal current, slight overload, severe overload, etc. The fault decision logic is traversed, and the records in each subset are matched and marked to determine the fault decision logic corresponding to each subset. The fault record is compared with the fault decision logic, and the fault record is marked or labeled according to the matching result. For example, if the current value of the fault record exceeds a certain threshold, it is marked as severe overload.

[0038] The remaining classification features are traversed in turn. For each classification feature, the construction process of the first classification layer is repeated until all features are considered. At each classification layer, the data is labeled according to the current classification features and fault decision logic. Each new classification layer will further refine the classification of faults based on the results of the previous classification layer. The first classification layer is associated with subsequent classification layers according to the order of feature importance or decision logic to form a hierarchical logical classifier, and each classification layer is part of the logical classifier. Hierarchical association refers to the process of associating multiple classification layers according to a certain hierarchical structure. Each classification layer is built based on a specific classification feature and arranged in order from coarse to fine to form a hierarchical classification system. The logical classifier can classify fault data according to multiple classification features and make corresponding decisions based on the fault decision logic.

[0039] In a specific example, current is randomly extracted as the first classification feature, and the data is divided into three subsets (for example, current <9A, 9A≤current<11A, current≥11A). Fault decision logic is applied to each subset, for example: current <9A: marked as low current fault risk; 9A≤current<11A: marked as medium current fault risk; current≥11A: marked as high current fault risk. Traverse the classification features to complete the construction of the Nth classification layer and the fault decision logic marking. Select temperature as the feature of the second classification layer and humidity as the feature of the third classification layer, and apply the corresponding fault decision logic on each subset. Associate the first classification layer with the second classification layer and the third classification layer, and the logical classifier will contain three classification layers: the first classification layer is based on current, the second classification layer is based on temperature, and the third classification layer is based on humidity. Through the hierarchical classification layer, the fault mode is finely identified and classified, and multiple features and the relationship between them are considered, thereby improving the accuracy and efficiency of fault classification.

[0040] Further, as attached Figure 2 As shown, the application also includes the following steps: Classify data streams and assign decision logic to determine grouped data streams; determine the life chain nodes of the grouped data streams; traverse the grouped data streams with the life chain nodes as constraints, and perform multi-layer concurrent data feature extraction and weight assignment on each group of data streams based on the target layer, rule layer and indicator layer in combination with the hierarchical comparison matrix to obtain a grouped feature set; perform triangular fuzzy fault decision positioning based on the grouped feature set.

[0041] Specifically, according to the output of the logical classifier, the data streams are divided into different groups. Each group of data streams represents a specific fault mode or category. The corresponding decision logic is assigned to each grouped data stream, and the data streams are divided into different groups. Each group contains data points with similar characteristics. The key nodes of each grouped data stream in the life chain of the electric energy meter are identified, including the time point of the fault, maintenance events, etc. The life chain nodes are used as constraints to ensure that the analysis focuses on the key stages of the life chain of the electric energy meter. The hierarchical analysis method is combined to extract features and assign weights to each grouped data stream. The hierarchical comparison matrix is ​​used to analyze the data stream features from different levels and angles, and the relative importance of each feature is determined. The extracted features and the assigned weights are combined into a group feature set. Each feature set represents the characteristics of a data stream group.

[0042] Based on the group feature set, triangular fuzzy fault decision location is performed to deal with the uncertainty and ambiguity in fault decision. Triangular fuzzy number is a mathematical tool for representing fuzzy information. It consists of three parameters: minimum value, possible value and maximum value. It is used to represent a fuzzy concept or numerical range, where the possible value is the most likely value, and the minimum value and maximum value represent the lower and upper bounds of the fuzzy number respectively. The uncertainty and ambiguity in fault decision are accurately handled by using triangular fuzzy numbers. The group feature set is analyzed by triangular fuzzy numbers to determine the fault risk level and specific fault location of each data stream group. For example, suppose there are three group data streams, corresponding to different fault modes: group data stream A corresponds to low current fault mode, group data stream B corresponds to high temperature fault mode, and group data stream C corresponds to high operating frequency fault mode. Target layer: ensure the normal operation of the electric energy meter, rule layer: fault prevention, fault detection, fault repair, indicator layer: current, temperature, operating frequency. The following weights are obtained using the hierarchical comparison matrix: current: 0.5, temperature: 0.3, operating frequency: 0.2. For each grouped data stream, triangular fuzzy numbers are used to represent its eigenvalues. The current characteristics of grouped data stream A are represented as (0.2, 0.4, 0.6), and the risk level is positioned as "low risk". Through triangular fuzzification, faults can be accurately located, even in the case of uncertain data. The grouped feature set and triangular fuzzification method make fault management more flexible and can adapt to different fault modes and scenarios.

[0043] Furthermore, the present application also includes the following steps: The indicator triples are traversed, the group feature set is converted, and the group feature triples are determined; based on the group feature triples, ternary value-based decision-making and interval analysis are performed to determine the operation and maintenance fault ticket, wherein the interval analysis is based on the interval range limited by the ternary value.

[0044] Specifically, the indicator triple (consisting of minimum value, possible value and maximum value) is applied to the group feature set for traversal to ensure that each feature is considered. The features in each group feature set are converted into triples to determine the group feature triples. The group feature triples consist of minimum value, possible value and maximum value, which are used to represent the uncertainty of the feature value. The feature triples are used to make decisions, considering the uncertainty of the feature value. The analysis is based on the interval range defined by the ternary value to determine the possibility of fault. For example, an interval range (4.8A, 5.2A) is set as the normal working range. If the feature value falls within this interval, it is considered that the device is operating normally; if it exceeds this interval, there may be a fault. Based on the results of decision-making and interval analysis, an operation and maintenance fault ticket is generated, which contains information such as fault type, possible cause, and recommended maintenance measures. For example, if the current center value of group data stream A continues to exceed 5.2A, a fault ticket is generated, and it is recommended to check the current sensor or related circuits. Through ternary value and interval analysis, faults can be accurately diagnosed, false alarms and missed alarms can be reduced, and decision-making based on ternary values ​​and interval analysis can handle the uncertainty of characteristic values, thereby improving the accuracy and robustness of fault diagnosis.

[0045] Furthermore, the present application also includes the following steps: The grouped data streams are identified, and static data vectors and dynamic data trends are determined group by group; based on the static data vectors and the dynamic data trends, fault decision locating is performed in a dynamic and static collaborative manner.

[0046] Specifically, identify the grouped data streams and analyze each grouped data stream separately. For each grouped data stream, extract the data features at a specific time point, such as current value, temperature, operating frequency, etc., to form a static data vector. The static data vector refers to the specific value of the data stream at a certain point in time. Analyze the changes of each grouped data stream over time, that is, the dynamic data trend, and identify the trends such as rising, falling or stable. Combined with the static data vector and the dynamic data trend, conduct a comprehensive analysis of each grouped data stream, and use the static and dynamic data analysis results to determine the fault risk level and specific fault location of each grouped data stream. For example, assume that there are the following data stream groups: Grouped data stream A: The current value fluctuates around 5A, but the trend shows that the current value is slowly rising; Grouped data stream B: The temperature is stable at 30°C, with no obvious change trend; Grouped data stream C: The operating frequency is 100 times / hour, but the trend shows that the operating frequency is gradually decreasing. Fault decision location: Group data stream A: The static data vector shows that the current value is normal, but the dynamic trend shows that the current value is rising, which may indicate a potential fault and requires further inspection; Group data stream B: Both static and dynamic data show normal, and the risk of fault is low; Group data stream C: The static data vector shows that the operating frequency is normal, but the dynamic trend shows a decrease, and the cause needs to be investigated. By combining static and dynamic data, the characteristics and fault risks of the data stream are fully understood, and the dynamic and static collaborative analysis helps to more accurately predict potential faults, so that preventive measures can be taken in advance.

[0047] In summary, the electric energy meter fault operation and maintenance method based on power consumption collection data analysis provided by this application has the following technical effects: By retrieving and retrieving the ammeter fault record, the ammeter fault record includes the self-test fault record and the data flow record; taking the ammeter assembly node as the starting node, based on the ammeter fault record, mining the survival function and the electric energy meter survival label under the covariate, and constructing the electric energy meter life chain information, the covariate is a strong correlation factor; introducing triangular fuzzy numbers, constructing the indicator triple of fault evaluation, the indicator triple consists of the minimum value, the possible value and the maximum value; based on the electric energy meter life chain information and the indicator triple, combining the hierarchical analysis principle and introducing the hierarchical comparison matrix, constructing the survival analysis model, wherein the analysis hierarchy includes the target layer, the rule layer and the indicator layer, the hierarchical comparison matrix is ​​used for indicator weighting; interactive power data flow, data flow classification and decision logic assignment, fault decision location in the survival analysis model, determine the operation and maintenance fault ticket; based on the operation and maintenance fault ticket, the electric energy meter operation and maintenance management under the power system is carried out. In other words, through comprehensive fault record retrieval and analysis, the introduction of triangular fuzzy numbers and hierarchical analysis method, data classification and decision logic assignment, flexible processing of complex fault modes, accurate identification of faults, and dynamic decision-making based on real-time data, the accuracy of fault handling is improved.

[0048] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0049] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. An electric energy meter fault operation and maintenance method based on power consumption data analysis, characterized in that: include: Retrieving and retrieving the ammeter fault record, wherein the ammeter fault record includes a self-test fault record and a data flow record; Taking the ammeter assembly node as the starting node, based on the ammeter fault record, mining the survival function and the electric energy meter survival label under the covariate, and constructing the electric energy meter life chain information, wherein the covariate is a strong correlation factor; Triangular fuzzy numbers are introduced to construct the indicator triplet of fault evaluation, which consists of minimum value, possible value and maximum value. Based on the energy meter life chain information and the indicator triplet, combined with the hierarchical analysis principle and introducing the hierarchical comparison matrix, a survival analysis model is constructed, wherein the analysis hierarchy includes the target layer, the rule layer and the indicator layer, and the hierarchical comparison matrix is ​​used to perform indicator weighting; Interact with power data streams, classify data streams and assign decision logic, locate fault decisions in the survival analysis model, and determine operation and maintenance fault tickets; Based on the operation and maintenance trouble ticket, the operation and maintenance management of the electric energy meter in the power system is carried out.

2. The electric energy meter fault operation and maintenance method based on power consumption data analysis according to claim 1, characterized in that: The survival function under the mining covariate and the survival label of the electric energy meter include: Determine a preset sampling method, wherein the preset sampling method is marked with a relative ratio and is a mixed sampling method based on oversampling and undersampling, and is used to eliminate class imbalance in samples; Based on the preset sampling method, the ammeter fault records are traversed to perform information mining.

3. The electric energy meter fault operation and maintenance method based on power consumption data analysis according to claim 2, characterized in that: The construction of the energy meter life cycle information includes: Based on the preset sampling method, determining a fault sample set; Traversing the fault sample set, constructing a life chain with potential fault phase nodes, wherein the fault phase nodes are constrained by a preset frequency; According to the life chain, the covariates under the link nodes are mined to determine the survival function and the survival label of the electric energy meter; The life chain, the survival function and the electric energy meter survival tag are associated to determine the electric energy meter life chain information.

4. The electric energy meter fault operation and maintenance method based on power consumption data analysis according to claim 1, characterized in that: The data flow classification and decision logic assignment include: Traversing the ammeter fault records, performing clustering processing based on decision dimensions, and determining N clusters; For the N clusters, mining classification features and fault decision logic, the classification features are distinguishing features; Based on the classification features and the fault decision logic, construct a logic classifier; Based on the logic classifier, data stream classification and decision logic assignment are performed.

5. The electric energy meter fault operation and maintenance method based on power consumption data analysis according to claim 4, characterized in that: The construction of a logical classifier comprises: Based on the classification feature, randomly extract one item, construct a first classification layer, and traverse the fault decision logic to perform matching marking; Traversing the classification features, completing the construction of the Nth classification layer and fault decision logic marking; The first classification layer up to the Nth classification layer are hierarchically associated to generate the logical classifier.

6. The electric energy meter fault operation and maintenance method based on power consumption data analysis according to claim 1, characterized in that: The survival analysis model is used to locate the fault decision and determine the operation and maintenance fault ticket, including: Perform data flow classification and decision logic assignment to determine the grouped data flow; Determine a life chain node of the packet data flow; Taking the life chain nodes as constraints, traverse the grouped data streams, and based on the target layer, rule layer and indicator layer, in combination with the hierarchical comparison matrix, perform multi-layer concurrent data feature extraction and weight assignment on each group of data streams to obtain a grouped feature set; Based on the grouping feature set, triangular fuzzy fault decision positioning is performed.

7. The electric energy meter fault operation and maintenance method based on power consumption data analysis according to claim 6, characterized in that: The triangular fuzzy fault decision location method includes: Traversing the indicator triples, converting the grouping feature set, and determining the grouping feature triples; Based on the grouping feature triples, a decision based on ternary values ​​and interval analysis are performed to determine the operation and maintenance fault ticket, wherein the interval analysis is based on an interval range limited by the ternary values.

8. The electric energy meter fault operation and maintenance method based on power consumption data analysis according to claim 6, characterized in that: After determining the packet data flow, including: Identifying the grouped data streams, and determining static data vectors and dynamic data trends group by group; Based on the static data vector and the dynamic data trend, the static and dynamic data are coordinated to make fault decision and locate.

Images