Operation monitoring method and system for electric energy meter

Abstract

The invention discloses an operation monitoring method and system for an electric energy meter, and relates to the related technical field of electric energy meter monitoring, and the method comprises the steps: building an electric energy collection parameter library which comprises current, voltage, power and power factors; performing data cleaning and compression on the acquisition parameters to obtain a to-be-analyzed database; searching and updating according to the to-be-analyzed database, and determining a clustering center; calculating the clustering probability of each data point in the database to be analyzed and the clustering center; and obtaining an abnormal clustering probability threshold value, and performing abnormal identification on each data point by using the abnormal clustering probability threshold value to obtain abnormal operation data. The technical problems that according to existing electric energy meter operation monitoring, subtle changes and abnormal states in the electric energy meter operation process are difficult to accurately recognize in the face of massive and complex electric energy data, so that the electric energy meter abnormality recognition timeliness is poor, and the monitoring efficiency is low are solved. The technical effect of improving the anomaly recognition timeliness and the monitoring efficiency is achieved.

Description

Technical Field

[0001] The present application relates to the technical field related to electric energy meter monitoring, and specifically to an operation monitoring method and system for an electric energy meter. Background Art

[0002] In the daily operation and management of the power system, the accurate operation and monitoring of the energy meter plays a vital role. As the core equipment for power metering and monitoring, the energy meter not only records the user's power consumption, but also provides key data support for the stable operation, energy efficiency analysis and troubleshooting of the power system. With the continuous advancement of smart grid technology and the widespread application of Internet of Things technology, the operation monitoring methods of the energy meter are also constantly innovating, aiming to achieve more efficient, accurate and real-time monitoring and analysis. However, the traditional operation monitoring of the energy meter mainly relies on manual meter reading and regular inspections. Faced with massive and complex energy data, this method is not only time-consuming and labor-intensive, but also difficult to efficiently process and analyze these data. It is impossible to capture abnormal operating conditions or subtle changes in the operation of the energy meter, which affects the timeliness and accuracy of the energy meter monitoring, and thus affects the stable operation of the power system.

[0003] Therefore, in the current technology related to the operation monitoring of electricity meters, there is a technical problem that it is difficult to accurately identify subtle changes and abnormal conditions in the operation of electricity meters in the face of massive and complex electricity data, which leads to poor timeliness in identifying abnormalities in electricity meters and low monitoring efficiency. Summary of the invention

[0004] The present application provides a method and system for monitoring the operation of an electric energy meter, thereby solving the technical problem that the existing operation monitoring of an electric energy meter is faced with massive and complex electric energy data, making it difficult to accurately identify subtle changes and abnormal conditions during the operation of the electric energy meter, which in turn leads to poor timeliness in identifying abnormalities in the electric energy meter and low monitoring efficiency. The application achieves the technical effect of improving the timeliness of abnormality identification and monitoring efficiency.

[0005] The present application provides an operation monitoring method for an electric energy meter, the method comprising: establishing an electric energy acquisition parameter library, the electric energy acquisition parameter library comprising current, voltage, power, and power factor; performing data cleaning and compression on the acquisition parameters according to the electric energy acquisition parameter library to obtain a database to be analyzed; performing search and update according to the database to be analyzed to determine a clustering center; based on the clustering center, calculating the clustering probability of each data point in the database to be analyzed and the clustering center; obtaining an abnormal clustering probability threshold, using the abnormal clustering probability threshold to perform abnormal identification on each data point, and obtaining abnormal operation data.

[0006] In a possible implementation, the operation monitoring method for the electric energy meter also performs the following processing: according to the electric energy acquisition parameter library, extract the acquisition timestamp and construct the electric energy meter data clock; based on the electric energy meter data clock, perform parameter classification extraction on the electric energy acquisition parameter library and construct a collection parameter category visual graph; based on the electric energy meter data clock, perform parameter classification alignment extraction on the database to be analyzed, construct a compressed parameter category visual graph, and establish an alignment mapping relationship between the compressed parameter category visual graph and the acquisition parameter category visual graph; use the clustering probability of the compressed parameter category visual graph and the cluster center to generate a normal parameter graph series and an abnormal parameter graph series; construct a hierarchical relationship visual graph series of the acquisition parameter category visual graph, the compressed parameter category visual graph, the cluster center, and the normal parameter graph series and the abnormal parameter graph series, and send it to the user display terminal.

[0007] In a possible implementation, according to the electric energy acquisition parameter library, the acquisition parameters are cleaned and compressed to obtain a database to be analyzed, and the following processing is performed: the electric energy acquisition parameter library is time-sequentially integrated according to the acquisition parameters using the acquisition timestamp to establish a time alignment relationship; according to the time alignment relationship, the missing abnormal parameter bits are identified; with the missing abnormal parameter bits as the center, the sliding time window is translated and slid to obtain the neighborhood time window mean; the missing abnormal parameter bits are filled with the neighborhood time window mean; the filled acquisition parameters are standardized and scaled to the same dimension to obtain the database to be analyzed.

[0008] In a possible implementation, with the missing abnormal parameter bit as the center, the sliding time window is translated and slid to obtain the neighborhood time window mean, and the following processing is also performed: according to the missing abnormal parameter bit, the missing time gradient is obtained; the time window is set according to the missing time gradient, and the setting size of the time window is not less than twice the missing time gradient; with the missing abnormal parameter bit as the center, the center is moved forward according to the setting size of the time window to obtain the leading neighborhood time window, and the center is moved backward according to the setting size of the time window to obtain the trailing neighborhood time window; the parameter mean is calculated according to the leading neighborhood time window and the trailing neighborhood time window to obtain the neighborhood time window mean.

[0009] In a possible implementation, the filled acquisition parameters are standardized and scaled to the same dimension to obtain the database to be analyzed, and the following processing is performed: a time dimension is set, and the time dimension includes a short-term dimension time period, a medium-term dimension time period, and a long-term dimension time period; based on each time dimension, acquisition parameter feature extraction is performed separately, including average value, peak value, minimum value, standard deviation, peak-to-valley difference, and trend feature; key feature extraction is performed on the parameter features of each time dimension, wherein the short-term dimension extracts high-frequency signals, the medium-term dimension extracts high-frequency signals and data trend features, and the long-term dimension retains data trend features; and the key features of each time dimension are dimensionally interacted to obtain the database to be analyzed.

[0010] In a possible implementation, the database to be analyzed is searched and updated to determine the cluster center, and the following processing is performed: according to the database to be analyzed, all data points to be clustered are divided into K clusters, the sum of square errors within the cluster is minimized, and the initial cluster center is obtained; N feature data individuals are randomly initialized, and the coordinates of the N feature data individuals are consistent with the coordinates of the K initial cluster centers; and the clustering target is used to construct a fitness function: , where Z is the characteristic data individual, Z={ }, are k cluster centers with feature dimensions consistent with the data points, is the i-th data point that needs to be clustered, There are K clusters; based on the dragonfly algorithm, the neighbor distribution of each feature data individual is calculated, the position and speed of each individual are updated, and the fitness of each feature data individual is calculated using the fitness function; through iterative update search, when the change value of the fitness calculation result is less than a preset threshold or the maximum number of iterations is reached, the cluster center is obtained.

[0011] In a possible implementation, based on the cluster center, the clustering probability of each data point in the database to be analyzed and the cluster center is calculated, and the following processing is also performed: for each data point, the distance to the cluster center is calculated: , m is the dimension of the data, and They are The value in the jth dimension; normalize all distance values ​​to obtain the normalization coefficient: ,in, is the maximum distance between all data points and the cluster center, is the minimum distance between all data points and the cluster center; according to the normalization coefficient, the clustering probability is calculated: ,in, The value is [0,1]. The value is [0,1].

[0012] The present application also provides an operation monitoring system for an electric energy meter, including: an electric energy acquisition parameter library establishment module, which is used to establish an electric energy acquisition parameter library, and the electric energy acquisition parameter library includes current, voltage, power, and power factor; a database to be analyzed acquisition module, which is used to clean and compress the acquisition parameters according to the electric energy acquisition parameter library to obtain the database to be analyzed; a cluster center determination module, which is used to search and update according to the database to be analyzed to determine the cluster center; a cluster probability calculation module, which is used to calculate the clustering probability of each data point in the database to be analyzed and the cluster center based on the cluster center; an abnormal operation data acquisition module, which is used to obtain an abnormal cluster probability threshold, and use the abnormal cluster probability threshold to identify abnormalities for each data point to obtain abnormal operation data.

[0013] Through the operation monitoring method and system for electric energy meters proposed in this application, it is planned to establish an electric energy acquisition parameter library, including current, voltage, power, and power factor; perform data cleaning and compression on the acquisition parameters to obtain a database to be analyzed; perform search and update based on the database to be analyzed to determine the cluster center; calculate the clustering probability of each data point in the database to be analyzed and the cluster center; obtain an abnormal clustering probability threshold, and use the abnormal clustering probability threshold to identify abnormalities for each data point to obtain abnormal operation data. This solves the technical problem that the existing operation monitoring of electric energy meters is faced with massive and complex electric energy data, making it difficult to accurately identify subtle changes and abnormal states during the operation of electric energy meters, which in turn leads to poor timeliness of abnormal identification of electric energy meters and low monitoring efficiency, and achieves the technical effect of improving the timeliness of abnormal identification and monitoring efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solution of the embodiment of the present disclosure, the accompanying drawings of the embodiment of the present disclosure will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the system according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0015] Figure 1 A schematic flow chart of a method for monitoring the operation of an electric energy meter provided in an embodiment of the present application.

[0016] Figure 2 A schematic diagram of the structure of an operation monitoring system for an electric energy meter provided in an embodiment of the present application.

[0017] Description of the reference numerals: power acquisition parameter library establishing module 10 , to-be-analyzed database obtaining module 20 , cluster center determining module 30 , cluster probability calculating module 40 , abnormal operation data obtaining module 50 . DETAILED DESCRIPTION

[0018] 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 in accordance with the contents of the specification. 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 listed below.

[0019] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0021] The present application provides an operation monitoring method for an electric energy meter, such as Figure 1 As shown, the method includes: Step S100, establishing an electric energy acquisition parameter library, wherein the electric energy acquisition parameter library includes current, voltage, power, and power factor.

[0022] Preferably, key electrical parameters of multiple electric energy meters, including current, voltage and power, are collected and acquired through sensors or other measuring devices, and the corresponding power factor is calculated. Specifically, current reflects the flow of electric energy in the circuit, which is directly related to the power consumption of the equipment and the transmission of electric energy; the higher the voltage, the stronger the ability of the current to flow in the circuit, which is used to calculate the working condition of the current under a specific voltage and to deduce the power consumption of the equipment; power refers to the amount of energy conversion completed per unit time, which is divided into active power (actual power doing work), reactive power (power generated by the phase difference between voltage and current) and apparent power (the vector sum of voltage and current, that is, the synthesis of active power and reactive power). The power factor The power factor is the ratio of active power to apparent power, usually expressed as a decimal between 0 and 1. The power factor reflects the efficiency of electric energy utilization, that is, the ratio of actual electric energy consumed to total electric energy transmitted. If the power factor is close to 1, it means that the energy conversion efficiency in the power system is high and the waste of electric energy is small; if the power factor is low, it means that the utilization rate of electric energy is low and there may be a large reactive power; then an electric energy acquisition parameter library is established with current, voltage, power and power factor to store and manage various parameters related to electric energy acquisition, and to monitor and analyze the use of electric energy and the operation of the power system. These data can be used to judge the consumption of electric energy, optimize the use of electricity and improve the efficiency of electric energy.

[0023] Step S200: Clean and compress the collected parameters according to the electric energy collection parameter library to obtain a database to be analyzed.

[0024] Preferably, the power acquisition parameters in the power acquisition parameter library are cleaned and compressed. Specifically, data cleaning includes denoising, for example, using a smoothing algorithm to remove random noise in the data and reduce irregular fluctuations; outlier processing and null value filling, for missing or vacant data, interpolation or other appropriate methods are used to fill in to ensure data integrity. The data compression process includes dimensionality reduction processing, such as reducing the dimension of the data through principal component analysis (PCA), feature selection and other methods to retain the most representative data features; data compression, retaining cross-scale key features through the CSMI mechanism, wherein the CSMI mechanism is a mechanism applied in the data compression process, which aims to improve compression efficiency and reduce data loss by identifying and retaining key parts of the data, and using compression algorithms such as Huffman coding and wavelet transform to reduce storage space and transmission bandwidth while maintaining key information. The cleaned and compressed data is integrated into a structured database, namely the database to be analyzed, which contains processed data points and can be directly used for tasks such as data mining, trend analysis, and predictive modeling.

[0025] Step S200 further includes step S210, using the acquisition timestamp to integrate the electric energy acquisition parameter library in time sequence according to the acquisition parameters to establish a time sequence alignment relationship; step S220, identifying missing abnormal parameter bits according to the time sequence alignment relationship; step S230, sliding the sliding time window with the missing abnormal parameter bit as the center to obtain the neighborhood time window mean; step S240, filling the missing abnormal parameter bits with the neighborhood time window mean; step S250, standardizing the filled acquisition parameters and scaling them to the same dimension to obtain the database to be analyzed.

[0026] Preferably, the data in the electric energy acquisition parameter library are time-series integrated using the acquisition timestamp, that is, the acquisition parameters of the same type (such as voltage, current, etc.) are integrated together according to the order of the timestamps to form a time series, that is, a timing alignment relationship is established, wherein the timing alignment relationship refers to the time correspondence between different parameters. Since the acquisition frequencies of different parameters may be different, or data at certain time points are missing due to some reasons (such as equipment failure), it is necessary to establish the time correspondence between these parameters through interpolation, filling and other methods. After the timing alignment relationship is established, those missing or abnormal parameter bits are identified. These missing or abnormal parameters may be caused by equipment failure, data transmission errors and other reasons. Specifically, the method of identifying missing abnormal parameter bits may include checking whether the data value is within a reasonable range, checking whether the time intervals between data points are consistent, etc.

[0027] Preferably, the missing or abnormal parameter bit is processed by using a sliding time window method with the missing abnormal parameter bit as the center. The sliding time window is a commonly used time series analysis technique. A fixed-size window is slid on the time series to calculate the statistics of the data in the window (such as mean, variance, etc.). Specifically, for each missing or abnormal parameter bit, a fixed-size sliding time window is set as the center, and the window is translated and slid on the time series to calculate the mean (or other statistics) of other normal parameter bits in the window to obtain the neighborhood time window mean, and then the neighborhood time window mean is used to perform Filling, that is, filling these missing or abnormal parameter bits, for example, directly assigning the mean to the missing or abnormal parameter bits, or using some interpolation method (such as linear interpolation, spline interpolation, etc.) to generate a smoother filling value, and then standardizing the filled acquisition parameters. Among them, standardization is a commonly used data preprocessing technology, which eliminates the dimensional differences between different parameters by scaling the data to the same dimension (usually between 0 and 1). For example, the filled acquisition parameters are standardized through Z-score standardization, Min-Max standardization, etc., and the acquisition parameters are scaled to the same dimension to obtain the database to be analyzed.

[0028] Step S230 further includes step S231, obtaining a missing time gradient according to the missing abnormal parameter bit; step S232, setting a time window according to the missing time gradient, and the setting size of the time window is not less than twice the missing time gradient; step S233, taking the missing abnormal parameter bit as the center, moving the center forward according to the time window setting size to obtain a leading neighbor time window, and moving the center backward according to the time window setting size to obtain a trailing neighbor time window; step S234, calculating the parameter mean according to the leading neighbor time window and the trailing neighbor time window to obtain the neighborhood time window mean.

[0029] Preferably, the missing abnormal parameter bits in the electric energy acquisition parameter library are identified, and the time intervals between them, i.e., the missing time gradient, are calculated, which represents the distribution of the missing abnormal parameters in the time series. Then, the size of the time window is set according to the missing time gradient, and the set size of the time window is ensured to be not less than twice the missing time gradient, so as to ensure that the time window can cover a sufficient range before and after the missing abnormal parameter bit, so as to more accurately calculate the mean of the neighborhood time window; after setting the time window size, the center is moved forward according to the set size of the time window, with the missing abnormal parameter bit as the center. And move backward to obtain two neighborhood time windows, the leading neighborhood time window and the trailing neighborhood time window, where the leading neighborhood time window is centered on the missing abnormal parameter position and the time range covered by the forward moving time window size, and the trailing neighborhood time window is centered on the missing abnormal parameter position and the time range covered by the backward moving time window size; finally, the parameter mean is calculated based on the leading neighborhood time window and the trailing neighborhood time window, that is, the mean of the parameters in the two time windows is calculated as the neighborhood time window mean, which is used to fill the missing abnormal parameter positions, thereby making the time series more complete and continuous.

[0030] Step S250 further includes step S251, setting the time dimension, wherein the time dimension includes a short-term dimension time period, a medium-term dimension time period, and a long-term dimension time period; step S252, extracting the acquisition parameter features based on each time dimension, including the average value, peak value, minimum value, standard deviation, peak-to-valley difference, and trend characteristics; step S253, extracting the key features of the parameter features of each time dimension, wherein the short-term dimension extracts high-frequency signals, the medium-term dimension extracts high-frequency signals and data trend features, and the long-term dimension retains data trend features, and the key features of each time dimension are dimensionally interacted to obtain the database to be analyzed.

[0031] Preferably, different time dimensions are set to perform feature extraction of acquisition parameters. Specifically, the time dimension is set to include a short-term dimension time period, a medium-term dimension time period and a long-term dimension time period. The short-term dimension time period may refer to a time range of several minutes, hours or days; the medium-term dimension time period may refer to a time range of several days to several months; and the long-term dimension time period may refer to a time range of several months to several years. Feature extraction of acquisition parameters is then performed based on each time dimension, and the extracted features include average value, peak value, minimum value, standard deviation, peak-to-valley difference and trend characteristics, wherein the average value reflects the average level of the data; the peak value represents the maximum value of the data within a period of time; the minimum value represents the minimum value of the data; the standard deviation measures the degree of dispersion of the data; the peak-to-valley difference represents the maximum fluctuation of the data within a period of time; and the trend characteristics describe the trend of data changes over time. ; After extracting the parameter features of each time dimension, further extract key features. The short-term dimension time period mainly extracts high-frequency signals to reflect the short-term fluctuations or changes of the data; the medium-term dimension time period extracts high-frequency signals and data trend features, taking into account both the short-term fluctuations and the long-term trends of the data; the long-term dimension time period mainly retains the data trend features, because the long-term trend is more important for understanding and predicting the long-term behavior of the data; in this way, the most representative information is extracted from the original data, for example, key points (such as peaks and anomalies) are generated for high-frequency data, and trends and overall changes are retained for low-frequency data; finally, these key features are interactively dimensioned, that is, the information between different time dimensions or different features is integrated and associated, such as merging the key features of the short-term, medium-term and long-term dimensions, such as joint encoding, to form a database to be analyzed containing multi-dimensional information.

[0032] Step S300: searching and updating the database to be analyzed to determine the cluster center.

[0033] Preferably, in the cluster analysis process, the data set to be analyzed (i.e., the database to be analyzed) is searched and updated to find the best cluster center. Specifically, the data in the database to be analyzed is searched to find potential cluster centers (i.e., points that can represent each data cluster), such as updating the positions of these center points through multiple iterations until the clustering effect meets the requirements and the cluster center is determined, wherein the cluster center refers to the center point of each cluster in the clustering algorithm, which is usually used to represent the average or typical characteristics of the data in the cluster. In the clustering algorithm (such as K-means), the determination of the cluster center is usually achieved by calculating the average value of all samples in the cluster. In general, through continuous updating and adjustment, the core position (cluster center) of each cluster in the cluster analysis is finally determined for effective classification or grouping.

[0034] Step S300 further includes step S310, dividing all data points to be clustered into K clusters according to the database to be analyzed, minimizing the sum of square errors within the clusters, and obtaining the initial cluster centers; step S320, randomly initializing N feature data individuals, the N feature data individuals are consistent with the coordinates of the K initial cluster centers; step S330, constructing a fitness function using the clustering target: , where Z is the characteristic data individual, Z={ }, are k cluster centers with feature dimensions consistent with the data points, is the i-th data point that needs to be clustered, There are K clusters; step S340, based on the dragonfly algorithm, calculate the neighbor distribution of each feature data individual, update the position and speed of each individual, and use the fitness function to calculate the fitness of each feature data individual; step S350, through iterative update search, when the change value of the fitness calculation result is less than a preset threshold or the maximum number of iterations is reached, obtain the cluster center.

[0035] Preferably, K-means clustering is used to divide all data points to be clustered from the database to be analyzed into K clusters. The goal is to minimize the sum of squared errors within the cluster (i.e., the sum of squared distances from each point to the center of its cluster), obtain K initial cluster centers, and then randomly initialize N feature data individuals. The coordinates of these individuals are consistent with the coordinates of the K initial cluster centers, that is, each feature data individual represents a possible cluster center position, where K and N are both positive integers, and then the clustering target is used to construct the fitness function: , where Z is the characteristic data individual, Z={ }, There are k cluster centers with the same feature dimensions as the data points. K represents the index of the cluster center and i represents the index of the data point. is the i-th data point that needs to be clustered, There are K clusters, and the fitness function is based on the relationship between the feature data individuals (i.e., potential cluster centers) and the data points. The distance from each data point to its nearest cluster center is considered, and then based on the dragonfly algorithm, the neighbor distribution of each feature data individual is calculated, the position and speed of each individual are updated, and the fitness function is used to calculate the fitness of each feature data individual. The dragonfly algorithm simulates the flight process of dragonflies, mainly including the selective flight behavior and group cooperation and competition behavior of dragonflies. When flying, dragonflies will randomly select positions in space, and update positions and adjust speeds according to certain rules to search for the optimal solution; through iterative update search, the position and speed of the feature data individuals are continuously adjusted to find a better cluster center position. When the change value of the fitness calculation result is less than the preset threshold, or when the maximum number of iterations is reached, the iteration is stopped. The cluster center obtained at this time is the optimized cluster center, which can better represent the distribution of data points, thereby minimizing the sum of square errors within the cluster.

[0036] Step S400: Based on the cluster center, the clustering probability between each data point in the database to be analyzed and the cluster center is calculated.

[0037] Preferably, in cluster analysis, the possibility or probability of each data point belonging to a certain cluster is calculated based on the determined cluster center. Specifically, cluster analysis is to divide a set of data into multiple subsets (clusters). The data points in each subset are highly similar, while the data points between different clusters are relatively different. The cluster center is usually the representative point of each cluster, which can be the mean, median, or centroid of all data points in the cluster. Calculating cluster probability generally refers to calculating the probability that a data point belongs to a certain cluster. Usually, a distance metric is used to calculate the distance between the data point to be classified and the cluster center (for example, Euclidean distance, Manhattan distance, etc.). The smaller the distance, the more likely the data point belongs to the cluster. Then all clustering probabilities are normalized so that their sum is 1, indicating that the data point must belong to a certain cluster. For example, if the K-means clustering algorithm is used, the cluster center is usually the mean of the points in the cluster. For each data point to be classified, its distance from all cluster centers is calculated, and finally the data point is assigned to the cluster where the cluster center with the smallest distance is located. However, in some more complex clustering algorithms (such as Gaussian mixture models), the probability of each data point belonging to a different cluster is calculated, and these probabilities can provide a more flexible and probabilistic explanation for the attribution of data points.

[0038] Step S400 further includes step S410, for each data point, calculating the distance to the cluster center: , m is the dimension of the data, and They are The value in the jth dimension; Step S420, normalize all distance values ​​to obtain a normalization coefficient: ,in, is the maximum distance between all data points and the cluster center, is the minimum distance between all data points and the cluster center; Step S430, calculate the clustering probability according to the normalization coefficient: ,in, The value is [0,1]. The value is [0,1].

[0039] Preferably, the Euclidean distance formula is used: , calculate the distance between each data point and the cluster center, where m is the dimension of the data, and They are The value in the jth dimension, then use the formula: , all distance values ​​are normalized to obtain the normalization coefficient, and the value range is limited to [0,1], where is the maximum distance between all data points and the cluster center, is the minimum distance between all data points and the cluster center. Finally, based on the normalization coefficient, the formula is used: , calculate the clustering probability of each data point, where The value is [0,1]. The value is [0,1].

[0040] Step S500, obtaining an abnormal clustering probability threshold, using the abnormal clustering probability threshold to perform abnormal identification on each data point, and obtaining abnormal operation data.

[0041] Preferably, in the analysis of the electric energy meter data, anomaly detection can help identify the abnormal operation data of the electric energy meter. Specifically, a probability threshold (i.e., the abnormal clustering probability threshold) is set to determine whether a data point is normal or abnormal. It is usually set according to the distribution of normal data points. The data points located outside the threshold in the edge area of ​​the normal data point distribution can be identified as abnormal data points by setting this threshold. After obtaining the abnormal clustering probability threshold, each data point can be evaluated, that is, the probability of each data point belonging to the abnormal cluster (or the distance from the center of the abnormal cluster) is calculated. If the probability of a data point belonging to the abnormal cluster exceeds the set threshold, the data point is considered abnormal. value (or the distance from the center of the abnormal cluster exceeds a certain threshold), the data point is identified as an abnormal data point, and then all abnormal data points are identified, representing the abnormal operation of the electric energy meter, such as abnormal power, abnormal current, abnormal voltage, etc. After identifying the abnormal data points, the relevant information of these abnormal data points can be extracted from the data set, such as timestamp, reading of the electric energy meter, abnormal type, etc. According to the analysis results of the abnormal data, the corresponding troubleshooting and processing plan is formulated. For example, for data points with abnormal power, check whether the reading of the electric energy meter is accurate; for data points with abnormal current or voltage, check whether the circuit has problems such as faults or short circuits.

[0042] Step S500 further includes step S510, extracting the collection timestamp and constructing the electric energy acquisition parameter library according to the electric energy acquisition parameter library; step S520, performing parameter classification extraction on the electric energy acquisition parameter library based on the electric energy meter data clock, and constructing a collection parameter category visual diagram; step S530, performing parameter classification alignment extraction on the database to be analyzed based on the electric energy meter data clock, constructing a compressed parameter category visual diagram, and establishing an alignment mapping relationship between the compressed parameter category visual diagram and the collection parameter category visual diagram; step S540, generating a normal parameter graph series and an abnormal parameter graph series by using the clustering probability of the compressed parameter category visual diagram and the cluster center; step S550, constructing a hierarchical relationship visual diagram series of the collection parameter category visual diagram, the compressed parameter category visual diagram, the cluster center, the normal parameter graph series, and the abnormal parameter graph series, and sending it to the user display terminal.

[0043] Preferably, the acquisition timestamp of each data point is extracted from the electric energy acquisition parameter library, and the electric energy meter data clock is constructed using these timestamps. The electric energy meter data clock reflects the time series of data acquisition. Based on the electric energy meter data clock, the parameters in the electric energy acquisition parameter library are classified and extracted, including classification and extraction of different types of parameters such as voltage, current, power, power factor, etc., and a collection parameter category visual diagram is constructed for each parameter, such as using a bar chart, a line chart, etc. to show the change of the parameter over time; the electric energy meter data clock is then used to perform parameter classification alignment extraction on the database to be analyzed, that is, the database to be analyzed may be data from another time period or under different conditions, and the same needs Perform parameter classification and alignment extraction. Specifically, use data compression technology to reduce the amount of data while retaining key information, such as aggregation based on time windows, dimensionality reduction algorithms (such as principal component analysis PCA), obtain multiple parameter sets through classification extraction, and build a compression parameter category visual graph based on the parameter sets, where the compression parameter category visual graph is consistent with the acquisition parameter category visual graph; then establish an alignment mapping relationship between the compression parameter category visual graph and the acquisition parameter category visual graph based on parameter classification and time series, that is, for each parameter classification, find the corresponding visual graph in the compression parameter category visual graph and the acquisition parameter category visual graph, and establish a mapping between them.

[0044] Preferably, the normal parameter graph series and the abnormal parameter graph series are generated by using the clustering probability of the compression parameter category visual graph and the clustering center. Specifically, the data points are divided into different clusters (categories) by using cluster analysis, that is, the power acquisition parameters in the compression parameter category visual graph have been clustered and the clustering center has been determined. By comparing the distance or similarity between each data point and the clustering center, the clustering probability of each data point is calculated, and then the normal parameters and abnormal parameters are divided according to the clustering probability. For normal parameters, a normal parameter graph series is generated to show the changes of these parameters. For abnormal parameters, an abnormal parameter graph series is also generated. Finally, a hierarchical relationship is constructed. Visual graph series, this visual graph series includes acquisition parameter category visual graph, compression parameter category visual graph, cluster center, normal parameter graph series and abnormal parameter graph series. These visual graph series are interconnected through hierarchical relationships to form a complete analysis and visualization system, which can help users better understand the operation of the electricity meter, discover abnormal parameters in time, and conduct troubleshooting and decision-making. This hierarchical relationship visual graph series is sent to the user display terminal (such as computer screen, mobile device, etc.) for user viewing and analysis. Users can browse different visual graph series through an interactive interface to obtain the required information, better understand the operation of the electricity meter, and discover potential problems in time.

[0045] In the above, refer to Figure 1The operation monitoring method for an electric energy meter according to an embodiment of the present invention is described in detail. Figure 2 An operation monitoring system for an electric energy meter according to an embodiment of the present invention is described.

[0046] The operation monitoring system for electric energy meters according to the embodiment of the present invention is used to solve the technical problem that the existing electric energy meter operation monitoring is difficult to accurately identify the subtle changes and abnormal states in the operation process of the electric energy meter in the face of massive and complex electric energy data, which leads to poor timeliness of abnormal identification of the electric energy meter and low monitoring efficiency, thereby achieving the technical effect of improving the timeliness of abnormal identification and monitoring efficiency. The operation monitoring system for electric energy meters includes: an electric energy acquisition parameter library establishment module 10, a database acquisition module 20 for analysis, a cluster center determination module 30, a cluster probability calculation module 40, and an abnormal operation data acquisition module 50.

[0047] An electric energy acquisition parameter library establishment module 10 is used to establish an electric energy acquisition parameter library, and the electric energy acquisition parameter library includes current, voltage, power, and power factor; a database acquisition module 20 to be analyzed is used to clean and compress the acquisition parameters according to the electric energy acquisition parameter library to obtain the database to be analyzed; a cluster center determination module 30 is used to search and update according to the database to be analyzed to determine the cluster center; a cluster probability calculation module 40 is used to calculate the clustering probability of each data point in the database to be analyzed and the cluster center based on the cluster center; an abnormal operation data acquisition module 50 is used to obtain an abnormal cluster probability threshold, and use the abnormal cluster probability threshold to identify abnormalities for each data point to obtain abnormal operation data.

[0048] The specific configuration of the electric energy acquisition parameter library establishment module 10 will be described in detail below. The electric energy acquisition parameter library establishment module 10 further includes: extracting the acquisition timestamp according to the electric energy acquisition parameter library and constructing the electric energy meter data clock; performing parameter classification extraction on the electric energy acquisition parameter library based on the electric energy meter data clock and constructing a collection parameter category visual graph; performing parameter classification alignment extraction on the database to be analyzed based on the electric energy meter data clock, constructing a compressed parameter category visual graph, and establishing an alignment mapping relationship between the compressed parameter category visual graph and the collection parameter category visual graph; using the clustering probability of the compressed parameter category visual graph and the cluster center to generate a normal parameter graph series and an abnormal parameter graph series; constructing a hierarchical relationship visual graph series of the collection parameter category visual graph, the compressed parameter category visual graph, the cluster center, and the normal parameter graph series and the abnormal parameter graph series, and sending it to the user display terminal.

[0049] The specific configuration of the module 20 for obtaining the database to be analyzed will be described in detail below. The module 20 for obtaining the database to be analyzed may further include: integrating the electric energy acquisition parameter library in time sequence according to the acquisition parameters with the acquisition timestamp to establish a time sequence alignment relationship; identifying the missing abnormal parameter bit according to the time sequence alignment relationship; sliding the sliding time window with the missing abnormal parameter bit as the center to obtain the neighborhood time window mean; filling the missing abnormal parameter bit with the neighborhood time window mean; standardizing the filled acquisition parameters and scaling them to the same dimension to obtain the database to be analyzed.

[0050] The specific configuration of the module 20 for obtaining the database to be analyzed will be described in detail below. The module 20 for obtaining the database to be analyzed further includes: obtaining the missing time gradient according to the missing abnormal parameter bit; setting the time window according to the missing time gradient, and the setting size of the time window is not less than twice the missing time gradient; taking the missing abnormal parameter bit as the center, moving the center forward according to the time window setting size, obtaining the leading neighborhood time window, and moving the center backward according to the time window setting size, obtaining the trailing neighborhood time window; performing parameter mean calculation according to the leading neighborhood time window and the trailing neighborhood time window, and obtaining the neighborhood time window mean.

[0051] The specific configuration of the module 20 for obtaining the database to be analyzed will be described in detail below. The module 20 for obtaining the database to be analyzed further includes: setting a time dimension, wherein the time dimension includes a short-term dimension time period, a medium-term dimension time period, and a long-term dimension time period; extracting acquisition parameter features based on each time dimension, including average value, peak value, minimum value, standard deviation, peak-to-valley difference, and trend features; extracting key features from the parameter features of each time dimension, wherein the short-term dimension extracts high-frequency signals, the medium-term dimension extracts high-frequency signals and data trend features, and the long-term dimension retains data trend features, and the key features of each time dimension are interactively dimensioned to obtain the database to be analyzed.

[0052] The specific configuration of the cluster center determination module 30 will be described in detail below. The cluster center determination module 30 may further include: dividing all data points that need to be clustered into K clusters according to the database to be analyzed, minimizing the sum of square errors within the clusters, and obtaining the initial cluster centers; randomly initializing N feature data individuals, and the N feature data individuals are consistent with the coordinates of the K initial cluster centers; using the clustering target to construct a fitness function: , where Z is the characteristic data individual, Z={ }, are k cluster centers with feature dimensions consistent with the data points, is the i-th data point that needs to be clustered, There are K clusters; based on the dragonfly algorithm, the neighbor distribution of each feature data individual is calculated, the position and speed of each individual are updated, and the fitness of each feature data individual is calculated using the fitness function; through iterative update search, when the change value of the fitness calculation result is less than a preset threshold or the maximum number of iterations is reached, the cluster center is obtained.

[0053] The specific configuration of the clustering probability calculation module 40 will be described in detail below. The clustering probability calculation module 40 may further include: for each data point, calculating the distance to the cluster center: , m is the dimension of the data, and They are The value in the jth dimension; normalize all distance values ​​to obtain the normalization coefficient: ,in, is the maximum distance between all data points and the cluster center, is the minimum distance between all data points and the cluster center; according to the normalization coefficient, the clustering probability is calculated: ,in, The value is [0,1]. The value is [0,1].

[0054] The operation monitoring system for an electric energy meter provided in the embodiment of the present invention can execute the operation monitoring method for an electric energy meter provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0055] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0056] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for monitoring the operation of an electric energy meter, characterized in that: The method comprises: Establishing an electric energy acquisition parameter library, wherein the electric energy acquisition parameter library includes current, voltage, power, and power factor; According to the electric energy acquisition parameter library, data cleaning and compression are performed on the acquisition parameters to obtain a database to be analyzed; Search and update the database to be analyzed to determine the cluster center; Based on the cluster center, calculating the clustering probability of each data point in the database to be analyzed and the cluster center; An abnormal clustering probability threshold is obtained, and each data point is identified as abnormal by using the abnormal clustering probability threshold to obtain abnormal operation data.

2. The operation monitoring method for an electric energy meter according to claim 1, characterized in that: The method further comprises: According to the electric energy acquisition parameter library, extract the acquisition timestamp and construct the electric energy meter data clock; Based on the electric energy meter data clock, the electric energy acquisition parameter library is subjected to parameter classification extraction, and a collection parameter category visual diagram is constructed; Based on the electric energy meter data clock, parameter classification alignment extraction is performed on the database to be analyzed, a compressed parameter category visual graph is constructed, and an alignment mapping relationship between the compressed parameter category visual graph and the acquisition parameter category visual graph is established; Generate a normal parameter graph and an abnormal parameter graph by using the compression parameter category visual graph and the clustering probability of the cluster center; Construct the acquisition parameter category visual graph, compression parameter category visual graph, cluster center, and hierarchical relationship visual graph series of normal parameter graph series and abnormal parameter graph series, and send them to the user display terminal.

3. The operation monitoring method for an electric energy meter according to claim 1, characterized in that: According to the electric energy acquisition parameter library, the acquisition parameters are cleaned and compressed to obtain a database to be analyzed, including: The electric energy acquisition parameter library is time-sequentially integrated according to the acquisition parameters using the acquisition timestamp to establish a time alignment relationship; According to the timing alignment relationship, identifying missing abnormal parameter bits; Taking the missing abnormal parameter position as the center, the sliding time window is shifted and slid to obtain the mean of the neighborhood time window; Filling the missing abnormal parameter bits with the neighborhood time window mean; The filled acquisition parameters are standardized and scaled to the same dimension to obtain the database to be analyzed.

4. The operation monitoring method for an electric energy meter according to claim 3, characterized in that: Taking the missing abnormal parameter position as the center, the sliding time window is shifted and slid to obtain the neighborhood time window mean, including: According to the missing abnormal parameter bit, obtaining the missing time gradient; Setting a time window according to the missing time gradient, wherein the setting size of the time window is not less than twice the missing time gradient; Taking the missing abnormal parameter bit as the center, the center is moved forward according to the time window setting size to obtain the leading neighboring time window, and the center is moved backward according to the time window setting size to obtain the trailing neighboring time window; Parameter mean calculation is performed according to the preceding neighborhood time window and the succeeding neighborhood time window to obtain the neighborhood time window mean.

5. The operation monitoring method for an electric energy meter according to claim 3, characterized in that: The filled acquisition parameters are standardized and scaled to the same dimension to obtain the database to be analyzed, including: Setting a time dimension, wherein the time dimension includes a short-term dimension time period, a medium-term dimension time period, and a long-term dimension time period; Based on each time dimension, the acquisition parameter features are extracted separately, including the average value, peak value, minimum value, standard deviation, peak-to-valley difference, and trend characteristics; The key features of the parameter features of each time dimension are extracted, wherein the short-term dimension extracts high-frequency signals, the medium-term dimension extracts high-frequency signals and data trend features, and the long-term dimension retains data trend features. The key features of each time dimension are interactively dimensioned to obtain the database to be analyzed.

6. The operation monitoring method for an electric energy meter according to claim 1, characterized in that: Searching and updating the database to be analyzed to determine the cluster center includes: According to the database to be analyzed, all data points that need to be clustered are divided into K clusters, the sum of square errors within the cluster is minimized, and the initial cluster center is obtained; Randomly initialize N feature data individuals, where the N feature data individuals are consistent with the coordinates of the K initial cluster centers; The fitness function is constructed using clustering objectives: , where Z is the characteristic data individual, Z={ }, are k cluster centers with feature dimensions consistent with the data points, is the i-th data point that needs to be clustered, There are K clusters; Based on the dragonfly algorithm, the neighbor distribution of each feature data individual is calculated, the position and speed of each individual are updated, and the fitness of each feature data individual is calculated using the fitness function; Through iterative update search, when the change value of the fitness calculation result is less than a preset threshold or reaches a maximum number of iterations, the cluster center is obtained.

7. The operation monitoring method for an electric energy meter according to claim 6, characterized in that: Based on the cluster center, calculating the clustering probability of each data point in the database to be analyzed and the cluster center, including: For each data point, calculate the distance to the cluster center: , m is the dimension of the data, and They are The value in the jth dimension; Normalize all distance values ​​to obtain the normalization coefficient: ,in, is the maximum distance between all data points and the cluster center, is the minimum distance between all data points and the cluster center; According to the normalization coefficient, the clustering probability is calculated: ,in, The value is [0,1]. The value is [0,1].

8. An operation monitoring system for an electric energy meter, characterized in that: The system is used to implement the operation monitoring method for an electric energy meter according to any one of claims 1 to 7, and the system comprises: An electric energy acquisition parameter library establishment module is used to establish an electric energy acquisition parameter library, wherein the electric energy acquisition parameter library includes current, voltage, power, and power factor; A module for obtaining a database to be analyzed, used for cleaning and compressing the collected parameters according to the electric energy collection parameter library to obtain a database to be analyzed; A cluster center determination module, used to search and update the database to be analyzed to determine the cluster center; A clustering probability calculation module, used to calculate the clustering probability of each data point in the database to be analyzed and the cluster center based on the cluster center; The abnormal operation data acquisition module is used to obtain an abnormal clustering probability threshold, and use the abnormal clustering probability threshold to perform abnormal identification on each data point to obtain abnormal operation data.

Images