A method and system for monitoring the operational error of an electricity meter

Abstract

This invention provides a method and system for monitoring the operational error of electricity meters, relating to the field of power monitoring technology. The method includes: acquiring target electricity consumption information through an electricity monitoring terminal; the electricity monitoring master station sequentially processing first and second electricity consumption data using a preprocessing strategy; inputting standard electricity consumption data into a backpropagation neural network to obtain a predicted error value; activating a metering chip to dynamically monitor the operation of individual electricity meters and obtain a self-monitoring error value; and issuing an operational error alarm for the individual electricity meter when the combined error value obtained by fusing the predicted error value and the self-monitoring error value is within a predetermined alarm threshold. This solves the technical problem in existing technologies where the single error monitoring method for electricity meters leads to the inability to accurately determine the operational error of electricity meters in real time. It achieves the technical effect of efficiently and accurately monitoring the operational error of electricity meters through a dual monitoring mechanism.

Description

Technical Field

[0001] This invention relates to the field of power monitoring technology, specifically to a method and system for monitoring the operational error of an electricity meter. Background Technology

[0002] With the increasing demands for smart grids and precise metering, the accuracy and reliability of electricity meters in power systems have become paramount. However, current technologies primarily rely on two methods for monitoring electricity meter operating errors: internal metering chip self-testing, which measures current and voltage using a standard signal source or sampling circuit integrated within the meter to calculate errors; and traditional periodic sampling or manual meter reading, which involves external data acquisition devices or maintenance personnel periodically recording and comparing electricity data. While internal metering chip self-testing provides some real-time capability, it typically only provides a localized assessment of instantaneous signals, making it difficult to reflect the long-term operating trends of the meter under different load conditions. Furthermore, it is susceptible to environmental noise and occasional fluctuations, leading to false alarms or missed alarms. Traditional sampling methods suffer from long sampling intervals and high data delays, making it difficult to capture short-term fluctuations or discontinuous anomalies, thus failing to detect changes in meter errors in a timely manner. These existing monitoring methods are simplistic and have significant limitations, making it difficult to accurately assess electricity meter operating errors, thereby reducing metering accuracy and the reliability of electricity billing.

[0003] It is evident that existing technologies suffer from a lack of simplistic error monitoring methods for electricity meters, resulting in the inability to accurately determine the operational errors of electricity meters in real time. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for monitoring the operating error of an electricity meter, which solves the technical problem that the existing technology has a single method for monitoring the error of electricity meters, resulting in the inability to accurately determine the operating error of electricity meters in real time.

[0005] In view of the above problems, this application provides a method and system for monitoring the operating error of an electricity meter.

[0006] The first aspect of this application provides a method for monitoring the operational error of an electricity meter. The method includes: acquiring target electricity consumption information through an electricity monitoring terminal, wherein the target electricity consumption information includes first electricity consumption data corresponding to a main electricity meter and second electricity consumption data corresponding to a sub-electricity meter; the electricity monitoring master station calls a preprocessing strategy to process the first electricity consumption data and the second electricity consumption data sequentially, and assembles standard electricity consumption data; inputting the standard electricity consumption data into a backpropagation neural network to obtain a predicted error value of the backpropagation neural network; activating a metering chip to perform dynamic operational monitoring of the sub-electricity meter to obtain a self-monitoring error value; and issuing an operational error alarm for the sub-electricity meter when the comprehensive error value obtained by fusing the predicted error value and the self-monitoring error value is at a predetermined alarm threshold.

[0007] Optionally, a BeiDou communication channel is established between the electricity monitoring terminal and the electricity monitoring master station, wherein the BeiDou communication channel realizes electricity data transmission using a message communication architecture.

[0008] Optionally, according to the data cleaning strategy, noise data in the first electricity consumption data and the second electricity consumption data are eliminated sequentially to obtain first processed data and second processed data, respectively; according to the scaling transformation strategy, the first processed data and the second processed data are subjected to maximum and minimum value normalization processing sequentially to obtain third processed data and fourth processed data, respectively; according to the data reduction strategy, the third processed data and the fourth processed data are subjected to dimensionality reduction processing sequentially to obtain fifth processed data and sixth processed data, respectively; based on the fifth processed data and the sixth processed data, the standard electricity consumption data is formed.

[0009] Optionally, historical electricity meter operation data is obtained as training samples; the training samples are input into a neural network for signal forward propagation and error backpropagation training to obtain the predicted operation error, and the preset connection weights and preset node thresholds of each layer of the neural network are updated according to the predicted operation error to obtain the backpropagation neural network; wherein, when the number of iterations reaches a preset maximum value or the predicted operation error is less than a preset accuracy threshold, the network training is completed and the backpropagation neural network is obtained.

[0010] Optionally, the initial connection weights and initial node thresholds are encoded into the position vectors of the optimization individual; a fitness factor is introduced to analyze the prediction error and obtain the fitness of the optimization individual; the fitness is used as an evaluation index to iteratively optimize the position vectors of the optimization individual and determine the optimal position vector corresponding to the highest fitness; the optimal position vector is decoded to obtain a decoding result, wherein the decoding result includes the preset connection weights and the preset node thresholds.

[0011] Optionally, the standard signal source is injected into the sampling circuit of the sub-energy meter at a predetermined frequency to obtain a test signal; the load signal of the sub-energy meter is acquired, and the load signal is superimposed and converted with the test signal to obtain a target test signal; the target change amount of the target test signal is obtained through the metering chip, and the self-monitoring error value corresponding to the target change amount is matched.

[0012] Optionally, the standard signal source is injected into the voltage sampling circuit of the sub-energy meter at a predetermined frequency to obtain a voltage test signal; the standard signal source is injected into the current sampling circuit of the sub-energy meter at a predetermined frequency to obtain a current test signal; the voltage test signal and the current test signal constitute the test signal.

[0013] Optionally, based on the comprehensive error value and its corresponding monitoring time, an operational error time series is constructed; a time window is introduced to analyze the operational error time series to obtain an alarm coverage rate; it is determined whether the alarm coverage rate reaches a predetermined coverage threshold. If it does, a continuous operational error alarm is issued to the sub-meter; wherein, the analysis of the operational error time series using a time window to obtain the alarm coverage rate includes: extracting a unit time series of the operational error time series according to the time window; randomly sampling the unit time series to obtain error value sampling points; when the error value sampling point is at the predetermined alarm threshold, the error value sampling point is recorded as an alarm sampling point; the ratio of the alarm sampling point to the error value sampling point is used as the alarm coverage rate.

[0014] Optionally, if the alarm coverage rate does not reach the predetermined coverage threshold, then obtain the scatter plot of the operating error corresponding to the operating error time sequence; fit the scatter plot of the operating error to obtain the operating error curve; perform variability weighting processing on the multidimensional feature parameters of the operating error curve to obtain the operating error feature coefficient; determine whether the operating error feature coefficient is within the predetermined threshold range, and if not, issue an electricity theft alarm to the sub-electricity meter.

[0015] A second aspect of this application provides an operational error monitoring system for electricity meters. The system includes: an electricity consumption information acquisition module, used to acquire target electricity consumption information through an electricity consumption monitoring terminal, wherein the target electricity consumption information includes first electricity consumption data corresponding to a main electricity meter and second electricity consumption data corresponding to individual electricity meters; an electricity consumption data processing module, used by an electricity consumption monitoring master station to call a preprocessing strategy to process the first electricity consumption data and the second electricity consumption data sequentially, and to form standard electricity consumption data; an error prediction module, used to input the standard electricity consumption data into a backpropagation neural network to obtain a predicted error value from the backpropagation neural network; an operational monitoring module, used to activate a metering chip to dynamically monitor the operation of the individual electricity meters and obtain a self-monitoring error value; and an error alarm module, used to issue an operational error alarm to the individual electricity meters when the combined error value obtained by fusing the predicted error value and the self-monitoring error value is at a predetermined alarm threshold.

[0016] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0017] The method provided in this application embodiment obtains target electricity consumption information through an electricity consumption monitoring terminal. The target electricity consumption information includes first electricity consumption data corresponding to the main electricity meter and second electricity consumption data corresponding to the sub-electricity meters. The electricity monitoring master station invokes a preprocessing strategy to process the first electricity consumption data and the second electricity consumption data sequentially, forming standard electricity consumption data. The standard electricity consumption data is input into a backpropagation neural network to obtain the estimated error value of the backpropagation neural network. The metering chip is activated to dynamically monitor the operation of the sub-electricity meters, obtaining a self-monitoring error value. When the comprehensive error value obtained by fusing the estimated error value and the self-monitoring error value is within a predetermined alarm threshold, an operation error alarm is issued to the sub-electricity meter. This achieves the technical effect of efficiently and accurately monitoring the operation error of electricity meters through a dual monitoring mechanism.

[0018] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 A flowchart illustrating a method for monitoring the operational error of an electricity meter provided in this application.

[0021] Figure 2 This is a schematic diagram of the structure of an operating error monitoring system for an electricity meter provided in this application.

[0022] Explanation of reference numerals in the attached diagram: 11 Electricity information acquisition module, 12 Electricity data processing module, 13 Error prediction module, 14 Operation monitoring module, 15 Error alarm module. Detailed Implementation

[0023] This application provides a method and system for monitoring the operational error of electricity meters, addressing the technical problem that existing technologies rely on a single error monitoring method, resulting in the inability to accurately determine the operational error of electricity meters in real time. It achieves the technical effect of efficiently and accurately monitoring the operational error of electricity meters through a dual monitoring mechanism.

[0024] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0025] Example 1, as Figure 1 As shown, this application provides a method for monitoring the operating error of an electricity meter, the method comprising: The target electricity consumption information is obtained through an electricity consumption monitoring terminal, wherein the target electricity consumption information includes the first electricity consumption data corresponding to the main electricity meter and the second electricity consumption data corresponding to the sub-electricity meters.

[0026] Specifically, target electricity consumption information is collected through smart meters in the electricity consumption monitoring terminal. The target electricity consumption information includes the first electricity consumption data corresponding to the main electricity meter and the second electricity consumption data corresponding to the sub-electricity meters. The main electricity meter and the sub-electricity meters are devices used to measure and record electricity consumption. The main electricity meter is usually installed at the main incoming line of the entire user-side power distribution system to measure the total electricity consumption of the entire area or building. The sub-electricity meters are installed at each branch circuit or electrical equipment of the user-side power system to measure the electricity consumption of a specific branch or equipment.

[0027] During data acquisition, the electricity monitoring terminal first sends data reading commands to the main electricity meter and individual electricity meters. These commands are based on specific communication protocols, such as the common Modbus protocol. After receiving the commands, the main and individual electricity meters, according to their own measurement data and storage mechanisms, transmit the corresponding first and second electricity consumption data back to the electricity monitoring terminal via communication lines. The first electricity consumption data includes various parameters of the total electricity consumption measured by the main electricity meter, such as active power, reactive power, voltage, current, and power factor. These parameters comprehensively reflect the total electricity consumption of the entire area or building. Active power is the electrical energy consumed in actual work. Reactive power is related to the reactive power of the power system and affects its stability and efficiency. Voltage and current are important indicators for measuring power quality; excessively high or low voltage, or excessively high or low current, can damage electrical equipment. The power factor reflects the efficiency of electrical equipment in utilizing electrical energy; the lower the power factor, the greater the proportion of reactive power and the more serious the energy waste. The second electricity consumption data also includes various parameters of electricity consumption for a specific branch or device measured by the sub-meter. The parameter types are similar to those of the first electricity consumption data, but they are for specific branch circuits or electrical devices.

[0028] By collecting data from individual electricity meters, we can understand the electricity consumption of each branch or device, facilitating electricity management and troubleshooting. Obtaining target electricity consumption information through electricity monitoring terminals provides comprehensive and reliable data support for monitoring the operational errors of electricity meters.

[0029] Furthermore, a BeiDou communication channel is established between the electricity monitoring terminal and the electricity monitoring master station, wherein the BeiDou communication channel realizes electricity data transmission using a message communication architecture.

[0030] Specifically, the BeiDou communication channel refers to the communication channel established between the electricity monitoring terminal and the electricity monitoring master station through the BeiDou satellite navigation system. The communication principle of the BeiDou satellite system is based on two-way communication. The electricity monitoring terminal is equipped with a BeiDou communication module, which establishes a communication channel with the electricity monitoring master station through a satellite link to realize the transmission and reception of signals.

[0031] During data transmission, the communication channel first establishes a connection with the satellite communication system through a ground station, and then transmits the data to the electricity monitoring master station via satellite. Furthermore, the BeiDou communication channel, based on a message communication architecture, ensures stable and reliable transmission of electricity data. Message communication architecture is a standard communication protocol, typically employing TCP / IP or UDP protocols, exchanging information through data packets. It offers advantages such as structure, flexibility, and reliability. Each data packet includes control information and a data portion. The control information is used for data packet transmission and integrity verification, while the data portion contains the actual transmitted electricity consumption data. The electricity monitoring terminal encapsulates the collected electricity information into a message and sends it to the monitoring master station via the BeiDou communication channel. The message contains information including the electricity meter number, collection timestamp, electricity consumption data, and a data checksum. The data checksum ensures that the data is not tampered with or lost during transmission.

[0032] By establishing a BeiDou communication channel and using a message communication architecture to achieve data transmission between the electricity monitoring terminal and the electricity monitoring master station, it is possible to ensure stable and timely transmission of electricity meter monitoring data in remote areas or areas with poor network conditions, thereby improving the effectiveness and reliability of electricity meter operation error monitoring.

[0033] The electricity monitoring master station invokes a preprocessing strategy to process the first electricity consumption data and the second electricity consumption data in sequence, and then combines them into standard electricity consumption data.

[0034] Furthermore, the preprocessing strategy includes a data cleaning strategy, a scaling transformation strategy, and a data reduction strategy. The electricity monitoring master station invokes the preprocessing strategy to process the first electricity consumption data and the second electricity consumption data sequentially, and assembles standard electricity consumption data. This includes: according to the data cleaning strategy, sequentially eliminating noise data in the first electricity consumption data and the second electricity consumption data to obtain first processed data and second processed data, respectively; according to the scaling transformation strategy, sequentially performing maximum and minimum value normalization processing on the first processed data and the second processed data to obtain third processed data and fourth processed data, respectively; according to the data reduction strategy, sequentially performing dimensionality reduction processing on the third processed data and the fourth processed data to obtain fifth processed data and sixth processed data, respectively; and assembling the standard electricity consumption data based on the fifth processed data and the sixth processed data.

[0035] Specifically, the electricity monitoring master station receives the first and second electricity consumption data collected by the electricity monitoring terminal through the BeiDou communication channel, and performs preprocessing through a preprocessing strategy. The preprocessing strategy includes a data cleaning strategy, a scaling transformation strategy, and a data reduction strategy. Among them, the data cleaning strategy is used to remove possible errors, outliers, or noise data in the first and second electricity consumption data. An outlier detection algorithm, such as the standard deviation method, is used to detect outliers in the first and second electricity consumption data. Electricity consumption data that deviates from the mean by more than 3 times the standard deviation is considered an outlier and removed. Filtering algorithms, such as median filtering or mean filtering, are used to clean the first and second electricity consumption data respectively, resulting in the first processed data and the second processed data.

[0036] Then, the first and second processed data are subjected to maximum-minimum value normalization according to the scaling transformation strategy. Maximum-minimum value normalization is a data scaling transformation method used to map data with different dimensions and value ranges to the [0,1] interval, eliminating the difference in data dimensions. Through the maximum-minimum value normalization formula, each data in the first and second processed data is normalized to obtain the third and fourth processed data.

[0037] Based on the data reduction strategy, principal component analysis (PCA) is used to sequentially reduce the dimensionality of the third and fourth processed data. PCA projects the third and fourth processed data into a new coordinate system through linear transformation, selecting several principal components with larger variances to represent the original data, thereby reducing the data dimensionality and achieving data reduction. After dimensionality reduction of the third and fourth processed data respectively, fifth and sixth processed data are obtained. Combining the fifth and sixth processed data yields the standard electricity consumption data.

[0038] Through data cleaning, scaling transformation, and data reduction strategies, the electricity monitoring master station can effectively process and optimize the electricity data collected from the electricity monitoring terminals, improving the accuracy and reliability of the data, and thus enhancing the precision and response speed of the entire power monitoring system.

[0039] The standard electricity consumption data is input into the backpropagation neural network to obtain the prediction error value of the backpropagation neural network.

[0040] Furthermore, the standard electricity consumption data is input into the backpropagation neural network to obtain the estimated error value of the backpropagation neural network. This process includes: acquiring historical electricity meter operation data as training samples; inputting the training samples into the neural network for signal forward propagation and error backpropagation training to obtain the predicted operating error; and updating the preset connection weights and preset node thresholds of each layer of the neural network based on the predicted operating error to obtain the backpropagation neural network. The network training is completed when the number of iterations reaches a preset maximum value or the predicted operating error is less than a preset accuracy threshold, thus obtaining the backpropagation neural network.

[0041] Specifically, historical electricity meter operation data is extracted from the electricity meter database. The historical electricity meter operation data consists of standardized total electricity meter and sub-electricity meter data, as well as the actual error at each time point. The actual error value can be determined through manual verification or on-site measurement. The historical electricity meter operation data is used as training samples. Each sample contains an input vector and a corresponding output label. The input vector is the standardized total electricity meter and sub-electricity meter data, and the corresponding output label is the actual error value.

[0042] Training samples are input into a neural network for training. The neural network consists of an input layer, hidden layers, and an output layer. The input layer receives training samples, with the number of input nodes matching the dimension of the electricity consumption data. For example, it might have one main electricity meter data point and three sub-electricity meter data points, totaling four input nodes. The hidden layer performs a nonlinear transformation, requiring at least one layer with 1.5 to 2 times the number of input nodes (e.g., six nodes). Each node undergoes a nonlinear transformation using the ReLU activation function, and all nodes are fully connected, meaning each input node has connection weights to every node in the hidden layer. The output layer outputs the predicted error value, representing the electricity meter's operating error.

[0043] For neural networks, small random numbers, such as a uniform distribution [-0.5, 0.5], are used to initialize the connection weights to ensure that the network output does not saturate at the start of training. The initial node threshold is set to 0 or a small random number to adjust the activation starting point of each node. The neural network training process first performs forward propagation: training samples are input into the input layer, passed through the input layer to the hidden layer, transformed by an activation function such as ReLU, and then output to the output layer. The output of the output layer is the predicted operating error of the electricity meter by the neural network.

[0044] Then, backpropagation of the error is performed: based on the difference between the predicted error value and the actual error value, the mean squared error (MSE) of the loss function is calculated. The preset connection weights and preset node thresholds in the neural network are adjusted and updated through the backpropagation algorithm, so that the prediction result gradually approaches the actual error. The updated connection weights and node thresholds can better capture the relationship between standard electricity consumption data and operational errors. During training, the number of iterations and the prediction accuracy threshold are set as training termination conditions. When the number of iterations reaches a preset maximum value, such as 1000 iterations, or when the predicted operational error is less than the preset accuracy threshold, such as 0.01, training stops. The neural network obtained at this point is the trained backpropagation neural network, which can accurately predict the operational error of the electricity meter.

[0045] Standard electricity consumption data is input into a backpropagation neural network. The backpropagation neural network processes the input data in a forward propagation manner and outputs a predicted error value. The predicted error value is used to reflect the possible error of the electricity meter under the current operating state.

[0046] By training a neural network using historical data, the backpropagation neural network can learn the intrinsic relationship between electricity meter operating data and operating errors, thus enabling it to predict errors in new electricity consumption data. Inputting standard electricity consumption data into the trained backpropagation neural network yields the predicted error value, allowing for real-time and accurate assessment of electricity meter operating errors and facilitating the timely detection of abnormal meter operation.

[0047] Furthermore, the training samples are input into the neural network for signal forward propagation and error backpropagation training to obtain the predicted running error. The preset connection weights and preset node thresholds of each layer of the neural network are then updated based on the predicted running error to obtain the backpropagation neural network. This process includes: encoding the initial connection weights and initial node thresholds into the position vector of the optimization individual; introducing a fitness factor to analyze the predicted running error and obtain the fitness of the optimization individual; using the fitness as an evaluation index, iteratively optimizing the position vector of the optimization individual to determine the optimal position vector corresponding to the highest fitness; and decoding the optimal position vector to obtain a decoding result, wherein the decoding result includes the preset connection weights and the preset node thresholds.

[0048] Specifically, before training the neural network, the connection weights and node thresholds of each layer are initially set. These are initialized using a small range of random numbers, such as initial connection weights randomly selected within a uniform distribution range of [-0.5, 0.5], and initial node thresholds randomly selected within a range of [0, 1]. All initial connection weights and initial node thresholds are combined to form a vector, called the position vector, which serves as the optimization target for the optimization algorithm. Each element corresponds to a connection weight or node threshold in the network. For example, if a neural network has 4 input nodes, 6 hidden nodes, and 1 output node, then there are 4 × 6 connection weights from the input layer to the hidden layer, and 6 × 1 connection weights from the hidden layer to the output layer, for a total of 4 × 6 + 6 × 1 = 30 connection weights. The node thresholds for the hidden layer and the output layer are 6 + 1 = 7. These parameters are combined to form a position vector of length 37.

[0049] A fitness factor is introduced to analyze the prediction runtime error. The fitness factor is an indicator that measures the quality of an individual in the search space, reflecting the network's prediction performance on training samples under the current combination of connection weights and node thresholds. The prediction runtime error is the difference between the neural network's output value and the actual value. The mean squared error (MSE) is used as the preset runtime error, and the network's performance can be evaluated by calculating the prediction runtime error. The fitness factor is inversely proportional to the prediction runtime error; that is, the smaller the prediction runtime error, the larger the fitness factor, indicating better performance of the individual. The fitness factor is calculated using the formula F = 1 / (1+MSE) to obtain the fitness of the individual.

[0050] Then, fitness is used as an evaluation metric to iteratively optimize the position vector of the candidate particle. During the iteration process, optimization algorithms, such as particle swarm optimization (PSO) and genetic algorithms, are used to update and adjust the position vector of the candidate particle. Taking PSO as an example, the position vector of each particle is initialized with random initial values ​​of the neural network parameters, including initial connection weights and initial node thresholds. An initial velocity vector is also set for each particle. The dimension of the particle position vector is the same as the total number of parameters in the neural network. For example, a network with 4 input nodes, 6 hidden nodes, and 1 output node has a total of 37 parameters, so the length of each particle vector is 37. For each particle, its current position is decoded into neural network weights and node thresholds, a corresponding neural network is constructed, training samples are input into the neural network for forward propagation, and the prediction error value is output. In each iteration, fitness is calculated based on the predicted error and the actual error. Each particle updates its velocity and position based on its historical best position and the global best position of the swarm. For each particle, if the fitness of its current position is higher than its historical best fitness, then the historical best position of that particle is updated. At the same time, the particle with the highest fitness in the swarm updates its global best position. The iteration terminates when the number of iterations reaches a preset maximum value, such as 100, or when the global best fitness reaches the accuracy requirement. After multiple iterations, the position vector of the individual with the highest fitness is selected as the optimal position vector. The optimal position vector represents the optimal combination of connection weights and node thresholds in the optimized neural network, enabling the neural network to achieve better performance and smaller prediction errors during training.

[0051] The optimal position vector is decoded to restore its values ​​to the preset connection weights and preset node thresholds of the neural network. The decoding process is the reverse of the encoding process. Following the rules set during encoding, each element of the optimal position vector is assigned to the corresponding connection weight and node threshold positions in the neural network, resulting in optimized neural network parameters. For example, following the encoding order, the vector elements are mapped back to the preset connection weights and preset node thresholds. The first 24 elements are assigned to the weights from the input layer to the hidden layer, the next 6 elements to the weights from the hidden layer to the output layer, and the last 7 elements to the node thresholds of the hidden layer and the output layer. The parsed connection weights and node thresholds are then written into the corresponding nodes of the neural network to obtain the optimized neural network parameters.

[0052] By introducing a fitness optimization algorithm to optimize the connection weights and node thresholds of the neural network, it is possible to avoid the neural network getting stuck in local optima during training, improve the training efficiency and performance of the neural network, and enable the neural network to learn more accurately the built-in relationship between the operating data and operating error of the electricity meter, thereby improving the accuracy and stability of electricity meter error prediction.

[0053] The metering chip is activated to dynamically monitor the operation of the sub-meters and obtain the self-monitoring error value.

[0054] Furthermore, the metering chip has a built-in standard signal source. Activating the metering chip allows for dynamic monitoring of the sub-energy meter to obtain a self-monitoring error value. This includes: injecting the standard signal source into the sampling circuit of the sub-energy meter at a predetermined frequency to obtain a test signal; acquiring the load signal of the sub-energy meter and superimposing and converting the load signal with the test signal to obtain a target test signal; obtaining the target change amount of the target test signal through the metering chip and matching the self-monitoring error value corresponding to the target change amount.

[0055] Specifically, the metering chip has a built-in standard signal source. This standard signal source, an integrated reference signal generator within the chip, outputs a signal with known characteristics, high accuracy, and stability, serving as a standard reference for measuring the energy meter's error. The standard signal source can be a sine wave, pulse, or modulated signal, and its amplitude, frequency, and phase can be precisely controlled. After activating the metering chip, the standard signal source is injected into the voltage and current sampling circuits of the sub-energy meter at a predetermined frequency. The predetermined frequency is determined based on the sub-energy meter's operating characteristics and monitoring requirements. For example, if the signal frequency range of the sub-energy meter during normal operation is between 50Hz and 60Hz, a predetermined frequency of 55Hz can be selected. This ensures that the test signal does not severely interfere with the normal operating signal while still being clearly identifiable in the sampled signal. To avoid affecting normal metering, the amplitude of the test signal is set to 1% to 5% of the load signal amplitude.

[0056] The injection process is implemented through the control circuit and signal transmission module inside the metering chip. The control circuit precisely controls the timing and duration of signal transmission, while the signal transmission module converts the standard signal source into an electrical signal suitable for transmission in the sampling loop and injects it to form the test signal. The load signal of the sub-meter is obtained through the current transformer and voltage transformer inside the sub-meter. This load signal reflects the electrical load experienced by the sub-meter during actual operation, including current and voltage. After obtaining the load signal from the sub-meter, the metering chip combines the load signal and the test signal through addition to form the target test signal.

[0057] The target test signal is precisely measured using a metering chip to obtain the target change. The metering chip integrates a high-precision analog-to-digital converter (ADC) and a digital signal processor (DSP). The ADC converts the target test signal from analog to digital, and the DSP performs frame-by-frame processing on the resulting discrete digital sequence, applying windowing operations such as Hanning or Blackman windows to reduce spectral leakage. Then, a Fast Fourier Transform is performed on each frame of data to convert the time-domain signal to a frequency-domain signal, obtaining the corresponding amplitude and phase spectra. In the frequency domain results, based on a predetermined frequency set by the standard signal source, such as 55Hz or a characteristic injection frequency, the corresponding frequency index point or its neighboring frequency band is located. The actual response amplitude at that frequency is extracted through the amplitude spectrum, and phase information is obtained through the phase spectrum. Simultaneously, the theoretical output amplitude and phase of the standard signal source are known parameters. The DSP calculates the change in the target test signal relative to the theoretical output value of the standard signal source, obtaining the target change, such as amplitude deviation or phase deviation. This target change reflects the response deviation of the sub-meter to the standard signal under the current operating state.

[0058] Based on a pre-defined error mapping relationship, the self-monitoring error value corresponding to the target change is matched. This error mapping relationship is established through a calibration process before the metering chip leaves the factory and is used to describe the correspondence between signal deviation and energy metering error, thereby achieving a quantitative assessment of the energy meter's operating error. For example, in a standard metering experimental environment, the energy meter to be calibrated is connected to a high-precision standard source device, such as a standard energy calibration bench, which outputs multiple sets of voltage, current, and power signals under different operating conditions. Simultaneously, a test signal with known parameters is injected into the sampling circuit by the metering chip's built-in standard signal source, and the target change, such as amplitude deviation, is calculated under each operating condition using the FFT method. Under each set of operating conditions, the correspondence between the standard energy error given by the standard calibration device and the target change is recorded, forming the original calibration dataset. Then, a discrete lookup table method is used to integrate the calibration data and generate an error mapping relationship table. By looking up the error mapping relationship table, the self-monitoring error value can be obtained quickly and accurately.

[0059] By activating the metering chip and utilizing its built-in standard signal source to dynamically monitor the sub-meter's operation, the self-monitoring error value of the sub-meter can be obtained in real time and accurately. This self-monitoring error value reflects the metering accuracy deviation during actual operation. In this way, self-diagnosis and error monitoring of the sub-meter can be achieved without relying on complex external testing equipment, improving the efficiency and convenience of meter monitoring and enabling timely alarms and anomaly detection for operational errors. For example, sub-meter A has a normal operating current range of 0A-100A and a voltage range of 220V±10%. The metering chip's built-in standard signal source generates a test signal with a frequency of 55Hz and an amplitude of 5V. This test signal is injected into the sub-meter's sampling circuit at a predetermined frequency. Simultaneously, the load signal of the sub-meter is obtained through current transformers and voltage transformers; at this time, the load current is 50A and the load voltage is 220V. Inside the metering chip, the load signal and the test signal are superimposed and converted to obtain the target test signal. After measurement by the metering chip, the change in current in the target test signal was found to be 0.5A, and the change in voltage was 0.2V. According to the pre-established error mapping table, when the current change is 0.5A, the corresponding self-monitoring current error value is 0.1%; when the voltage change is 0.2V, the corresponding self-monitoring voltage error value is 0.05%. Combining this information, the self-monitoring error value of the sub-meter under the current operating state is obtained, providing real-time and accurate data support for evaluating the metering accuracy of the sub-meter.

[0060] Furthermore, injecting the standard signal source into the sampling circuit of the sub-energy meter at a predetermined frequency to obtain a test signal includes: injecting the standard signal source into the voltage sampling circuit of the sub-energy meter at a predetermined frequency to obtain a voltage test signal; injecting the standard signal source into the current sampling circuit of the sub-energy meter at a predetermined frequency to obtain a current test signal; the voltage test signal and the current test signal constitute the test signal.

[0061] Specifically, in order to achieve online detection of the metering accuracy of the sub-energy meter, the standard signal source built into the metering chip injects test signals of predetermined frequencies into the voltage sampling circuit and the current sampling circuit respectively. The predetermined frequency is set according to the working characteristics of the sub-energy meter and the detection requirements to avoid complete overlap with the normal power frequency signal, while ensuring that it can be effectively distinguished in the frequency domain.

[0062] When a standard signal source is injected into the voltage sampling circuit of the energy meter, the control circuit inside the metering chip precisely controls the timing and duration of signal transmission. Through a voltage coupling circuit, the electrical signal generated by the standard signal source is coupled to the voltage sampling circuit at a predetermined frequency. The voltage sampling circuit mainly consists of a voltage transformer and a signal conditioning circuit. The voltage transformer is responsible for proportionally converting high voltage to low voltage, while the signal conditioning circuit filters, amplifies, and processes the converted voltage signal to meet the requirements of subsequent acquisition and analysis. After the standard signal source is injected, a signal component with predetermined frequency characteristics is superimposed on the original voltage sampling signal, forming a voltage test signal.

[0063] Similarly, when a standard signal source is injected into the current sampling circuit of the sub-meter, it is precisely controlled by the metering chip's control circuit. Through a current coupling device, the electrical signal from the standard signal source is coupled to the current sampling circuit. The current sampling circuit includes a current transformer and corresponding signal processing circuitry. The current transformer proportionally converts a large current into a smaller current, and the signal processing circuitry conditions the smaller current signal. After the standard signal source is injected, a signal component with a predetermined frequency characteristic is superimposed on the current sampling signal, forming a current test signal. In this way, controllable standard disturbance signals are introduced into the voltage and current channels respectively without affecting the normal metering of the sub-meter. Finally, the voltage test signal and the current test signal together constitute a test signal, used for target test signal generation and error analysis.

[0064] By injecting standard signal sources into the voltage sampling circuit and the current sampling circuit respectively, a test signal with known characteristics can be constructed without affecting the normal operation of the energy meter, thereby improving the accuracy and real-time performance of error monitoring.

[0065] When the combined error value obtained by integrating the estimated error value and the self-monitoring error value is within a predetermined alarm threshold, an operational error alarm is issued for the sub-meter.

[0066] Specifically, after obtaining the predicted error value and the self-monitoring error value, the two types of errors are fused. The predicted error value refers to the prediction result of the energy meter error under the current operating state by the neural network trained based on historical data, which reflects a data-driven trend error. The self-monitoring error value comes from standard signal injection and real-time detection, which reflects the real-time error at the physical measurement level.

[0067] A weighted fusion method is used to calculate the comprehensive error value. Weighting coefficients are set based on the reliability and accuracy of the estimated and self-monitored error values. If the self-monitored error value is obtained through a high-precision standard signal source and rigorous testing procedures, its reliability is high, and its weight can be set larger. The estimated error value is based on historical data and model estimation, so its weight is set relatively smaller. For example, if the weight of the self-monitored error value is set to 0.6 and the weight of the estimated error value is set to 0.4, the comprehensive error value = self-monitored error value × 0.6 + estimated error value × 0.4.

[0068] The obtained comprehensive error value is compared with a preset alarm threshold. The preset alarm threshold is set according to the metering accuracy level of the sub-meter and the actual application scenario. Different industries and different uses of sub-meters have different requirements for metering accuracy, so the alarm threshold will also vary. For example, in scenarios with extremely high metering accuracy requirements, such as electricity trading, the alarm threshold is set at ±0.2%, while in some general industrial electricity monitoring scenarios, the alarm threshold is set at ±0.5%. When the comprehensive error value is within the preset alarm threshold range, that is, when the comprehensive error value is greater than or equal to the positive alarm threshold or less than or equal to the negative alarm threshold, it is determined that the current sub-meter has an abnormal operating error, triggering the operating error alarm mechanism and generating alarm information. The alarm information typically includes: meter number, current timestamp, comprehensive error value, and error type. The error type refers to positive deviation or negative deviation. The operating error is alarmed through various means such as audible and visual alarms and sending alarm information to the monitoring center, reminding relevant personnel to check and maintain the sub-meter in a timely manner.

[0069] By integrating data-driven prediction errors with real-time errors obtained from hardware detection, a dual verification and comprehensive evaluation of the operating status of electricity meters can be achieved. This can effectively improve the accuracy and reliability of error judgment, avoid misjudgments caused by a single error source, and thus provide timely and reliable alarm basis for abnormal operation of electricity meters, ensuring the safety and stability of power operation.

[0070] Furthermore, when the combined error value obtained by fusing the estimated error value and the self-monitoring error value is at a predetermined alarm threshold, an operational error alarm is issued to the sub-energy meter. This process further includes: constructing an operational error time series based on the combined error value and its corresponding monitoring time; analyzing the operational error time series using a time window to obtain an alarm coverage rate; determining whether the alarm coverage rate reaches a predetermined coverage rate threshold; if so, issuing a continuous operational error alarm to the sub-energy meter. The process of analyzing the operational error time series using a time window to obtain the alarm coverage rate includes: extracting a unit time series from the operational error time series according to the time window; randomly sampling the unit time series to obtain error value sampling points; when the error value sampling point is at the predetermined alarm threshold, recording the error value sampling point as an alarm sampling point; and using the ratio of the alarm sampling point to the error value sampling point as the alarm coverage rate.

[0071] Specifically, the monitoring time is precisely recorded by the clock module inside the sub-meter. This clock module refers to a high-precision electronic clock circuit or real-time clock chip integrated within the sub-meter, possessing high-precision time synchronization capabilities and providing a continuous and stable time reference. The clock module supports periodic synchronization with external standard time sources, such as BeiDou, GPS, or Network Time Protocol (NTP), ensuring the accuracy and reliability of the time stamp for each sampling point or comprehensive error value. The comprehensive error values ​​are arranged sequentially according to the monitoring time, forming an operational error time series. This operational error time series reflects the changes in the sub-meter's operational error at different times.

[0072] A time window is introduced, which is a pre-defined time range whose size can be determined based on actual monitoring needs and the operating characteristics of the individual energy meters. For example, for energy meters whose operating status changes frequently, the time window can be set relatively small, such as 10 minutes; for energy meters with more stable operation, the time window can be set larger, such as 1 hour. Based on the set time window, the operating error time series is divided into several unit time series segments. For each unit time series segment, random sampling is performed, that is, several error value points are randomly selected from the unit time series as sampling points to calculate the alarm coverage. The number of samples can be determined according to the length of the unit time series and the monitoring accuracy requirements; for example, 10 error value sampling points are randomly sampled every 10 minutes.

[0073] Within each unit time segment, when the number of error value sampling points exceeds a predetermined alarm threshold, these error value sampling points are marked as alarm sampling points. The ratio of alarm sampling points to error value sampling points is used as the alarm coverage rate, i.e., alarm coverage rate = (alarm sampling points / total number of error value sampling points) × 100%. The alarm coverage rate reflects the proportion of the electricity meter error in an alarm state within that time window. The alarm coverage rate is compared with a predetermined coverage threshold, which is set based on actual monitoring needs and the importance of the individual electricity meters, using expert experience (e.g., 30%). When the alarm coverage rate is greater than or equal to the predetermined coverage threshold, it indicates that the individual electricity meter frequently experiences operational errors exceeding the standard within a certain period. In this case, a continuous operational error alarm is issued for the individual electricity meter, reminding relevant personnel to take timely measures for inspection and maintenance to avoid greater losses.

[0074] By constructing an operational error time series and introducing a time window for analysis, we can gain a more comprehensive and dynamic understanding of the operational error of the sub-meters. Furthermore, the calculation of alarm coverage provides a quantitative basis for determining whether the sub-meters need continuous alarms, effectively avoiding false alarms caused by single instantaneous anomalies or occasional fluctuations. At the same time, it can respond promptly to continuous operational errors, achieving stable and reliable alarms for abnormal operation of the sub-meters.

[0075] Further, after determining whether the alarm coverage rate has reached a predetermined coverage threshold, the method further includes: if the alarm coverage rate has not reached the predetermined coverage threshold, obtaining a scatter plot of the operating error corresponding to the operating error time sequence; fitting the operating error scatter plot to obtain an operating error curve; performing variability weighting processing on the multidimensional feature parameters of the operating error curve to obtain operating error feature coefficients; determining whether the operating error feature coefficients are within a predetermined threshold range; if not, issuing an electricity theft alarm for the sub-meter.

[0076] Specifically, after performing time window analysis on the operating error of the sub-meters, if the alarm coverage rate does not reach the predetermined coverage threshold (i.e., is less than the predetermined coverage threshold), the operating error time series for the current time period is obtained, i.e., the comprehensive error value corresponding to each sampling time. A scatter plot is then drawn with the monitoring time as the horizontal axis and the comprehensive error value in the operating error time series as the vertical axis, forming an operating error scatter plot. The operating error scatter plot can intuitively display the distribution characteristics of the error over time.

[0077] The scatter plot of the operating error is fitted using methods such as polynomial fitting, spline curve fitting, or least squares fitting. For example, the least squares method finds the best function match for the data by minimizing the sum of squares of the errors, resulting in an operating error curve. This curve more smoothly reflects the changing trend of the operating error and removes some random fluctuations from the scatter plot. Multidimensional feature parameters are extracted from the fitted operating error curve, including at least the slope, curvature, and amplitude of the fluctuation. The slope reflects the rate of change of the operating error over time, the curvature reflects the degree of bending of the operating error change, and the amplitude of the fluctuation represents the dispersion of the operating error.

[0078] The multidimensional feature parameters undergo variability-weighted processing. This process involves first normalizing the multidimensional feature parameters extracted from the operating error curve, such as slope, curvature, and fluctuation amplitude, and then assigning different weights to each parameter to reflect its importance in anomaly detection. Simultaneously, a small random perturbation, such as ±1% to 2%, is introduced to enhance sensitivity to intermittent or discontinuous anomalies. The weighted sum of these parameters yields the operating error characteristic coefficient, which serves as a comprehensive index for evaluating the energy meter's operating status. For example, after normalizing the extracted feature parameters slope, curvature, and fluctuation amplitude, the slope is normalized to S=0.85, curvature to K=0.75, and fluctuation amplitude to A=0.95. A small random perturbation is applied to each feature parameter to increase sensitivity to nonlinear anomalies. r =S + (S × Rand(±1%)) = 0.85 + 0.85 × 0.01 = 0.8585. Similarly, small perturbations are introduced into the curvature and fluctuation amplitude to enhance the sensitivity of the error characteristic coefficients to sudden events, such as load changes or electricity theft. Then, different weights are assigned to each feature to represent the importance of each feature in judging anomalies. The slope is more important in terms of the rate of change, so it is assigned a larger weight: slope weight 0.5, curvature weight 0.2, and fluctuation amplitude weight 0.3. Then, these weights are used to perform a weighted summation of the perturbated normalized features to obtain the final operating error characteristic coefficients.

[0079] Based on historical calibration data, typical normal operating error distribution, and known characteristics of electricity theft, a predetermined threshold range is set. For example, by analyzing the statistical distribution of characteristic coefficients under a large number of normal loads, a 95% confidence interval is used as the safety threshold, i.e., the predetermined threshold range. By analyzing data from a large number of normally operating sub-meters, the statistical distribution of the operating error characteristic coefficients is calculated to have a sample mean of 0.75 and a sample standard deviation of 0.1. A 95% confidence interval is set as the safety threshold range for normal operation, i.e., the predetermined threshold range is 0.55~0.95. The operating error characteristic coefficients are compared with the predetermined threshold range. If the calculated operating error characteristic coefficients are not within the predetermined threshold range, it indicates that the current error change trend and amplitude deviate significantly from the normal operating mode, and there may be abnormal energy consumption behavior. Therefore, it is determined that the sub-meter has abnormal consumption behavior, triggering an electricity theft alarm so that maintenance personnel can intervene in time or conduct further analysis. If the operating error characteristic coefficients are within the predetermined threshold range, it indicates that the error change is within the normal fluctuation range, no alarm is triggered, and periodic monitoring can continue.

[0080] By fitting the scatter plot of operating error and extracting multidimensional feature parameters, abnormal energy consumption behavior of the electricity meter, such as electricity theft, can still be identified even when the alarm coverage does not exceed the threshold, thus improving the detection capability and accuracy of abnormal conditions of the electricity meter.

[0081] Example 2, based on the same inventive concept as the method for monitoring the operating error of an electricity meter in the foregoing examples, such as... Figure 2 As shown, this application provides an operational error monitoring system for an electricity meter, wherein the operational error monitoring system for an electricity meter includes: The electricity consumption information acquisition module 11 is used to acquire target electricity consumption information through the electricity consumption monitoring terminal, wherein the target electricity consumption information includes first electricity consumption data corresponding to the main electricity meter and second electricity consumption data corresponding to the sub-electricity meters; the electricity consumption data processing module 12 is used for the electricity consumption monitoring master station to call a preprocessing strategy to process the first electricity consumption data and the second electricity consumption data in sequence, and form standard electricity consumption data; the error prediction module 13 is used to input the standard electricity consumption data into the backpropagation neural network to obtain the predicted error value of the backpropagation neural network; the operation monitoring module 14 is used to activate the metering chip to perform dynamic operation monitoring of the sub-electricity meters to obtain the self-monitoring error value; the error alarm module 15 is used to issue an operation error alarm to the sub-electricity meters when the comprehensive error value obtained by fusing the predicted error value and the self-monitoring error value is at a predetermined alarm threshold.

[0082] Furthermore, a BeiDou communication channel is established between the electricity monitoring terminal and the electricity monitoring master station, wherein the BeiDou communication channel realizes electricity data transmission using a message communication architecture.

[0083] Furthermore, the electricity consumption data processing module 12 is also used to: according to the data cleaning strategy, sequentially eliminate noise data in the first electricity consumption data and the second electricity consumption data to obtain first processed data and second processed data respectively; according to the scaling transformation strategy, sequentially perform maximum and minimum value normalization processing on the first processed data and the second processed data to obtain third processed data and fourth processed data respectively; according to the data reduction strategy, sequentially perform dimensionality reduction processing on the third processed data and the fourth processed data to obtain fifth processed data and sixth processed data respectively; and based on the fifth processed data and the sixth processed data, form the standard electricity consumption data.

[0084] Furthermore, the error prediction module 13 is also used to: acquire historical electricity meter operation data as training samples; input the training samples into the neural network for signal forward propagation and error backpropagation training to obtain the predicted operation error, and update the preset connection weights and preset node thresholds of each layer of the neural network according to the predicted operation error to obtain the backpropagation neural network; wherein, when the number of iterations reaches a preset maximum value or the predicted operation error is less than a preset accuracy threshold, the network training is completed and the backpropagation neural network is obtained.

[0085] Furthermore, the error prediction module 13 is also used to: encode the initial connection weights and initial node thresholds into the position vector of the optimization individual; introduce a fitness factor to analyze the prediction error and obtain the fitness of the optimization individual; use the fitness as an evaluation index to iteratively optimize the position vector of the optimization individual and determine the optimal position vector corresponding to the highest fitness; decode the optimal position vector to obtain a decoding result, wherein the decoding result includes the preset connection weights and the preset node thresholds.

[0086] Furthermore, the operation monitoring module 14 is also used to: inject the standard signal source into the sampling circuit of the sub-energy meter at a predetermined frequency to obtain a test signal; acquire the load signal of the sub-energy meter, and superimpose and convert the load signal with the test signal to obtain a target test signal; obtain the target change amount of the target test signal through the metering chip, and match the self-monitoring error value corresponding to the target change amount.

[0087] Furthermore, the operation monitoring module 14 is also used to: inject the standard signal source into the voltage sampling circuit of the sub-energy meter at a predetermined frequency to obtain a voltage test signal; inject the standard signal source into the current sampling circuit of the sub-energy meter at a predetermined frequency to obtain a current test signal; the voltage test signal and the current test signal constitute the test signal.

[0088] Furthermore, the error alarm module 15 is also used to: construct an operating error time series based on the comprehensive error value and its corresponding monitoring time; analyze the operating error time series using a time window to obtain an alarm coverage rate; determine whether the alarm coverage rate reaches a predetermined coverage threshold, and if so, issue a continuous operating error alarm to the sub-energy meter; wherein, analyzing the operating error time series using a time window to obtain the alarm coverage rate includes: extracting a unit time series of the operating error time series according to the time window; randomly sampling the unit time series to obtain error value sampling points; when the error value sampling point is at the predetermined alarm threshold, recording the error value sampling point as an alarm sampling point; and using the ratio of the alarm sampling point to the error value sampling point as the alarm coverage rate.

[0089] Furthermore, the error alarm module 15 is also used to: if the alarm coverage rate does not reach the predetermined coverage threshold, obtain the scatter plot of the operating error corresponding to the operating error time sequence; fit the scatter plot of the operating error to obtain the operating error curve; perform variability weighting processing on the multidimensional feature parameters of the operating error curve to obtain the operating error feature coefficient; determine whether the operating error feature coefficient is within the predetermined threshold range, and if not, issue an electricity theft alarm to the sub-electricity meter.

[0090] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The method and specific example for monitoring the operating error of an electricity meter in the first embodiment described above are also applicable to the operating error monitoring system for an electricity meter in this embodiment. Through the detailed description of the operating error monitoring method for an electricity meter described above, those skilled in the art can clearly understand the operating error monitoring system for an electricity meter in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0091] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily 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 this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0092] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for monitoring the operational error of an electricity meter, characterized in that, include: Target electricity consumption information is obtained through an electricity consumption monitoring terminal, wherein the target electricity consumption information includes first electricity consumption data corresponding to the main electricity meter and second electricity consumption data corresponding to the sub-electricity meters; The electricity monitoring master station invokes a preprocessing strategy to process the first electricity consumption data and the second electricity consumption data in sequence, and then combines them into standard electricity consumption data. The standard electricity consumption data is input into the backpropagation neural network to obtain the prediction error value of the backpropagation neural network; The metering chip is activated to dynamically monitor the operation of the sub-electricity meter and obtain the self-monitoring error value. When the combined error value obtained by integrating the estimated error value and the self-monitoring error value is within a predetermined alarm threshold, an operational error alarm is issued for the sub-meter.

2. The method for monitoring the operating error of an electricity meter according to claim 1, characterized in that, A BeiDou communication channel is established between the power consumption monitoring terminal and the power consumption monitoring master station. The BeiDou communication channel uses a message communication architecture to transmit power consumption data.

3. The method for monitoring the operating error of an electricity meter according to claim 1, characterized in that, The preprocessing strategy includes a data cleaning strategy, a scaling transformation strategy, and a data reduction strategy. The electricity monitoring master station calls the preprocessing strategy to process the first electricity consumption data and the second electricity consumption data in sequence, and forms standard electricity consumption data, including: According to the data cleaning strategy, noise data in the first power consumption data and the second power consumption data are eliminated in sequence to obtain the first processed data and the second processed data, respectively. According to the ratio transformation strategy, the first processed data and the second processed data are sequentially subjected to maximum and minimum value normalization processing to obtain the third processed data and the fourth processed data, respectively. According to the data reduction strategy, the third processed data and the fourth processed data are sequentially subjected to dimensionality reduction processing to obtain the fifth processed data and the sixth processed data, respectively. The standard electricity consumption data is composed based on the fifth and sixth processed data.

4. The method for monitoring the operating error of an electricity meter according to claim 1, characterized in that, The standard electricity consumption data is input into the backpropagation neural network to obtain the prediction error value of the backpropagation neural network, which includes the following steps: Historical electricity meter operation data was used as training samples. The training samples are input into the neural network for signal forward propagation and error backpropagation training to obtain the predicted running error. The preset connection weights and preset node thresholds of each layer of the neural network are updated according to the predicted running error to obtain the backpropagation neural network. Specifically, network training is completed when the number of iterations reaches a preset maximum value or the prediction error is less than a preset accuracy threshold, thus obtaining the backpropagation neural network.

5. The method for monitoring the operating error of an electricity meter according to claim 4, characterized in that, The training samples are input into the neural network for signal forward propagation and error backpropagation training to obtain the predicted running error. Based on the predicted running error, the preset connection weights and preset node thresholds of each layer of the neural network are updated to obtain the backpropagation neural network. This process includes: The initial connection weights and initial node thresholds are encoded into the position vectors of the individuals being optimized. A fitness factor is introduced to analyze the predicted running error, and the fitness of the optimized individual is obtained. Using fitness as an evaluation index, the position vector of the optimization individual is iteratively optimized to determine the optimal position vector corresponding to the highest fitness. The optimal position vector is decoded to obtain a decoding result, wherein the decoding result includes the preset connection weight and the preset node threshold.

6. The method for monitoring the operating error of an electricity meter according to claim 1, characterized in that, The metering chip has a built-in standard signal source. Activating the metering chip allows for dynamic monitoring of the sub-meter's operation, resulting in a self-monitoring error value, including: The standard signal source is injected into the sampling circuit of the energy meter at a predetermined frequency to obtain a test signal; The load signal of the sub-electricity meter is acquired, and the load signal is superimposed and converted with the test signal to obtain the target test signal; The target change amount of the target test signal is obtained through the metering chip, and the self-monitoring error value corresponding to the target change amount is matched.

7. The method for monitoring the operating error of an electricity meter according to claim 6, characterized in that, The standard signal source is injected into the sampling circuit of the energy meter at a predetermined frequency to obtain a test signal, including: The standard signal source is injected into the voltage sampling circuit of the energy meter at a predetermined frequency to obtain a voltage test signal; The standard signal source is injected into the current sampling circuit of the sub-energy meter at a predetermined frequency to obtain a current test signal; The voltage test signal and the current test signal together constitute the test signal.

8. The method for monitoring the operating error of an electricity meter according to claim 1, characterized in that, When the combined error value obtained by fusing the estimated error value and the self-monitoring error value is at a predetermined alarm threshold, an operational error alarm is issued for the sub-meter, followed by: Based on the comprehensive error value and its corresponding monitoring time, an operational error time series is constructed. A time window is introduced to analyze the timing of the operational errors, and the alarm coverage rate is obtained. Determine whether the alarm coverage rate has reached a predetermined coverage threshold. If it has, issue a continuous alarm for the operating error of the sub-electricity meter. Specifically, a time window is introduced to analyze the timing of the operational errors to obtain the alarm coverage rate, including: The unit timing sequence of the running error timing is extracted according to the time window; Randomly sample the unit time series to obtain error value sampling points; When the error value sampling point is at the predetermined alarm threshold, the error value sampling point is recorded as an alarm sampling point; The ratio of the alarm sampling points to the error value sampling points is used as the alarm coverage rate.

9. The method for monitoring the operating error of an electricity meter according to claim 8, characterized in that, After determining whether the alarm coverage rate has reached a predetermined coverage threshold, the process further includes: If the alarm coverage rate does not reach the predetermined coverage rate threshold, then obtain the scatter plot of the running error corresponding to the running error time sequence. The running error curve is obtained by fitting the scatter plot of the running error. The multidimensional characteristic parameters of the operating error curve are subjected to variability weighting to obtain the operating error characteristic coefficients; Determine whether the operating error characteristic coefficient is within a predetermined threshold range. If not, issue an electricity theft alarm to the sub-electricity meter.

10. A system for monitoring the operational error of an electricity meter, characterized in that, The steps for implementing the method for monitoring the operating error of an electricity meter according to any one of claims 1 to 9 include: The electricity consumption information acquisition module is used to acquire target electricity consumption information through an electricity consumption monitoring terminal, wherein the target electricity consumption information includes first electricity consumption data corresponding to the main electricity meter and second electricity consumption data corresponding to the sub-electricity meters; The electricity consumption data processing module is used by the electricity consumption monitoring master station to call the preprocessing strategy to process the first electricity consumption data and the second electricity consumption data in sequence, and to form standard electricity consumption data. An error prediction module is used to input the standard electricity consumption data into a backpropagation neural network to obtain the prediction error value of the backpropagation neural network. The operation monitoring module is used to activate the metering chip to dynamically monitor the operation of the sub-electricity meter and obtain the self-monitoring error value; The error alarm module is used to issue an operational error alarm to the sub-energy meter when the comprehensive error value obtained by fusing the estimated error value and the self-monitoring error value is at a predetermined alarm threshold.

Images