A method for maintaining power metering

Abstract

The present invention belongs to the technical field of power systems. The present invention discloses a method for maintaining power metering, including collecting power metering-related data; preprocessing the power metering-related data to obtain a comprehensive power metering feature data set; training to obtain a power metering error prediction model based on the comprehensive power metering feature data set, and predicting the power metering error based on the power metering error prediction model; judging whether error correction is required according to the predicted power metering error; if error correction is required, generating a warning message and automatically collecting power metering error data; training to obtain an error type identification model based on the power metering error data, and predicting the power metering error type data based on the error type identification model; correcting the power metering error by using a distribution calibration algorithm based on the power metering error type data; reducing the need for manual intervention and improving the working efficiency of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and more specifically, to a method for maintaining electric energy metering. Background Art

[0002] A patent with the patent publication number CN101871967A discloses a metering method for an electric energy meter, including the following steps: a. The microprocessor monitors the voltage of the working power supply. Once the voltage value is lower than a voltage set value, the electric energy meter enters the electricity theft metering mode; b. After a first set time has passed, the microprocessor starts current detection, performs current detection within a second set time, multiplies the measured current value by the rated voltage value of the electric energy meter and then by the total time length of the first set time and the second set time as the electric energy for the time period, and accumulates the electric energy for the time period to the total electric energy; c. The microprocessor monitors the voltage of the working power supply. If the voltage value remains lower than the voltage set value, return to step b; otherwise, multiply the measured current value by the rated voltage value of the electric energy meter and then by the total time length from the previous metering moment to the present as the electric energy for the time period, and return to the normal working mode. Thus, the electric energy in the case of electricity theft will not be under-recorded.

[0003] Existing methods for maintaining electric energy metering have the following defects:

[0004] There is a lack of targeted correction strategies and it is unable to select appropriate correction strategies based on the error type, which may lead to the use of a single correction method when facing different types of errors. This will make the error correction inaccurate and unable to handle complex electric energy metering error situations; no appropriate correction function is defined for each node, which may lead to a too rough correction process;

[0005] The error propagation path is not clear. The electric energy metering system is not transformed into a graph structure and lacks an effective error propagation path; the correction information between nodes cannot be propagated through directed edges, resulting in the errors in the system not being gradually propagated and corrected globally, unable to accurately track the error propagation process of each node, and reducing the effect and efficiency of correction; without a clear error propagation path, it is difficult for the system to trace back to the root cause of the error, resulting in difficulty in locating the problem nodes and low correction efficiency; the correction amount is not dynamically adjusted according to the error values of adjacent nodes, and the correction amount of the node may be fixed or inaccurate; the fixed correction amount cannot reflect the mutual influence between nodes, is unable to flexibly handle the error type and the actual situation of the system, and thus leads to overcorrection or undercorrection, affecting the accuracy of the system;

[0006] Without dynamic adjustment based on the errors of actual adjacent nodes, some nodes with large errors cannot be fully corrected, or some nodes with small errors may be over-corrected, affecting the accuracy of the entire power metering system; without considering the topological structure, the interaction between nodes is not effectively reflected. Each node in the system may not be able to fully utilize the information of adjacent nodes, resulting in poor performance of the correction strategy; without a mechanism to dynamically adjust the correction factor, some nodes may experience excessive changes in their power metering errors due to over-correction, leading to system oscillations; especially when the errors of some nodes in the system are large, failure to adjust the correction amplitude according to the actual errors will result in over-correction of the errors, even interfering with the entire system and affecting stability; without achieving error consistency through a global error correction formula, the error correction of each node cannot be synchronized;

[0007] Without transmitting the correction information to other nodes through PLC communication, the errors cannot be effectively propagated and corrected, thus affecting global consistency; the errors may gradually accumulate in some nodes, ultimately leading to a decline in the power metering accuracy of the entire system; if there is a lack of mutual correction mechanism between nodes, some nodes may not be able to obtain the correction information of other nodes, resulting in local accumulation of errors; especially when the network topology is complex or the number of nodes is large, the errors of isolated nodes may continuously expand, thus affecting the overall performance of the entire system; without a global error synchronization mechanism, the correction processes between nodes may not have a proper timing and pace, resulting in an unbalanced distribution of errors; without mutual correction and information transmission between nodes, simply relying on the initial correction of individual nodes may ignore the influence of other nodes; without dynamically synchronizing the errors between nodes through an information propagation mechanism, the errors may continuously accumulate in some nodes, affecting the correction processes of other nodes; without a dynamic correction factor and information propagation mechanism, the system cannot adaptively adjust according to the changes in the network topology of the power metering system;

[0008] Without a mechanism to gradually make the error values approach global consistency, the error correction of each node may remain at a local optimum and cannot effectively integrate global information; without an error correction synchronization formula, the final correction errors of nodes may be inconsistent, resulting in unbalanced correction effects between different nodes; as the power metering equipment and network environment change, the errors may change; without a dynamic adjustment mechanism, the system cannot optimize in real time according to the changes in the errors, leading to a decline in system accuracy.

[0009] In view of this, the present invention proposes a method for maintaining power metering to solve the above problems. Summary of the Invention

[0010] To overcome the above-mentioned defects of the prior art and to achieve the above object, the present invention provides the following technical solution: A method for maintaining power metering, comprising:

[0011] S1. Collect power metering related data;

[0012] S2. Preprocess the power metering related data to obtain a power metering comprehensive feature dataset;

[0013] S3. Train a power metering error prediction model based on the power metering comprehensive feature dataset, and predict the power metering error based on the power metering error prediction model;

[0014] S4. Determine whether error correction is required according to the predicted power metering error; if error correction is required, generate a warning message and automatically collect power metering error data;

[0015] S5. Train an error type identification model based on the power metering error data, and predict the power metering error type data based on the error type identification model; based on the power metering error type data, use a distribution calibration algorithm to correct the power metering error, and transmit the power metering error correction result to other nodes through distributed topology and PLC communication, so that the power metering of all nodes is consistent;

[0016] S6. Verify all the power metering error correction results. If there are still errors, the power metering control terminal issues a secondary correction instruction until the corrected power metering error reaches the preset power metering error threshold and then stops.

[0017] Furthermore, the power metering related data includes electrical parameter data, equipment status data and network parameter data; the electrical parameter data includes line voltage, phase voltage, line current, phase current, power transmission power, grid frequency and total power; the equipment status data includes equipment operation status, equipment temperature and equipment operation time; the network parameter data includes communication network bandwidth, network delay, communication protocol, packet loss rate, signal strength, network load.

[0018] Furthermore, the method for preprocessing the power metering related data to obtain a power metering comprehensive feature dataset includes:

[0019] Identify and remove the outliers existing in the electrical parameter data, equipment status data and network parameter data included in the power metering related data through a density clustering algorithm to obtain an electrical parameter feature dataset, an equipment status feature dataset and a network parameter feature dataset;

[0020] Denoise the electrical parameter feature dataset, device status feature dataset, and network parameter feature dataset through Kalman filtering, and perform standard deviation normalization on the denoised electrical parameter feature dataset, device status feature dataset, and network parameter feature dataset to convert them into a standard normal distribution with a mean of 0 and a standard deviation of 1, obtaining the normalized electrical parameter feature dataset, device status feature dataset, and network parameter feature dataset; fuse the normalized electrical parameter feature dataset, device status feature dataset, and network parameter feature dataset through a weighted model to obtain the comprehensive power metering feature dataset.

[0021] Further, the method for fusing the normalized electrical parameter feature dataset, device status feature dataset, and network parameter feature dataset through a weighted model to obtain the comprehensive power metering feature dataset includes:

[0022] Denote the electrical parameter feature dataset as and the device status feature dataset as and the network parameter feature dataset as ; the weighted model is: ; where is the comprehensive power metering feature dataset; is the weight coefficient of the electrical parameter feature dataset; is the weight coefficient of the device status feature dataset; is the weight coefficient of the network parameter feature dataset.

[0023] Further, the training method of the power metering error prediction model includes:

[0024] Divide the dataset into a training set, a validation set, and a test set, and construct a power metering error prediction model; the sample set is a subset of the dataset, and each sample set includes the historical comprehensive power metering feature dataset and the corresponding power metering error; the power metering error prediction model includes an input layer, a GRU layer, a fully connected layer, and an output layer; the input layer of the model is used to input the historical comprehensive power metering feature dataset, and the number of neurons in the input layer matches the number of features of the historical comprehensive power metering feature dataset;

[0025] Use the GRU layer to process the historical comprehensive power metering feature dataset, and adjust the number of GRU layers and the number of neurons according to the task complexity; provide additional non-linear transformation through the fully connected layer; the output layer of the model is used to output the power metering error, and a neuron is used to output the predicted value, and the identity function is used as the activation function; the power metering error prediction model is a gated recurrent unit model;

[0026] Use the mean absolute error as the loss function to measure the error between the predicted value and the actual value of the model; use the training set data to train the model, and minimize the loss function through the Adam optimizer; use the validation set to evaluate the performance of the model, tune the hyperparameters of the model until the model performance no longer improves or reaches the preset stop condition and then stop;

[0027] Use the test set to evaluate the performance of the electric energy metering error prediction model in the prediction task, input the current comprehensive electric energy metering feature dataset into the trained electric energy metering error prediction model, and obtain the electric energy metering error.

[0028] Furthermore, the method for judging whether error correction is needed according to the predicted electric energy metering error includes:

[0029] If the predicted electric energy metering error is less than or equal to the preset electric energy metering error threshold, it is determined that error correction is not needed;

[0030] If the predicted electric energy metering error is greater than the preset electric energy metering error threshold, it is determined that error correction is needed.

[0031] Furthermore, the training method of the error type recognition model includes:

[0032] Divide the dataset into a training set, a test set and a validation set, and construct an error type recognition model. The error type recognition model includes an input layer, a hidden layer and an output layer; the input layer of the model is used to input historical electric energy metering error data, and the output layer is used to output electric energy metering error type data; the output layer is set with neurons equal in number to the number of electric energy metering error types, and each neuron corresponds to the prediction probability of an electric energy metering error type. Use the softmax function as the activation function; the error type recognition model is a multi-layer perceptron MLP model;

[0033] Use multi-class cross-entropy as the loss function of the model to measure the difference between the predicted value and the actual value of the model; use the training set to train the error type recognition model, and update the model parameters through the backpropagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the error type recognition model by calculating the accuracy index;

[0034] Select the SGD optimization algorithm as the optimizer, tune the model according to the performance feedback of the validation set, adjust the model parameters until the performance no longer improves or reaches the preset stop condition and then stop; use the test set to evaluate the performance of the model in the prediction task, and use the trained error type recognition model to predict the current electric energy metering error data to obtain the electric energy metering error type data.

[0035] Further, the electric energy metering error type data includes measurement deviation error, equipment failure error, environmental impact error, network communication error, algorithm error, clock synchronization error, and power grid fluctuation error.

[0036] Further, the method for correcting the electric energy metering error by using the distribution calibration algorithm based on the electric energy metering error type data includes:

[0037] S91. Denote the electric energy metering error type data as a data set ; where is the th electric energy metering error type data; is the index of the electric energy metering error type data, ; is the total number of the electric energy metering error type data;

[0038] S92. Based on the electric energy metering error type data , select the corresponding correction strategy; set the correction functions corresponding to different electric energy metering error type data : ; where is the correction function corresponding to the th electric energy metering error type data; is the electric energy metering error value of the th node;

[0039] S93. Construct a distributed topology graph for each node corresponding to the electric energy metering error type data in the electric energy metering system ; where is the node set, representing the electric energy metering nodes in the electric energy metering system; is the edge set, representing the directed communication relationship between nodes;

[0040] The node set ; where is the th node; is the number of nodes; the edge set ; where is the th node; is the th node; represents that node sends error correction information to the adjacent node ; and are the indexes of the nodes;

[0041] S94. For each node , calculate the error correction amount of the node through the error correction amount calculation formula; the error correction amount calculation formula is: ; where is the error correction amount of the node; is the adjacent node set of the node, that is, all the parent node sets pointing to the node ; is the th power metering error of the adjacent node; is the correction amount influence factor; is the weight factor of the correction function is the index of the adjacent node;

[0042] S95. Dynamically adjust and design the correction amount influence factor through the correction amount influence factor adjustment formula. The correction amount influence factor adjustment formula is: ; where is the number of adjacent nodes of the node; is a constant that controls the size of the overall weight factor; is the influence degree of the number of adjacent nodes on the weight factor;

[0043] S96. Constrain and limit the weight factor of the correction function through the error sensitivity adjustment formula. The error sensitivity adjustment formula is: ; where is the weight factor of the restricted correction function ; is the maximum value of the power metering errors of all nodes in the power metering system; is the factor that controls the influence of the error size on the weight factor ;

[0044] Further, the method of transmitting the power metering error correction result to other nodes through distributed topology and PLC communication to make the power metering of all nodes consistent includes:

[0045] Through PLC communication, transmit the corrected power metering error of each node to its adjacent nodes through the signal transmission formula. The preliminary correction error of the preset node is , and the signal transmission formula is: ; where is the preliminary correction error of the node;

[0046] In each round of correction, the node transmits its preliminary correction error through PLC Send it to its downstream node, namely , while node receives the correction error from the upstream node and updates its own power metering error according to the error correction information formula; the error correction information formula is: ; where is the power metering error of node after update; is the power metering error of node before receiving the error correction information;

[0047] As the error correction information propagates in the power metering system, while node updates its own power metering error according to the error correction information from other nodes, the global error consistency is gradually achieved through the global error correction formula; the global error correction formula is: ; where is the global error value of node ;

[0048] All nodes continue to synchronize the error through the error correction synchronization formula, and finally the power metering errors of all nodes tend to be consistent; the error correction synchronization formula is: ; where is the final correction error of node ; is the correction factor for adjusting the size of the final correction amount;

[0049] The correction factor for adjusting the size of the final correction amount is dynamically adjusted through the correction factor self-adaptation formula, and the correction factor self-adaptation formula is: ; where is the correction factor for adjusting the size of the final correction amount at the th iteration; is the coefficient affecting the influence degree of the error mean on the correction factor; is the mean value of the power metering errors of all nodes in the system; is the attenuation factor affecting the number of iterations in the correction process; is the current number of iterations; is the maximum number of iterations.

[0050] The technical effects and advantages of a method for maintaining power metering according to the present invention:

[0051] By collecting the power metering error data of all nodes into a set, the present invention can uniformly manage and analyze the errors of the entire system, providing a basis for subsequent data processing and analysis, ensuring that the errors of each node can be independently identified and corrected, selecting corresponding correction strategies based on the error types, setting correction functions for data of different error types, and correcting according to the power metering error values of the nodes; defining appropriate correction functions for each node to make the correction process more accurate and effective.

[0052] Transform the power metering system into a graph structure so that the correction information of each node can be propagated through the directed edges in the graph, promoting the mutual influence between nodes; the directed graph provides a structured way to track the error propagation paths of each node and its adjacent nodes, improving the efficiency and accuracy of error correction; the correction amount calculation formula can dynamically adjust the correction amount of the target node according to the error values of adjacent nodes, ensuring that the correction strategy of the entire system can be adjusted according to the actual situation; by calculating the correction amount of each node, gradually reduce the error of each node and improve the overall accuracy of the system; the influence factor of the number of adjacent nodes can be dynamically adjusted according to the system topology structure to ensure that under different network environments, the adjustment of the correction factor can reflect the interaction between nodes in the network; the dynamic adjustment of the correction amount influence factor helps to balance the stability of the system and avoid oscillations or unnecessary corrections caused by overcorrection of some nodes; through error sensitivity adjustment, it is possible to avoid overcorrection during the correction of the power metering system, resulting in an unstable situation of the system.

[0053] Each node transmits the preliminarily corrected power metering error to adjacent nodes through PLC communication. This process ensures that the correction information of each node can be propagated throughout the system, thereby affecting the error correction of other nodes; through this information transmission, the system can gradually converge to global error consistency; in each round of correction, after receiving the error correction information from the upstream node, the node updates its own power metering error; this mechanism ensures the gradual adjustment and correction of the errors of each node in the system, avoiding the locality problem of single-point correction; through the mutual correction between nodes, ensure that the errors of the entire power metering system gradually tend to be consistent; through PLC signal transmission, the errors of all nodes are effectively synchronized, reducing error accumulation or local deviation; through multiple rounds of information propagation and update, each node can be updated according to the correction information of other nodes in the system, rather than simply relying on its own preliminary correction, avoiding the amplification effect of the errors of isolated nodes; this method is distributed and there is no single control center, so it can avoid the impact of single-point failures and improve the reliability and robustness of the system.

[0054] Based on information transmission and node update, each node gradually makes the error value approach global consistency through the global error correction formula. This ensures that as the correction progresses, the system error can be globally adjusted, thereby improving the overall accuracy of the system; through the error correction synchronization formula, the final correction errors of each node tend to be consistent, thus achieving the goal of error synchronization; finally, the power measurement errors of all nodes tend to be consistent, ensuring the overall accuracy of the system; according to the mean value of the power measurement errors of all nodes in the system and the number of iterations, the correction factor for each iteration is dynamically adjusted; this adjustment mechanism takes into account factors such as the distribution of errors and the attenuation of the correction factor, making the error correction process more flexible and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 FIG. is a schematic flowchart of a method for maintaining power measurement;

[0056] Figure 2 FIG. is a schematic structural diagram of a system for maintaining power measurement. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0057] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] Embodiment 1

[0059] Please refer to Figure 1 As shown, a method for maintaining power measurement in this embodiment includes:

[0060] S1. Collect power measurement-related data;

[0061] S2. Preprocess the power measurement-related data to obtain a power measurement comprehensive feature dataset;

[0062] S3. Train a power measurement error prediction model based on the power measurement comprehensive feature dataset, and predict the power measurement error based on the power measurement error prediction model;

[0063] S4. Determine whether error correction is required according to the predicted power measurement error; if error correction is required, generate a warning message and automatically collect power measurement error data;

[0064] S5. Train and obtain an error type recognition model based on the power metering error data. Based on the error type recognition model, predict the power metering error type data. Based on the power metering error type data, use the distribution calibration algorithm to correct the power metering error, and transmit the power metering error correction result to other nodes through the distributed topology and PLC communication, so that the power metering of all nodes is consistent.

[0065] S6. Verify all the power metering error correction results. If there are still errors, the power metering control terminal issues a secondary correction instruction until the corrected power metering error reaches the preset power metering error threshold and then stops.

[0066] The power metering related data includes electrical parameter data, equipment status data, and network parameter data. The electrical parameter data includes line voltage, phase voltage, line current, phase current, power transmission power, grid frequency, and total electric energy. The equipment status data includes equipment operation status, equipment temperature, and equipment operation time. The network parameter data includes communication network bandwidth, network delay, communication protocol, packet loss rate, signal strength, and network load.

[0067] The methods for preprocessing the power metering related data to obtain the power metering comprehensive feature dataset include:

[0068] Identify and remove the outliers in the electrical parameter data, equipment status data, and network parameter data included in the power metering related data through the density clustering algorithm, and obtain the electrical parameter feature dataset, equipment status feature dataset, and network parameter feature dataset.

[0069] Denoise the electrical parameter feature dataset, equipment status feature dataset, and network parameter feature dataset through the Kalman filter, and perform standard deviation normalization on the denoised electrical parameter feature dataset, equipment status feature dataset, and network parameter feature dataset, and convert them into a standard normal distribution with a mean of 0 and a standard deviation of 1 to obtain the normalized electrical parameter feature dataset, equipment status feature dataset, and network parameter feature dataset. Fuse the normalized electrical parameter feature dataset, equipment status feature dataset, and network parameter feature dataset through a weighted model to obtain the power metering comprehensive feature dataset.

[0070] The methods for fusing the normalized electrical parameter feature dataset, equipment status feature dataset, and network parameter feature dataset through a weighted model to obtain the power metering comprehensive feature dataset include:

[0071] Denote the electrical parameter feature dataset as , denote the equipment status feature dataset as , denote the network parameter feature dataset as ; The weighted model is: ; among them, is the comprehensive feature dataset of electric energy metering; is the weight coefficient of the electrical parameter feature dataset; is the weight coefficient of the equipment status feature dataset; is the weight coefficient of the network parameter feature dataset.

[0072] The training method of the electric energy metering error prediction model includes:

[0073] Divide the dataset into a training set, a validation set, and a test set, and construct an electric energy metering error prediction model; the sample set is a subset of the dataset, and each sample set includes the historical comprehensive feature dataset of electric energy metering and the corresponding electric energy metering error; the electric energy metering error prediction model includes an input layer, a GRU layer, a fully connected layer, and an output layer; the input layer of the model is used to input the historical comprehensive feature dataset of electric energy metering, and the number of neurons in the input layer matches the number of features in the historical comprehensive feature dataset of electric energy metering;

[0074] Use the GRU layer to process the historical comprehensive feature dataset of electric energy metering, and adjust the number of GRU layers and the number of neurons according to the task complexity; provide additional non-linear transformation through the fully connected layer; the output layer of the model is used to output the electric energy metering error, and a neuron is used to output the predicted value, and the identity function is used as the activation function; the electric energy metering error prediction model is a gated recurrent unit model;

[0075] Use the mean absolute error as the loss function to measure the error between the predicted value and the actual value of the model; use the training set data to train the model, and minimize the loss function through the Adam optimizer; use the validation set to evaluate the performance of the model, and tune the hyperparameters of the model until the model performance no longer improves or reaches the preset stop condition;

[0076] Use the test set to evaluate the performance of the electric energy metering error prediction model in the prediction task, input the current comprehensive feature dataset of electric energy metering into the trained electric energy metering error prediction model, and obtain the electric energy metering error.

[0077] The method for judging whether error correction is needed according to the predicted electric energy metering error includes:

[0078] If the predicted electric energy metering error is less than or equal to the preset electric energy metering error threshold, it is determined that error correction is not needed;

[0079] If the predicted electric energy metering error is greater than the preset electric energy metering error threshold, it is determined that error correction is needed.

[0080] The training method of the error type recognition model includes:

[0081] The dataset is divided into a training set, a test set, and a validation set, and an error type recognition model is constructed. The error type recognition model includes an input layer, a hidden layer, and an output layer. The input layer of the model is used to input historical power metering error data, and the output layer is used to output power metering error type data. The output layer is set with neurons equal in number to the number of power metering error types, and each neuron corresponds to the prediction probability of a power metering error type. The softmax function is used as the activation function. The error type recognition model is a multi-layer perceptron (MLP) model.

[0082] The multi-class cross-entropy is used as the loss function of the model to measure the difference between the predicted value and the actual value of the model. The training set is used to train the error type recognition model, and the model parameters are updated through the backpropagation algorithm to minimize the loss function. The validation set is used to evaluate the performance of the error type recognition model by calculating the accuracy metric.

[0083] The SGD optimization algorithm is selected as the optimizer, and the model is tuned according to the performance feedback of the validation set. The model parameters are adjusted until the performance no longer improves or reaches the preset stop condition. The test set is used to evaluate the performance of the model in the prediction task, and the trained error type recognition model is used to predict the current power metering error data to obtain the power metering error type data.

[0084] The power metering error type data includes measurement deviation error, equipment failure error, environmental impact error, network communication error, algorithm error, clock synchronization error, and power grid fluctuation error.

[0085] The measurement deviation error is usually caused by problems in the design, manufacture, or calibration of the power metering equipment. The equipment may have nonlinear errors, zero drift, sensitivity errors, etc. This kind of error generally accumulates gradually during long-term use, or the calibration error of the equipment itself is not fully corrected at the time of factory shipment.

[0086] The equipment failure error is caused by hardware failures, abnormal operation, or aging of the power metering equipment itself. For example, sensor damage, loose wiring, internal circuit failures of the equipment, high temperature, etc. may all affect the accuracy of the equipment. The equipment status data is monitored, including the operating status, temperature, operating time, etc. of the equipment. When the equipment fails, the monitoring system can record information such as the failure time, type, and failure code. In addition, the failure history records and alarm logs of the equipment can also provide such data.

[0087] When the electrical energy metering device operates under different environmental conditions, environmental factors such as temperature, humidity, and vibration will affect its accuracy. For example, too high a temperature may cause drift of electronic components, and a humid environment may lead to poor electrical contact, etc. The environmental sensors are used to monitor the environmental data such as temperature, humidity, and vibration around the device, and at the same time record the working environmental conditions of the electrical energy metering device. These environmental data can be compared with the electrical energy metering data to analyze whether there is environmental impact error.

[0088] Network communication errors are usually caused by the instability or unreliability of the communication network. For example, factors such as insufficient bandwidth, excessive network latency, data packet loss, and signal attenuation may all cause data transmission errors from the electrical energy metering device to the data concentration system. By monitoring network communication parameters such as bandwidth, latency, packet loss rate, signal strength, network load, etc. data, to diagnose whether there are communication problems. These data can be collected through regular network quality inspections, network diagnostic tools, and logs.

[0089] Algorithm errors are usually caused by the assumptions or calculation biases of the algorithm itself during the processing and calibration of electrical energy metering data. For example, limitations in data preprocessing, signal filtering, distribution calibration algorithms, etc. may lead to incompletely accurate calibration results. By comparing and analyzing different algorithms, monitoring the data changes before and after algorithm processing, or through a validation dataset (such as a standard dataset) to evaluate algorithm errors. The collected data includes the input and output data of the algorithm, model parameters, and debug logs.

[0090] The timestamp of the electrical energy metering device is usually provided by a built-in clock or a synchronization system. If the clocks between multiple devices are not synchronized, or the clock drifts, it may cause timestamp errors, thus affecting the accuracy of the electrical energy metering data. By collecting the time synchronization signal of the device, monitor the clock error situation. The clock synchronization error can be detected by comparing with a standard time source (such as GPS, NTP server, etc.), and the accuracy of the clock can be monitored through clock drift and error records. By collecting the time synchronization signal of the device, monitor the clock error situation. The clock synchronization error can be detected by comparing with a standard time source (such as GPS, NTP server, etc.), and the accuracy of the clock can be monitored through clock drift and error records.

[0091] Power system problems such as frequency fluctuations, voltage instability, and harmonics in the power grid may affect the readings of power metering devices. For example, frequency fluctuations may cause power errors in the power meter, and voltage instability may lead to inaccurate voltage readings of the metering device. Monitor the fluctuations of the power grid by collecting data such as voltage, frequency, and power factor of the power grid. Specialized power monitoring equipment (such as power quality analyzers) can be used to obtain detailed data on power grid fluctuations. In addition, real-time collection of data such as current, voltage, and frequency of the power grid and comparison with power metering data can identify the impact of power grid fluctuations on metering results.

[0092] A method for correcting power metering errors using a distributed calibration algorithm based on power metering error type data includes:

[0093] S91. Denote the power metering error type data as a data set ; where is the th power metering error type data; is the index of the power metering error type data, ; is the total number of power metering error type data;

[0094] S92. Based on the power metering error type data , select the corresponding correction strategy; set the correction functions corresponding to different power metering error type data : ; where is the correction function corresponding to the th power metering error type data; is the power metering error value of the th node;

[0095] S93. Construct a distributed topology graph for each node corresponding to the power metering error type data in the power metering system through a directed graph ; where is the node set, representing the power metering nodes in the power metering system; is the edge set, representing the directed communication relationship between nodes;

[0096] Node set ; where is the th node; is the number of nodes; Edge set ; where is the th node; is the th node; Represents a node Send error correction information to adjacent nodes ; And Is the index of the node;

[0097] S94. For each node , calculate the error correction amount of the node through the error correction amount calculation formula; the error correction amount calculation formula is: ; where Is the error correction amount of the node; Is the node 's set of adjacent nodes, that is, the set of all parent nodes pointing to the node ; Is the th power of the power metering error of the adjacent node; Is the correction amount influence factor, indicating the weight factor of the influence of the power metering error of the adjacent node on the correction amount of the node ; Is the weight factor of the correction function ; Is the index of the adjacent node;

[0098] S95. Dynamically adjust the correction amount influence factor through the correction amount influence factor adjustment formula. The correction amount influence factor adjustment formula is: ; where Is the number of adjacent nodes of the node ; Is a constant that controls the size of the overall weight factor; Is the degree of influence of the number of adjacent nodes on the weight factor;

[0099] As the system scale increases (that is, the total number of nodes increases), the influence of each individual node on the global error correction usually decreases. Because the information transfer between nodes is carried out through adjacent nodes, as the system scale increases, the error correction effect of local nodes will be interfered by more nodes. Therefore, appears in the denominator, meaning that the larger the system scale, the smaller the value, and the correction weight of local nodes is also weakened accordingly, avoiding the excessive influence of the error correction of a single node on the overall system in a large-scale system and ensuring global consistency and robustness; the number of adjacent nodes of each node reflects the connection strength of the node in the network. If the number of adjacent nodes of the node is large, it means that the node may exchange information with more other nodes. Therefore, the error correction of this node should not be determined only by itself, but the influence of adjacent nodes should be considered.

[0100] For example, the total number of nodes is 10, and the number of adjacent nodes of a node is 3. The number of adjacent nodes of other nodes in the power metering system may be different. Suppose a node is connected to 5 adjacent nodes; the constant controlling the size of the overall weight factor is 10, and the degree of influence of the number of adjacent nodes on the weight factor is 2;

[0101] Calculate the weight factor of node ; Therefore, the weight factor of node is ; ;

[0102] Calculate the weight factor of node ; Therefore, the weight factor of node is ; ;

[0103] The number of adjacent nodes of node is small ( = 3), so its weight factor is 0.625, which is relatively large. This means that the error correction of node will rely more on the errors of its adjacent nodes. The number of adjacent nodes of node is large ( = 5), so its weight factor is 0.5, which is relatively small, indicating that node has a relatively small influence on the error correction of adjacent nodes.

[0104] S96. Constrain and limit the weight factor of the correction function through the error sensitivity adjustment formula. The error sensitivity adjustment formula is: ; where is the weight factor of the corrected function after constraint; is the maximum value of the power metering errors of all nodes in the power metering system; is the factor controlling the influence of the error magnitude on the weight factor ;

[0105] When the error of a certain node is large, it means that there may be more serious problems in the system (such as equipment failure, system deviation, etc.). In this case, the influence of this node should be greater, so its correction weight Increase, so that the correction function can correct it more powerfully. For nodes with smaller errors, the intensity of correction should be relatively reduced because these nodes may already be relatively accurate, and strong correction may lead to overcorrection. Therefore, will be smaller to avoid unnecessary correction; the dynamic adjustment mechanism can adaptively reflect the error situation of each node in the system, ensuring that important nodes (nodes with large errors) receive more correction without excessive interference to nodes with smaller errors.

[0106] For example, assume there is an electric energy metering system composed of 5 electric energy metering nodes, and the electric energy metering errors of each node are as follows:

[0107] Node : = 5 (error value is 5);

[0108] Node : = 3 (error value is 3);

[0109] Node : = 7 (error value is 7);

[0110] Node : = 4 (error value is 4);

[0111] Node : = 6 (error value is 6);

[0112] The weight factor of the correction function is 1, and the factor controlling the influence of the error magnitude on the weight factor is 0.5, and the maximum value of all node errors in the electric energy metering system

[0113] Calculate the weight factor of the correction function after each limit:

[0114] ;

[0115] ;

[0116] ;

[0117] ;

[0118] .

[0119] A method for transmitting the power metering error correction result to other nodes through distributed topology and PLC communication to make the power metering of all nodes consistent includes:

[0120] Through PLC communication, the corrected power metering error of each node is transmitted to its adjacent nodes according to the signal transmission formula, and the preset node The preliminary correction error is , and the signal transmission formula is: ; where is the preliminary correction error of node ;

[0121] In each round of correction, node sends its preliminary correction error to its downstream node, that is , and at the same time node receives the correction error from the upstream node and updates its own power metering error according to the error correction information formula; the error correction information formula is: ; where is the power metering error of the updated node ; is the power metering error of node before receiving the error correction information;

[0122] As the error correction information propagates in the power metering system, while node updates its own power metering error according to the error correction information from other nodes, the global error consistency is gradually achieved through the global error correction formula; the global error correction formula is: ; where is the global error value of node , representing the power metering error after information propagation and error correction;

[0123] For example, assuming there are 3 nodes in the power metering system, and the power metering errors of nodes , and are , , 3 respectively, and these error values become , , and after correction;

[0124] Then, the global error value of node is: ; this means that node After considering the error information corrected by other nodes, its global error value is updated to 5.17.

[0125] All nodes continue to synchronize errors through the error correction synchronization formula, and finally the power metering errors of all nodes tend to be consistent; the error correction synchronization formula is: ; where is the final corrected error of node ; is the correction factor that adjusts the size of the final correction amount; it determines the influence degree of the difference between the preliminary correction error and the global error. The larger the value, the smaller the difference between the final correction error and the global error; the smaller the value, the more the correction of the node will depend on its own correction result.

[0126] The correction factor that adjusts the size of the final correction amount is dynamically adjusted and designed through the correction factor adaptive formula. The correction factor adaptive formula is: ; where is the correction factor that adjusts the size of the final correction amount at the th iteration; is the coefficient that affects the influence degree of the error mean on the correction factor. The larger the control error mean, the stronger the adjustment of the correction factor; is the mean value of the power metering errors of all nodes in the system; is the attenuation factor that affects the number of iterations in the correction process, controlling the reduction speed of the correction factor during the iteration process; is the current number of iterations; is the maximum number of iterations.

[0127] Combining the error mean and the iteration attenuation mechanism, the whole formula realizes adaptive adjustment. This adaptive mechanism can adjust the value of the correction factor according to the current system error situation and the progress of the correction. When the error is large, the correction factor will increase to strengthen the correction process. As the number of iterations increases, the correction factor will gradually decrease to avoid overcorrection and system oscillation. In this way, the correction factor will not be too strong throughout the process, avoiding the risk of overcorrection. Instead, it will correct larger errors in the early stage and gradually decrease in the later stage to ensure the gradual convergence of the system;

[0128] For example, assume that there are 5 nodes in the power metering system, and the power metering errors of each node have different values at different time points. The correction factor that adjusts the size of the final correction amount is 0.5, the total maximum error of the power metering system is 100, the maximum number of iterations is 10 times, the coefficient that affects the influence degree of the error mean on the correction factor is 0.1, and the attenuation factor is 0.05;

[0129] Suppose the initial power metering errors of 5 nodes in the power metering system are as follows:

[0130] Node has an error of ;

[0131] Node has an error of ;

[0132] Node has an error of ;

[0133] Node has an error of ;

[0134] Node has an error of ;

[0135] The mean value of the power metering errors of all nodes ;

[0136] Calculate the correction factor for adjusting the size of the final correction amount:

[0137] The first round of correction ( ),

[0138] ;

[0139] The second round of correction ( ),

[0140] Suppose after the first round of correction, the node errors become:

[0141] Node has an error of ;

[0142] Node has an error of ;

[0143] Node has an error of ;

[0144] Node has an error of ;

[0145] Node has an error of ;

[0146] The mean value of the power metering errors of all nodes ;

[0147] Recalculate the correction factor that adjusts the magnitude of the final correction amount: ; At this time, the correction factor increases, indicating that the power metering system error is relatively large and the correction intensity is stronger.

[0148] The preset power metering error threshold is set by the staff. Different power metering errors are collected through the power metering control terminal, and the average value of multiple power metering errors is taken as the preset power metering error threshold.

[0149] In this embodiment, by collecting the power metering error data of all nodes into a set, the errors of the entire system can be uniformly managed and analyzed; it provides a basis for subsequent data processing and analysis, ensuring that the errors of each node can be independently identified and corrected; based on the error type, the corresponding correction strategy is selected, the correction functions for data of different error types are set, and the correction is performed according to the power metering error value of the node; appropriate correction functions are defined for each node, making the correction process more accurate and effective;

[0150] Convert the power metering system into a graph structure so that the correction information of each node can be propagated through the directed edges in the graph, promoting the mutual influence between nodes; the directed graph provides a structured way to track the error propagation paths of each node and its adjacent nodes, improving the efficiency and accuracy of error correction; the correction amount calculation formula can dynamically adjust the correction amount of the target node according to the error values of adjacent nodes, ensuring that the correction strategy of the entire system can be adjusted according to the actual situation; by calculating the correction amount of each node, the error of each node is gradually reduced, improving the overall accuracy of the system; the influence factor of the number of adjacent nodes can be dynamically adjusted according to the system topology structure, ensuring that under different network environments, the adjustment of the correction factor can reflect the interaction between nodes in the network; the dynamic adjustment of the correction amount influence factor helps to balance the stability of the system and avoid oscillations or unnecessary corrections caused by excessive correction of some nodes; through error sensitivity adjustment, it is possible to avoid the power metering system from being over-corrected during correction, resulting in an unstable situation of the system;

[0151] Each node transmits the preliminarily corrected power metering error to adjacent nodes through PLC communication. This process ensures that the correction information of each node can be propagated throughout the system, thereby affecting the error correction of other nodes; through this information transmission, the system can gradually converge to global error consistency; in each round of correction, after receiving the error correction information from the upstream node, the node updates its own power metering error; this mechanism ensures the gradual adjustment and correction of the errors of each node in the system, avoiding the locality problem of single-point correction; through the mutual correction between nodes, it is ensured that the errors of the entire power metering system gradually tend to be consistent; through PLC signal transmission, the errors of all nodes are effectively synchronized, reducing error accumulation or local deviation; through multiple rounds of information dissemination and update, each node can be updated according to the correction information of other nodes in the system, rather than simply relying on its own preliminary correction, avoiding the amplification effect of the errors of isolated nodes; this method is distributed and there is no single control center, so it can avoid the impact of single-point failures and improve the reliability and robustness of the system;

[0152] Based on information transmission and node update, each node gradually makes the error value approach global consistency through the global error correction formula. This ensures that as the correction progresses, the system error can be globally adjusted, thereby improving the overall accuracy of the system; through the error correction synchronization formula, the final correction errors of each node tend to be consistent, thus achieving the goal of error synchronization; finally, the power metering errors of all nodes tend to be consistent, ensuring the overall accuracy of the system; according to the average value of the power metering errors of all nodes in the system and the number of iterations, the correction factor for each iteration is dynamically adjusted; this adjustment mechanism takes into account factors such as the distribution of errors and the attenuation of the correction factor, making the error correction process more flexible and efficient.

[0153] Embodiment 2

[0154] Please refer to Figure 2 as shown. For the parts not described in detail in this embodiment, refer to the description content of Embodiment 1. A system for maintaining power metering is provided, including:

[0155] A data acquisition module for acquiring power metering related data;

[0156] A data processing module for preprocessing the power metering related data to obtain a power metering comprehensive feature data set;

[0157] An error prediction module for training a power metering error prediction model based on the power metering comprehensive feature data set and predicting the power metering error based on the power metering error prediction model;

[0158] A calibration evaluation module, which is used to judge whether error correction is required according to the predicted power metering error; if error correction is required, it generates a warning message and automatically collects power metering error data;

[0159] An error calibration module, which is used to train and obtain an error type recognition model according to the power metering error data, predict the power metering error type data based on the error type recognition model; based on the power metering error type data, use the distribution calibration algorithm to correct the power metering error, and transmit the power metering error correction result to other nodes through distributed topology and PLC communication, so that the power metering of all nodes is consistent;

[0160] A feedback verification module, which is used to verify all power metering error correction results. If there are still errors, the power metering control terminal issues a secondary correction instruction until the corrected power metering error reaches a preset power metering error threshold and then stops; each module is connected by wired and / or wireless means.

[0161] Since the electronic device introduced in this embodiment is the electronic device used to implement a method for maintaining power metering in an embodiment of the present application, based on the method for maintaining power metering introduced in the embodiment of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device realizes the method in the embodiment of the present application will not be described in detail here. As long as the electronic device used by those skilled in the art to implement the method for maintaining power metering in the embodiment of the present application belongs to the scope protected by the present application.

[0162] The above formulas are all calculated by taking the numerical value after removing the dimension. The formula is a formula obtained by software simulation of collecting a large amount of data to be closest to the real situation. The preset parameters and threshold selection in the formula are set by those skilled in the art according to the actual situation.

[0163] The above is only the preferred implementation manner of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the idea of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in the technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for maintaining electricity metering, characterized in that, Including: S1. Collect power metering related data; S2. Preprocess the power metering related data to obtain a power metering comprehensive feature dataset; S3. Train a power metering error prediction model based on the power metering comprehensive feature dataset, and predict the power metering error based on the power metering error prediction model; S4. Determine whether error correction is required according to the predicted power metering error; if error correction is required, generate a warning message and automatically collect power metering error data; S5. Train an error type identification model based on the power metering error data, predict the power metering error type data based on the error type identification model; based on the power metering error type data, use a distribution calibration algorithm to correct the power metering error, and transmit the power metering error correction result to other nodes through distributed topology and PLC communication, so that the power metering of all nodes is consistent; The method of using a distribution calibration algorithm to correct the power metering error based on the power metering error type data includes: S91. Denote the power metering error type data as the data set y = {y1, y2,..., y j ,..., y m}; where y j is the j-th power metering error type data; j is the index of the power metering error type data, j ∈ {1, 2,..., m}; m is the total number of power metering error type data; S92. Based on the power metering error type data y j , select the corresponding correction strategy; set the correction functions corresponding to different power metering error type data Among them, is the correction function corresponding to the j-th power metering error type data; e j is the power metering error value of the j-th node; S93. Construct each node corresponding to the power metering error type data in the power metering system into a distributed topology graph G=(V, E) through a directed graph; where V is the node set, representing N power metering nodes in the power metering system; E is the edge set, representing the directed communication relationship between nodes; S94. For each node v j , calculate the error correction amount of this node through the error correction amount calculation formula; the error correction amount calculation formula is: where Δe j is the error correction amount of the node; N j is the set of adjacent nodes of node v j , that is, the set of all parent nodes pointing to node v j ; e k is the power metering error of the k-th adjacent node; α is the correction amount influence factor; β is the weight factor of the correction function ; k is the index of the adjacent node; S95. Dynamically adjust and design the correction factor influence factor through the correction factor influence factor adjustment formula; the correction factor influence factor adjustment formula is: where Z j is the number of adjacent nodes of node v j ; c is a constant that controls the magnitude of the overall weight factor; d is the degree of influence of the number of adjacent nodes on the weight factor. S96. Constraining and limiting the weight factor of the calibration function through the error sensitivity adjustment formula; S6. Verify all the power metering error correction results. If there are still errors, the power metering control terminal issues a secondary correction instruction until the corrected power metering error reaches a preset power metering error threshold and then stops.

2. The method for maintaining power metering according to claim 1, characterized in that, The power metering related data includes electrical parameter data, equipment status data and network parameter data; the electrical parameter data includes line voltage, phase voltage, line current, phase current, power transmission power, grid frequency and total power; the equipment status data includes equipment operation status, equipment temperature and equipment operation time; the network parameter data includes communication network bandwidth, network delay, communication protocol, packet loss rate, signal strength, network load.

3. The method for maintaining power metering according to claim 2, characterized in that, The method of preprocessing the power metering related data to obtain a power metering comprehensive feature dataset includes: Identify and remove outliers in the electrical parameter data, equipment status data and network parameter data included in the power metering related data through a density clustering algorithm to obtain an electrical parameter feature dataset, an equipment status feature dataset and a network parameter feature dataset; Denoise the electrical parameter feature dataset, equipment status feature dataset and network parameter feature dataset through Kalman filtering, and perform standard deviation normalization processing on the denoised electrical parameter feature dataset, equipment status feature dataset and network parameter feature dataset, and convert them into a standard normal distribution with a mean of 0 and a standard deviation of 1 to obtain a normalized electrical parameter feature dataset, an equipment status feature dataset and a network parameter feature dataset; fuse the normalized electrical parameter feature dataset, equipment status feature dataset and network parameter feature dataset through a weighted model to obtain a power metering comprehensive feature dataset.

4. The method for maintaining power metering according to claim 3, characterized in that, The method of fusing the normalized electrical parameter feature dataset, device status feature dataset, and network parameter feature dataset through a weighted model to obtain a comprehensive power metering feature dataset includes: Denote the electrical parameter feature dataset as E1, the device status feature dataset as E2, and the network parameter feature dataset as E3; the weighted model is: RL = E1·δ1 + E2·δ2 + E3·δ3; where RL is the comprehensive power metering feature dataset; δ1 is the weight coefficient of the electrical parameter feature dataset; δ2 is the weight coefficient of the device status feature dataset; δ3 is the weight coefficient of the network parameter feature dataset.

5. The method for maintaining power metering according to claim 4, characterized in that, The training method of the power metering error prediction model includes: Divide the dataset into a training set, a validation set, and a test set, and construct a power metering error prediction model; the sample set is a subset of the dataset, and each sample set includes a historical comprehensive power metering feature dataset and the corresponding power metering error; the power metering error prediction model includes an input layer, a GRU layer, a fully connected layer, and an output layer; the input layer of the model is used to input the historical comprehensive power metering feature dataset, and the number of neurons in the input layer matches the number of features in the historical comprehensive power metering feature dataset; Use the GRU layer to process the historical comprehensive power metering feature dataset, and adjust the number of GRU layers and the number of neurons according to the task complexity; provide additional non-linear transformation through the fully connected layer; the output layer of the model is used to output the power metering error, and a single neuron is used to output the predicted value, and the identity function is used as the activation function; the power metering error prediction model is a gated recurrent unit model; Use the mean absolute error as the loss function to measure the error between the predicted value and the actual value of the model; use the training set data to train the model, and minimize the loss function through the Adam optimizer; use the validation set to evaluate the performance of the model, and tune the hyperparameters of the model until the model performance no longer improves or reaches the preset stop condition and then stop; Use the test set to evaluate the performance of the power metering error prediction model in the prediction task, input the current comprehensive power metering feature dataset into the trained power metering error prediction model, and obtain the power metering error.

6. The method for maintaining power metering according to claim 5, characterized in that, The method of judging whether error correction is needed according to the predicted power metering error includes: If the predicted power metering error is less than or equal to the preset power metering error threshold, it is determined that error correction is not needed; If the predicted power metering error is greater than the preset power metering error threshold, it is determined that error correction is needed.

7. The method for maintaining power metering according to claim 6, wherein The training method of the error type recognition model includes: Divide the dataset into a training set, a test set, and a validation set, and construct an error type recognition model. The error type recognition model includes an input layer, a hidden layer, and an output layer. The input layer of the model is used to input historical power metering error data, and the output layer is used to output power metering error type data. The output layer is set with neurons equal in number to the number of power metering error types, and each neuron corresponds to the prediction probability of a power metering error type. The softmax function is used as the activation function. The error type recognition model is a multi-layer perceptron (MLP) model. Use multi-class cross-entropy as the loss function of the model to measure the difference between the predicted value and the actual value of the model. Use the training set to train the error type recognition model, and update the model parameters through the backpropagation algorithm to minimize the loss function. Use the validation set to evaluate the performance of the error type recognition model by calculating the accuracy metric. Select the SGD optimization algorithm as the optimizer, and tune the model according to the performance feedback of the validation set. Adjust the model parameters until the performance no longer improves or reaches the preset stopping condition and then stop. Use the test set to evaluate the performance of the model in the prediction task, and use the trained error type recognition model to predict the current power metering error data to obtain the power metering error type data.

8. The method for maintaining power metering according to claim 7, wherein The power metering error type data includes measurement deviation error, equipment failure error, environmental impact error, network communication error, algorithm error, clock synchronization error, and power grid fluctuation error.

9. The method for maintaining power metering according to claim 8, wherein The method for correcting the power metering error by using the distribution calibration algorithm based on the power metering error type data further includes: The set of nodes \(V = \{v_1, v_2, \ldots, v N \}\); where \(v N \) is the \(N\)th node; \(N\) is the number of nodes; the set of edges \(E=\{v j , v k \in V, v j \to v k \}\); where \(v j \) is the \(j\)th node; \(v k \) is the \(k\)th node; \(v j \to v k \) means that the node \(v j \) sends error correction information to the adjacent node \(v k \); \(j\) and \(k\) are the indexes of the nodes; the error sensitivity adjustment formula is: where \(\beta'\) is the weight factor of the restricted correction function ; \(\max(e)\) is the maximum value of the power measurement errors of all nodes in the power measurement system; \(\lambda\) is the factor that controls the influence of the error magnitude on the weight factor \(\beta\).

10. The method for maintaining power metering according to claim 9, wherein, The method for transmitting the power metering error correction result to other nodes through the distributed topology and PLC communication to make the power metering of all nodes consistent includes: Through PLC communication, the corrected power metering error of each node is transmitted to its adjacent nodes according to the signal transmission formula, and the preset node v j The preliminary correction error is The signal transmission formula is: Among them, is the preliminary correction error of node v j ; In each round of calibration, node v j sends its preliminary calibration error to its downstream node, i.e., v j →v k . At the same time, node v j receives the calibration error from the upstream node and updates its own power metering error according to the error calibration information formula; the error calibration information formula is: where, is the power metering error of node v j after update; is the power metering error of node v j before receiving the error calibration information; As the error correction information propagates in the power metering system, node v j While updating its own power metering error according to the error correction information from other nodes, gradually achieves the consistency of the global error through the global error correction formula; the global error correction formula is: Wherein, Is the global error value of node v j ; All nodes continue to synchronize the error through the error correction synchronization formula, and finally the power measurement errors of all nodes tend to be consistent; the error correction synchronization formula is: Among them, is the final correction error of node v j ; γ is the correction factor for adjusting the magnitude of the final correction amount; The correction factor for adjusting the magnitude of the final correction amount is dynamically adjusted and designed through the correction factor self - adaptation formula; the correction factor self - adaptation formula is: where γ t+1 is the correction factor for adjusting the magnitude of the final correction amount at the (t + 1)-th iteration; γ t represents the correction factor for adjusting the magnitude of the final correction amount at the t - th iteration; ω is the coefficient affecting the influence degree of the error mean value on the correction factor; is the mean value of the power metering errors of all nodes in the system; η is the attenuation factor affecting the number of iterations in the correction process; t is the current number of iterations; T is the maximum number of iterations.

Images