An electric energy meter informationization evaluation calibration model based on district electric energy conservation

Abstract

The application discloses a kind of based on the informationization evaluation calibration model of electric energy meter of district electric energy conservation.The present application is based on the principle of district electric energy conservation, realizes hierarchical classification precision decoupling measurement acquisition and multilayer node collaborative calculation, constructs the mathematical model of non-convex sparse multi-objective optimization, proposes global adaptive optimization algorithm considering dynamic and static data, and solves the district line loss and electric energy meter evaluation calibration result in a coordinated manner.

Description

Technical Field

[0001] This invention belongs to the field of power supply management technology, and in particular relates to an information-based evaluation and calibration model for electricity meters based on the conservation of electrical energy in transformer substations. Background Technology

[0002] Currently, the meter condition assessment technology based on the energy conservation of distribution areas has been widely adopted by power companies, and these companies have deployed and applied meter condition assessment models, achieving good results in improving meter operation and maintenance. However, the meter condition assessment technology still has the following problems: In terms of meter assessment model construction, the current models lack quality prediction and early warning based on multi-dimensional data, and lack hierarchical and accurate calculation of energy loss. Existing assessment models do not comprehensively analyze and evaluate the meter condition and predict its trend based on multi-dimensional data analysis of historical, current performance, and causal reasoning; they only consider the current state of the meter in a simple and isolated way, which limits the scientific validity of the assessment results and their guidance for future operation and use. Therefore, it is necessary to accurately establish a physical model of the actual electrical energy of the distribution area to achieve accurate measurement of key parameters through hierarchical and graded decoupling of measurement and acquisition and multi-level node collaborative calculation of parameters.

[0003] Non-convex sparse optimization problems refer to optimization problems where feasible solutions are sparse. These problems typically exhibit characteristics such as non-convexity and discontinuity, and have NP complexity. Currently, no domestic or international scholars have constructed non-convex sparse optimization models using the sparsity of electricity meter errors. Summary of the Invention

[0004] The purpose of this invention is to provide an information-based evaluation and calibration model for electricity meters based on the principle of energy conservation in transformer substations. Based on this principle, the invention studies hierarchical and graded precise decoupling measurement and multi-layer node collaborative computing technologies to construct a physical model for information-based evaluation and calibration of electricity meters. It also studies hybrid network communication technology to achieve efficient high-frequency data compression and transmission and automatic clock synchronization, improving the real-time performance and time-stamp accuracy of high-frequency data acquisition and transmission. Using the physical mechanism model as an equation constraint, and based on generative adversarial network data quality enhancement and non-convex sparse multi-objective optimization techniques, a mathematical model for electricity meter evaluation and calibration is constructed. Finally, the invention studies a global adaptive optimization algorithm that considers prior data such as initial inspection errors to achieve a collaborative and coupled solution of transformer substation line losses and electricity meter evaluation and calibration results.

[0005] To achieve the objectives of this invention, the technical solution provided by this invention is as follows:

[0006] An information-based evaluation and calibration model for electricity meters based on the conservation of electrical energy in transformer substations is constructed as follows:

[0007] Step 1: By using a hierarchical and graded method based on the principle of energy conservation and a precise decoupling method for transformer area line loss, the impact of transformer area line loss measurement error on the model calculation results is reduced, ensuring that the uncertainty of the model calculation results meets the requirements of the specific field; by using the multi-level node collaborative calculation technology of energy meter information evaluation and calibration, a top-down online dynamic error calibration method is constructed to evaluate and correct the operating status of the total node at different levels in real time.

[0008] Step 2: Utilize HPLC-based hybrid networking technology to achieve effective compression and efficient transmission of high-frequency data in the electricity meter box; utilize HPLC-based NTB synchronization technology to achieve automatic clock synchronization; ensure the efficiency and accuracy of electricity meter data acquisition.

[0009] Step 3: Using a non-convex sparse multi-objective optimization model construction method with the energy meter error as the variable and data priority, system stability, and energy conservation as multiple optimization objectives, the multi-objective optimization model is simplified and relaxed using mathematical techniques, and an efficient numerical optimization algorithm is used to achieve accurate estimation of the energy meter error.

[0010] Step 4: Using a multivariate optimization model construction method with transformer area line loss and electricity meter operating error as variables, design a local jump-out mechanism and hyperparameter adaptive selection technology, and design a global adaptive algorithm based on alternating direction multiplier method and block coordinate descent method to achieve coupled solution of transformer area line loss and electricity meter operating error.

[0011] Specifically, step 1 includes the following:

[0012] Step 1.1: By analyzing the typical structure of existing low-voltage distribution areas, a hierarchical structure model suitable for typical power consumption scenarios such as residential buildings is described using JP cabinets, box-type substations, distribution cabinets, branch boxes, and metering boxes as basic elements.

[0013] Step 1.2: Based on different hierarchical models, and combined with key measurement data such as time-series load voltage, current, and power, establish loss calculation models applicable to different levels. The problem of estimating line loss by the difference between the total node and the sum of each sub-node under the general case is transformed into loss calculation models with specific physical meanings under different levels, so as to achieve accurate decoupling of loss.

[0014] Step 1.3: Through multi-level node collaborative calculation, the total node error and uncertainty at different levels are transmitted and traced step by step to the end node and the total node, further optimizing the uncertainty caused by the total node error at different levels, and realizing the information-based evaluation and calibration of the electricity meter.

[0015] Specifically, step 2 includes the following:

[0016] Step 2.1: Utilize HPLC-based hybrid networking technology to achieve physical topology identification of the transformer substation. By superimposing characteristic current signals on HPLC data, physical topology identification is performed under hybrid networking mode. When the transformer-side management terminal calls the meter box-side metering modules one by one, the meter box-side metering modules generate characteristic current signals sequentially after receiving the call information. After receiving the signals, the branch-side metering modules record the information and forward the characteristic signals to the management terminal, thereby constructing a physical topology model.

[0017] Step 2.2: Utilizing HPLC-based NTB synchronization and data synchronization acquisition technology, firstly, the management terminal on the transformer side acts as the CCO, and the metering modules on the transformer branch and meter box sides act as STA modules. The CCO module sends broadcast time synchronization commands to the STA modules on the branch and meter box sides to achieve precise clock synchronization. The metering modules on the branch and meter box sides are checked one by one. If there is a clock deviation at the second level, the metering module with the clock deviation is re-calibrated. Secondly, research is conducted on synchronous sampling technology based on HPLC characteristic signals. When the voltage is close to zero, the voltage characteristic signal is superimposed to represent the information. Within the voltage cycle, the metering modules at each level start synchronous sampling after receiving the voltage characteristic signal.

[0018] Step 2.3: Utilize high-frequency data compression transmission technology based on hybrid networking to achieve efficient data transmission to the management terminal and main station for analysis and calculation. Firstly, research is conducted on high-frequency data compression transmission based on hybrid networking technology. This involves using numbers instead of strings and constructing a Huffman binary tree based on the frequency of character occurrences. Data appearing more frequently is placed at the upper level of the tree, and data appearing less frequently is placed at the lower level. The path from the root node to each data point is encoded and lossless compression is achieved, thus enabling efficient data transmission based on the physical topology. Secondly, synchronous sampling technology is used to sample electrical parameter curve data at the same time point.

[0019] Specifically, step 3 includes the following:

[0020] Step 3.1: Utilize the non-convex sparse multi-objective optimization mathematical model construction method for electricity meters to accurately map the electricity meter information evaluation and calibration problem; based on the principle of energy conservation in distribution areas, construct a multi-objective optimization model with electricity meter error as the variable and multiple optimization objectives such as data priority, system stability, and energy conservation. A metric function with non-convex and sparse mathematical properties is used to characterize the data priority objective; a metric function with non-smooth mathematical properties is used to characterize the system stability objective; and a metric function with convex and smooth mathematical properties is used to characterize the energy conservation objective. Finally, a non-convex sparse multi-objective optimization model that accurately maps the electricity meter information evaluation and calibration problem is constructed.

[0021] Step 3.2: Utilize effective relaxation methods for the non-convex sparse multi-objective optimization mathematical model of the electricity meter to reasonably simplify the original problem; considering the non-convex, non-smooth, and multi-objective mathematical characteristics of the aforementioned non-convex sparse multi-objective optimization model, study various convexification methods for non-convex objectives and the corresponding convex approximation theory to characterize the error between the solution to the original problem and the solution to the relaxed problem; utilize smoothing methods for non-smooth sparse objectives and deep learning methods for various smoothing model parameters to characterize the error between the solution to the original problem and the solution to the relaxed problem; based on existing scalarization methods, design an autonomous learning strategy for the parameters between various optimization objectives and develop new scalarization methods;

[0022] Step 3.3: Utilize an efficient adaptive algorithm based on the non-convex sparse multi-objective optimization mathematical model of the energy meter to achieve accurate identification of out-of-tolerance energy meters; for the special sparse structure of the model, combine machine learning techniques to develop an efficient solution algorithm adapted to multi-objective optimization problems with special sparse structures, including drawing on the computational experience of existing global optimization algorithms to design a local escape mechanism for the global adaptive optimization algorithm for non-convex multi-objective optimization problems, seeking better local or global optimal solutions; improve the accuracy and efficiency of the algorithm by designing the search direction and step size using approximation algorithms, conjugate gradient algorithms, and trust region algorithms; further conduct theoretical analysis of the algorithm's convergence based on existing theoretical analysis results.

[0023] Specifically, step 4 includes the following:

[0024] Step 4.1: Utilizing a dual-driven method of dynamic and static data and mathematical-physical models, this method accurately describes the nonlinear functional relationship between power consumption, voltage, current, transformer topology, line parameters, dynamic and static data, and transformer line loss. Using real-time dynamic data of voltage and current, and static data of voltage impedance, current impedance, and line parameters, the method calculates the voltage and current losses of the current transformer area using a physical model based on the power formula. Combined with the calculated energy meter error data, machine learning techniques are employed to study the nonlinear functional relationship between voltage loss, current loss, energy meter error, power supply / consumption difference, transformer topology data, and transformer line loss, thereby achieving accurate estimation of transformer line loss.

[0025] Step 4.2: Using a multivariate optimization model design method with transformer substation line loss and electricity meter operating error as variables, combined with a non-convex sparse optimization model for electricity meter error estimation and a machine learning estimation model for transformer substation line loss, a multivariate coupled optimization model for transformer substation line loss and electricity meter operating error is constructed.

[0026] Step 4.3: Utilize a separable alternating iterative algorithm to collaboratively and coupledly solve for transformer line loss and electricity meter operating errors; for the multivariate optimization model, utilize its augmented Lagrangian function, gradient function, and dual function; leverage the separability of the objective function to decompose the original problem into several optimization subproblems; construct a distributed alternating direction multiplier method and a distributed block coordinate descent method to solve the optimization subproblems concerning transformer line loss and electricity meter errors in parallel; construct an adaptive shutdown criterion to accelerate the algorithm's convergence speed, and analyze the algorithm's computational complexity and convergence.

[0027] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0028] 1. This invention proposes for the first time an evaluation and calibration method for electricity meters based on non-convex sparse multi-objective optimization. Considering the objective fact that most electricity meters are in good operating condition, it innovatively proposes to use the sparsity of electricity meter operating error as the objective function, and to use the difference between the current error and the historical error as the objective function to reduce the volatility of the solution. Based on this, a non-convex sparse multi-objective optimization model is constructed to accurately map the electricity meter evaluation and calibration problem.

[0029] 2. This invention proposes for the first time an optimized model for the coupled solution of transformer substation line loss and electricity meter operating error. It suggests a research approach using dynamic and static data and a physical model to calculate transformer substation line loss, and constructs a multivariate optimization model coupling electricity meter error and transformer substation line loss. A numerical iterative algorithm is designed to alternately solve for line loss and error step by step. Solving this model not only helps in accurately estimating electricity meter error but also aids in calculating transformer substation line loss. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the energy meter information evaluation and calibration model based on the conservation of electrical energy in the distribution area, as proposed in this invention.

[0031] Figure 2 This is a schematic diagram of the non-convex sparse multi-objective optimization mathematical model for energy meter evaluation and calibration in this invention.

[0032] Figure 3 This is a schematic diagram of the coupling optimization model of line loss in the distribution area and operation error of the electricity meter in this invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0034] like Figure 1As shown in the figure, this embodiment provides an information-based evaluation and calibration model for electricity meters based on the conservation of electrical energy in distribution areas, including:

[0035] First, by employing a hierarchical and graded method based on the principle of energy conservation and a precise decoupling method for transformer substation line loss, the impact of transformer substation line loss measurement errors on model calculation results is reduced, ensuring that the uncertainty of model calculation results meets the requirements of specific fields. Second, by utilizing multi-level node collaborative computing technology for information-based evaluation and calibration of electricity meters, a top-down online dynamic error calibration method is constructed to evaluate and correct the operating status of the total node at different levels in real time.

[0036] It should be noted that the goal of this phase of work is to construct a physical mechanism model under the hierarchical structure of electricity meter information calibration, eliminating or reducing the absolute value and volatility of distribution area line loss rate at the physical level, thus providing strong theoretical support for subsequent model solving. Specifically, this includes:

[0037] By analyzing the typical structure of existing low-voltage distribution areas, a hierarchical structure model suitable for typical power consumption scenarios such as residential buildings is described using JP cabinets, box-type substations, distribution cabinets, branch (connection) boxes, and metering boxes as basic elements.

[0038] Based on different hierarchical models, and combined with key measurement data such as time-series load voltage, current, and power, loss calculation models applicable to different levels are established. The problem of estimating line loss by the difference between the total node and the sum of each sub-node under normal circumstances is transformed into loss calculation models with specific physical meanings under different levels, thereby achieving accurate decoupling of losses.

[0039] Through multi-level node collaborative computing, the total node error and uncertainty at different levels are transmitted and traced step by step to the end node and the total node, further optimizing the uncertainty caused by the total node error at different levels, and realizing the information-based evaluation and calibration of electricity meters.

[0040] Second, by utilizing HPLC-based hybrid networking technology, effective compression and efficient transmission of high-frequency data in the electricity meter box are achieved; and based on HPLC-based NTB synchronization technology, automatic clock synchronization is realized; ensuring the efficiency and accuracy of electricity meter data acquisition.

[0041] It should be noted that the specifics include the following:

[0042] By utilizing HPLC-based hybrid networking technology, physical topology identification of transformer substations is achieved. This is accomplished by superimposing characteristic current signals onto the HPLC data, enabling physical topology identification in a hybrid networking mode. When the transformer-side management terminal calls the meter box-side metering modules one by one, the meter box-side metering modules receive the call information and sequentially generate characteristic current signals. Upon receiving the signals, the branch-side metering modules record the information and forward the characteristic signals to the management terminal, thereby constructing a physical topology model.

[0043] Utilizing HPLC-based NTB synchronization and data synchronization acquisition technology, the following two approaches are employed: First, the management terminal on the transformer side acts as the Control Center (CCO), while the metering modules on the transformer branch and meter box sides act as Stabilizer (STA) modules. The CCO module sends broadcast time synchronization commands to the STA modules on the branch and meter box sides for precise clock synchronization. Each metering module on the branch and meter box sides is individually checked; if a second-level clock deviation exists, the deviating metering module is recalibrated. Second, research is conducted on HPLC-based characteristic signal synchronization sampling technology. When the voltage approaches zero, a voltage characteristic signal is superimposed to represent the information. Within the voltage cycle, each metering module begins synchronous sampling upon receiving the voltage characteristic signal.

[0044] This study utilizes high-frequency data compression transmission technology based on hybrid networking to achieve efficient data transmission to management terminals and the main station for analysis and computation. The research focuses on two main aspects: First, high-frequency data compression transmission is achieved using hybrid networking technology. This involves replacing strings with numbers and constructing a Huffman binary tree based on the frequency of character occurrences. Data appearing more frequently is placed at the upper levels of the tree, and less frequently appearing data at the lower levels. The path from the root node to each data point is encoded and lossless compression is achieved, thus enabling efficient data transmission based on the physical topology. Second, synchronous sampling technology is used to sample electrical parameter curves at the same time point.

[0045] Third, such as Figure 2 As shown, a non-convex sparse multi-objective optimization model is constructed using the energy meter error as the variable and data priority, system stability, and energy conservation as multiple optimization objectives. Mathematical techniques such as convexity, smoothing, and scalarization are used to reasonably simplify and relax the multi-objective optimization model, and an efficient numerical optimization algorithm is designed to achieve accurate estimation of energy meter error.

[0046] It should be noted that the specifics include the following:

[0047] First, a non-convex sparse multi-objective optimization mathematical model for electricity meters is constructed to accurately map the problem of electricity meter information evaluation and calibration. Based on the principle of energy conservation in power distribution areas, a multi-objective optimization model is constructed with electricity meter error as the variable and multiple optimization objectives such as data priority, system stability, and energy conservation. The data priority objective is characterized by metrics such as the L0 norm and Lp norm of electricity meter error, which have non-convex and sparse mathematical properties. The system stability objective is characterized by metrics such as the absolute value function between current and historical errors and the SCAD function, which have non-smooth mathematical properties. The energy conservation objective is characterized by metrics such as the quadratic function and polynomial function of the difference between power supply and power consumption, which have convex and smooth mathematical properties. Finally, a non-convex sparse multi-objective optimization model is constructed to accurately map the problem of electricity meter information evaluation and calibration.

[0048] Secondly, an effective relaxation method is used to simplify the original problem by utilizing the non-convex sparse multi-objective optimization mathematical model of the electricity meter. Considering the non-convex, non-smooth, and multi-objective mathematical characteristics of the aforementioned non-convex sparse multi-objective optimization model, various convexification methods for non-convex objectives and the corresponding convex approximation theory are studied to characterize the error between the solution to the original problem and the solution to the relaxed problem. Smoothing methods for non-smooth sparse objectives and deep learning methods for various smoothing model parameters are used to characterize the error between the solution to the original problem and the solution to the relaxed problem. Based on existing scalarization methods, an autonomous learning strategy for the parameters between various optimization objectives is designed to develop new scalarization methods.

[0049] Finally, an efficient adaptive algorithm based on a non-convex sparse multi-objective optimization mathematical model of electricity meters is used to achieve accurate identification of out-of-tolerance electricity meters. For the special sparse structure of the model, combined with machine learning techniques, efficient solution algorithms adapted to multi-objective optimization problems with special sparse structures are developed. This includes drawing on the computational experience of existing global optimization algorithms to design a local escape mechanism for a global adaptive optimization algorithm for non-convex multi-objective optimization problems, seeking better local or global optimal solutions; researching the search direction and step size design of methods such as approximation algorithms, conjugate gradient algorithms, and trust region algorithms to improve algorithm accuracy and efficiency; and further conducting theoretical analysis of algorithm convergence based on existing theoretical analysis results.

[0050] Fourth, such as Figure 3 As shown, a multivariate optimization model construction method with transformer area line loss and electricity meter operating error as variables is used to study local jump-out mechanism, hyperparameter adaptive selection and other technologies. A global adaptive algorithm based on alternating direction multiplier method and block coordinate descent method is designed to realize the coupled solution of transformer area line loss and electricity meter operating error.

[0051] It should be noted that the specifics include the following:

[0052] First, a method for estimating transformer substation line losses using a dual-driven approach of dynamic and static data and mathematical-physical models is employed to accurately describe the nonlinear functional relationship between dynamic and static data such as power consumption, voltage, current, transformer substation topology, and line parameters, and transformer substation line losses. Then, using real-time dynamic data such as voltage and current, and static data such as voltage impedance, current impedance, and line parameters, physical models such as the power formula are used to calculate the current voltage and current losses of the transformer substation. Finally, combining the electricity meter error data calculated in this study, machine learning techniques are used to investigate the nonlinear functional relationship between data such as voltage loss, current loss, electricity meter error, power supply / consumption difference, and transformer substation topology, and transformer substation line losses, thereby achieving accurate estimation of transformer substation line losses.

[0053] Secondly, a multivariate optimization model with transformer substation line loss and electricity meter operating error as variables is used to design a scheme. Combining the non-convex sparse optimization model for electricity meter error estimation with the machine learning estimation model for transformer substation line loss, a multivariate coupled optimization model for transformer substation line loss and electricity meter operating error is constructed.

[0054] Finally, separable alternating iterative algorithms such as the alternating direction multiplier method and the block coordinate descent method are used to synergistically couple and solve for transformer line loss and electricity meter operating errors. For the multivariate optimization model, its augmented Lagrangian function, gradient function, and dual function are studied; utilizing the separability of the objective function, the original problem is decomposed into several optimization subproblems; a distributed alternating direction multiplier method and a distributed block coordinate descent method are constructed to solve the optimization subproblems concerning transformer line loss and electricity meter errors in parallel; an adaptive shutdown criterion is constructed to accelerate the convergence speed of the algorithm, and the computational complexity and convergence of the algorithm are analyzed.

[0055] Finally, it should be noted that the above embodiments are merely illustrative and explanatory of the present invention, and are not intended to limit the present invention to the scope of the described embodiments. Furthermore, those skilled in the art will understand that the present invention is not limited to the above embodiments, and many more variations and modifications can be made based on the teachings of the present invention, all of which fall within the scope of protection claimed by the present invention.

Claims

1. An information-based evaluation and calibration model for electricity meters based on the conservation of electrical energy in distribution areas, characterized in that, Construct it in the following way: Step 1: By using a hierarchical and graded method based on the principle of energy conservation and a precise decoupling method for transformer area line loss, the impact of transformer area line loss measurement error on the model calculation results is reduced, ensuring that the uncertainty of the model calculation results meets the requirements of the field; by using the multi-level node collaborative calculation technology of energy meter information evaluation and calibration, a top-down online dynamic error calibration method is constructed to evaluate and correct the operating status of the total node at different levels in real time. Step 2: Utilize HPLC-based hybrid networking technology to achieve effective compression and efficient transmission of high-frequency data in the electricity meter box; utilize HPLC-based NTB synchronization technology to achieve automatic clock synchronization; Ensure the efficiency and accuracy of electricity meter data collection; Step 3: Using a non-convex sparse multi-objective optimization model construction method with the energy meter error as the variable and data priority, system stability, and energy conservation as multiple optimization objectives, the multi-objective optimization model is simplified and relaxed using mathematical techniques, and an efficient numerical optimization algorithm is used to achieve accurate estimation of the energy meter error. Step 4: Using a multivariate optimization model construction method with transformer area line loss and electricity meter operating error as variables, design a local escape mechanism and hyperparameter adaptive selection technology, and design a global adaptive algorithm based on alternating direction multiplier method and block coordinate descent method to achieve coupled solution of transformer area line loss and electricity meter operating error; Step 1 specifically includes the following: Step 1.1: By analyzing the typical structure of existing low-voltage distribution areas, a hierarchical structure model suitable for typical electricity consumption scenarios in residential buildings is described using JP cabinets, box-type substations, distribution cabinets, branch boxes, and metering boxes as basic elements. Step 1.2: Based on different hierarchical models, and combined with key measurement data based on time-series load voltage, current, and power, establish loss calculation models applicable to different levels. Transform the problem of estimating line loss by the difference between the total node and the sum of each sub-node into loss calculation models with physical meaning under different levels, thereby achieving accurate decoupling of losses. Step 1.3: Through multi-level node collaborative calculation, the total node error and uncertainty at different levels are transmitted and traced step by step to the end node and the total node, further optimizing the uncertainty caused by the total node error at different levels, and realizing the information-based evaluation and calibration of the electricity meter.

2. The information-based evaluation and calibration model for electricity meters based on the conservation of electrical energy in transformer substations, as described in claim 1, is characterized in that... Step 2 specifically includes the following: Step 2.1: Utilize HPLC-based hybrid networking technology to achieve physical topology identification of the transformer area, through... Based on HPLC, a characteristic current signal is superimposed to perform physical topology identification in a hybrid networking mode; when the transformer-side management terminal calls the meter box-side metering module one by one, the meter box-side metering module generates a characteristic current signal in sequence after receiving the call information. After receiving the signal, the branch-side metering module records the information and forwards the characteristic signal to the management terminal, thereby constructing a physical topology model; Step 2.2: Utilizing HPLC-based NTB synchronization and data synchronization acquisition technology, firstly, the management terminal on the transformer side acts as the CCO, and the metering modules on the transformer branch and meter box sides act as STA modules. The CCO module sends broadcast time synchronization commands to the STA modules on the branch and meter box sides to achieve precise clock synchronization. The metering modules on the branch and meter box sides are checked one by one. If there is a clock deviation at the second level, the metering module with the clock deviation is re-calibrated. Secondly, research is conducted on synchronous sampling technology based on HPLC characteristic signals. When the voltage is close to zero, the voltage characteristic signal is superimposed to represent the information. Within the voltage cycle, the metering modules at each level start synchronous sampling after receiving the voltage characteristic signal. Step 2.3: Utilize high-frequency data compression transmission technology based on hybrid networking to achieve efficient data transmission to the management terminal and main station for analysis and calculation. Firstly, research is conducted on high-frequency data compression transmission based on hybrid networking technology. This involves using numbers instead of strings and constructing a Huffman binary tree based on the frequency of character occurrences. Data appearing more frequently is placed at the upper level of the tree, and data appearing less frequently is placed at the lower level. The path from the root node to each data point is encoded and lossless compression is achieved, thus enabling efficient data transmission based on the physical topology. Secondly, synchronous sampling technology is used to sample electrical parameter curve data at the same time point.

3. The information-based evaluation and calibration model for electricity meters based on the conservation of electrical energy in transformer substations, as described in claim 1, is characterized in that... Step 3 specifically includes the following: Step 3.1: Utilize the non-convex sparse multi-objective optimization mathematical model construction method for electricity meters to accurately map the electricity meter data. This paper addresses the problem of energy meter information-based evaluation and calibration. Based on the principle of energy conservation in power distribution areas, a multi-objective optimization model is constructed, with energy meter error as the variable and multiple optimization objectives including data priority, system stability, and energy conservation. A metric function with non-convex and sparse mathematical properties is used to characterize the data priority objective; a metric function with non-smooth mathematical properties is used to characterize the system stability objective; and a metric function with convex and smooth mathematical properties is used to characterize the energy conservation objective. Finally, a non-convex sparse multi-objective optimization model that accurately maps the energy meter information-based evaluation and calibration problem is constructed. Step 3.2: Utilize effective relaxation methods for the non-convex sparse multi-objective optimization mathematical model of the electricity meter to reasonably simplify the original problem; considering the non-convex, non-smooth, and multi-objective mathematical characteristics of the aforementioned non-convex sparse multi-objective optimization model, study various convexification methods for non-convex objectives and the corresponding convex approximation theory to characterize the error between the solution to the original problem and the solution to the relaxed problem; utilize smoothing methods for non-smooth sparse objectives and deep learning methods for various smoothing model parameters to characterize the error between the solution to the original problem and the solution to the relaxed problem. Based on existing scalarization methods, we design autonomous learning strategies for parameters among various optimization objectives and develop new scalarization methods. Step 3.3: Utilize an efficient adaptive algorithm for the non-convex sparse multi-objective optimization mathematical model of the electricity meter to implement... The project aims to accurately identify out-of-range energy meters; address the sparse structure of the model by developing efficient algorithms for solving multi-objective optimization problems adapted to sparse structures using machine learning techniques. This includes drawing on the computational experience of existing global optimization algorithms to design a local escape mechanism for a globally adaptive optimization algorithm for non-convex multi-objective optimization problems, seeking better local or global optimal solutions; improving algorithm accuracy and efficiency by designing the search direction and step size using approximation algorithms, conjugate gradient algorithms, and trust region algorithms; and further conducting theoretical analysis of algorithm convergence based on existing theoretical analysis results.

4. The information-based evaluation and calibration model for electricity meters based on the conservation of electrical energy in transformer substations, as described in claim 1, is characterized in that... Step 4 specifically includes the following: Step 4.1: Utilizing a dual-driven method of dynamic and static data and mathematical-physical models, this method accurately describes the nonlinear functional relationship between power consumption, voltage, current, transformer topology, line parameters, dynamic and static data, and transformer line loss. Using real-time dynamic data of voltage and current, and static data of line parameters, the method calculates the voltage and current losses of the current transformer area using a physical model based on the power formula. Combined with the calculated energy meter error data, machine learning techniques are employed to study the nonlinear functional relationship between voltage loss, current loss, energy meter error, power supply / consumption difference, transformer topology data, and transformer line loss, thereby achieving accurate estimation of transformer line loss. Step 4.2: Using a multivariate optimization model design method with transformer substation line loss and electricity meter operating error as variables, combined with a non-convex sparse optimization model for electricity meter error estimation and a machine learning estimation model for transformer substation line loss, a multivariate coupled optimization model for transformer substation line loss and electricity meter operating error is constructed. Step 4.3: Utilize a separable alternating iterative algorithm to collaboratively and coupledly solve for transformer line loss and electricity meter operating errors; for the multivariate optimization model, utilize its augmented Lagrangian function, gradient function, and dual function; leverage the separability of the objective function to decompose the original problem into several optimization subproblems; construct a distributed alternating direction multiplier method and a distributed block coordinate descent method to solve the optimization subproblems concerning transformer line loss and electricity meter errors in parallel; construct an adaptive shutdown criterion to accelerate the algorithm's convergence speed, and analyze the algorithm's computational complexity and convergence.

Images