Data quality assessment method and system

Abstract

The invention provides a data quality evaluation method and system. The method comprises the following steps: firstly, collecting three types of data sets of three-phase unbalance degree, free charge density distribution and line impedance decomposition; and then, carrying out space topology mapping on the charge density gradient extreme point of the cable joint area and the voltage phase angle, considering charge migration effect compensation, and calculating a corrected three-phase voltage unbalance parameter. Secondly, establishing a coupling factor matrix based on the separated line inherent impedance component and the load fluctuation impedance component, and performing time domain convolution operation on the coupling factor matrix and a voltage unbalance parameter to generate a distortion feature vector; and finally, calculating data credibility weight distribution of the global monitoring nodes according to the nonlinear correlation degree between the three-phase unbalance degree data and the distortion feature vector, and using the data credibility weight distribution to evaluate the quality of the low-voltage treatment data. According to the technical scheme provided by the invention, the accuracy and reliability of data quality evaluation can be improved.

Description

Technical Field

[0001] This application relates to the technical field of quality assessment, and particularly to a data quality assessment method and system. Background Art

[0002] With the continuous expansion of the scale and increasing complexity of the power system, ensuring data quality has become a key factor in improving the operation efficiency and reliability of the power grid. In modern smart grids, a large number of sensors and monitoring devices are deployed to collect various electrical parameters in real time, such as voltage, current, and power. However, due to reasons such as equipment aging, environmental interference, or communication failures, the collected data often has problems such as noise, missing values, or errors. In order to accurately evaluate the data quality of all monitoring nodes in the distribution network line and formulate effective maintenance strategies accordingly, an efficient and accurate data quality assessment method is needed. This method should not only be able to identify abnormal data but also quantify the data credibility of each monitoring node, providing a reliable basis for subsequent data processing and analysis.

[0003] Existing solutions usually use statistical-based methods for data quality assessment. For example, outliers are detected by calculating the mean and standard deviation, or abnormal patterns in trend changes are identified using time series analysis techniques. In addition, some studies have attempted to combine machine learning algorithms, such as support vector machines and neural networks, to train historical data to predict possible future problems. These methods can, to a certain extent, discover abnormal situations in the data and make a preliminary judgment on the data quality. At the same time, some solutions also introduce multi-source data fusion technology, comprehensively considering datasets from different sources to improve the accuracy of the evaluation results.

[0004] Although the above solutions have achieved good results in specific scenarios, there are still some obvious defects. First, statistical-based methods often rely on the assumption that the data follows a certain specific distribution, which may not always hold in practical applications, resulting in a high misjudgment rate. Second, machine learning models require a large amount of high-quality labeled data for training, and obtaining such data in the power system is costly and time-consuming. Finally, although multi-source data fusion can improve the evaluation accuracy, it also increases the complexity and computational burden of the system. Especially in a large-scale power grid environment, it may lead to problems of insufficient real-time performance. Therefore, there is an urgent need to develop a more flexible, efficient, and adaptable data quality assessment method to overcome the limitations of existing solutions and meet the growing needs of the power system. Summary of the Invention

[0005] This application provides a data quality assessment method and system to solve the problems of poor accuracy and insufficient reliability in data quality assessment in the prior art.

[0006] In a first aspect, the present application provides a data quality assessment method, including: Under the operating state of the distribution network line, obtain a three-phase unbalance degree data set, a free charge density distribution data set generated based on the millimeter-wave radar space charge detection technology, and a line impedance decomposition data set generated through the impedance spectrum dynamic reconstruction technology; Perform spatial topological mapping on the extreme points of the charge density gradient in the cable joint area in the free charge density distribution data set and the voltage phase angle in the three-phase unbalance degree data set, and generate three-phase voltage unbalance parameters with corrected phase deviation through charge migration effect compensation calculation; Based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition data set, establish an impedance parameter coupling factor matrix, and perform a time-domain convolution operation on the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameters to generate a distortion feature vector of the combined line impedance and voltage; According to the non-linear correlation degree between the three-phase unbalance degree data set and the distortion feature vector, calculate the data credibility weight distribution of the global monitoring nodes of the distribution network line, and the data credibility weight distribution is used to characterize the quality assessment result of the low-voltage governance data.

[0007] Optionally, the establishing an impedance parameter coupling factor matrix based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition data set includes: Perform frequency-domain decomposition on the line inherent impedance component and extract the maximum points of the steady-state amplitude-frequency response curve within a preset frequency band, and generate a dominant resonance frequency set based on the distribution range of the maximum points; Divide the time domain interval of the load fluctuation impedance component according to the load switching timestamp, and calculate the composite parameter of the dynamic change rate and the phase offset of the load fluctuation impedance component within each time domain interval; Construct an initial coupling factor matrix with the dominant resonance frequency set as the row index and the composite parameter as the column index, and apply dynamic constraints to the initial coupling factor matrix, including generating a spatial correction coefficient based on the impedance transfer function of adjacent monitoring nodes, and generating a statistical correction coefficient based on the historical probability density kurtosis and skewness of the load fluctuation impedance component; Superimpose the correlation degree of the spatial correction coefficient and the statistical correction coefficient on the diagonal elements of the matrix to establish an impedance parameter coupling factor matrix.

[0008] Optionally, the constructing an initial coupling factor matrix with the dominant resonance frequency set as the row index and the composite parameter as the column index includes: In the dominant resonance frequency set, extract the frequency interval and the amplitude attenuation slope between adjacent maximum points on the steady-state amplitude-frequency response curve of each resonance frequency point to generate a resonance intensity factor; Segment and discretize the composite parameter according to the time domain interval corresponding to the load switching timestamp, and construct the anti-aliasing weight coefficient of the composite parameter according to the cosine similarity of the dynamic change rate and the phase offset in adjacent time domain intervals; Take the resonance intensity factor as the row vector element, and the composite parameter weighted by the anti-aliasing weight coefficient as the column vector element, and generate the initial matrix framework through the outer product operation of the row vector element and the column vector element; Based on the initial matrix framework, for the row vector elements corresponding to each resonance frequency point, superimpose the amplitude-frequency integral value of the maximum value point within the preset frequency band to which the corresponding resonance frequency point belongs, and perform element-by-element correlation with the cumulative energy distribution of the composite parameter of the corresponding column vector element in the time domain interval to construct the initial coupling factor matrix.

[0009] Optionally, dividing the time domain interval of the load fluctuation impedance component according to the load switching timestamp, and calculating the composite parameter of the dynamic change rate and the phase offset of the load fluctuation impedance component in each time domain interval, including: Based on the trigger interval of the load switching timestamp, use the time window between adjacent timestamps as an independent time domain interval, and extract the extreme value point of the load fluctuation impedance component within the independent time domain interval as the sampling reference point of the dynamic change rate; Construct a local time window centered on the sampling reference point, and calculate the mean value of the time domain curvature change rate of the load fluctuation impedance component to generate the original parameter pair of the dynamic change rate and the phase offset; Perform cross-validation on the original parameter pair, impose a time window scaling constraint on the dynamic change rate through the load switching timestamp interval length of adjacent time domain intervals, and correct the phase polarity of the phase offset according to the continuity characteristics of adjacent time domain intervals; Non-linearly superimpose the scaled dynamic change rate and the corrected phase offset, and weight them by the energy ratio coefficient of the load fluctuation impedance component in the corresponding time domain interval to obtain the composite parameter.

[0010] Optionally, the time domain convolution operation of the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameter to generate the distortion feature vector of the line impedance and voltage combination includes: Based on the orthogonal decomposition relationship between the inherent impedance component and the load fluctuation impedance component in the impedance parameter coupling factor matrix, construct a three-dimensional convolution kernel function with time-varying characteristics; Expand the three-phase voltage unbalance parameter into a multi-channel pulse sequence according to the preset time granularity, and map each channel to the differential increment of the voltage amplitude corresponding to the phase; Within the sliding time window, slice the three-dimensional convolution kernel function along the time axis, and perform a cyclic convolution with asymmetric boundary constraints on the multi-channel pulse sequence. When the impedance parameter mutates, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant decay function; Extract the interaction strength coefficients of the phase voltage trajectories and impedance trajectories in the convolution output result, and construct a multi-dimensional tensor containing the combined spectral characteristics of impedance and voltage. Perform an energy threshold screening on the multi-dimensional tensor along the impedance coupling main axis, and use the projection components exceeding the energy threshold as the distortion eigenvectors.

[0011] Optionally, the step where when the impedance parameter mutates, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant decay function includes: Set a mutation detection window on the time axis of the three-dimensional convolution kernel function, and calculate the instantaneous change rate of the load fluctuation impedance component in the impedance parameter coupling factor matrix and the trigger expansion mechanism of the hyperbolic secant decay function in real time; Apply an energy conservation constraint to the frequency domain components of the trigger expansion mechanism, and maintain the total energy of the trigger expansion mechanism before and after mutation through the energy integration of the load fluctuation impedance component; Dynamically correct the constrained trigger expansion mechanism according to the phase difference change trajectory of the impedance parameter and the three-phase voltage unbalance parameter within the mutation detection window. When the instantaneous change rate exceeds the corrected trigger expansion mechanism, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant decay function.

[0012] Optionally, the step of calculating the data credibility weight distribution of the global monitoring nodes of the distribution network line according to the non-linear correlation degree between the three-phase unbalance dataset and the distortion eigenvector includes: Perform multi-dimensional coupling of the phase voltage unbalance parameters in the three-phase unbalance dataset and the distortion eigenvector in the time domain dimension, and establish a non-linear correlation degree index for each monitoring node through the phase sensitivity kernel function; Based on the spatial distribution pattern of the non-linear correlation degree index in the three-dimensional charge density gradient field, construct a probability density diffusion surface reflecting the line impedance coupling strength; Perform multi-scale convolution fusion of the probability density diffusion surface and the inherent impedance spectrum characteristics in the line impedance decomposition dataset to generate the dynamic weight base value of each monitoring node; Establish a weight correction equation based on the double constraints of impedance and charge, perform path integral optimization on the dynamic weight base value, solve the steady-state solution set of the weight correction equation under the action of the charge migration effect through an iterative convergence algorithm, and obtain the data credibility weight distribution according to the projection amplitude of the steady-state solution set in the monitoring node space coordinate system.

[0013] Second aspect, the present application provides a data quality assessment system, including: An acquisition module, which acquires a three-phase unbalance degree data set, a free charge density distribution data set generated based on millimeter-wave radar space charge detection technology, and a line impedance decomposition data set generated by impedance spectrum dynamic reconstruction technology under the operating state of the distribution network line; A mapping module, which performs spatial topological mapping on the extreme points of the charge density gradient in the cable joint area in the free charge density distribution data set and the voltage phase angle in the three-phase unbalance degree data set, and generates three-phase voltage unbalance parameters with corrected phase deviation through charge migration effect compensation calculation; A building module, which builds an impedance parameter coupling factor matrix based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition data set, and performs time-domain convolution operation on the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameters to generate a distortion feature vector of the joint of line impedance and voltage; A characterization module, which calculates the data credibility weight distribution of the global monitoring nodes of the distribution network line according to the non-linear correlation degree between the three-phase unbalance degree data set and the distortion feature vector, and the data credibility weight distribution is used to characterize the quality assessment result of the low voltage governance data.

[0014] Third aspect, an embodiment of the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a data quality assessment method as described in the first aspect above.

[0015] Fourth aspect, an embodiment of the present application provides a computer storage medium, storing a computer program, and when the computer program is executed by a computer, it implements a data quality assessment method as described in the first aspect.

[0016] In the embodiments of the present application, under the operating state of the distribution network line, a three-phase unbalance degree data set, a free charge density distribution data set generated based on the millimeter-wave radar space charge detection technology, and a line impedance decomposition data set generated through the impedance spectrum dynamic reconstruction technology are obtained; the extreme points of the charge density gradient in the cable joint area in the free charge density distribution data set are subjected to spatial topological mapping with the voltage phase angle in the three-phase unbalance degree data set, and the three-phase voltage unbalance parameters after phase deviation correction are generated through compensation calculation of the charge migration effect; based on the line inherent impedance component and the load fluctuation impedance component separated in the line impedance decomposition data set, an impedance parameter coupling factor matrix is established, and the impedance parameter coupling factor matrix is subjected to time-domain convolution operation with the three-phase voltage unbalance parameters to generate a distortion feature vector jointly of the line impedance and the voltage; according to the non-linear correlation degree between the three-phase unbalance degree data set and the distortion feature vector, the data credibility weight distribution of the whole-region monitoring nodes of the distribution network line is calculated, and the data credibility weight distribution is used to characterize the quality evaluation result of the low-voltage governance data.

[0017] The technical solution of the present application has the following beneficial effects: By obtaining the data sets of the three-phase unbalance degree, the free charge density distribution, and the line impedance decomposition, the present application realizes the comprehensive monitoring and evaluation of the operating state of the distribution network. This provides rich basic data support for subsequent analysis. The extreme points of the charge density gradient in the cable joint area are subjected to spatial topological mapping with the voltage phase angle, and the three-phase voltage unbalance parameters are corrected through the compensation of the charge migration effect, improving the accuracy of the detection of the voltage unbalance problem. Based on the separated line inherent impedance component and the load fluctuation impedance component, a coupling factor matrix is established, and a distortion feature vector is generated through time-domain convolution operation, which helps to deeply understand the complex relationship between the line impedance and the voltage and improves the accuracy of fault diagnosis. According to the non-linear correlation degree between the three-phase unbalance degree data set and the distortion feature vector, the data credibility weight distribution is calculated, thereby providing an effective method for evaluating the quality of the low-voltage governance data and enhancing the reliability of the power grid management decision-making.

[0018] Furthermore, first, the inherent impedance components of the line are decomposed in the frequency domain to extract the set of dominant resonance frequencies, and the load fluctuation impedance components are divided into time intervals to calculate the composite parameters of the dynamic change rate and the phase offset. Then, based on this information, an initial coupling factor matrix is constructed and dynamic constraints are imposed, including introducing spatial correction coefficients and statistical correction coefficients to enhance the accuracy and stability of the matrix. The finally established impedance parameter coupling factor matrix can more accurately reflect the interaction between the line impedance and the load fluctuation, improving the detail level and prediction ability of the monitoring of the operating state of the distribution network. By comprehensively considering the spatial and statistical characteristics, the understanding and analysis of the grid impedance characteristics are deepened, effectively improving the accuracy and reliability of grid fault diagnosis and state assessment.

[0019] These aspects or other aspects of the present application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0021] Figure 1 The flowchart of a data quality assessment method provided by the present application is shown; Figure 2 The structural schematic diagram of a data quality assessment system provided by the present application is shown; Figure 3 The structural schematic diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0023] In some processes described in the specification, claims, and the above-mentioned drawings of this application, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations can be executed not in the order in which they appear herein or in parallel. The operation numbers such as 101, 102, etc. are only used to distinguish different operations, and the numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and these operations can be executed in sequence or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent a sequence, and do not limit that "first" and "second" are of different types.

[0024] This project aims to comprehensively analyze various key data under the operating state of the distribution network, including three-phase unbalance degree, free charge density distribution, and line impedance decomposition data. First, the three-phase voltage unbalance parameters are compensated and adjusted by combining spatial topology mapping with charge migration effect compensation. Then, a coupling factor matrix is established based on the separated line inherent and load fluctuation impedance components. Finally, the data credibility weight distribution is calculated according to the non-linear correlation between the three-phase unbalance degree and the distortion eigenvector, so as to evaluate the quality of low-voltage governance data. This process effectively improves the accuracy and reliability of data processing at distribution network monitoring nodes, providing solid data support for power grid management.

[0025] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0026] Figure 1 A flowchart of a data quality evaluation method is provided for the embodiments of the present application, as Figure 1 shown, the method includes: 101. Under the operating state of the distribution network line, obtain a three-phase unbalance degree data set, a free charge density distribution data set generated based on millimeter-wave radar space charge detection technology, and a line impedance decomposition data set generated by impedance spectrum dynamic reconstruction technology; In this step, the three-phase unbalance degree data set contains voltage and current parameters of each monitoring point in the power grid, and is used to evaluate the balance state between the three phases of the power system.

[0027] The millimeter-wave radar space charge detection technology is a technology that detects the charge density in the cable joint area by transmitting and receiving millimeter waves. The generated free charge density distribution data set can reveal local charge anomalies.

[0028] The line impedance decomposition dataset is a data set obtained by dynamically reconstructing the impedance spectrum of the distribution network, which can separate the inherent impedance (determined by the line material and structure) and the load fluctuation impedance component (changing with the load), so as to reflect the line operation status.

[0029] In the embodiments of this application, first, when the distribution network is operating normally, the sensors deployed in the power grid start to collect three-phase unbalance data, including voltage and current parameters. Then, use the millimeter-wave radar equipment to scan the selected key cable joint areas and record the spatial distribution of the charge density. Next, apply the impedance spectrum dynamic reconstruction technology to process the data collected from the distribution line, and extract the inherent impedance component representing the line's own attributes and the impedance fluctuation component caused by load changes. Finally, integrate all these data to form a complete line impedance decomposition dataset, providing basic support for subsequent analysis.

[0030] In an intelligent distribution network project in a certain city, technicians set up multiple sensor nodes in the substation to monitor the changes in voltage and current of each phase in real time. At the same time, use the millimeter-wave radar to conduct a detailed scan of a cable joint with frequent known problems, and obtain a detailed charge density map. In addition, advanced software tools are used to deeply analyze the collected data, and the inherent impedance and load fluctuation impedance components are successfully separated, providing an important basis for subsequent optimization of the power grid performance.

[0031] 102. Perform a spatial topological mapping between the extreme points of the charge density gradient in the cable joint area of the free charge density distribution dataset and the voltage phase angle in the three-phase unbalance dataset, and generate the three-phase voltage unbalance parameters with corrected phase deviation through charge migration effect compensation calculation; In this step, the extreme point of the charge density gradient refers to the position of the maximum change rate of the charge density in its spatial distribution, usually indicating potential fault points.

[0032] The voltage phase angle is an important parameter describing the relative time offset between the three-phase voltages and is crucial for evaluating voltage unbalance.

[0033] Spatial topological mapping refers to the process of correlating and analyzing the above two points, and charge migration effect compensation is to adjust the voltage phase angle on this basis to more accurately correct the voltage unbalance problem.

[0034] The three-phase voltage unbalance parameters are a set of indicators used to quantify the differences between the three-phase voltages in the power system. These differences may be due to the asymmetry of the power supply system, the imbalance of the load, or other factors.

[0035] In the embodiments of the present application, first, the extreme points of the charge density gradient in the cable joint area are identified from the free charge density distribution dataset. Then, the position information of these extreme points is corresponded to the voltage phase angles in the three-phase unbalance data to establish a spatial mapping relationship between the two. Next, based on the charge migration theory, the phase deviation value to be corrected is calculated, and the original three-phase voltage unbalance parameters are adjusted accordingly. Finally, through a series of precise calculations, the three-phase voltage unbalance parameters after phase deviation correction are obtained, improving the evaluation accuracy.

[0036] Based on the data collected in the early stage, the technician first identifies the extreme points of the gradient in the free charge density distribution, which usually indicate potential problem areas. Then, the position information is corresponded to the voltage phase angles in the three-phase unbalance data to establish a spatial mapping relationship between the two. Through a complex algorithm model, the phase deviation value to be corrected is calculated according to the charge migration theory, and the original three-phase voltage unbalance parameters are adjusted accordingly. This step significantly improves the accuracy of the original data and lays a foundation for more accurate subsequent analysis.

[0037] 103. Based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition dataset, an impedance parameter coupling factor matrix is established, and the impedance parameter coupling factor matrix is subjected to a time-domain convolution operation with the three-phase voltage unbalance parameters to generate a distortion feature vector jointly representing the line impedance and voltage; In this step, the line inherent impedance component is the resistance, inductive reactance and other characteristics determined by the material and structure of the line itself.

[0038] The load fluctuation impedance component is the impedance change caused by the load change.

[0039] The impedance parameter coupling factor matrix aims to describe the interaction relationship between the line inherent impedance component and the load fluctuation impedance component.

[0040] The time-domain convolution operation refers to combining the three-phase voltage unbalance parameters with this matrix to generate a comprehensive feature vector for comprehensively characterizing the relationship between the line impedance and voltage.

[0041] The distortion feature vector is a mathematical representation for characterizing the complex interaction between voltage and impedance in a power system, especially in the distribution network line.

[0042] In the embodiments of the present application, first, the inherent impedance and the load fluctuation impedance components are separated from the line impedance decomposition dataset. Then, a coupling factor matrix is constructed by using frequency-domain and time-domain analysis methods, taking into account the influence of the dominant resonance frequency set and the composite parameters. Next, a three-dimensional convolution kernel function is designed to convert the three-phase voltage unbalance parameters into a pulse sequence form. Finally, a circular convolution operation is performed, combined with the impedance parameter coupling factor matrix, to generate a distortion feature vector containing the combined characteristics of impedance and voltage for in-depth analysis of the line health status.

[0043] Using the previously obtained data, the technical personnel further analyzed and constructed a detailed impedance parameter coupling factor matrix. First, the inherent impedance and the load fluctuation impedance components are separated from the line impedance decomposition dataset, and then a coupling factor matrix is constructed by using frequency-domain and time-domain analysis methods, taking into account the influence of the dominant resonance frequency set and the composite parameters. Next, a three-dimensional convolution kernel function is designed to convert the three-phase voltage unbalance parameters into a pulse sequence form. Finally, a circular convolution operation is performed, combined with the impedance parameter coupling factor matrix, to generate a distortion feature vector containing the combined characteristics of impedance and voltage. This process helps to deeply understand the line health status and provides a basis for further optimization.

[0044] 104. Calculate the data credibility weight distribution of the global monitoring nodes of the distribution network line according to the non-linear correlation degree between the three-phase unbalance dataset and the distortion feature vector, and the data credibility weight distribution is used to characterize the quality evaluation result of the low-voltage governance data.

[0045] In this step, the three-phase unbalance dataset is a key data set used to evaluate the balance state between the three-phase voltages or currents in the power system.

[0046] The non-linear correlation degree is a quantitative index used to measure the complex relationship between the three-phase unbalance dataset and the distortion feature vector.

[0047] The data credibility weight distribution is based on this correlation degree and is used to evaluate the data quality of each monitoring node to ensure the effectiveness and scientificity of the low-voltage governance measures.

[0048] The low-voltage governance data refers to a series of related data collected and processed in the power system, especially in the distribution network, to identify, analyze and solve the low-voltage problems.

[0049] In the embodiments of the present application, first, the non-linear correlation degree between the three-phase unbalance dataset and the distortion feature vector is calculated. Then, based on this correlation degree, a probability density diffusion surface is constructed to reflect the line impedance coupling strength. Next, through multi-scale convolution fusion, the dynamic weight base values of each monitoring node are generated. Finally, through path integral optimization, the data credibility weight distribution result is obtained to guide the formulation of the low-voltage governance strategy.

[0050] After obtaining the detailed distortion feature vectors, the technicians calculated the non-linear correlation degree between the three-phase unbalance degree data set and these feature vectors. Based on this correlation degree, a probability density diffusion surface was constructed to reflect the line impedance coupling strength. Subsequently, through multi-scale convolution fusion, the dynamic weight base values of each monitoring node were generated. After path integral optimization, the data credibility weight distribution result was obtained. These analyses not only helped improve the data quality assessment level, but also provided a scientific basis for more effective low-voltage governance measures, ensuring the safe and stable operation of the entire intelligent distribution network.

[0051] In summary, steps 101 to 104 achieve full-process coverage from basic data collection, in-depth analysis to final data quality assessment. This method can not only accurately capture the subtle changes in the operation of the distribution network, but also effectively guide the low-voltage governance work, greatly improving the grid management efficiency and power supply reliability. Each step is closely linked to form a complete technical chain, providing strong support for the development of the smart grid.

[0052] To solve the problem of accurately describing the complex dynamic behavior in the distribution network, first, the line inherent impedance component is decomposed in the frequency domain and the dominant resonance frequency set is extracted. At the same time, the load fluctuation impedance component is divided into time domain intervals according to the load switching timestamp. Then, dynamic constraints are applied, including the spatial correction coefficient based on the impedance transfer function of adjacent monitoring nodes and the statistical correction coefficient based on the historical probability density characteristics. Finally, these correction coefficients are superimposed on the diagonal elements of the matrix to form an optimized impedance parameter coupling factor matrix to more accurately reflect the actual operating conditions. In some embodiments, step 103 of establishing an impedance parameter coupling factor matrix based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition data set includes: 201. Decompose the line inherent impedance component in the frequency domain and extract the maximum points of the steady-state amplitude-frequency response curve within a preset frequency band, and generate a dominant resonance frequency set based on the distribution range of the maximum points; In step 201, the line inherent impedance component refers to the resistance, inductive reactance and other characteristics determined by the material and structure of the line itself. Frequency domain decomposition is a technology that converts a time-domain signal into a frequency-domain representation, and identifies the dominant resonance frequency set by extracting the maximum points of the steady-state amplitude-frequency response curve within a preset frequency band. The dominant resonance frequency set is those frequency points that exhibit significant vibration characteristics within a specific frequency range and are used to reflect the main vibration modes of the line.

[0053] In the embodiments of the present application, first, the inherent impedance component of the line is decomposed in the frequency domain and converted into a frequency-domain representation using the fast Fourier transform technique. Then, the maximum points on the steady-state amplitude-frequency response curve are searched within a preset frequency range, and a set of dominant resonance frequencies is generated based on the distribution range of these maximum points. Finally, a frequency set that can accurately reflect the main vibration modes of the line is obtained.

[0054] 202. Divide the time domain interval of the load fluctuation impedance component according to the load switching timestamp, and calculate the composite parameter of the dynamic change rate and phase offset of the load fluctuation impedance component within each time domain interval; In step 202, the load fluctuation impedance component refers to the impedance change caused by the load change. Dividing the time domain interval according to the load switching timestamp means splitting the data into multiple independent time periods based on the actual load switching time points. The composite parameter includes the dynamic change rate and phase offset, which are used to quantify the load fluctuation characteristics within each time period.

[0055] In the embodiments of the present application, first, each independent time domain interval is determined according to the load switching timestamp. Then, the dynamic change rate and phase offset of the load fluctuation impedance component are calculated within each interval to form a composite parameter. Specifically, the differential algorithm is used to calculate the dynamic change rate, and the phase offset is obtained through the phase detection method. Finally, these composite parameters are summarized to form a data set that comprehensively reflects the load fluctuation characteristics.

[0056] 203. Construct an initial coupling factor matrix with the set of dominant resonance frequencies as the row index and the composite parameter as the column index, and apply dynamic constraints to the initial coupling factor matrix, including generating a spatial correction coefficient based on the impedance transfer function between adjacent monitoring nodes and generating a statistical correction coefficient based on the historical probability density kurtosis and skewness of the load fluctuation impedance component; In step 203, the initial coupling factor matrix is a matrix structure with the set of dominant resonance frequencies as the row index and the composite parameter as the column index. The spatial correction coefficient is generated based on the impedance transfer function between adjacent monitoring nodes, while the statistical correction coefficient is generated according to the historical probability density kurtosis and skewness of the load fluctuation impedance component. These two types of correction coefficients are used to adjust the matrix elements to make them more accurately reflect the actual situation.

[0057] In the embodiments of the present application, first, an initial coupling factor matrix is constructed, with the set of dominant resonance frequencies as the row vector and the composite parameter as the column vector. Then, a spatial correction coefficient is generated using the impedance transfer function between adjacent monitoring nodes, and a statistical correction coefficient is generated by analyzing the probability density characteristics in the historical data. Then, these correction coefficients are superimposed on the diagonal elements of the matrix to complete the adjustment of the initial matrix, and finally, a more accurate impedance parameter coupling factor matrix is formed.

[0058] 204. Superimpose the correlation degree between the spatial correction coefficient and the statistical correction coefficient on the diagonal elements of the matrix to establish an impedance parameter coupling factor matrix.

[0059] In step 204, the spatial correction coefficient and the statistical correction coefficient are generated based on the impedance transfer function between adjacent monitoring nodes and the historical probability density characteristics of the load fluctuation impedance component. These correction coefficients are used to adjust the elements in the initial coupling factor matrix to more accurately reflect the actual situation. The impedance parameter coupling factor matrix is a mathematical model used to describe the complex interaction relationship between different physical quantities and the line impedance characteristics in the power system. The correlation degree refers to the mutual relationship between the spatial correction coefficient and the statistical correction coefficient. Superimposing it on the diagonal elements of the matrix can further optimize the matrix structure and ensure that it can more accurately describe the dynamic behavior of the system.

[0060] In the embodiment of the present application, first, calculate the correlation degree between the spatial correction coefficient and the statistical correction coefficient, and quantify the relationship strength between the two through the correlation analysis method. Then, superimpose the obtained correlation degree value on the diagonal elements of the initial coupling factor matrix, thereby realizing the final adjustment of the matrix. Finally, the impedance parameter coupling factor matrix formed after this series of adjustments not only includes the inherent characteristics of the line and the load fluctuation characteristics, but also considers the influence between different nodes and the statistical characteristics of historical data, enabling the matrix to more accurately describe the actual operating state of the distribution network.

[0061] The following is a specific example: In an intelligent distribution network project in a certain city, the technical personnel first completed the identification of the dominant resonance frequency set, the calculation of the composite parameters, and the construction of the initial coupling factor matrix. Subsequently, the spatial correction coefficient was generated, and the statistical correction coefficient was generated through historical data analysis. Next, the technical personnel calculated the correlation degree between the spatial correction coefficient and the statistical correction coefficient, and superimposed these correlation degree values on the diagonal elements of the initial coupling factor matrix. Finally, an optimized impedance parameter coupling factor matrix was obtained, significantly improving the understanding accuracy of the cable segment health status and providing solid data support for subsequent power grid maintenance and low-voltage governance.

[0062] In summary, through in-depth analysis of the inherent impedance component and load fluctuation impedance component of the line, combined with frequency-domain decomposition, time-domain interval division, the application of multi-dimensional correction factors, and finally the superposition adjustment of the correlation degree, steps 201 to 204 successfully construct an impedance parameter coupling factor matrix that accurately describes the complex dynamic behavior inside the power system. This method not only improves the accuracy of understanding the health status of the power grid but also provides scientific support for formulating effective low-voltage governance strategies, greatly improving the power grid management efficiency and power supply reliability. In addition, by introducing the superposition of spatial and statistical correction factors and their correlation degrees, the adaptability and accuracy of the model are further enhanced, making the assessment of the power grid operating state more comprehensive and reliable.

[0063] To solve the problem of accurately describing the complex dynamic behavior in the power system, the solution extracts the frequency interval and amplitude attenuation slope of each frequency point in the dominant resonance frequency set to generate a resonance intensity factor, and constructs an anti-aliasing weight coefficient for the composite parameter based on the dynamic change rate and phase offset amount between adjacent time-domain intervals. Subsequently, an initial matrix framework is generated through the outer product operation of the row vector elements and the column vector elements, and a more accurate initial coupling factor matrix is constructed, improving the ability to capture subtle changes. In some embodiments, constructing the initial coupling factor matrix with the dominant resonance frequency set as the row index and the composite parameter as the column index in step 203 includes: 301. In the dominant resonance frequency set, extract the frequency interval and amplitude attenuation slope between adjacent maximum value points on the steady-state amplitude-frequency response curve of each resonance frequency point to generate a resonance intensity factor; In step 301, the resonance intensity factor refers to the frequency interval and amplitude attenuation slope between adjacent maximum value points of each resonance frequency point on the steady-state amplitude-frequency response curve. The frequency interval represents the frequency difference between two adjacent maximum value points, and the amplitude attenuation slope reflects the speed of amplitude change between these maximum value points. The resonance intensity factor is used to quantify the importance of each resonance frequency point and its impact on the overall system vibration characteristics.

[0064] In the embodiments of the present application, first, the adjacent maximum value points of each resonance frequency point on the steady-state amplitude-frequency response curve are extracted from the dominant resonance frequency set. Then, the frequency interval and amplitude attenuation slope between these maximum value points are calculated to generate a resonance intensity factor. Specifically, frequency-domain analysis techniques (such as fast Fourier transform) are used to identify the maximum value points, and the slope is calculated by numerical differentiation methods. Finally, a set of resonance intensity factors that can accurately reflect the importance of each resonance frequency point is obtained.

[0065] 302. Segment and discretize the composite parameter according to the time-domain interval corresponding to the load switching timestamp, and construct the anti-aliasing weight coefficient of the composite parameter based on the cosine similarity of the dynamic change rate and phase offset amount between adjacent time-domain intervals; In step 302, the composite parameter is segmented and discretized according to the time domain interval corresponding to the load switching timestamp, which means that the data is divided into multiple independent time periods based on the actual load switching time points. The anti-aliasing weight coefficient is constructed based on the cosine similarity of the dynamic change rate and the phase offset within adjacent time domain intervals, and is used to reduce the aliasing error that may occur during the data sampling process and ensure the accuracy of the composite parameter. The cosine similarity of the phase offset refers to the degree of association between two voltage or current signals by calculating the similarity of the phase angle differences between different time points (or time periods) in power system analysis. Specifically, the mathematical method of cosine similarity is used to compare the phase offsets to determine the similarity between them. The dynamic change rate within adjacent time domain intervals refers to the speed or amplitude at which the composite parameter changes with time in continuous time intervals.

[0066] In the embodiment of the present application, first, each independent time domain interval is determined according to the load switching timestamp, and the composite parameter is segmented and discretized according to these intervals. Then, the dynamic change rate and the phase offset are calculated within each interval, and the anti-aliasing weight coefficient is constructed using the cosine similarity algorithm. Specifically, the differential algorithm is used to calculate the dynamic change rate, and the phase offset is obtained through the phase detection method. Finally, these anti-aliasing weight coefficients are weighted to the composite parameter to form a more accurate data set.

[0067] 303. Use the resonance intensity factor as the row vector element and the composite parameter weighted by the anti-aliasing weight coefficient as the column vector element, and generate an initial matrix framework through the outer product operation of the row vector element and the column vector element; In step 303, the row vector element refers to the vector composed of the resonance intensity factors, and the column vector element is the vector composed of the composite parameters weighted by the anti-aliasing weight coefficient. The outer product operation is a mathematical operation used to generate a matrix form between two vectors and is used here to construct the initial matrix framework. The resonance intensity factor as an element of the row vector reflects the response intensity or sensitivity of the system at different frequencies. The initial matrix framework is a preliminary data analysis model constructed based on the resonance intensity factor and the weighted composite parameter, and is used to reveal the complex correlations within the data.

[0068] In the embodiment of the present application, first, the resonance intensity factor is used as the row vector element, and the composite parameter weighted by the anti-aliasing weight coefficient is used as the column vector element. Then, the initial matrix framework is generated through the outer product operation. Specifically, the outer product formula in linear algebra is used to perform the multiplication operation on the row vector and the column vector to obtain a preliminary matrix structure. This process not only integrates data from different sources but also provides a basis for subsequent optimization.

[0069] 304. Based on the initial matrix framework, for the row vector elements corresponding to each resonance frequency point, superimpose the amplitude-frequency integral value of the maximum value point within the preset frequency band to which the resonance frequency point belongs, and perform element-by-element correlation with the cumulative energy distribution of the composite parameter of the corresponding column vector element within the time domain interval to construct an initial coupling factor matrix.

[0070] In step 304, the initial matrix framework is based on the matrix structure generated in the previous step. The amplitude-frequency integral value is the result of integrating the amplitudes of the maximum value points within the preset frequency band, and the cumulative energy distribution is the total energy of the composite parameter within the time domain interval. Element-by-element correlation means corresponding and combining these values one by one to complete the initial coupling factor matrix. The resonance frequency point refers to the specific frequency position where resonance occurs in the power system. Resonance refers to the phenomenon that when the input frequency of the system matches the natural frequency of the system, the system response is amplified to a significant level. In power system analysis, identifying these resonance frequency points is crucial for evaluating system stability, optimizing performance, and fault detection. The initial coupling factor matrix is a matrix calculated based on the initial matrix framework, aiming to capture the interactions or correlations between different parameters.

[0071] In the embodiment of this application, first, for the row vector elements corresponding to each resonance frequency point in the initial matrix framework, superimpose the amplitude-frequency integral value of the maximum value point within the preset frequency band to which it belongs. Then, calculate the cumulative energy distribution of the composite parameter of the corresponding column vector element within the time domain interval and perform element-by-element correlation. Specifically, by integrating the amplitude-frequency response of each frequency point and combining the energy distribution in the time domain interval, the matrix elements are adjusted. Finally, a more accurate initial coupling factor matrix is constructed.

[0072] The following is a specific example: In an intelligent power distribution network project in a certain city, technicians first extract the adjacent maximum value points of each resonance frequency point from the set of dominant resonance frequencies of the selected cable section, calculate the frequency interval and amplitude attenuation slope, and generate a resonance intensity factor. Then, divide the time domain interval according to the load switching timestamp, and use the cosine similarity algorithm to construct an anti-aliasing weight coefficient. Then, use the resonance intensity factor as the row vector element and the composite parameter weighted by the anti-aliasing weight coefficient as the column vector element to generate an initial matrix framework. Finally, superimpose the amplitude-frequency integral value for each resonance frequency point and combine the cumulative energy distribution in the time domain interval to construct an initial coupling factor matrix that accurately describes the operating state of this cable section, providing solid data support for subsequent power grid maintenance and low voltage management.

[0073] In summary, through in-depth analysis of the dominant resonance frequency set and composite parameters in steps 301 to 304, combined with methods such as frequency-domain decomposition, time-domain interval division, outer product operation, and element-by-element correlation, an initial coupling factor matrix that accurately describes the complex dynamic behavior within the power system is successfully constructed. This method not only improves the accuracy of understanding the health status of the power grid but also provides scientific support for formulating effective low-voltage governance strategies, greatly enhancing the power grid management efficiency and power supply reliability. In addition, by introducing the anti-aliasing weight coefficient and amplitude-frequency integral value, the adaptability and accuracy of the model are further enhanced, making the assessment of the power grid operating state more comprehensive.

[0074] To solve the problem of accurately quantifying the load fluctuation impedance component in the power system, first, the data is divided into independent time-domain intervals based on the load switching timestamp, and the extreme points of the load fluctuation impedance component are extracted as sampling reference points within each interval. Then, local time windows are constructed around these reference points, and the mean value of the time-domain curvature change rate is calculated to generate the original parameter pairs. Finally, the scaled dynamic change rate and the corrected phase offset are non-linearly superimposed to obtain the composite parameter, which significantly improves the quantization accuracy of the load fluctuation characteristics. In some embodiments, step 202 of dividing the time domain interval of the load fluctuation impedance component according to the load switching timestamp and calculating the composite parameter of the dynamic change rate and phase offset of the load fluctuation impedance component within each time domain interval includes: 401. Based on the trigger interval of the load switching timestamp, the time window between adjacent timestamps is used as an independent time-domain interval, and the extreme points of the load fluctuation impedance component are extracted as the sampling reference points for the dynamic change rate within the independent time-domain interval; In step 401, the independent time-domain interval refers to the time period determined based on the trigger interval of the load switching timestamp. Each independent time-domain interval is used to analyze the change of the load fluctuation impedance component. The load switching timestamp is a data marker that records the exact time point when a specific load (such as industrial equipment, commercial buildings, or household appliances, etc.) is connected to or disconnected from the power grid in the power system. The sampling reference point is the extreme point of the load fluctuation impedance component extracted within this interval, which serves as the basic data point for calculating the dynamic change rate. These reference points can accurately reflect the maximum or minimum change points of the load fluctuation.

[0075] In the embodiments of the present application, first, the time window between adjacent timestamps is determined as an independent time-domain interval according to the load switching timestamp. Then, the extreme points of the load fluctuation impedance component are extracted within each independent time-domain interval and used as the sampling reference points for the dynamic change rate. Specifically, a peak detection algorithm is used to identify the extreme points and record their corresponding moments and values. Finally, a set of sampling reference point sets that can accurately describe the load fluctuation characteristics is obtained.

[0076] 402. Construct a local time window centered on the sampling reference point, calculate the mean value of the time-domain curvature change rate of the load fluctuation impedance component, and generate the original parameter pair of the dynamic change rate and the phase offset; In step 402, the local time window is a time period constructed centered on the sampling reference point, which is used to analyze the change trend of the load fluctuation impedance component in more detail. The mean value of the time-domain curvature change rate reflects the average change speed of the load fluctuation impedance component within this time period. The original parameter pair consists of the dynamic change rate and the phase offset, which is used to initially describe the load fluctuation characteristics.

[0077] In the embodiment of the present application, first, a local time window is constructed centered on the sampling reference point. Then, the mean value of the time-domain curvature change rate of the load fluctuation impedance component is calculated within each local time window. Finally, the original parameter pair of the dynamic change rate and the phase offset is generated. Specifically, a differential algorithm is used to calculate the curvature change rate, and the phase offset is obtained through a phase detection method. These parameters are combined into an original parameter pair to form a preliminary description of the load fluctuation.

[0078] 403. Perform cross-validation on the original parameter pair, apply a time window scaling constraint to the dynamic change rate through the load switching timestamp interval length of adjacent time domains, and perform phase polarity correction on the phase offset according to the continuity characteristics of adjacent time domains; In step 403, cross-validation is a process of performing multiple tests on the original parameter pair to ensure the accuracy and consistency of the data. The time window scaling constraint is to adjust the time window size of the dynamic change rate according to the load switching timestamp interval length of adjacent time domains. The phase polarity correction is to correct the phase offset according to the continuity characteristics of adjacent time domains to ensure the accuracy of the phase information. The load switching timestamp interval length refers to the time difference between the time points when different loads are connected (switched on) or disconnected (switched off) in the power system. These time points are called load switching timestamps, and the interval length of adjacent time domains reflects the frequency and pattern of load changes. The phase offset refers to the phase angle difference of voltage or current signals in an AC power system.

[0079] In the embodiment of the present application, first, cross-validation is performed on the original parameter pair, and a time window scaling constraint is applied to the dynamic change rate through the load switching timestamp interval length of adjacent time domains. Then, phase polarity correction is performed on the phase offset according to the continuity characteristics of adjacent time domains. Specifically, a time series analysis method is used to adjust the time window of the dynamic change rate, and the phase offset is corrected through a phase continuity check. Finally, a set of corrected dynamic change rate and phase offset is obtained.

[0080] 404. Nonlinearly superimpose the scaled dynamic change rate and the corrected phase offset, and weight them by the energy proportion coefficient of the load fluctuation impedance component within the corresponding time domain interval to obtain a composite parameter.

[0081] In step 404, the non - linear superposition is a process of combining the scaled dynamic change rate and the corrected phase offset, aiming to integrate the information of both. The energy proportion coefficient is the energy ratio of the load fluctuation impedance component within the corresponding time domain interval, which is used to weight the composite parameter to ensure that it can accurately reflect the impact of actual load fluctuations. The scaled dynamic change rate refers to the speed at which electrical parameters such as voltage or current change over time after adjustment. The corrected phase offset is the phase difference between voltage or current signals after phase polarity correction. In an AC power system, the phase offset describes the relative position relationship between different signals or the same signal at different times.

[0082] In the embodiment of this application, first, non - linearly superimpose the scaled dynamic change rate and the corrected phase offset. Then, weight them by the energy proportion coefficient of the load fluctuation impedance component within the corresponding time domain interval to obtain a composite parameter. Specifically, use a non - linear function to combine the dynamic change rate and the phase offset, and adjust the weight by the energy proportion coefficient. Finally, obtain a set of composite parameters that can comprehensively describe the load fluctuation characteristics.

[0083] The following is a specific example: In an intelligent power distribution network project in a certain city, technicians first determine the time window between adjacent timestamps as an independent time domain interval according to the load switching timestamps, and extract the extreme points of the load fluctuation impedance component within each interval as the sampling reference points of the dynamic change rate. Then, construct local time windows centered on these reference points to generate the original parameter pairs of the dynamic change rate and the phase offset. Next, perform cross - validation on the original parameter pairs and correct the phase offset according to the continuity characteristics. Finally, non - linearly superimpose the scaled dynamic change rate and the corrected phase offset, and weight them by the energy proportion coefficient to obtain a set of composite parameters that can comprehensively describe the load fluctuation characteristics, providing solid data support for subsequent power grid maintenance and low - voltage governance.

[0084] In summary, steps 401 to 404, through in-depth analysis of the impedance component of load fluctuations, combined with methods such as time-domain interval division, local time window construction, cross-validation, and non-linear superposition, successfully generated a set of composite parameters that accurately describe the characteristics of load fluctuations. This method not only improves the accuracy of understanding load fluctuations but also provides a scientific basis for formulating effective low-voltage governance strategies, greatly improving the efficiency of power grid management and power supply reliability. In addition, by introducing time-window scaling constraints and phase polarity corrections, the adaptability and accuracy of the model are further enhanced.

[0085] To solve the problem of accurately describing complex dynamic behaviors in the power system, first, expand the three-phase voltage unbalance parameters into multi-channel pulse sequences. Then, perform circular convolution on the multi-channel pulse sequences within a sliding time window, and automatically expand the convolution kernel function when impedance parameter mutations are detected. Finally, through energy threshold screening of the convolution output results, a distortion feature vector containing the combined spectral characteristics of impedance and voltage is obtained, effectively characterizing the line health state. In some embodiments, step 103 of performing time-domain convolution operation on the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameters to generate a distortion feature vector of the combined line impedance and voltage includes: 501. Based on the orthogonal decomposition relationship between the inherent impedance component and the load fluctuation impedance component in the impedance parameter coupling factor matrix, construct a three-dimensional convolution kernel function with time-varying characteristics; In step 501, the inherent impedance component and the load fluctuation impedance component are two main components separated based on the line impedance decomposition dataset. The orthogonal decomposition relationship refers to the characteristic that these two impedance components are independent of each other and do not affect each other. The three-dimensional convolution kernel function is a mathematical tool used to process multi-dimensional data (such as time, frequency, and spatial dimensions), with time-varying characteristics, capable of adapting to data changes in different time periods.

[0086] In the embodiments of the present application, first analyze the orthogonal decomposition relationship between the inherent impedance component and the load fluctuation impedance component in the impedance parameter coupling factor matrix. Then, construct a three-dimensional convolution kernel function with time-varying characteristics based on these relationships. Specifically, use tensor decomposition methods to extract the time series characteristics of the inherent impedance and load fluctuations, and construct the corresponding convolution kernel. Finally, a three-dimensional convolution kernel function that can dynamically adapt to grid changes is obtained.

[0087] 502. Expand the three-phase voltage unbalance parameters into multi-channel pulse sequences according to a preset time granularity, and map each channel to the differential increment of the voltage amplitude corresponding to the phase; In step 502, the preset time granularity refers to dividing time into time periods of fixed length for facilitating data analysis. The multi-channel pulse sequence is the result of expanding the three-phase voltage imbalance parameters according to the time granularity, and each channel corresponds to the differential increment of the voltage amplitude of one phase. This representation helps to capture the change trend of the voltage over time and its subtle differences. The differential increment of the voltage amplitude refers to the change rate of the change amount of the voltage amplitude of a certain phase with respect to time within a specific time interval, reflecting the increase or decrease of the voltage within a very short time period.

[0088] In the embodiment of the present application, first, the three-phase voltage imbalance parameters are expanded into a multi-channel pulse sequence according to the preset time granularity. Then, each channel is mapped to the differential increment of the voltage amplitude of the corresponding phase. Specifically, the differential algorithm is used to calculate the differential increment of each phase voltage and organize it into a multi-channel pulse sequence. Finally, a group of multi-channel pulse sequences that can accurately describe the characteristics of the three-phase voltage imbalance is obtained.

[0089] 503. Within the sliding time window, perform tensor slicing on the three-dimensional convolution kernel function along the time axis, and perform a cyclic convolution with asymmetric boundary constraints on the multi-channel pulse sequence. When the impedance parameter mutates, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant decay function; In step 503, the sliding time window is a time period of fixed length that moves in the time domain and is used to analyze data segment by segment. Tensor slicing is to extract data segments of a specific time period from the three-dimensional convolution kernel function. The cyclic convolution with asymmetric boundary constraints is a method for dealing with boundary effects to ensure that the convolution operation can also be correctly executed at the boundary. The hyperbolic secant decay function is a special mathematical function used to describe the behavior when the impedance parameter mutates.

[0090] In the embodiment of the present application, first, within the sliding time window, perform tensor slicing on the three-dimensional convolution kernel function along the time axis. Then, perform a cyclic convolution with asymmetric boundary constraints on the multi-channel pulse sequence. When it is detected that the impedance parameter mutates, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant decay function. Specifically, set a mutation detection window to monitor the change of the impedance parameter in real time and adjust the convolution kernel form when a mutation occurs. Finally, a group of output results after convolution processing is obtained.

[0091] 504. Extract the interaction strength coefficient between the voltage trajectory and the impedance trajectory in the convolution output result, and construct a multi-dimensional tensor containing the joint spectral characteristics of the impedance and the voltage. Perform energy threshold screening on the multi-dimensional tensor along the impedance coupling main axis, and use the projection components exceeding the energy threshold as the distortion eigenvectors.

[0092] In step 504, the interaction strength coefficient is an index to measure the degree of mutual influence between the phase voltage trajectories and the impedance trajectories. A multi-dimensional tensor is a high-dimensional array structure used to store data in multiple dimensions. Energy threshold screening is a method of selectively retaining data components with larger energy to highlight important feature information. The distortion feature vector is a multi-dimensional data structure used to characterize the complex interaction relationship between line impedance and voltage in a power system.

[0093] In the embodiment of the present application, first, the interaction strength coefficients of the phase voltage trajectories and the impedance trajectories are extracted from the convolution output results. Then, a multi-dimensional tensor containing the joint spectral characteristics of impedance and voltage is constructed. Next, energy threshold screening is performed on the multi-dimensional tensor along the main axis of impedance coupling to retain the projection components exceeding the energy threshold. Specifically, the joint spectral characteristics are extracted through spectral analysis techniques, and the energy threshold screening algorithm is applied to highlight important features. Finally, a set of distortion feature vectors that can comprehensively describe the joint characteristics of line impedance and voltage is obtained.

[0094] The following is a specific example: In an intelligent power distribution network project in a certain city, the technical personnel first constructed a three-dimensional convolution kernel function with time-varying characteristics based on the orthogonal decomposition relationship between the inherent impedance component and the load fluctuation impedance component in the impedance parameter coupling factor matrix. Then, the three-phase voltage unbalance parameters are expanded into a multi-channel pulse sequence according to a preset time granularity. Next, within a sliding time window, circular convolution with asymmetric boundary constraints is performed on the multi-channel pulse sequence. When a mutation in the impedance parameter is detected, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant type attenuation function. Finally, the interaction strength coefficients of the phase voltage trajectories and the impedance trajectories are extracted from the convolution output results, a multi-dimensional tensor is constructed and energy threshold screening is performed to obtain a set of distortion feature vectors that can comprehensively describe the joint characteristics of line impedance and voltage, providing solid data support for subsequent power grid maintenance and low-voltage management.

[0095] In summary, steps 501 to 504, through in-depth analysis of the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameters, combined with methods such as three-dimensional convolution kernel functions and time-domain convolution operations, successfully generated a set of distortion feature vectors that accurately describe the joint characteristics of line impedance and voltage. This method not only improves the accuracy of understanding the health status of the power grid but also provides a scientific basis for formulating effective low-voltage management strategies, greatly improving the power grid management efficiency and power supply reliability. In addition, by introducing the hyperbolic secant type attenuation function and energy threshold screening, the adaptability and accuracy of the model are further enhanced.

[0096] To solve the problem of accurately describing the dynamic behavior of the power system when impedance parameters mutate, a mutation detection window is set on the time axis of the three-dimensional convolution kernel function to calculate the instantaneous change rate in real time, and the morphology of the convolution kernel function is adjusted according to the triggered expansion mechanism. In addition, an energy conservation constraint is imposed on the frequency domain components of the triggered expansion mechanism to ensure that the total energy remains consistent. According to the phase difference change trajectory between the impedance parameters and the three-phase voltage unbalance parameters, the triggered expansion mechanism is dynamically corrected, enabling the system to adapt to rapidly changing grid conditions and providing stable and reliable evaluation results. In some embodiments, when the impedance parameters mutate in step 503, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant type attenuation function, including: 601. Set a mutation detection window on the time axis of the three-dimensional convolution kernel function, and calculate the instantaneous change rate of the load fluctuation impedance component in the impedance parameter coupling factor matrix and the triggered expansion mechanism of the hyperbolic secant type attenuation function in real time; In step 601, the mutation detection window refers to a specific time period set on the time axis for real-time monitoring of the change in impedance parameters. The instantaneous change rate is the speed at which the load fluctuation impedance component changes with time, used to identify the occurrence of mutations. The triggered expansion mechanism is a predefined rule or algorithm for determining when and how to adjust the morphology of the three-dimensional convolution kernel function to adapt to mutation situations.

[0097] In the embodiments of the present application, first, a mutation detection window is set on the time axis of the three-dimensional convolution kernel function, and the instantaneous change rate of the load fluctuation impedance component is calculated in real time. Then, according to the preset triggered expansion mechanism, it is determined whether the three-dimensional convolution kernel function needs to be expanded. Specifically, the sliding window technique is used to monitor the data in real time, and the numerical differentiation method is used to calculate the instantaneous change rate. Once a significant change is detected, the triggered expansion mechanism is activated to prepare for adjusting the convolution kernel function. Finally, a set of data sets that can accurately reflect the mutation situation of the impedance parameters is obtained.

[0098] 602. Impose an energy conservation constraint on the frequency domain components of the triggered expansion mechanism, and keep the total energy of the triggered expansion mechanism consistent before and after the mutation through the energy integration of the load fluctuation impedance component; In step 602, the energy conservation constraint refers to the principle of ensuring that the total energy of the system remains unchanged before and after the mutation. The frequency domain components are the various frequency components after the time domain signal is transformed into the frequency domain. Keeping the total energy of the triggered expansion mechanism consistent before and after the mutation through energy integration ensures the stability of the system. The energy conservation constraint means that when analyzing or processing certain frequency domain components (such as frequency changes caused by load fluctuations), the total energy of these components remains unchanged before and after the entire process. The triggered expansion mechanism refers to a program or algorithm mechanism that can be activated when detecting specific conditions or events (such as voltage dips, surges, or other power quality problems).

[0099] In the embodiments of the present application, first, an energy conservation constraint is imposed on the frequency-domain components that trigger the expansion mechanism. Then, the consistency of the total energy before and after the mutation is maintained by integrating the energy of the load fluctuation impedance component. Specifically, the fast Fourier transform is used to convert the time-domain signal into a frequency-domain representation, and the energy integration formula is applied to calculate the energy of each frequency band. Then, according to the principle of energy conservation, the parameters in the trigger expansion mechanism are adjusted to ensure that the total energy of the system remains unchanged before and after the mutation. Finally, a trigger expansion mechanism corrected by energy conservation is obtained.

[0100] 603. Dynamically correct the constrained trigger expansion mechanism according to the phase difference change trajectory of the impedance parameter and the three-phase voltage unbalance parameter within the mutation detection window. When the instantaneous change rate exceeds the corrected trigger expansion mechanism, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant type attenuation function.

[0101] In step 603, the phase difference change trajectory refers to the change of the phase difference between the impedance parameter and the three-phase voltage unbalance parameter within the mutation detection window over time. Dynamically correcting the constrained trigger expansion mechanism means further optimizing the trigger expansion mechanism according to the actually observed change trajectory to improve its accuracy. The three-dimensional convolution kernel function can be used to detect and analyze the interaction between different parameters (such as the impedance parameter and the three-phase voltage unbalance parameter) and its change over time. The hyperbolic secant type attenuation function is used to adjust the analysis model to more accurately simulate the influence of these drastic changes and may help better understand and predict the abnormal behavior in the power system.

[0102] In the embodiments of the present application, first, the constrained trigger expansion mechanism is dynamically corrected according to the phase difference change trajectory of the impedance parameter and the three-phase voltage unbalance parameter within the mutation detection window. Then, the instantaneous change rate is monitored in real time and compared with the corrected trigger expansion mechanism. Once the instantaneous change rate exceeds the threshold, the three-dimensional convolution kernel function is immediately adjusted to expand into a hyperbolic secant type attenuation function. Specifically, a phase detection algorithm is used to track the phase difference change and the trigger expansion mechanism is adjusted according to the actual change situation. Finally, a three-dimensional convolution kernel function that can adapt to the mutation situation is obtained, so as to more accurately describe the dynamic behavior of the power grid.

[0103] The following is a specific example: In an intelligent power distribution network project in a certain city, technicians first set a mutation detection window on the time axis of the three-dimensional convolution kernel function and calculated the instantaneous change rate of the load fluctuation impedance component in real time. The technicians used the sliding window technique and numerical differentiation method to calculate the instantaneous change rate and initiated the trigger expansion mechanism. Then, an energy conservation constraint was imposed on the frequency domain components of the trigger expansion mechanism, and the total energy before and after the mutation was maintained consistent through the energy integration of the load fluctuation impedance component. Once the instantaneous change rate exceeded the corrected trigger expansion mechanism, the three-dimensional convolution kernel function automatically expanded into a hyperbolic secant decay function. This method not only improved the accuracy of understanding the power grid health status but also provided a scientific basis for formulating effective low-voltage governance strategies, greatly improving the power grid management efficiency and power supply reliability.

[0104] In summary, steps 601 to 603, through the real-time monitoring and analysis of impedance parameter mutations, combined with the energy conservation constraint and the dynamic correction mechanism, successfully achieved the adaptive expansion of the three-dimensional convolution kernel function. This method can not only accurately capture the mutation phenomena in the power grid but also effectively maintain the stability of the system, avoiding abnormal fluctuations caused by mutations. In addition, by introducing the hyperbolic secant decay function, the adaptability and accuracy of the model are further enhanced, making the evaluation of the power grid operation status more efficient.

[0105] To solve the accuracy and reliability problems of data quality assessment in the system, first, a non-linear correlation index of each monitoring node is established through the phase sensitivity kernel function, and a probability density diffusion surface is constructed. Then, this surface is convolved and fused with the inherent impedance spectrum characteristics at multiple scales to generate a dynamic weight base value. Finally, a weight correction equation based on the double constraints of impedance and charge is established to obtain the final data credibility weight distribution, providing a scientific basis for low-voltage governance measures. In some embodiments, calculating the data credibility weight distribution of all monitoring nodes on the distribution network line according to the non-linear correlation between the three-phase unbalance dataset and the distortion feature vector in step 104 includes: 701. Multidimensionally couple the phase voltage unbalance parameters in the three-phase unbalance dataset with the distortion feature vector in the time domain dimension, and establish a non-linear correlation index of each monitoring node through the phase sensitivity kernel function; In step 701, the phase voltage unbalance parameters refer to information such as the effective value and phase angle deviation of each phase voltage recorded in the three-phase unbalance dataset. The distortion feature vector is a multi-dimensional data structure that describes the combined characteristics of line impedance and voltage. The phase sensitivity kernel function is a mathematical model used to quantify the non-linear correlation between different monitoring nodes. The non-linear correlation index reflects the complex relationship between voltage unbalance and distortion characteristics among each monitoring node.

[0106] In the embodiments of the present application, first, the voltage imbalance parameters of each phase in the three-phase imbalance degree dataset and the distortion feature vectors are multi-dimensionally coupled in the time domain. Then, a non-linear correlation degree index of each monitoring node is established through a phase sensitivity kernel function. Specifically, time series analysis technology is used to synchronize the voltage imbalance parameters of each phase and the distortion feature vectors, and the phase sensitivity kernel function is applied to calculate their non-linear correlation degree. Finally, a set of non-linear correlation degree indexes that can accurately reflect the dynamic relationship between each monitoring node are obtained.

[0107] 702. Based on the spatial distribution pattern of the non-linear correlation degree index in the three-dimensional charge density gradient field, construct a probability density diffusion surface reflecting the line impedance coupling strength; In step 702, the three-dimensional charge density gradient field is a spatial distribution pattern describing the change of charge density in the cable joint area generated based on the space charge detection technology. The distribution pattern of the non-linear correlation degree index in this field can reveal the spatial characteristics of the line impedance coupling strength. The probability density diffusion surface is a probability distribution model used to represent the change of these characteristics with spatial position.

[0108] In the embodiments of the present application, first, based on the spatial distribution pattern of the non-linear correlation degree index in the three-dimensional charge density gradient field, construct a probability density diffusion surface reflecting the line impedance coupling strength. Specifically, a spatial interpolation method (such as Kriging interpolation) is used to construct the charge density gradient field, and the probability density diffusion surface is generated in combination with the non-linear correlation degree index. Finally, a surface model that can comprehensively describe the spatial distribution of the line impedance coupling strength is obtained.

[0109] 703. Perform multi-scale convolution fusion on the probability density diffusion surface and the inherent impedance spectrum characteristics in the line impedance decomposition dataset to generate the dynamic weight base values of each monitoring node; In step 703, the inherent impedance spectrum characteristics are the frequency response characteristics extracted from the line impedance decomposition dataset. The multi-scale convolution fusion of the probability density diffusion surface and the inherent impedance spectrum characteristics refers to performing convolution operations on the two at different scales to generate the dynamic weight base values of each monitoring node. The dynamic weight base value is a basic weight value reflecting the importance of the data of each monitoring node.

[0110] In the embodiments of the present application, first, perform multi-scale convolution fusion on the probability density diffusion surface and the inherent impedance spectrum characteristics in the line impedance decomposition dataset. Specifically, a multi-scale convolutional neural network is used to process the two and generate the dynamic weight base values of each monitoring node. The convolution operation is performed at different scales to capture various features from local to global. Finally, a set of dynamic weight base values that can accurately reflect the importance of the data of each monitoring node are obtained.

[0111] 704. Establish a weight correction equation based on the double constraints of impedance and charge, perform path integral optimization on the dynamic weight base value, solve the steady-state solution set of the weight correction equation under the action of the charge migration effect through an iterative convergence algorithm, and obtain the data credibility weight distribution according to the projection amplitude of the steady-state solution set in the monitoring node space coordinate system.

[0112] In step 704, the double constraints of impedance and charge refer to the influencing factors that simultaneously consider the line impedance and the change in space charge density. The weight correction equation is a mathematical model for optimizing the dynamic weight base value. Path integral optimization is a method for solving the weight correction equation, and the steady-state solution set is found through an iterative convergence algorithm. The steady-state solution set is the optimal weight allocation scheme under the action of the charge migration effect. The projection amplitude refers to the numerical size obtained by mapping multi-dimensional data to a specific dimension (usually one or several axes of the space coordinate system).

[0113] In the embodiment of the present application, first, a weight correction equation based on the double constraints of impedance and charge is established. Then, path integral optimization is performed on the dynamic weight base value, and the weight correction equation is solved through an iterative convergence algorithm. Specifically, the path integral optimization algorithm is used to adjust the parameters in the weight correction equation, and the steady-state solution set is found through multiple iterations. Finally, the data credibility weight distribution is obtained according to the projection amplitude of the steady-state solution set in the monitoring node space coordinate system. Finally, a set of weight distributions that can accurately reflect the data credibility of each monitoring node is obtained.

[0114] The following is a specific example: In an intelligent power distribution network project in a certain city, technicians first perform multi-dimensional coupling of the phase voltage imbalance parameters and the distortion feature vectors in the time domain dimension in the three-phase imbalance degree dataset, and establish a non-linear correlation index. Then, a probability density diffusion surface reflecting the line impedance coupling strength is constructed. Then, the dynamic weight base value of each monitoring node is generated. Finally, the steady-state solution set of the weight correction equation under the action of the charge migration effect is solved through an iterative convergence algorithm to obtain the data credibility weight distribution. For example, in a certain cable section, whenever a large device starts or stops, it will cause significant voltage imbalance and impedance changes. Technicians use the above method to generate the data credibility weight distribution and find that the data of some monitoring nodes is more reliable, thus providing a scientific basis for subsequent maintenance work.

[0115] In summary, steps 701 to 704, through in-depth analysis of the three-phase unbalance dataset and the distortion feature vector, combined with methods such as the phase sensitivity kernel function, probability density diffusion surface, multi-scale convolution fusion, and path integral optimization, successfully generated a set of data credibility weight distributions that accurately describe the global monitoring nodes of the distribution network lines. This method not only improves the understanding accuracy of the power grid health status but also provides a scientific basis for formulating effective low-voltage governance strategies.

[0116] Figure 2 The following is a schematic structural diagram of a data quality assessment system provided by an embodiment of the present application. As Figure 2 shown, the system includes: An acquisition module 21, which acquires a three-phase unbalance dataset, a free charge density distribution dataset generated based on millimeter-wave radar space charge detection technology, and a line impedance decomposition dataset generated through impedance spectrum dynamic reconstruction technology during the operation state of the distribution network line; A mapping module 22 that performs spatial topological mapping between the extreme points of the charge density gradient in the cable joint area of the free charge density distribution dataset and the voltage phase angle in the three-phase unbalance dataset, and generates three-phase voltage unbalance parameters with corrected phase deviation through charge migration effect compensation calculation; A establishment module 23 that establishes an impedance parameter coupling factor matrix based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition dataset, and performs a time-domain convolution operation on the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameters to generate a distortion feature vector of the combined line impedance and voltage; A characterization module 24 that calculates the data credibility weight distribution of the global monitoring nodes of the distribution network line according to the non-linear correlation degree between the three-phase unbalance dataset and the distortion feature vector, and the data credibility weight distribution is used to characterize the quality assessment result of the low-voltage governance data.

[0117] Figure 2 The above-mentioned data quality assessment system can execute Figure 1 the data quality assessment method described in the embodiment shown. Its implementation principle and technical effects will not be elaborated again. For each module and unit in the above-mentioned data quality assessment system for the specific manner of performing operations, it has been described in detail in the embodiment related to the method, and will not be elaborated in detail here.

[0118] In a possible design, Figure 2 the data quality assessment system described in the embodiment shown can be implemented as a computing device. As Figure 3 shown, the computing device can include a storage component 31 and a processing component 32; The storage component 31 stores one or more computer instructions, and the one or more computer instructions are called and executed by the processing component 32.

[0119] The processing component 32 is used for the above Figure 1 A data quality assessment method of the above embodiment.

[0120] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0121] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0122] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0123] The input / output interface provides an interface between the processing component and the peripheral interface module, and the peripheral interface module may be an output device, an input device, etc.

[0124] The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc.

[0125] Among them, the computing device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device may refer to a cloud server, and the above processing component, storage component, etc. may be basic server resources leased or purchased from a cloud computing platform.

[0126] The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above Figure 1 A data quality assessment method of the above embodiment.

[0127] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0128] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0129] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A data quality assessment method, characterized in that Including: Under the operating state of the distribution network line, obtaining a three-phase unbalance degree data set, a free charge density distribution data set generated based on the millimeter-wave radar space charge detection technology, and a line impedance decomposition data set generated through the impedance spectrum dynamic reconstruction technology; Performing spatial topological mapping on the extreme points of the charge density gradient in the cable joint area in the free charge density distribution data set and the voltage phase angle in the three-phase unbalance degree data set, and generating three-phase voltage unbalance parameters with corrected phase deviation through charge migration effect compensation calculation; Based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition data set, establishing an impedance parameter coupling factor matrix, and performing a time-domain convolution operation on the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameters to generate a distortion feature vector of the joint of the line impedance and voltage; According to the non-linear correlation degree between the three-phase unbalance degree data set and the distortion feature vector, calculating the data credibility weight distribution of the global monitoring nodes of the distribution network line, and the data credibility weight distribution is used to characterize the quality evaluation result of the low voltage governance data.

2. The method according to claim 1, wherein The establishing an impedance parameter coupling factor matrix based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition data set includes: Performing frequency-domain decomposition on the line inherent impedance component and extracting the maximum value points of the steady-state amplitude-frequency response curve within a preset frequency band, and generating a set of dominant resonance frequencies based on the distribution range of the maximum value points; Dividing the time domain interval of the load fluctuation impedance component according to the load switching time stamp, and calculating the composite parameter of the dynamic change rate and the phase offset amount of the load fluctuation impedance component within each time domain interval; Constructing an initial coupling factor matrix with the set of dominant resonance frequencies as the row index and the composite parameter as the column index, and applying dynamic constraints to the initial coupling factor matrix, including generating a spatial correction coefficient based on the impedance transfer function of adjacent monitoring nodes, and generating a statistical correction coefficient based on the historical probability density kurtosis and skewness of the load fluctuation impedance component; Superimposing the correlation degree of the spatial correction coefficient and the statistical correction coefficient on the diagonal elements of the matrix to establish an impedance parameter coupling factor matrix.

3. The method according to claim 2, wherein The constructing an initial coupling factor matrix with the set of dominant resonance frequencies as the row index and the composite parameter as the column index includes: In the set of dominant resonance frequencies, extracting the frequency interval and the amplitude attenuation slope between adjacent maximum value points of each resonance frequency point on the steady-state amplitude-frequency response curve, and generating a resonance intensity factor; Segmenting and discretizing the composite parameter according to the time domain interval corresponding to the load switching time stamp, and constructing an anti-aliasing weight coefficient of the composite parameter according to the cosine similarity of the dynamic change rate and the phase offset amount within adjacent time domain intervals; Taking the resonance intensity factor as the row vector element and the composite parameter weighted by the anti-aliasing weight coefficient as the column vector element, and generating an initial matrix framework through the outer product operation of the row vector element and the column vector element; Based on the initial matrix framework, for the row vector elements corresponding to each resonant frequency point, the amplitude-frequency integral value of the maximum value point within the preset frequency band to which the corresponding resonant frequency point belongs is superimposed, and element-by-element correlation is performed with the cumulative energy distribution of the composite parameter of the corresponding column vector element within the time domain interval to construct an initial coupling factor matrix.

4. The method according to claim 2, wherein The division of the load fluctuating impedance component into time domain intervals according to the load switching timestamp and the calculation of the composite parameter of the dynamic change rate and phase offset of the load fluctuating impedance component within each time domain interval include: Based on the trigger interval of the load switching timestamp, the time window between adjacent timestamps is used as an independent time domain interval, and the extreme value point of the load fluctuating impedance component within the independent time domain interval is extracted as the sampling reference point of the dynamic change rate; A local time window is constructed centered on the sampling reference point, and the mean value of the time domain curvature change rate of the load fluctuating impedance component is calculated to generate the original parameter pair of the dynamic change rate and phase offset; Cross-validation is performed on the original parameter pair. A time window scaling constraint is imposed on the dynamic change rate through the load switching timestamp interval length of adjacent time domain intervals, and the phase polarity of the phase offset is corrected according to the continuity characteristics of adjacent time domain intervals; The scaled dynamic change rate and the corrected phase offset are non-linearly superimposed, and weighted by the energy proportion coefficient of the load fluctuating impedance component within the corresponding time domain interval to obtain the composite parameter.

5. The method according to claim 1, wherein The time domain convolution operation of the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameter to generate the distortion feature vector of the line impedance and voltage joint includes: Based on the orthogonal decomposition relationship between the inherent impedance component and the load fluctuating impedance component in the impedance parameter coupling factor matrix, a three-dimensional convolution kernel function with time-varying characteristics is constructed; The three-phase voltage unbalance parameter is expanded into a multi-channel pulse sequence according to a preset time granularity, and each channel is mapped to the differential increment of the voltage amplitude corresponding to the phase; Within a sliding time window, the three-dimensional convolution kernel function is sliced along the time axis in a tensor manner, and an asymmetric boundary-constrained circular convolution is performed on the multi-channel pulse sequence. When the impedance parameter mutates, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant type attenuation function; The interaction intensity coefficient of each phase voltage trajectory and impedance trajectory in the convolution output result is extracted, and a multi-dimensional tensor containing the joint spectral characteristics of impedance and voltage is constructed. Energy threshold screening is performed on the multi-dimensional tensor along the impedance coupling main axis, and the projection component exceeding the energy threshold is used as the distortion feature vector.

6. The method according to claim 5, wherein When the impedance parameter mutates, the three-dimensional convolution kernel function automatically expands into a hyperbolic secant type attenuation function, including: A mutation detection window is set on the time axis of the three-dimensional convolution kernel function, and the instantaneous change rate of the load fluctuating impedance component in the impedance parameter coupling factor matrix and the trigger expansion mechanism of the hyperbolic secant type attenuation function are calculated in real time; An energy conservation constraint is imposed on the frequency domain component of the trigger expansion mechanism, and the total energy of the trigger expansion mechanism before and after mutation is maintained through the energy integral of the load fluctuating impedance component. According to the phase difference change trajectory between the impedance parameter and the three-phase voltage unbalance parameter within the mutation detection window, the constrained trigger expansion mechanism is dynamically corrected. When the instantaneous change rate exceeds the corrected trigger expansion mechanism, the three-dimensional convolution kernel function is automatically expanded into a hyperbolic secant type attenuation function.

7. The method according to claim 1, wherein Calculating the data credibility weight distribution of all monitoring nodes on the distribution network line according to the non-linear correlation degree between the three-phase unbalance degree data set and the distortion feature vector includes: Performing multi-dimensional coupling of the phase voltage unbalance parameters in the three-phase unbalance degree data set and the distortion feature vector in the time domain dimension, and establishing a non-linear correlation degree index for each monitoring node through a phase sensitivity kernel function; Based on the spatial distribution form of the non-linear correlation degree index in the three-dimensional charge density gradient field, constructing a probability density diffusion surface reflecting the line impedance coupling strength; Performing multi-scale convolution fusion of the probability density diffusion surface and the inherent impedance spectrum characteristics in the line impedance decomposition data set to generate the dynamic weight base value of each monitoring node; Establishing a weight correction equation based on the double constraints of impedance and charge, performing path integral optimization on the dynamic weight base value, solving the steady-state solution set of the weight correction equation under the action of the charge migration effect through an iterative convergence algorithm, and obtaining the data credibility weight distribution according to the projection amplitude of the steady-state solution set in the monitoring node space coordinate system.

8. A data quality assessment system, characterized in that, Including: An acquisition module that, under the operating state of the distribution network line, acquires a three-phase unbalance degree data set, a free charge density distribution data set generated based on the millimeter-wave radar space charge detection technology, and a line impedance decomposition data set generated through the impedance spectrum dynamic reconstruction technology; A mapping module that performs spatial topological mapping of the charge density gradient extreme points in the cable joint area in the free charge density distribution data set and the voltage phase angle in the three-phase unbalance degree data set, and generates the three-phase voltage unbalance parameter after phase deviation correction through charge migration effect compensation calculation; A establishment module that, based on the line inherent impedance component and the load fluctuation impedance component separated from the line impedance decomposition data set, establishes an impedance parameter coupling factor matrix, and performs time domain convolution operation on the impedance parameter coupling factor matrix and the three-phase voltage unbalance parameter to generate a distortion feature vector of the joint of the line impedance and the voltage; A characterization module that calculates the data credibility weight distribution of all monitoring nodes on the distribution network line according to the non-linear correlation degree between the three-phase unbalance degree data set and the distortion feature vector, and the data credibility weight distribution is used to characterize the quality evaluation result of the low voltage governance data.

9. A computing device, characterized in that, Including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a data quality evaluation method according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that, A computer program is stored, and when the computer program is executed by the computer, it implements a data quality evaluation method according to any one of claims 1 to 7.

Images