Power supply reliability analysis method and system based on operation and maintenance characteristics of power distribution network

Abstract

The invention discloses a power supply reliability analysis method and system based on operation and maintenance characteristics of a power distribution network, and particularly relates to the technical field of power supply analysis of the power distribution network. Harmonic data, including harmonic current, harmonic voltage and frequency components, of each node in a power distribution network system are collected in real time; in combination with the topological structure of the power distribution network, the sparse linear equation set is solved through L1 norm optimization, and the main source of harmonic waves is identified; the contribution degree of each nonlinear load to the harmonic distortion of a power distribution network system is calculated, a list of key harmonic sources is screened out through Pareto analysis, a filter configuration optimization model is constructed based on the identified key harmonic sources, and the harmonic distortion level and the filter installation cost are used as constraint functions to optimize the filter configuration. The installation position and the parameter configuration of the filter serve as decision variables, an optimization algorithm is adopted for solving, and the recommended installation position and the parameter configuration of the filter are output; the harmonic suppression effects under different load conditions are evaluated, and the problem that the power supply reliability is reduced due to harmonic distortion is effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply analysis of distribution networks, and more specifically, to a power supply reliability analysis method and system based on distribution network operation and maintenance characteristics. Background Art

[0002] In modern distribution network systems, the widespread use of nonlinear loads such as industrial inverters, electric vehicle charging piles and household electronic devices has significantly increased the probability of harmonic distortion. Harmonic distortion can cause equipment overheating, insulation aging, power system oscillation and even failure, seriously threatening power supply reliability. For example, in a 10kV distribution network, the access of electric vehicle charging piles may cause the total harmonic distortion of local nodes to increase from 3% to 15%, directly affecting the power supply quality of nearby residential areas, highlighting the urgent need for harmonic control.

[0003] Existing harmonic control technologies have many deficiencies and are unable to meet the requirements of dynamic operation scenarios of distribution networks. First, traditional methods mostly adopt passive monitoring and post-repair strategies, such as locating problems through power quality monitors after a fault occurs, but lack the ability to actively identify and prevent harmonic sources. Existing technologies mainly rely on static models for filter configuration optimization and fail to fully consider the dynamic nature of distribution network operation characteristics. For example, filter parameters are not adjusted according to real-time load fluctuations, which can easily lead to resonance risks or over-compensation.

[0004] Therefore, how to accurately identify the main sources of harmonics in the distribution network through data analysis technology, and dynamically optimize the layout of the filter to reduce the impact of harmonic distortion on power supply reliability has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a power supply reliability analysis method and system based on the operation and maintenance characteristics of the distribution network. By constructing a harmonic contribution matrix and adopting an L1 norm optimization method, it can effectively identify key harmonic sources, and optimize the filter configuration in combination with a dynamic adjustment strategy, thereby significantly improving power supply reliability and solving the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a power supply reliability analysis method based on distribution network operation and maintenance characteristics, comprising the following steps: Step S11: Using smart meters and power quality monitors, real-time collection of harmonic data of each node in the distribution network system, including harmonic current, harmonic voltage and frequency components; Furthermore, the collected data needs to be cleaned and standardized to remove outliers and noise to ensure data quality; Step S12. Identify the main sources of harmonics: Combine the topology of the distribution network and optimize the solution of the sparse linear equations through the L1 norm to identify the main sources of harmonics; Calculate the contribution degree of each nonlinear load to the harmonic distortion of the distribution network system, and use Pareto analysis to screen out the list of key harmonic sources, including the node numbers and the corresponding contribution degrees; Step S13. Optimize the filter configuration: Based on the identified key harmonic sources, construct an optimization model for filter configuration, use the harmonic distortion level and the filter installation cost as constraint functions, use the installation location and parameter configuration of the filter as decision variables, and use an optimization algorithm to solve, and output the recommended installation location and parameter configuration of the filter; Step S14. Verify and dynamically adjust the optimization scheme: Verify the filter configuration and evaluate the harmonic suppression effect under different load conditions; If the verification result shows that the harmonic distortion level does not meet the standard or there is overcompensation, adjust the parameters of the filter configuration optimization model and solve again; Combine the real-time monitoring data and update the filter configuration optimization model regularly to achieve dynamic optimization.

[0007] Preferably, based on the located harmonic sources, construct a harmonic contribution degree matrix to quantify the influence of each harmonic source on the harmonic distortion of each node in the distribution network; Use i and j to represent the sequential numbers of the nodes in the distribution network system, and calculate the contribution degree of harmonic distortion through the following formula: ; where represents the contribution degree of node j to the harmonic voltage distortion of node i, is the harmonic impedance between node i and node j, is the harmonic current of node j, is the harmonic voltage of node i; Based on the contribution degree matrix, calculate the cumulative contribution degree of each harmonic source to the total harmonic distortion of the distribution network system; Use Pareto analysis (80 / 20 principle) to screen out the harmonic sources whose cumulative contribution degree to the total harmonic distortion accounts for the top 80%, and generate a list of key harmonic sources.

[0008] Preferably, in step S11, introduce an adaptive data cleaning method based on wavelet transform to perform wavelet decomposition on the harmonic data and extract high-frequency and low-frequency components; Use information entropy analysis to distinguish normal mutations and noise; Filter out the noise components, retain the mutation characteristics, and reconstruct the cleaned harmonic data set; The cleaned harmonic data set is directly input into step S12 for harmonic source location.

[0009] Preferably, it includes an anomaly detection step based on the time series characteristics of harmonic data, including: Obtain the time series data of the distribution network harmonics to be detected; Identify abnormal features in harmonic time-series data and generate a time-series anomaly profile corresponding to the abnormal features; Determine whether the harmonic data contains time-series anomalies according to the time-series anomaly profile; If the harmonic data contains time-series anomalies, an anomaly detection result is generated based on the long short-term memory network model, and the input data of the filter configuration optimization scheme is adjusted to enhance the system's response ability to dynamic anomalies.

[0010] Preferably, based on the dynamic changes in the distribution network topology, a graph neural network is used to update the harmonic contribution matrix. Based on the change in the edge weights of the dynamic collaboration network graph, the harmonic contribution matrix is updated in real time to generate a list of key harmonic sources adapted to the topology change, enhancing the adaptability of harmonic source identification.

[0011] Preferably, when the load of the distribution network mutates or the topology changes, a dynamic harmonic source identification model is started to identify the dynamic influence of the harmonic source and the collaborative effect between the harmonic source and the load; Input the real-time harmonic data of each node in the distribution network system, including harmonic voltage, harmonic current, and the distribution network topology; Output an enhanced list of key harmonic sources, including node numbers and phase-time delay comprehensive contribution degrees, for subsequent filter configuration optimization.

[0012] Preferably, the operation process of the dynamic harmonic source identification model includes the following steps: Step S21: Collect real-time harmonic data, extract voltage phase, current phase, and time-delay features, and construct a spatio-temporal feature matrix to ensure the data accuracy of dynamic collaborative analysis; by constructing a spatio-temporal feature matrix, ensure that subsequent analysis can accurately capture the dynamic propagation characteristics of harmonic sources; use and to represent the harmonic current sequences of nodes i and j, and determine the time-delay feature of harmonic propagation from node i to node j by comparing the similarity of and at different time offsets; Step S22: Analyze the spatio-temporal feature matrix, and generate a dynamic collaboration network graph containing node collaboration relationships and propagation paths through phase-time delay joint clustering, revealing the spatio-temporal propagation path of harmonic sources and providing a basis for contribution degree calculation; Calculate the distance between node i and node j through the following distance function :

[0013] where , are the weights of phase and time delay respectively, is the maximum value of the normalized time-delay feature; represents the voltage phase difference; An improved K-means algorithm is adopted. The spatio-temporal feature matrix and the distance between nodes are input, the number of clusters K is set, and through iterative grouping, the nodes are divided into several collaborative groups; According to the clustering results, the nodes are divided into several collaborative groups, which reflect the dynamic interaction mode between harmonic sources and loads. Based on the collaborative groups and the topology of the distribution network, a dynamic collaborative network diagram is constructed. The nodes in the diagram are the nodes of the distribution network, the edges in the diagram are the node collaborative relationships, and weight values are assigned to each edge based on the distance; Step S23: Process the dynamic collaborative network diagram, calculate the phase-time delay comprehensive contribution degree, output an enhanced list of key harmonic sources, and improve the pertinence of filter configuration; Construct a dynamic collaborative network diagram, and simultaneously collect the topology data of each node and the corresponding line in the distribution network. Calculate the phase-time delay comprehensive contribution degree of each node through the following formula :

[0014] where is the set of adjacent nodes of node i, is the harmonic impedance between nodes; is the time-delay attenuation factor, is the attenuation coefficient; respectively represent the voltage phases of node i and node j; Step S24: Use Pareto analysis to screen and obtain an enhanced list of key harmonic sources.

[0015] Preferably, the acquisition method of the time-delay feature is as follows: Use the fast Fourier transform to perform frequency-domain decomposition on the harmonic voltage and current sequences of each node, extract the phase angles of the main harmonics, denoted as the voltage phase and the current phase, where i and j are node numbers; Adopt the cross-correlation function Calculate the time-delay feature T of harmonic propagation between (topologically connected) adjacent nodes ij , defined as:

[0016] where represents the continuous time series of the harmonic current value of node i at time s (such as the 5th harmonic current), represents the time-lag variable; is the function to take the value that makes reach the maximum value.

[0017] Preferably, it includes steps for optimizing filter configuration based on harmonic propagation between nodes, including: Obtain the harmonic data of the distribution network nodes to be analyzed; Identify the harmonic propagation relationship between nodes in the distribution network and generate a non-linear characteristic profile of harmonic propagation between nodes; Determine whether there is a significant non-linear impact on harmonic propagation between nodes according to the non-linear characteristic profile; If there is a significant non-linear impact on harmonic propagation between nodes, generate a non-linear impact analysis result based on the support vector machine model, optimize the filter configuration scheme, and enhance the suppression ability of the distribution network against complex harmonic propagation.

[0018] Preferably, the optimization of the filter configuration in step S13 includes: through harmonic impedance frequency response analysis, adding resonance risk constraint conditions to ensure that the filter parameters do not cause system resonance, outputting a recommended configuration that meets the constraints, and improving the safety of the filter configuration.

[0019] Preferably, use a long short-term memory network to predict future load and harmonic trends, adjust the filter configuration based on the prediction results, generate a dynamic scheduling scheme and verify it through simulation, including: Use historical load and harmonic data to train a long short-term memory network to predict the load and harmonic trends within a future time period (such as within 24 hours); According to the prediction results, adjust and optimize the model parameters to generate a filter configuration suitable for future loads; Verify the effectiveness of the predicted configuration through simulation, update the scheduling scheme regularly, and optimize the dynamic performance of the filter.

[0020] Preferably, under the condition of periodic fluctuations in the distribution network load, perform the following steps to optimize harmonic suppression: Step S31: Decompose the harmonic data in step S11, use the empirical mode decomposition method to extract harmonic components on short, medium, and long time scales, form a multi-time scale harmonic data set, and enhance the understanding of the dynamic harmonic propagation; Step S32: Analyze the multi-time scale harmonic data set, construct a state transition model based on a Markov chain to predict the harmonic propagation trend, generate a multi-scale harmonic propagation prediction map, and reveal potential distortion risks; Step S33: Process the multi-scale harmonic propagation prediction map, adjust the constraint conditions of the filter configuration optimization model in combination with the key harmonic source list in step S12, and output a multi-time scale filter configuration strategy to enhance the adaptability of the suppression effect; Step S34: Verify the multi-time scale filter configuration strategy, use the simulation software in step S14 to evaluate the harmonic suppression effect at each time scale, generate a dynamically optimized filter scheduling scheme, and ensure the stability of the distribution network system.

[0021] Preferably, the method includes the following steps: Step S41: Collect the harmonic data of each node in the distribution network system in real time, including harmonic current, harmonic voltage, frequency components, and load power fluctuation data; Step S42: Build a dynamic harmonic source identification model, including: Spatiotemporal feature extraction: Extract the phase difference of harmonic voltage / current through the Fast Fourier Transform (FFT), and calculate the harmonic propagation time delay feature between adjacent nodes using the cross-correlation function; Dynamic collaborative network modeling: Based on the phase difference and time delay features between nodes, construct a spatiotemporal feature matrix, and combine the Graph Neural Network (GNN) to dynamically update the harmonic propagation path weights; Calculation of phase-time delay comprehensive contribution degree: Quantify the contribution degree of node i to the system harmonic distortion through a formula; Step S43: Based on the output of the dynamic harmonic source identification model, screen the list of key harmonic sources, and adopt an adaptive reinforcement learning algorithm to optimize the filter configuration to minimize the harmonic distortion rate and installation cost; Step S44: Verify the harmonic suppression effect through the digital twin platform, dynamically update the model parameters in combination with real-time monitoring data, and form a closed loop.

[0022] To achieve the above object, the present invention provides the following technical solution: A power supply reliability analysis system based on the operation and maintenance characteristics of the distribution network, including: A data collection module, which uses smart meters and power quality monitors to collect the harmonic data of each node in the distribution network system in real time, including harmonic current, harmonic voltage, and frequency components; A harmonic source identification module, which receives the real-time harmonic data set output by the data collection module, combines the distribution network topology structure, optimizes and solves the sparse linear equations through the L1 norm, models the harmonic current injection as a sparse vector and identifies the main sources of harmonics; calculates the contribution degree of each nonlinear load to the harmonic distortion of the distribution network system, and uses Pareto analysis to screen out the list of key harmonic sources, including node numbers and corresponding contribution degrees, and outputs the list of key harmonic sources; A filter configuration optimization module, which receives the list of key harmonic sources output by the harmonic source identification module, constructs a filter configuration optimization model based on this, uses the harmonic distortion level and filter installation cost as constraint functions, uses the installation location and parameter configuration of the filter as decision variables, and solves using an optimization algorithm, and outputs the recommended installation location and parameter configuration of the filter, and outputs data: filter configuration scheme; A verification and adjustment module, which receives the filter configuration scheme output by the filter configuration optimization module, verifies it in combination with the distribution network simulation software, and evaluates the harmonic suppression effect under different load conditions; if the verification result shows that the harmonic distortion level does not meet the standard or there is over-compensation, the parameters of the filter configuration optimization model are adjusted, re-solved, and combined with real-time monitoring data, and the dynamic optimization is realized by setting a regular update frequency.

[0023] Technical effects and advantages of the present invention: (1) The power supply reliability analysis method based on the operation and maintenance characteristics of the distribution network provided by the present invention combines the distribution network topology structure and real-time harmonic data, optimizes the harmonic contribution matrix using the L1 norm, and updates it with a graph neural network to generate a list of key harmonic sources adapted to dynamic changes. By jointly clustering the phase-time delay, it identifies the synergistic effect between harmonic sources and loads, effectively solving the problem of insufficient identification accuracy of traditional methods in scenarios of topological changes and load mutations. (2) The power supply reliability analysis method based on the operation and maintenance characteristics of the distribution network provided by the present invention introduces a multi-objective optimization algorithm (such as the non-dominated sorting genetic algorithm) and resonance risk constraints in the filter configuration optimization, and optimizes the configuration scheme through nonlinear impact analysis and time-series anomaly detection, significantly improving the harmonic suppression effect and system stability, and solving the problems of low filter configuration efficiency and potential safety hazards in the prior art.

[0024] (3) The power supply reliability analysis method based on the operation and maintenance characteristics of the distribution network provided by the present invention can adapt to the changes in the operation characteristics of the distribution network through long short-term memory network to predict future load trends, multi-time scale analysis to optimize harmonic suppression, and a dynamic adjustment mechanism with regular updates (such as step S14), ensuring long-term power supply reliability and completely overcoming the limitations of traditional static analysis methods in dealing with complex dynamic environments. Brief Description of the Drawings

[0025] Figure 1 It is a flowchart of the power supply reliability analysis method of the present invention.

[0026] Figure 2 It is a flowchart of the power supply reliability analysis method based on the dynamic harmonic source identification model of the present invention. Detailed Embodiments

[0027] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0028] At the same time, it should be understood that for the sake of description, the dimensions of the various parts shown in the drawings are not drawn in actual proportional relationships.

[0029] The following description of at least one exemplary embodiment is merely illustrative in nature and in no way serves as a limitation to the present application and its application or use.

[0030] Techniques, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such techniques, methods, and devices shall be regarded as part of the specification.

[0031] Explanation: The distribution network topology structure described in the embodiments of the present invention refers to the connection relationship and layout mode among various electrical components (such as transformers, lines, switches, loads, etc.) in the distribution network, which is represented as a set of nodes and edges in graph theory in mathematics; each node in the distribution network refers to a specific point with electrical significance in the distribution network, which is the connection point of the line, the installation point of the equipment, or the access point of the load. The node can be a power source point, a load point, or an intermediate connection point; the distribution network system refers to the overall composed of the distribution network topology structure, nodes, electrical equipment (such as transformers, switches, filters), and control systems, which is used to distribute electrical energy from the power source to users.

[0032] Embodiment 1, refer to Figure 1 the flowchart of the power supply reliability analysis method. The embodiments of the present invention provide a power supply reliability analysis method based on the operation and maintenance characteristics of the distribution network, including the following steps: Step S11: Use smart meters and power quality monitors to collect harmonic data of each node in the distribution network system in real time, including harmonic current, harmonic voltage, and frequency components; the collected data needs to be cleaned and standardized to remove outliers and noise to ensure data quality; Step S12: Identify the main sources of harmonics: Combine the distribution network topology structure and solve the sparse linear equations through L1 norm optimization to identify the main sources of harmonics (model the injection of harmonic current as a sparse vector and solve using L1 norm regularization); by calculating the contribution degree of each nonlinear load to the harmonic distortion of the distribution network system, use Pareto analysis to screen out a list of key harmonic sources, including node numbers and corresponding contribution degrees; Explanation: Based on the located harmonic sources, construct a harmonic contribution matrix to quantify the impact of each harmonic source on the harmonic distortion of each node in the distribution network; use i and j to represent the sequential numbers of nodes in the distribution network system, and calculate the contribution degree of harmonic distortion through the following formula: ; where represents the contribution degree of node j to the harmonic voltage distortion of node i, is the harmonic impedance between node i and node j, is the harmonic current of node j, is the harmonic voltage of node i; Based on the contribution matrix, calculate the cumulative contribution degree of each harmonic source to the total harmonic distortion of the distribution network system; use Pareto analysis (80 / 20 principle) to screen out the harmonic sources whose cumulative contribution degree to the total harmonic distortion accounts for the top 80%, and generate a list of key harmonic sources; Step S13. Optimize filter configuration: Based on the key harmonic sources identified in step S12, construct an optimization model for filter configuration, with the harmonic distortion level and filter installation cost as constraint functions, and the installation location and parameter configuration of the filter (such as capacitance and inductance values) as decision variables, and use an optimization algorithm (such as genetic algorithm, particle swarm optimization) to solve, and output the recommended installation location and parameter configuration of the filter; Use the non-dominated sorting genetic algorithm for multi-objective optimization to generate a set of non-inferior solution sets that simultaneously optimize the total harmonic distortion level and installation cost; Step S14. Verify and dynamically adjust the optimization scheme: Verify the filter configuration obtained in step S13, and evaluate the harmonic suppression effect under different load conditions; if the verification result shows that the harmonic distortion level does not meet the standard or there is over-compensation, adjust the parameters of the filter configuration optimization model and solve again; combine real-time monitoring data, and update the filter configuration optimization model regularly to achieve dynamic optimization; For example, use a distribution network simulation software for simulation verification, such as using MATLAB / Simulink distribution network simulation software to simulate and verify the filter configuration obtained in step S13. If the total harmonic distortion (THD) under any load condition exceeds 5% or over-compensation is detected, trigger the subsequent adjustment mechanism; combine real-time monitoring data, set the frequency of automatically updating the filter configuration optimization model to adapt to the changes in the operating characteristics of the distribution network, and achieve dynamic optimization.

[0033] Foreground summary: The harmonic data of the distribution network may generate instantaneous outliers due to sudden load changes (such as rapid connection or disconnection of electric vehicle chargers) or external interferences (such as lightning strikes). Traditional cleaning methods (such as mean filtering) are difficult to distinguish normal mutations from noise, resulting in data distortion or omission of key features; existing technologies mostly focus on removing outliers in the long-term trend and ignore the impact of instantaneous harmonic mutations on data quality, and such mutations may be important signals for the dynamic changes of harmonic sources. Based on this: In a possible embodiment, in step S11, introduce an adaptive data cleaning method based on wavelet transform, perform wavelet decomposition on the harmonic data, and extract high-frequency and low-frequency components; Use information entropy analysis to distinguish normal mutations (high information entropy) from noise (low information entropy); Filter out the noise components, retain the mutation features, and reconstruct the cleaned harmonic data set; The cleaned harmonic data set is directly input into step S12 for harmonic source location.

[0034] In the embodiments of the present invention, it needs to be further explained that the method includes an outlier detection step based on the time series characteristics of harmonic data, including: Obtain the time series data of the harmonic of the distribution network to be detected; Identify abnormal features in harmonic time-series data and generate a time-series anomaly profile corresponding to the abnormal features; Determine whether the harmonic data contains time-series anomalies according to the time-series anomaly profile; If the harmonic data contains time-series anomalies, generate an anomaly detection result based on a long short-term memory network model, adjust the input data of the filter configuration optimization scheme, and enhance the system's response ability to dynamic anomalies; The anomaly detection result includes the time point of anomaly occurrence, the duration, the degree of anomaly (such as the exceeded value of the total harmonic distortion level THD), and the possible cause (for example, abnormal injection from a specific harmonic source).

[0035] Foreground summary: In step S12 (harmonic source identification), the distribution network topology may change dynamically due to switch operations or fault isolation, resulting in distorted analysis results of the contribution degree of harmonic sources; traditional methods assume a fixed topology, do not consider the impact of dynamic changes on harmonic source localization, and ignore the common dynamic adjustments (such as load transfer) in the operation and maintenance of the distribution network system, which are common in actual distribution networks; based on this, in step S12, a graph neural network is introduced to track harmonic sources under dynamic topologies.

[0036] In the embodiments of the present invention, it needs to be further explained that based on the dynamic changes of the distribution network topology, a graph neural network is used to update the harmonic contribution matrix. Based on the change of the edge weights of the dynamic collaborative network graph, the harmonic contribution matrix is updated in real time to generate a list of key harmonic sources adapted to the topological changes, enhancing the adaptability of harmonic source identification.

[0037] In a possible embodiment, the rapid access of nonlinear loads (such as electric vehicle chargers, industrial frequency converters) and the dynamic adjustment of the distribution network topology (such as switch operation to isolate a faulty line) in the distribution network system will cause complex changes in the spatio-temporal distribution of harmonic sources; traditional methods mostly identify harmonic sources based on static features (such as amplitude) or a single dimension (such as spatial topology), and it is difficult to adapt to the dual impacts brought by load mutations and topological dynamics; by integrating phase features and time-delay features, a spatio-temporal collaborative analysis framework is constructed to accurately identify key harmonic sources and optimize filter configuration, improving the power supply reliability of the distribution network; When there are load mutations (such as a sharp increase in power demand during the peak period of an electric vehicle charging station) or topological changes (such as a switch cutting off a certain line after a fault) in the distribution network, start the dynamic harmonic source identification model to identify the dynamic impact of harmonic sources and the collaborative effect between harmonic sources and loads; Input the real-time harmonic data of each node in the distribution network system (collected by the smart meters and power quality monitors in step S11), including harmonic voltage, harmonic current, and the distribution network topology; Output an enhanced list of key harmonic sources, including node numbers and phase-time-delay comprehensive contribution degrees, for subsequent filter configuration optimization.

[0038] In the embodiments of the present invention, it needs to be further explained that the operation process of the dynamic harmonic source identification model includes the following steps: Step S21: Collect real-time harmonic data, extract voltage phase, current phase and time-delay characteristics, construct a spatio-temporal feature matrix to ensure the data accuracy of dynamic collaborative analysis; by constructing the spatio-temporal feature matrix, ensure that subsequent analysis can accurately capture the dynamic propagation characteristics of the harmonic source; use and to represent the harmonic current sequences of nodes i and j, and determine the time-delay characteristics of the harmonic propagating from node i to node j by comparing the and similarity at different time offsets; Step S22: Analyze the spatio-temporal feature matrix, generate a dynamic collaborative network diagram containing node collaborative relationships and propagation paths through phase-time-delay joint clustering, reveal the spatio-temporal propagation path of the harmonic source, and provide a basis for contribution calculation; Calculate the distance between node i and node j through the following distance function :

[0039] where , are the weights of phase and time-delay respectively (the weights are adjusted according to the load mutation degree), is the maximum value of the normalized time-delay feature; represents the voltage phase difference; Adopt an improved K-means algorithm, input the spatio-temporal feature matrix and the distance between nodes, set the number of clusters K, and after iterative grouping, divide the nodes into several collaborative groups; According to the clustering results, divide the nodes into several collaborative groups, which reflect the dynamic action mode of the harmonic source and the load. Based on the collaborative groups and the distribution network topology, construct a dynamic collaborative network diagram. The nodes in the diagram are the distribution network nodes, the edges in the diagram are the node collaborative relationships, and assign weight values to each edge based on the distance; Step S23: Process the dynamic collaborative network diagram, calculate the phase-time-delay comprehensive contribution degree, and output an enhanced list of key harmonic sources to improve the pertinence of filter configuration; Explanation: Processing the dynamic collaborative network diagram includes calculating the comprehensive contribution degree between each node based on the latest phase difference and time-delay characteristics, and determining the weight of each connection in the network accordingly; pruning the connections with weights lower than the preset threshold to remove weak associations and obtain a sparsified network structure; normalizing the weights of the remaining connections to eliminate the interference of magnitude differences on the analysis results; applying graph filtering or spectral clustering methods to the normalized network diagram to evaluate the collaborative contribution degree of each node to identify the most influential nodes in the network; outputting the processed dynamic collaborative network diagram and the results of the collaborative contribution degree of each node to provide a basis for key harmonic source localization and subsequent optimization; Construct a dynamic collaborative network diagram, and simultaneously collect the topological structure data of each node and the corresponding lines in the distribution network, and calculate the phase-time delay comprehensive contribution degree of each node through the following formula : ; where is the set of adjacent nodes of node i, is the harmonic impedance between nodes; is the time-delay attenuation factor, is the attenuation coefficient, and its value range is from 0.1 to 0.5, which is used to adjust the weight of the influence of time delay on harmonic propagation; respectively represent the voltage phases of node i and node j; Step S24: Use Pareto analysis to screen and obtain an enhanced list of key harmonic sources.

[0040] In a possible embodiment, under the condition of a sharp increase in non-linear loads in the distribution network, the following steps are performed to identify the collaborative influence of harmonic sources and loads: Collect the harmonic data in step S11, extract the harmonic voltage phase and current phase characteristics of each node to form a phase characteristic data set, and improve the characterization accuracy of the dynamic relationship between harmonic sources and loads; Analyze the phase characteristic data set, apply the phase difference clustering algorithm to identify the collaborative action mode of harmonic sources and load nodes, generate a source-load collaborative relationship diagram, and reveal the hidden harmonic propagation path; Process the source-load collaborative relationship diagram, calculate the phase collaborative contribution degree in combination with the distribution network topology structure, output an enhanced list of key harmonic sources, and optimize the pertinence of subsequent filter configurations.

[0041] In a possible embodiment, based on the integrated learning method of multi-source heterogeneous data, solve the problem that the identification of harmonic sources in the distribution network is affected by environmental and operation and maintenance characteristics; for example, based on the harmonic data, real-time meteorological data (such as temperature, humidity) and load fluctuation data of each node in the distribution network; construct an integrated learning model, fuse multi-source data through feature weighting, output an enhanced harmonic contribution degree matrix, and generate a more robust list of key harmonic sources.

[0042] In a possible embodiment, referring to Figure 2 the flowchart of the power supply reliability analysis method based on the dynamic harmonic source identification model, the method includes the following steps: Step S41: Real-time collect the harmonic data of each node in the distribution network system, including harmonic current, harmonic voltage, frequency components, and load power fluctuation data; Step S42: Build a dynamic harmonic source identification model, including: Space-time feature extraction: Extract the phase difference of harmonic voltage / current through fast Fourier transform, and calculate the harmonic propagation time-delay feature between adjacent nodes using the cross-correlation function; Dynamic collaborative network modeling: Based on the phase difference and time-delay feature between nodes, construct a space-time feature matrix, and dynamically update the harmonic propagation path weight in combination with the graph neural network; Further, the space-time feature matrix S is denoted as , where Q represents the total number of node indices, represents the harmonic phase difference feature between node i and node j, represents the harmonic propagation time-delay feature between node i and node j.

[0043] Phase-time delay comprehensive contribution degree calculation: Quantify the contribution degree of node i to the system harmonic distortion through a formula; Step S43: Based on the output of the dynamic harmonic source identification model, screen the list of key harmonic sources, and adopt an adaptive reinforcement learning algorithm to optimize the filter configuration to minimize the harmonic distortion rate and installation cost; Step S44: Verify the harmonic suppression effect through the digital twin platform, and dynamically update the model parameters in combination with the real-time monitoring data to form a closed loop.

[0044] Embodiment 2. The difference between the embodiment of the present invention and Embodiment 1 is that the method further includes a filter configuration optimization step based on the harmonic propagation between nodes, including: Obtain the harmonic data of the distribution network nodes to be analyzed; Identify the harmonic propagation relationship between each node in the distribution network, and generate a non-linear feature profile of the harmonic propagation between nodes; According to the non-linear feature profile, determine whether the harmonic propagation between nodes has a significant non-linear effect; If the harmonic propagation between nodes has a significant non-linear effect, then generate a non-linear effect analysis result based on the support vector machine model, optimize the filter configuration scheme, and improve the harmonic suppression ability of the distribution network for complex harmonic propagation.

[0045] In a possible embodiment, the filter configuration optimization in step S13 includes: through harmonic impedance frequency response analysis, adding resonance risk constraint conditions to ensure that the filter parameters do not cause system resonance, outputting a recommended configuration that meets the constraints, and improving the safety of the filter configuration.

[0046] Explanation: After completing the harmonic impedance frequency response analysis, for the frequency range that may cause system resonance, set resonance risk constraint conditions; the constraint conditions at least include: setting an upper limit on the impedance amplitude at each frequency point to avoid generating overly amplified harmonics; monitoring the impedance phase mutation points to identify possible resonance phenomena; imposing restrictions on the frequency segments exceeding the threshold to ensure that the resonance risk is controllable during system operation.

[0047] In a possible embodiment, use a long short-term memory network to predict future load and harmonic trends, adjust the filter configuration based on the prediction results, generate a dynamic scheduling scheme and verify it through simulation, including: Use historical load and harmonic data to train a long short-term memory network to predict the load and harmonic trends within a future time period (such as within 24 hours); According to the prediction results, adjust and optimize the model parameters to generate a filter configuration adapted to the future load; Verify the effectiveness of the predicted configuration through simulation, update the scheduling scheme regularly, and optimize the dynamic performance of the filter.

[0048] In a possible embodiment, under the condition of periodic fluctuations in the distribution network load, perform the following steps to optimize harmonic suppression: Step S31: Decompose the harmonic data in step S11, use the empirical mode decomposition method to extract short-term, medium-term, and long-term scale harmonic components, form a multi-time scale harmonic data set, and enhance the understanding of the harmonic propagation dynamics; Step S32: Analyze the multi-time scale harmonic data set, construct a state transition model based on Markov chain to predict the harmonic propagation trend, generate a multi-scale harmonic propagation prediction map, and reveal potential distortion risks; Step S33: Process the multi-scale harmonic propagation prediction map, combine the key harmonic source list in step S12 to adjust the constraint conditions of the filter configuration optimization model, and output a multi-time scale filter configuration strategy to improve the adaptability of the suppression effect; Explanation: The filter configuration optimization model includes: Define the installation positions and corresponding parameters (such as capacitance, inductance, etc.) of each filter as decision variables; Construct an objective function by weighted combination of the total harmonic distortion level and the total filter installation cost, where the harmonic distortion level and the installation cost are respectively assigned different weight coefficients to balance power quality and investment expenditure; Set two types of constraints on the objective function: one is to ensure that the total harmonic distortion level does not exceed the preset maximum allowable value; the other is to ensure that the total cost of filter installation does not exceed the budget ceiling. Set boundary constraints on the minimum and maximum values of each filter parameter to limit the range of optional device specifications.

[0049] Step S34: Verify the filter configuration strategy for multiple time scales, use the simulation software in Step S14 to evaluate the harmonic suppression effect of each time scale, generate a dynamically optimized filter scheduling plan, and ensure the stability of the distribution network system.

[0050] The embodiment of the present invention provides a power supply reliability analysis system based on the operation and maintenance characteristics of the distribution network, including: A data acquisition module, which uses smart meters and power quality monitors to collect harmonic data of each node in the distribution network system in real time, including harmonic current, harmonic voltage and frequency components; the collected data is cleaned and standardized, outliers and noise are removed, and a cleaned real-time harmonic data set (including harmonic current, harmonic voltage and frequency components) is output. A harmonic source identification module, which receives the real-time harmonic data set output by the data acquisition module, combines the distribution network topology structure, solves the sparse linear equations through L1 norm optimization, models the harmonic current injection as a sparse vector and identifies the main sources of harmonics; calculates the contribution degree of each nonlinear load to the harmonic distortion of the distribution network system, and uses Pareto analysis to screen out a list of key harmonic sources, including node numbers and corresponding contribution degrees, and outputs the list of key harmonic sources. A filter configuration optimization module, which receives the list of key harmonic sources output by the harmonic source identification module, constructs a filter configuration optimization model based on this, uses the harmonic distortion level and the filter installation cost as constraint functions, uses the installation location and parameter configuration of the filter (such as capacitance and inductance values) as decision variables, and uses an optimization algorithm (such as genetic algorithm or particle swarm optimization) to solve, and outputs the recommended installation location and parameter configuration of the filter, and outputs data: filter configuration plan (recommended installation location and parameter configuration). A verification and adjustment module, which receives the filter configuration plan output by the filter configuration optimization module, combines the distribution network simulation software (such as MATLAB / Simulink) for verification, and evaluates its harmonic suppression effect under different load conditions; if the verification result shows that the harmonic distortion level does not meet the standard (for example, THD exceeds 5%) or there is an overcompensation phenomenon, then adjust the parameters of the filter configuration optimization model, solve again, and set a regular update frequency (such as every 24 hours) in combination with real-time monitoring data to achieve dynamic optimization.

[0051] Finally, the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A power supply reliability analysis method based on the operation and maintenance characteristics of the distribution network, characterized in that: The following steps are involved: Real-time collection of harmonic data of each node in the distribution network system, including harmonic current, harmonic voltage and frequency components; Identify the main sources of harmonics: Combine the topology of the distribution network and solve the sparse linear equations through L1 norm optimization to identify the main sources of harmonics; extract the phase difference of harmonic voltage / current through fast Fourier transform, and use the cross-correlation function to calculate the harmonic propagation time lag characteristics between adjacent nodes; Based on the phase difference and time lag characteristics between nodes, a spatiotemporal feature matrix is ​​constructed, and the weight of the harmonic propagation path is dynamically updated in combination with the graph neural network. By calculating the contribution of each nonlinear load to the harmonic distortion of the distribution network system, a list of key harmonic sources is screened out using Pareto analysis, including node numbers and corresponding contributions. Optimize filter configuration: Based on key harmonic sources, build a filter configuration optimization model, use harmonic distortion level and filter installation cost as constraint functions, use filter installation location and parameter configuration as decision variables, use optimization algorithm to solve, and output the recommended installation location and parameter configuration of the filter; The harmonic suppression effect is verified through simulation on the digital twin platform, and the model parameters are dynamically updated based on real-time monitoring data to form a closed loop.

2. The power supply reliability analysis method based on distribution network operation and maintenance characteristics according to claim 1 is characterized in that: Based on the dynamic changes of the distribution network topology, the graph neural network is used to update the harmonic contribution matrix. Based on the changes in the edge weights of the dynamic collaborative network graph, the harmonic contribution matrix is ​​updated in real time to generate a list of key harmonic sources that adapt to the topology changes, thereby enhancing the adaptability of harmonic source identification.

3. The power supply reliability analysis method based on distribution network operation and maintenance characteristics according to claim 2 is characterized in that: When the load of the distribution network changes suddenly or the topology changes, the dynamic harmonic source identification model is started to identify the dynamic impact of the harmonic source and the synergy between the harmonic source and the load; Input the real-time harmonic data of each node in the distribution network system, including harmonic voltage, harmonic current and distribution network topology; Output enhanced key harmonic source list, including node number and phase-delay comprehensive contribution, for subsequent filter configuration optimization.

4. The power supply reliability analysis method based on distribution network operation and maintenance characteristics according to claim 3 is characterized in that: The operation process of the dynamic harmonic source identification model includes the following steps: Step S21, collecting real-time harmonic data, extracting voltage phase, current phase and time lag characteristics, constructing a time-space characteristic matrix, and determining the time lag characteristics of harmonic propagation from node i to node j by comparing the similarity of harmonic current sequences under different time offsets; Step S22, analyzing the spatiotemporal feature matrix, and generating a dynamic collaborative network diagram including node collaborative relationships and propagation paths through phase-time lag joint clustering; Step S23: Process the dynamic collaborative network diagram, calculate the phase-time lag comprehensive contribution, and output the enhanced key harmonic source list: Construct a dynamic collaborative network diagram and collect the topological data of each node and the corresponding line of the distribution network at the same time. Calculate the phase-delay comprehensive contribution of each node by the following formula: : ;in, is the set of adjacent nodes of node i, is the inter-node harmonic impedance; is the time-delay attenuation factor, is the attenuation coefficient; Represent the voltage phases of node i and node j respectively.

5. The power supply reliability analysis method based on distribution network operation and maintenance characteristics according to claim 4 is characterized in that: The time-lag feature is obtained as follows: The frequency domain decomposition of the harmonic voltage and current sequence of each node is performed using fast Fourier transform to extract the voltage phase and current phase of the main harmonics, where i and j are the node numbers; Using the cross-correlation function Calculate the time lag characteristic T of harmonic propagation between adjacent nodes ij , defined as: ;in, represents the continuous time series of the harmonic current value of node i at time s, represents the time lagged variable; To obtain The function reaches its maximum value.

6. The power supply reliability analysis method based on distribution network operation and maintenance characteristics according to claim 1 is characterized in that: The filter configuration optimization step based on the harmonic propagation between nodes is included, including: Obtaining the harmonic data of the distribution network nodes to be analyzed; Identify the harmonic propagation relationship between nodes in the distribution network and generate the nonlinear characteristic profile of harmonic propagation between nodes; According to the nonlinear characteristic profile, determine whether the harmonic propagation between nodes has a significant nonlinear effect; If the harmonic propagation between nodes has significant nonlinear impact, the nonlinear impact analysis results are generated based on the support vector machine model to optimize the filter configuration scheme.

7. The power supply reliability analysis method based on distribution network operation and maintenance characteristics according to claim 6 is characterized in that: Filter configuration optimization includes: adding resonance risk constraints through harmonic impedance frequency response analysis to ensure that filter parameters do not cause system resonance, outputting recommended configurations that meet the constraints, and improving the safety of filter configuration.

8. The power supply reliability analysis method based on distribution network operation and maintenance characteristics according to claim 6 is characterized in that: Use long short-term memory networks to predict future load and harmonic trends, adjust filter configurations based on the prediction results, generate dynamic scheduling plans and verify them through simulation, including: Using historical load and harmonic data, the long short-term memory network is trained to predict the load and harmonic trends in the future. According to the prediction results, the optimization model parameters are adjusted to generate the filter configuration that adapts to the future load; The effectiveness of the predicted configuration is verified through simulation, and the scheduling scheme is updated regularly to optimize the dynamic performance of the filter.

9. The power supply reliability analysis method based on distribution network operation and maintenance characteristics according to claim 8 is characterized in that: Under the condition of periodic fluctuation of distribution network load, perform the following steps to optimize harmonic suppression: Step S31, extracting the short-time, medium-time and long-time scale harmonic components from the decomposed harmonic data using an empirical mode decomposition method to form a multi-time scale harmonic data set to enhance the understanding of the harmonic propagation dynamics; Step S32, analyzing the multi-time scale harmonic data set, constructing a state transition model based on a Markov chain to predict the harmonic propagation trend, generating a multi-scale harmonic propagation prediction graph, and revealing potential distortion risks; Step S33, processing the multi-scale harmonic propagation prediction graph, adjusting the constraint conditions of the filter configuration optimization model in combination with the key harmonic source list, outputting the multi-time scale filter configuration strategy, and improving the adaptability of the suppression effect; Step S34: verify the filter configuration strategy of multiple time scales, use simulation software to evaluate the harmonic suppression effect of each time scale, generate a dynamically optimized filter scheduling plan, and ensure the stability of the distribution network system.

10. The power supply reliability analysis system based on the distribution network operation and maintenance characteristics is characterized by: include: The data acquisition module uses smart meters and power quality monitors to collect real-time harmonic data of each node in the distribution network system, including harmonic current, harmonic voltage and frequency components; The harmonic source identification module receives the real-time harmonic data set output by the data acquisition module, combines the topology of the distribution network, solves the sparse linear equations through L1 norm optimization, models the harmonic current injection as a sparse vector and identifies the main sources of harmonics; calculates the contribution of each nonlinear load to the harmonic distortion of the distribution network system, uses Pareto analysis to screen out a list of key harmonic sources, including node numbers and corresponding contributions, and outputs a list of key harmonic sources; The filter configuration optimization module receives the key harmonic source list output by the harmonic source identification module, and builds a filter configuration optimization model based on it. It takes the harmonic distortion level and the filter installation cost as constraint functions, and the installation position and parameter configuration of the filter as decision variables. It uses the optimization algorithm to solve and output the recommended installation position and parameter configuration of the filter. The output data is: filter configuration plan; The verification and adjustment module receives the filter configuration scheme output by the filter configuration optimization module, verifies it in combination with the distribution network simulation software, and evaluates the harmonic suppression effect under different load conditions; if the verification result shows that the harmonic distortion level does not meet the standard or there is over-compensation, the parameters of the filter configuration optimization model are adjusted and re-solved, and combined with real-time monitoring data, a regular update frequency is set to achieve dynamic optimization.

Images