Bus duct insulation performance evaluation system

Abstract

The invention discloses a bus duct insulation performance evaluation system, and the system comprises a microscopic collection module which is used for collecting a joint interface microscopic image through a scanning electron microscope, and obtaining material distribution and defect parameters; the finite element analysis module is used for performing finite element analysis by utilizing the material distribution and defect parameters and calculating an electric field intensity gradient value; the electric field distortion identification module is used for identifying an electric field distortion area according to the electric field intensity gradient value and determining a partial discharge position coordinate; the dielectric parameter measurement module is used for measuring a dielectric constant and a loss factor at the partial discharge position coordinates to generate a change curve; the aging index calculation module is used for extracting peak value and slope characteristics from the change curve and calculating a material aging index; the correlation model building module is used for building a correlation model of the material aging index and the insulation strength and generating a prediction coefficient; and the evaluation optimization module is used for correcting the insulation performance evaluation value by using the prediction coefficient and outputting an optimization evaluation report.

Description

Technical Field

[0001] This invention relates to the field of information technology, specifically a busbar insulation performance evaluation system. Background Technology

[0002] Busbar trunking, as a critical component of power transmission, directly impacts the safe and stable operation of the power system through its insulation performance. With increasing power loads and more complex operating environments, the insulation design and performance evaluation of busbar trunking have become core aspects of ensuring system reliability. Traditional evaluation systems primarily rely on single electrical tests, such as withstand voltage tests or partial discharge detection. However, these systems struggle to comprehensively capture the dynamic performance of insulation materials under complex operating conditions, particularly the microscopic changes at the interfaces between different materials. This can lead to evaluation results that may not accurately reflect the insulation reliability of the busbar trunking during long-term operation, especially in harsh environments such as high temperature, high humidity, or frequent load changes.

[0003] The limitation of existing systems lies in their tendency to overlook the crucial role of the interface in insulation performance and their lack of in-depth analysis of the interface's microscopic properties. As the bonding region between different insulating materials, the interface is prone to micro-gaps or stress concentrations due to differences in the thermal expansion coefficients of the materials or manufacturing defects. These microscopic defects can induce local electric field distortions under the influence of an electric field, thereby accelerating insulation aging or breakdown. For example, during the operation of busbar trunking, uneven electric field distribution at the interface can lead to localized high temperatures, accelerating material aging; however, existing systems struggle to quantify the correlation between this electric field distribution and material aging.

[0004] Uneven electric field distribution at the interface poses a primary technical challenge. Due to the difference in dielectric constants between different insulating materials, the electric field is easily distorted at the interface, leading to stress concentration. This stress concentration further exacerbates the change in loss factor at the interface, causing a decline in insulation performance. For example, during high-voltage operation, the electric field distortion at the interface may trigger micro-discharges, which, after long-term accumulation, lead to insulation failure. Accurately measuring the changes in dielectric constant and loss factor at the interface and establishing a quantitative relationship between these changes and insulation strength is a key issue in improving the evaluation of busbar insulation performance. Summary of the Invention

[0005] This invention provides a busbar insulation performance evaluation system to solve the above-mentioned problems existing in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A busbar trunking insulation performance evaluation system, characterized in that it comprises:

[0008] The microscopic acquisition module is used to acquire microscopic images of the bonding interface using a scanning electron microscope to obtain material distribution and defect parameters.

[0009] The finite element analysis module is used to perform finite element analysis using material distribution and defect parameters, and to calculate the electric field intensity gradient value.

[0010] The electric field distortion identification module is used to identify electric field distortion regions based on electric field intensity gradient values ​​and determine the coordinates of partial discharge locations.

[0011] The dielectric parameter measurement module is used to measure the dielectric constant and loss factor at the partial discharge location coordinates and generate variation curves.

[0012] The aging index calculation module is used to extract peak and slope features from the change curve and calculate the material aging index.

[0013] The correlation model building module is used to establish a correlation model between the material aging index and insulation strength, and generate prediction coefficients.

[0014] The evaluation and optimization module is used to correct the insulation performance evaluation value using prediction coefficients and output an optimized evaluation report.

[0015] The microscopic acquisition module includes:

[0016] An image acquisition unit is used to acquire microscopic images of the interface using a scanning electron microscope.

[0017] The image filtering unit is used to perform Gaussian filtering noise reduction on the microscopic image;

[0018] The edge detection unit is used to extract material boundaries from the denoised microscopic image using Canny edge detection.

[0019] The uniformity analysis unit is used to perform distribution uniformity analysis on material boundaries and calculate distribution parameters;

[0020] The defect feature extraction unit is used to calculate the geometric features of the defect region based on the distribution parameters.

[0021] Defect classification unit, used to classify defects by comparing their geometric features with a reference standard;

[0022] The report generation unit is used to generate a report on the microscopic characteristics of the interface based on the comparison and classification results.

[0023] The finite element analysis module includes:

[0024] The material parameter acquisition unit is used to acquire material distribution and defect parameters, and extract the material dielectric constant and spatial coordinates.

[0025] The three-dimensional model building unit is used to establish a three-dimensional electric field distribution finite element model based on the material dielectric constant and spatial coordinates.

[0026] Mesh generation unit, used to perform mesh generation and optimize mesh density based on geometric complexity;

[0027] The electric field distribution calculation unit is used to calculate the electric field distortion distribution and obtain the nodal electric field strength values.

[0028] Gradient calculation unit, used to calculate the electric field intensity gradient at the interface;

[0029] Mesh optimization cells are used to adjust the mesh density and recalculate to optimize gradient values.

[0030] The result verification unit is used to verify the consistency between the simulation results and the experimental data, and to determine the final gradient value.

[0031] The electric field distortion identification module includes:

[0032] The data processing unit is used to process electric field intensity distribution data and construct a three-dimensional grid data structure.

[0033] The gradient distribution calculation unit is used to calculate the electric field intensity gradient value of each node and form a gradient distribution dataset.

[0034] The distortion region marking unit is used to mark the electric field distortion region nodes according to the set gradient threshold;

[0035] The clustering analysis unit is used to perform distance clustering analysis on nodes in the electric field distortion region.

[0036] The gradient trend calculation unit is used to calculate the gradient change trend of nodes around the cluster center point;

[0037] The risk assessment unit is used to assess the discharge risk level of each center point based on the gradient trend.

[0038] The coordinate output unit is used to output the center point of the high-risk level as the coordinate set of the partial discharge location.

[0039] The dielectric parameter measurement module includes:

[0040] The measurement point selection unit is used to select the coordinates of the measurement point based on the partial discharge location list.

[0041] The parameter acquisition unit is used to acquire dielectric constant and loss factor data at the measurement point using a dielectric spectrum analyzer.

[0042] The frequency domain transformation unit is used to perform a fast Fourier transform on the acquired data and convert it into frequency domain features;

[0043] The peak identification unit is used to identify peak points in the frequency domain characteristics and determine the trend of dielectric parameter variation.

[0044] The curve generation unit is used to perform cubic spline interpolation calculations to generate continuously changing curves.

[0045] The stability analysis unit is used to calculate the slope of the curve and the location of the inflection point to analyze the stability of dielectric properties.

[0046] The curve sequence generation unit is used to generate a sequence of dielectric property change curves for each measurement point.

[0047] The aging index calculation module includes:

[0048] Peak marking unit, used to mark the point of maximum value in the change curve as the peak feature point;

[0049] The slope calculation unit is used to calculate the slope value between adjacent peak points, forming a slope dataset.

[0050] The aging trend calculation unit is used to extract the rate of change of the absolute value of the slope and calculate the aging trend index.

[0051] The aging level determination unit is used to compare the aging trend index with a reference value to determine the aging level.

[0052] The time series processing unit is used to perform time series smoothing on aging level data;

[0053] The aging rate calculation unit is used to calculate the slope and peak distribution of the smoothed curve and obtain the aging rate value.

[0054] The evaluation result generation unit is used to generate material aging degree evaluation results based on the aging rate value.

[0055] The association model construction module includes:

[0056] The data acquisition unit is used to acquire historical data on material aging index, dielectric constant, and insulation strength.

[0057] The outlier handling unit is used to remove outliers in the dataset that deviate from the mean by three times the standard deviation.

[0058] The dimensionality reduction unit is used to perform principal component analysis dimensionality reduction and extract feature vectors.

[0059] The regression model building unit is used to establish a linear regression model of dielectric constant change and insulation strength attenuation.

[0060] The coefficient calculation unit is used to calculate regression coefficients and form a prediction coefficient matrix.

[0061] The insulation strength calculation unit is used to calculate the insulation strength attenuation value based on the prediction coefficient matrix.

[0062] The grading result generation unit is used to generate material aging grading results based on the insulation strength attenuation value.

[0063] The evaluation and optimization module includes:

[0064] The strength correction unit is used to obtain the standard insulation strength test value and apply the prediction coefficient matrix for correction;

[0065] The standard comparison unit is used to compare the corrected insulation strength value with the operating standard value.

[0066] The feature extraction unit is used to extract the statistical features of the corrected insulation strength value;

[0067] The classification processing unit is used to perform K-means classification processing on the insulation strength characteristics;

[0068] The parameter adjustment unit is used to adjust the insulation strength assessment parameters according to the classification results.

[0069] The index calculation unit is used to calculate the adjusted insulation performance index value;

[0070] The report generation unit is used to generate structured reports containing evaluation results.

[0071] The gradient calculation unit is equipped with a gradient vector calculator, which calculates the electric field intensity gradient value according to the following formula:

[0072] gradient vector

[0073] Where E represents the electric field strength scalar, This represents the partial derivative of the electric field intensity in the x-direction. This represents the partial derivative of the electric field intensity in the y-direction. denoted by , where x, y, and z represent the partial derivative of the electric field intensity in the z-direction, and x, y, and z represent the spatial coordinates.

[0074] The clustering analysis unit includes: a distance calculation subunit, used to calculate the Euclidean distance between nodes in the electric field distortion region; a cluster initialization subunit, used to initialize K cluster centers; an iterative update subunit, used to iteratively assign nodes to the nearest center and update the center position; and a termination judgment subunit, used to complete clustering when the change in the center position is less than a specified value.

[0075] The risk assessment unit includes: a feature construction subunit, used to construct node feature vectors; a hyperplane setting subunit, used to set hyperplane parameters; a distance calculation subunit, used to calculate the distance from the node to the hyperplane; and a risk determination subunit, used to determine the risk level of the node based on the distance.

[0076] Compared with the prior art, the present invention has the following advantages:

[0077] This invention discloses a busbar trunking insulation performance evaluation system. Addressing the issues of partial discharge and insulation strength attenuation at high-voltage equipment interface caused by uneven material distribution, defects, and aging, the system uses scanning electron microscopy to collect interface microscopic characteristics, obtaining material distribution and defect parameters. Finite element analysis is then employed to simulate electric field distortion, calculate the electric field intensity gradient, and mark high-risk areas. Subsequently, dielectric spectroscopy is used to obtain local dielectric constant and loss factor variation curves, extracting peak values ​​and slope characteristics to quantify the degree of material aging. If aging indicators exceed standards, regression analysis is used to establish a predictive model of dielectric constant change and insulation strength attenuation, generating a coefficient matrix to correct traditional evaluation results and determine whether operational standards are met. This invention significantly improves the accuracy and reliability of insulation performance evaluation through multi-dimensional data fusion and precise modeling, providing an optimized decision-making basis for safe equipment operation.

[0078] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention.

[0079] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0080] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0081] Figure 1 This is a structural diagram of a busbar insulation performance evaluation system according to an embodiment of the present invention;

[0082] Figure 2 This is a structural diagram of the microscopic acquisition module in an embodiment of the present invention. Detailed Implementation

[0083] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0084] The embodiments of the present invention provide, as follows Figure 1 As shown, a busbar insulation performance evaluation system includes:

[0085] The microscopic acquisition module is used to acquire microscopic images of the bonding interface using a scanning electron microscope to obtain material distribution and defect parameters.

[0086] The finite element analysis module is used to perform finite element analysis using material distribution and defect parameters, and to calculate the electric field intensity gradient value.

[0087] The electric field distortion identification module is used to identify electric field distortion regions based on electric field intensity gradient values ​​and determine the coordinates of partial discharge locations.

[0088] The dielectric parameter measurement module is used to measure the dielectric constant and loss factor at the partial discharge location coordinates and generate variation curves.

[0089] The aging index calculation module is used to extract peak and slope features from the change curve and calculate the material aging index.

[0090] The correlation model building module is used to establish a correlation model between the material aging index and insulation strength, and generate prediction coefficients.

[0091] The evaluation and optimization module is used to correct the insulation performance evaluation value using prediction coefficients and output an optimized evaluation report.

[0092] The working principle and beneficial effects of the above technical solution are as follows: A microscopic acquisition module is used to acquire microscopic characteristic image data of the bonding interface using a scanning electron microscope, obtaining material distribution and defect feature parameters at the interface; original image data is obtained by acquiring microscopic image data of the bonding interface using a scanning electron microscope. Image processing algorithms are used to denoise the original image data, resulting in a denoised image. If the pixel grayscale value of the denoised image exceeds a preset threshold, an edge detection algorithm is used to extract the material distribution boundary, obtaining boundary features. Based on the boundary features, a convolutional neural network is applied to analyze the uniformity of the material distribution, obtaining distribution parameters. For the distribution parameters, the geometric characteristics of the defect region are calculated, obtaining defect parameters. By comparing the defect parameters with a preset standard, the defect type is determined, obtaining a classification result. Based on the classification result, microscopic characteristic analysis data of the bonding interface is generated, obtaining the final analysis data.

[0093] The finite element analysis module is used to simulate the electric field distortion distribution based on the collected material distribution and defect characteristic parameters, and to determine the electric field intensity gradient value at the interface. It acquires material distribution and defect parameter data, and extracts the material dielectric constant and the spatial coordinates of the defects from a preset database. An electric field distribution model is established using the finite element system, and a three-dimensional geometric structure is generated based on the material distribution and defect parameters. If the geometric structure complexity exceeds a preset threshold, the mesh density is optimized using an adaptive mesh generation algorithm to obtain a discretized model. The electric field distortion distribution is calculated through finite element analysis to determine the electric field intensity value at each mesh node. Based on the electric field intensity value, the gradient calculation formula (…) is used. Where E is the electric field strength, and x, y, z are spatial coordinates, the gradient value at the calculation interface is determined. If the gradient value deviates from the preset verification standard deviation by more than a threshold, the mesh density is adjusted and the calculation is repeated to obtain the optimized gradient value. The result verification algorithm is used to compare the agreement between the simulation results and the experimental data to determine the final electric field strength gradient value.

[0094] The electric field distortion identification module acquires the simulated electric field intensity gradient values. If the gradient value exceeds a preset threshold, it marks the area as a high-risk region, generating a list of potential partial discharge locations. It also acquires electric field intensity distribution data calculated by electromagnetic field simulation software, generating a 3D mesh data containing the electric field intensity values ​​of each node. Based on the 3D mesh data, it calculates the electric field intensity gradient value of each node, obtaining a gradient distribution dataset. If the gradient value in the gradient distribution dataset exceeds a preset threshold, the corresponding node is marked as a high-risk region, generating a set of high-risk region nodes. Through cluster analysis of the high-risk region node set, using the K-means algorithm, it determines the coordinates of the center point of the potential partial discharge location. It acquires the neighboring node data of the center point coordinates, calculates the electric field intensity gradient change trend, and generates a feature distribution dataset of partial discharge risk. Based on the feature distribution dataset, it uses a support vector machine algorithm to determine whether each center point is a high-risk partial discharge location, obtaining a list of potential partial discharge locations. Through spatial geometric analysis of the potential partial discharge location list, it determines the spatial distribution characteristics of each location, generating the final set of partial discharge location coordinates.

[0095] The dielectric parameter measurement module is used to measure the local dielectric constant and loss factor in real time using a dielectric spectrum analyzer, based on the obtained list of high-risk area locations, to obtain a sequence of variation curves. It also retrieves the coordinates of the target area from the list of high-risk area locations to determine the distribution of measurement points. The module collects local dielectric constant and loss factor data at the measurement points in the target area using the dielectric spectrum analyzer, obtaining the original data sequence. A fast Fourier transform algorithm is used to perform frequency domain transformation on the original data sequence, generating a frequency domain feature sequence. If significant peaks exist in the frequency domain feature sequence, the variation trends of the dielectric constant and loss factor are determined based on the peak positions. Based on the variation trends, an interpolation algorithm is used to generate a continuous sequence of variation curves. For the variation curve sequence, the slope and inflection points of the curves are calculated to determine the stability of the dielectric properties within the region. Based on the stability determination results, a sequence of dielectric property distribution for the high-risk area is generated.

[0096] The aging index calculation module extracts peak and slope features from the change curve sequence to determine the material aging degree index. Peak features are obtained from the sequence curves by using a signal processing system to detect the maximum value points, resulting in a peak feature set. For this set, the slope between adjacent peak points is calculated using a linear regression system to obtain a slope feature set. Based on the slope feature set, the slope change trend is extracted; if the absolute value of the slope continuously increases, it is judged as an accelerating aging trend, resulting in an aging trend index. Using the aging trend index and a preset threshold range, a decision tree algorithm is used to determine the material aging degree, obtaining an aging degree level. Time series features are extracted from the aging degree level, and the data is smoothed using a moving average system to obtain an aging degree change curve. Based on the aging degree change curve, the time distribution characteristics of the curve slope and peak values ​​are calculated to obtain an aging rate index. Using the aging rate index and time series analysis, an autoregressive model is used to predict future aging trends, obtaining aging prediction results.

[0097] The correlation model construction module is used to establish a quantitative relationship between dielectric constant change and insulation strength attenuation through regression analysis if the material aging index exceeds the warning line, thus obtaining a prediction coefficient matrix. If the material aging index exceeds a preset warning threshold, it retrieves material aging index, dielectric constant, and insulation strength data from the database to generate an initial dataset. Preprocessing techniques are used to clean the initial dataset, removing outliers to obtain a standardized dataset. Principal component analysis is used to extract the main features of dielectric constant and insulation strength from the standardized dataset, resulting in a dimensionality-reduced feature set. Based on the dimensionality-reduced feature set, a linear regression model is applied to fit the relationship between dielectric constant change and insulation strength attenuation, generating a prediction coefficient matrix. If the fitting error of the prediction coefficient matrix is ​​below a preset threshold, the trend of dielectric constant change is calculated using the matrix to obtain the attenuation prediction value. Based on the attenuation prediction value, a support vector machine classifier is used to determine the material aging level and generate aging level labels. The aging level labels and the prediction coefficient matrix are used to update the material aging status in the database, generating an updated material status record.

[0098] The evaluation and optimization module is used to correct the traditional system evaluation results using the obtained prediction coefficient matrix, determine whether the corrected insulation strength value meets the operating standards, and obtain an optimized performance evaluation report. It obtains the initial insulation strength value from the traditional evaluation results, corrects it using a pre-established prediction coefficient matrix, and obtains the corrected insulation strength value. Based on the corrected insulation strength value, it compares it with a preset operating standard threshold to determine whether it meets the operating standards, obtaining a compliance result. If the compliance result indicates that the operating standards are met, key features are extracted from the corrected insulation strength value to generate preliminary performance evaluation data. A support vector machine algorithm is used to classify the preliminary performance evaluation data to determine the optimization direction, obtaining classification results. Based on the classification results, a random forest algorithm is used to further optimize the performance evaluation data, generating optimized performance evaluation data. Performance indicators are obtained from the optimized performance evaluation data, analyzed using preset threshold judgment rules, and the final performance evaluation result is obtained. Based on the final performance evaluation result, structured data output is generated and stored in a preset database, completing the business process.

[0099] In another embodiment, such as Figure 2 As shown, the microscopic acquisition module includes:

[0100] An image acquisition unit is used to acquire microscopic images of the interface using a scanning electron microscope.

[0101] The image filtering unit is used to perform Gaussian filtering noise reduction on the microscopic image;

[0102] The edge detection unit is used to extract material boundaries from the denoised microscopic image using Canny edge detection.

[0103] The uniformity analysis unit is used to perform distribution uniformity analysis on material boundaries and calculate distribution parameters;

[0104] The defect feature extraction unit is used to calculate the geometric features of the defect region based on the distribution parameters.

[0105] Defect classification unit, used to classify defects by comparing their geometric features with a reference standard;

[0106] The report generation unit is used to generate a report on the microscopic characteristics of the interface based on the comparison and classification results.

[0107] The working principle and beneficial effects of the above technical solution are as follows: The image acquisition unit acquires microscopic image data of the bonding interface through a scanning electron microscope to obtain raw image data; wherein, the scanning electron microscope scans the sample surface with an electron beam, generating secondary electrons and backscattered electrons, which are collected by a detector and converted into electrical signals to form an image. The bonding interface refers to the connection area between the insulating material and the supporting structure or different insulating materials, and is a key part for insulation performance evaluation.

[0108] The image filtering unit employs image processing algorithms to denoise the original image data, resulting in a denoised image. These algorithms include techniques such as Gaussian filtering, median filtering, and wavelet transform, used to eliminate random noise generated during the scanning process. The denoising process improves the accuracy of subsequent analysis by preserving image structural features while reducing noise interference.

[0109] When the pixel grayscale value of the denoised image exceeds a preset threshold, the edge detection unit extracts the material distribution boundary using an edge detection algorithm to obtain boundary features. The pixel grayscale value reflects the brightness of each point in the image, typically ranging from 0 to 255, representing the electron emission capability of different materials or structures. The preset threshold is determined based on material properties and image statistical features, used to distinguish between valid signals and background. The edge detection algorithm includes operators such as Sobel, Canny, or LoG, which identify regions with abrupt changes in grayscale value by calculating pixel gradients. Boundary features include parameters such as boundary position, length, curvature, and connectivity, reflecting the geometric characteristics of the material distribution.

[0110] The uniformity analysis unit, based on boundary features, applies a convolutional neural network to analyze the uniformity of material distribution and obtain distribution parameters. The convolutional neural network is a deep learning architecture containing convolutional layers, pooling layers, and fully connected layers, specifically designed for image feature extraction and analysis. The material distribution uniformity evaluation network extracts boundary texture, shape, and arrangement features through multi-layer convolution, quantifying the regularity of material distribution. Distribution parameters include quantitative indices such as uniformity, clustering coefficient, dispersion, and directionality, comprehensively describing the spatial distribution characteristics of the material at the interface.

[0111] The defect feature extraction unit calculates the geometric characteristics of the defect region based on distribution parameters to obtain defect parameters. The defect region refers to abnormal structures such as voids, cracks, and impurities in the material, which are the main causes of decreased insulation performance. Geometric characteristic calculations include area calculation, perimeter measurement, shape factor analysis, and orientation assessment. Defect parameters include quantitative indicators such as size distribution, density, shape factor, and spatial orientation, comprehensively describing the geometric and statistical characteristics of the defect.

[0112] The defect classification unit determines the defect type and obtains the classification result by comparing defect parameters with preset standards. The preset standards are a defect feature database established based on theoretical analysis and experimental data, containing parameter ranges for various typical defects. Defect types include various forms such as bubbles, microcracks, impurity particles, and interface separation, and different types have different mechanisms of impact on insulation performance. The classification result includes defect type identification, quantity statistics, and hazard level assessment, providing category information for subsequent insulation performance analysis. The judgment process employs pattern recognition and multi-parameter comparison technology, determining the specific type and hazard level of each defect region by comparing it with templates in the standard database.

[0113] Based on the classification results, the report generation unit generates microscopic characteristic analysis data of the interface, resulting in the final analysis data. This microscopic characteristic analysis data provides a comprehensive description of the interface structure and defect characteristics, including quantitative parameters and qualitative evaluations. The final analysis data is stored in a structured format, including material distribution characteristics, defect type statistics, hazardous area identification, and overall quality assessment. The data generation process comprehensively considers the analysis results, using feature fusion and weight allocation to form a comprehensive assessment of the interface's microstructure.

[0114] In another embodiment, the finite element analysis module includes:

[0115] The material parameter acquisition unit is used to acquire material distribution and defect parameters, and extract the material dielectric constant and spatial coordinates.

[0116] The three-dimensional model building unit is used to establish a three-dimensional electric field distribution finite element model based on the material dielectric constant and spatial coordinates.

[0117] Mesh generation unit, used to perform mesh generation and optimize mesh density based on geometric complexity;

[0118] The electric field distribution calculation unit is used to calculate the electric field distortion distribution and obtain the nodal electric field strength values.

[0119] Gradient calculation unit, used to calculate the electric field intensity gradient at the interface;

[0120] Mesh optimization cells are used to adjust the mesh density and recalculate to optimize gradient values.

[0121] The result verification unit is used to verify the consistency between the simulation results and the experimental data, and to determine the final gradient value.

[0122] The working principle and beneficial effects of the above technical solution are as follows: Material distribution and defect parameter data are acquired, and the material dielectric constant and spatial coordinates of defects are extracted from a pre-set database. The material distribution and defect parameter data originate from the microscopic analysis results of the previous stage, including spatial distribution information of various materials at the interface and characteristic data such as the shape, size, and location of defects. The pre-set database is a structured storage system containing electrical characteristic parameters of various common insulating materials, storing key parameters such as the dielectric constant and conductivity of different materials. The material dielectric constant is a physical quantity characterizing the material's ability to store electric field energy, directly affecting the electric field distribution.

[0123] An electric field distribution model was established using a finite element method (FEM), and a three-dimensional geometric structure was generated based on material distribution and defect parameters. The electric field distribution model is based on Maxwell's equations and considers the dielectric properties of the materials and boundary conditions. The three-dimensional geometric structure is a digital representation of the physical structure of the busbar interface, including the shape, thickness, and joint area of ​​different material layers, as well as the geometric features of defects. The construction process of the geometric structure must consider the continuity of the material interface and the true morphology of defects, and high-precision modeling is achieved using computer-aided design (CAD) techniques.

[0124] When the geometric complexity exceeds a preset threshold, an adaptive meshing algorithm optimizes the mesh density to obtain a discretized model. Geometric complexity refers to a comprehensive indicator of the number of features, rate of change of dimensions, and shape complexity in the model, typically evaluated by calculating parameters such as feature curvature and the number of boundaries. The preset threshold is a complexity judgment criterion determined in advance based on computational resources and accuracy requirements. The adaptive meshing algorithm automatically adjusts the mesh size based on local geometric features and field gradients, increasing mesh density in critical regions such as defect edges and material interfaces, while using a coarser mesh in regions with gradual changes. The discretized model consists of a large number of finite element elements, each using a specific interpolation function to approximate the physical field distribution.

[0125] The electric field distortion distribution is calculated using finite element analysis (FEM) to determine the electric field intensity at each mesh node. FEM, based on variational principles or the weighted residual method, calculates the potential distribution by solving a large set of sparse matrix equations, and then obtains the electric field intensity from the potential gradient. Electric field distortion refers to the non-uniform distribution of the electric field in space, particularly noticeable distortions at material interfaces and around defects. Mesh nodes are calculation points in the discretized model, each with unique spatial coordinates. The electric field intensity value is the vector magnitude of the electric field at each point, expressed in V / m (volts per meter), representing the magnitude of the electric force per unit charge.

[0126] Based on the electric field strength value, the electric field strength gradient value at the interface is calculated; the gradient calculation uses a central difference or higher-order difference system to ensure calculation accuracy. The larger the gradient value, the more uneven the electric field distribution, and the higher the risk of insulation breakdown.

[0127] When the gradient value deviates from the preset verification standard by a threshold, the mesh density is adjusted and the calculation is repeated to obtain the optimized gradient value. The preset verification standard is a gradient distribution reference value determined based on theoretical analysis and experimental data, used to evaluate the reliability of the calculation results. The deviation threshold is the acceptable upper limit of error; exceeding this value indicates insufficient computational accuracy. Adjusting the mesh density is a key method to improve computational accuracy, typically employing a gradient adaptive refinement strategy, increasing the mesh density in regions with high gradient values. The recalculation process includes updating the geometric discretization, reconstructing the finite element equations, and solving the updated equation set. The optimized gradient value has higher computational accuracy and more accurately reflects the actual electric field distribution characteristics.

[0128] The final electric field intensity gradient value is determined by comparing the agreement between the simulation results and experimental data using a result verification algorithm. The experimental data, derived from electric field tests of actual busbar samples, serves as a reference standard for verifying the simulation accuracy. The agreement assessment includes point-to-point comparison, trend consistency analysis, and statistical significance testing. The final electric field intensity gradient value is a verified and reliable calculation result, serving as a crucial basis for subsequent insulation performance evaluation.

[0129] In another embodiment, the electric field distortion identification module includes:

[0130] The data processing unit is used to process electric field intensity distribution data and construct a three-dimensional grid data structure.

[0131] The gradient distribution calculation unit is used to calculate the electric field intensity gradient value of each node and form a gradient distribution dataset.

[0132] The distortion region marking unit is used to mark the electric field distortion region nodes according to the set gradient threshold;

[0133] The clustering analysis unit is used to perform distance clustering analysis on nodes in the electric field distortion region.

[0134] The gradient trend calculation unit is used to calculate the gradient change trend of nodes around the cluster center point;

[0135] The risk assessment unit is used to assess the discharge risk level of each center point based on the gradient trend.

[0136] The coordinate output unit is used to output the center point of the high-risk level as the coordinate set of the partial discharge location.

[0137] The working principle and beneficial effects of the above technical solution are as follows: Electric field intensity distribution data is acquired, and three-dimensional mesh data containing the electric field intensity values ​​of each node is generated. The electric field intensity distribution data is obtained through the finite element analysis calculation in the previous stage and contains the electric field intensity vector information of each point in space. Electric field intensity is a physical quantity characterizing the strength of an electric field, with units of V / m (volts per meter). The three-dimensional mesh data is a spatial structure composed of a large number of discrete points. Each mesh node has unique spatial coordinates (x, y, z) and a corresponding electric field intensity value. The node electric field intensity value typically contains three components (Ex, Ey, Ez), representing the components of the electric field in the three coordinate axes, respectively.

[0138] Based on 3D mesh data, the electric field intensity gradient value of each node is calculated, resulting in a gradient distribution dataset. The electric field intensity gradient value is the rate of change of the electric field intensity in space, reflecting the degree of non-uniformity in the electric field distribution. The gradient distribution dataset contains the gradient vector and its magnitude for each node, forming a complete description of the spatial variation characteristics of the electric field. Regions with larger gradient values ​​have more non-uniform electric field distributions and a higher risk of partial discharge.

[0139] When the gradient values ​​in the gradient distribution dataset exceed a preset threshold, the corresponding nodes are marked as high-risk regions, generating a high-risk region node set. The preset threshold is a critical gradient value determined based on the material breakdown field strength and a safety factor. Regions with gradient values ​​exceeding this threshold are considered to have significant electric field distortion and are potential points for partial discharge. The high-risk region node set is a set of spatial points that meet the risk conditions, each point having explicit spatial coordinates and a corresponding gradient value. The marking process uses a threshold comparison method, setting the risk mark of points exceeding the threshold to 1 and points below the threshold to 0, forming a binary risk distribution map.

[0140] Cluster analysis of nodes in high-risk areas determines the coordinates of the center points of potential partial discharge locations. Cluster analysis is a data mining technique that groups similar objects; the K-means clustering algorithm is used to spatially group high-risk nodes. Through an iterative optimization process, the K-means algorithm groups adjacent high-risk nodes into one cluster, each cluster representing a potential partial discharge region. The number of clusters K is automatically determined using the silhouette coefficient or the elbow rule to ensure the rationality of the classification.

[0141] The algorithm acquires neighboring node data of the center point coordinates, calculates the electric field intensity gradient variation trend, and generates a characteristic distribution dataset of partial discharge risk. Neighboring node data refers to the electric field and gradient data of all grid nodes within a preset radius r, with the center point as the reference. The electric field intensity gradient variation trend describes the spatial distribution of gradient values, which is obtained by calculating the directional derivative. The expression is represented as follows, where n is the direction vector of interest. The feature distribution dataset contains the gradient statistical characteristics within the neighborhood of each center point, including indicators such as maximum value, average value, standard deviation, and rate of change, forming a multidimensional description of the electric field characteristics of the local region. The trend analysis considers the spatial continuity and directionality of the gradient, enabling a more comprehensive assessment of the possibility of partial discharge.

[0142] Based on the feature distribution dataset, it is determined whether each center point is a high-risk partial discharge location, resulting in a list of potential partial discharge locations. The feature distribution dataset is multi-dimensional data describing the electric field characteristics of the neighborhood of each center point. The determination process uses a support vector machine algorithm to construct an optimal classification hyperplane that divides the sample points into high-risk and low-risk categories.

[0143] By performing spatial geometric analysis on a list of potential partial discharge locations, the spatial distribution characteristics of each location are determined, generating a final set of partial discharge location coordinates. This spatial geometric analysis studies the distribution patterns of discharge locations in three-dimensional space. The analysis includes distance calculation between locations, density analysis, clustering assessment, and directionality determination. Spatial distribution characteristics describe patterns such as clustering areas, linear arrangements, or random distributions of discharge points, helping to understand the distribution patterns of weak points in the insulation structure. The final set of partial discharge location coordinates is a spatially optimized set of high-risk points, with each point accompanied by a detailed risk rating and geometric description. The coordinate set is stored in a standardized format for easy positioning and measurement by subsequent testing equipment.

[0144] In another embodiment, the dielectric parameter measurement module includes:

[0145] The measurement point selection unit is used to select the coordinates of the measurement point based on the partial discharge location list.

[0146] The parameter acquisition unit is used to acquire dielectric constant and loss factor data at the measurement point using a dielectric spectrum analyzer.

[0147] The frequency domain transformation unit is used to perform a fast Fourier transform on the acquired data and convert it into frequency domain features;

[0148] The peak identification unit is used to identify peak points in the frequency domain characteristics and determine the trend of dielectric parameter variation.

[0149] The curve generation unit is used to perform cubic spline interpolation calculations to generate continuously changing curves.

[0150] The stability analysis unit is used to calculate the slope of the curve and the location of the inflection point to analyze the stability of dielectric properties.

[0151] The curve sequence generation unit is used to generate a sequence of dielectric property change curves for each measurement point.

[0152] The working principle and beneficial effects of the above technical solution are as follows: The coordinates of the target area are obtained from the list of high-risk area locations to determine the distribution of measurement points. The list of high-risk area locations is a set of coordinates of potential partial discharge locations obtained from electric field intensity gradient analysis. After the target area coordinates are determined, the system sets up multiple measurement points within that area. The distribution of measurement points typically adopts a grid layout or a denser layout based on areas with significant changes in electric field intensity gradient, to ensure accurate capture of changes in local dielectric properties.

[0153] The dielectric constant and loss factor data of the target area are collected by a dielectric spectrum analyzer to obtain the original data sequence. The dielectric spectrum analyzer obtains the dielectric constant and loss factor of the material by applying an electric field to the measurement point at different frequencies and recording the material's response to the electric field.

[0154] The original data sequence is transformed into a frequency domain using the Fast Fourier Transform (FFT) algorithm, generating a frequency domain feature sequence. Specifically, the FFT algorithm converts the original time-domain data sequence into a frequency-domain feature sequence, allowing for more intuitive identification of periodic variations and characteristic frequencies within the data. This frequency-domain feature sequence can more effectively characterize the response properties of dielectric materials at different frequencies.

[0155] When significant peaks exist in the frequency domain characteristic sequence, the changing trends of the dielectric constant and loss factor can be determined based on the peak positions. These significant peaks typically correspond to characteristic changes in the material's dielectric properties. The position (frequency) of the peak reflects the characteristic frequency of the dielectric response, and the magnitude of the peak reflects the intensity of the response. By analyzing the position and magnitude of these peaks, the changing trends of the material's dielectric constant and loss factor can be determined, thereby evaluating the insulation performance.

[0156] Based on the changing trend, an interpolation algorithm is used to generate a continuous sequence of changing curves. The interpolation algorithm usually employs systems such as cubic spline interpolation or Bézier curves to connect discrete frequency domain feature points into a continuous changing curve.

[0157] For a sequence of changing curves, the slope and inflection points of the curves are calculated to determine the stability of the dielectric properties within a region. The slope of the curve represents the rate of change of the dielectric parameter, and the inflection point indicates the location where the rate of change changes significantly. By calculating the slope values ​​of the curves in different frequency ranges and identifying the locations of inflection points, the stability of the dielectric properties within a region can be determined.

[0158] Based on the stability assessment results, a dielectric property distribution sequence for high-risk areas is generated. The stability assessment results include stability indices for dielectric constant and loss factor, as well as comparisons between these indices and preset thresholds. The dielectric property distribution sequence is a dataset indexed by spatial coordinates, containing dielectric property stability indices for each measurement point. This sequence visually displays the spatial distribution of dielectric properties within high-risk areas.

[0159] In another embodiment, the aging index calculation module includes:

[0160] Peak marking unit, used to mark the point of maximum value in the change curve as the peak feature point;

[0161] The slope calculation unit is used to calculate the slope value between adjacent peak points, forming a slope dataset.

[0162] The aging trend calculation unit is used to extract the rate of change of the absolute value of the slope and calculate the aging trend index.

[0163] The aging level determination unit is used to compare the aging trend index with a reference value to determine the aging level.

[0164] The time series processing unit is used to perform time series smoothing on aging level data;

[0165] The aging rate calculation unit is used to calculate the slope and peak distribution of the smoothed curve and obtain the aging rate value.

[0166] The evaluation result generation unit is used to generate material aging degree evaluation results based on the aging rate value.

[0167] The working principle and beneficial effects of the above technical solution are as follows: Peak features are obtained from the sequence curves, and the maximum value points in the curves are detected using a signal processing system to obtain a set of peak features. Peak features refer to local maxima in the sequence of changing curves; these peaks typically correspond to anomalies or changes in dielectric properties. The signal processing system includes algorithms such as the window sliding comparison method and the derivative zero-point method to accurately locate the maximum value points in the curves. The set of peak features contains information such as the frequency position and amplitude of each peak point; these features are important indicators for judging the degree of material aging.

[0168] Based on the peak feature set, the slope between adjacent peak points is calculated using a linear regression system to obtain the slope feature set; the formula for calculating the slope between adjacent peak points is (y 2 -y 1 ) / (x 2 -x 1), where y represents the peak value and x represents the corresponding frequency position. A linear regression system is used to fit the changing trends of these peak points to obtain a more stable slope feature. The slope feature set contains information on the rate of change of the material's dielectric properties within each frequency range, and is a key parameter for assessing the material's aging rate.

[0169] Based on the slope feature set, the slope change trend is extracted. When the absolute value of the slope continuously increases, it is judged as an accelerated aging trend, resulting in an aging trend index. The slope change trend is obtained by calculating the first difference of continuous slope values ​​in the slope feature set. When the absolute value of the slope continuously increases, i.e., the difference result is consistently positive, it indicates that the rate of change of the material's dielectric properties is accelerating. The aging trend index is a numerical parameter obtained through statistical analysis of the slope change trend.

[0170] Based on aging trend indicators and a preset threshold range, the degree of material aging is determined to obtain an aging level. The preset threshold range is typically determined based on experimental data and engineering experience, categorizing the aging trend indicator into multiple levels, such as slight aging, moderate aging, and severe aging. The determination of material aging level employs a decision tree algorithm, comparing the aging trend indicator with the preset threshold and determining the aging level based on the comparison result.

[0171] Time series features are extracted from the aging level, and the data is smoothed using a moving average system to obtain the aging level change curve; where time series features refer to the changes in the aging level over time.

[0172] Based on the aging rate curve, the temporal distribution characteristics of the curve slope and peak value are calculated to obtain the aging rate index. The curve slope represents the rate of change in aging, while the temporal distribution characteristics of the peak value reflect key time points in the aging process. The aging rate index comprehensively considers both the magnitude of the slope and the frequency of peak occurrence, making it a comprehensive parameter that can quantitatively describe the aging rate of materials. A higher aging rate index indicates a faster material aging rate, requiring more timely intervention measures.

[0173] Based on the aging rate index and combined with time series analysis, future aging trends are predicted, yielding aging prediction results. Specifically, the time series analysis employs an autoregressive model, analyzing patterns and regularities in historical aging data to predict aging trends over a future period.

[0174] In another embodiment, the association model building module includes:

[0175] The data acquisition unit is used to acquire historical data on material aging index, dielectric constant, and insulation strength.

[0176] The outlier handling unit is used to remove outliers in the dataset that deviate from the mean by three times the standard deviation.

[0177] The dimensionality reduction unit is used to perform principal component analysis dimensionality reduction and extract feature vectors.

[0178] The regression model building unit is used to establish a linear regression model of dielectric constant change and insulation strength attenuation.

[0179] The coefficient calculation unit is used to calculate regression coefficients and form a prediction coefficient matrix.

[0180] The insulation strength calculation unit is used to calculate the insulation strength attenuation value based on the prediction coefficient matrix.

[0181] The grading result generation unit is used to generate material aging grading results based on the insulation strength attenuation value.

[0182] The working principle and beneficial effects of the above technical solution are as follows: When the material aging index exceeds the preset warning threshold, the material aging index, dielectric constant, and insulation strength data are retrieved from the database to generate an initial dataset. The material aging index is determined by the aging rate index and aging prediction results. The preset warning threshold is a critical value determined according to the safety operation standards for busbar insulation materials; exceeding this value indicates that the material has entered the accelerated aging stage. The initial dataset contains the corresponding relationship data between the material aging index, dielectric constant, and insulation strength. This data is extracted from historical operation and test data from a database specifically established for busbar insulation materials.

[0183] Preprocessing techniques are used to clean the initial dataset, remove outliers, and obtain a standardized dataset. These preprocessing techniques include systems for outlier detection, data standardization, and data normalization.

[0184] Principal component analysis (PCA) is used to extract the main features of dielectric constant and insulation strength from a standardized dataset, resulting in a dimensionality-reduced feature set. PCA is a dimensionality reduction technique that projects high-dimensional data into a low-dimensional space composed of principal components by calculating the eigenvalues ​​and eigenvectors of the data covariance matrix.

[0185] Based on the dimensionality-reduced feature set, a linear regression model is applied to fit the relationship between the change in dielectric constant and the attenuation of insulation strength, generating a prediction coefficient matrix. The linear regression model uses the least squares method to estimate the regression coefficients, establishing a linear relationship between the change in dielectric constant and the attenuation of insulation strength. The mathematical expression for the linear regression is y = β0 + β1x1 + β2x2 + ... + β n x n+ε, where y represents the insulation strength attenuation, x represents the principal components of the dielectric constant, β represents the regression coefficient, and ε represents the random error term. The prediction coefficient matrix contains the regression coefficients corresponding to each feature, and these coefficients quantify the degree of influence of changes in dielectric properties on insulation strength.

[0186] When the fitting error of the prediction coefficient matrix is ​​lower than a preset threshold, the dielectric constant change trend is calculated through the matrix to obtain the attenuation prediction value; where the fitting error is usually expressed as the root mean square error (RMSE) or the coefficient of determination (R²). 2 RMSE is measured by RMSE, and the formula for calculating RMSE is: Where yi represents the actual value. The value represents the predicted value, and n represents the number of samples. When the fitting error is below a preset threshold, it indicates that the established model has good predictive ability. The attenuation prediction value is the expected attenuation of future insulation strength calculated by substituting the current dielectric constant change data into the prediction coefficient matrix.

[0187] Based on the predicted degradation value, the aging level of the material is determined and an aging level label is generated.

[0188] Based on aging level labels and prediction coefficient matrices, the aging status of materials in the database is updated, generating updated material status records. These records include information such as the aging level at the current time point, dielectric property parameters, and predicted insulation strength attenuation values. The update process is implemented through a database management system, which stores the newly generated status records in chronological order and establishes a correlation with historical data.

[0189] In another embodiment, the evaluation and optimization module includes:

[0190] The strength correction unit is used to obtain the standard insulation strength test value and apply the prediction coefficient matrix for correction;

[0191] The standard comparison unit is used to compare the corrected insulation strength value with the operating standard value.

[0192] The feature extraction unit is used to extract the statistical features of the corrected insulation strength value;

[0193] The classification processing unit is used to perform K-means classification processing on the insulation strength characteristics;

[0194] The parameter adjustment unit is used to adjust the insulation strength assessment parameters according to the classification results.

[0195] The index calculation unit is used to calculate the adjusted insulation performance index value;

[0196] The report generation unit is used to generate structured reports containing evaluation results.

[0197] The working principle and beneficial effects of the above technical solution are as follows: An initial insulation strength value is obtained from traditional evaluation results, and then corrected using a pre-established prediction coefficient matrix to obtain the corrected insulation strength value. The traditional evaluation results refer to preliminary insulation strength assessment values ​​obtained using conventional systems (such as periodic insulation resistance testing, dielectric loss testing, etc.). These systems often fail to fully consider the influence of material microstructure and electric field distribution on insulation performance. The prediction coefficient matrix contains a quantitative relationship between changes in dielectric constant and insulation strength attenuation.

[0198] Based on the corrected insulation strength value, a preset operating standard threshold is used for comparison to determine whether it meets the operating standard, thus obtaining a compliance result. The operating standard threshold is the minimum insulation strength value determined according to national standards, industry specifications, and equipment safety operation requirements. The comparison process uses a simple threshold judgment method: when the corrected insulation strength value is higher than the operating standard threshold, it is determined to meet the operating standard; otherwise, it is determined to not meet the operating standard.

[0199] When the compliance result meets the operating standards, key features are extracted from the corrected insulation strength values ​​to generate preliminary performance evaluation data. These key features include parameters such as the absolute magnitude of the insulation strength value, the margin ratio relative to the standard threshold, and historical trends. Feature extraction employs a statistical analysis system to calculate the mean, standard deviation, and rate of change of these parameters.

[0200] The preliminary performance evaluation data is classified to determine the optimization direction of the performance evaluation data and obtain the classification results. The classification process uses the support vector machine algorithm, which takes the preliminary performance evaluation data as feature input and outputs the corresponding performance category.

[0201] Based on the classification results, the performance evaluation data is further optimized to generate optimized performance evaluation data. The optimization process employs the Random Forest algorithm, which improves evaluation accuracy by constructing multiple decision trees and averaging their predictions. The Random Forest works by randomly selecting multiple subsets with replacement from the original training set, training a decision tree for each subset, and then voting on or averaging the predictions from all decision trees to obtain the final prediction result.

[0202] Performance indicators are obtained from optimized performance evaluation data, and analyzed using preset threshold judgment rules to obtain the final performance evaluation results. These performance indicators include multiple aspects such as insulation strength margin coefficient, expected remaining service life, and maintenance recommendations. The threshold judgment rules are a set of decision rules based on empirical data and safety standards, used to convert continuous performance indicator values ​​into discrete evaluation levels.

[0203] Based on the final performance evaluation results, structured data output is generated, completing the performance evaluation report. The structured data output organizes the evaluation results into a structured data file according to a predefined format, typically using standard formats such as JSON and XML. The performance evaluation report includes a summary of the evaluation results, detailed analysis data, risk warnings, and maintenance recommendations, presented in a combination of charts and text for easy understanding and decision-making by management. After the evaluation report is generated, the system stores it in a preset database and can push notifications to relevant personnel as needed, ensuring that the evaluation results can be used promptly to guide actual work.

[0204] In another embodiment, the gradient calculation unit is equipped with a gradient vector calculator, which calculates the electric field intensity gradient value according to the following formula:

[0205]

[0206] Where E represents the electric field strength scalar, This represents the partial derivative of the electric field intensity in the x-direction. This represents the partial derivative of the electric field intensity in the y-direction. denoted by , where x, y, and z represent the partial derivative of the electric field intensity in the z-direction, and x, y, and z represent the spatial coordinates.

[0207] The working principle and beneficial effects of the above technical solution are as follows: The electric field intensity gradient value at the calculation interface is calculated using a gradient formula, where the gradient value includes the partial derivatives of the electric field intensity in the x, y, and z directions, where x, y, and z are spatial coordinates. The electric field intensity gradient value is the rate of change of the electric field intensity E in each spatial direction, used to evaluate the uniformity of the local electric field distribution. The gradient calculation formula is specifically expressed as follows: Where E is the electric field intensity, and x, y, and z are three-dimensional spatial coordinates, representing the three orthogonal directions of the busbar interface space.

[0208] In the process of evaluating the insulation performance of busbar trunking, calculating the electric field intensity gradient is crucial for accurately identifying regions of concentrated electric field. Because the interfaces of insulating materials often contain material discontinuities and microscopic defects, resulting in uneven electric field distribution, calculating the electric field gradient can precisely locate these high-risk areas.

[0209] In another embodiment, the clustering analysis unit includes: a distance calculation subunit for calculating the Euclidean distance between nodes in the electric field distortion region; a clustering initialization subunit for initializing K cluster centers; an iterative update subunit for iteratively assigning nodes to the nearest center and updating the center position; and a termination judgment subunit for completing clustering when the change in the center position is less than a specified value.

[0210] The risk assessment unit includes: a feature construction subunit, used to construct node feature vectors; a hyperplane setting subunit, used to set hyperplane parameters; a distance calculation subunit, used to calculate the distance from the node to the hyperplane; and a risk determination subunit, used to determine the risk level of the node based on the distance.

[0211] The working principle and beneficial effects of the above technical solution are as follows: the K-means algorithm is used to perform cluster analysis on the set of nodes in the high-risk area, and the support vector machine algorithm is used to determine whether each center point is a high-risk partial discharge location. Among them, the K-means algorithm is a commonly used cluster analysis system used to divide the points in the set of nodes in the high-risk area into K clusters. First, K points are randomly selected as the initial cluster centers, and then the following steps are repeated until convergence: (1) each point is assigned to the cluster to which the nearest cluster center belongs; (2) the center point of each cluster is recalculated.

[0212] Support Vector Machine (SVM) is a supervised learning system used to determine whether each cluster center is a high-risk partial discharge location. SVM separates data points of different categories by constructing an optimal separating hyperplane, mathematically expressed as f(x) = sign(w·x+b), where w represents the weight vector, b represents the bias term, and x represents the feature vector. During the training phase, the algorithm solves a quadratic programming problem to find the hyperplane parameters w and b that maximize the inter-class margin. During the prediction phase, new sample points are substituted into the function f(x), and their category is determined based on the output sign.

[0213] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from the spirit and scope of this invention.

Claims

1. A busbar trunking insulation performance evaluation system, characterized in that, include: The microscopic acquisition module is used to acquire microscopic images of the bonding interface using a scanning electron microscope to obtain material distribution and defect parameters. The finite element analysis module is used to perform finite element analysis using material distribution and defect parameters, and to calculate the electric field intensity gradient value. The electric field distortion identification module is used to identify electric field distortion regions based on electric field intensity gradient values ​​and determine the coordinates of partial discharge locations. The dielectric parameter measurement module is used to measure the dielectric constant and loss factor at the partial discharge location coordinates and generate variation curves. The aging index calculation module is used to extract peak and slope features from the change curve and calculate the material aging index. The correlation model building module is used to establish a correlation model between the material aging index and insulation strength, and generate prediction coefficients. The evaluation and optimization module is used to correct the insulation performance evaluation value using prediction coefficients and output an optimized evaluation report.

2. The busbar insulation performance evaluation system according to claim 1, characterized in that, The microscopic acquisition module includes: An image acquisition unit is used to acquire microscopic images of the interface using a scanning electron microscope. The image filtering unit is used to perform Gaussian filtering noise reduction on the microscopic image; The edge detection unit is used to extract material boundaries from the denoised microscopic image using Canny edge detection. The uniformity analysis unit is used to perform distribution uniformity analysis on material boundaries and calculate distribution parameters; The defect feature extraction unit is used to calculate the geometric features of the defect region based on the distribution parameters. Defect classification unit, used to classify defects by comparing their geometric features with a reference standard; The report generation unit is used to generate a report on the microscopic characteristics of the interface based on the comparison and classification results.

3. The busbar insulation performance evaluation system according to claim 1, characterized in that, The finite element analysis module includes: The material parameter acquisition unit is used to acquire material distribution and defect parameters, and extract the material dielectric constant and spatial coordinates. The three-dimensional model building unit is used to establish a three-dimensional electric field distribution finite element model based on the material dielectric constant and spatial coordinates. Mesh generation unit, used to perform mesh generation and optimize mesh density based on geometric complexity; The electric field distribution calculation unit is used to calculate the electric field distortion distribution and obtain the nodal electric field strength values. Gradient calculation unit, used to calculate the electric field intensity gradient at the interface; Mesh optimization cells are used to adjust the mesh density and recalculate to optimize gradient values. The result verification unit is used to verify the consistency between the simulation results and the experimental data, and to determine the final gradient value.

4. The busbar insulation performance evaluation system according to claim 1, characterized in that, The electric field distortion identification module includes: The data processing unit is used to process electric field intensity distribution data and construct a three-dimensional grid data structure. The gradient distribution calculation unit is used to calculate the electric field intensity gradient value of each node and form a gradient distribution dataset. The distortion region marking unit is used to mark the electric field distortion region nodes according to the set gradient threshold; The clustering analysis unit is used to perform distance clustering analysis on nodes in the electric field distortion region. The gradient trend calculation unit is used to calculate the gradient change trend of nodes around the cluster center point; The risk assessment unit is used to assess the discharge risk level of each center point based on the gradient trend. The coordinate output unit is used to output the center point of the high-risk level as the coordinate set of the partial discharge location.

5. The busbar insulation performance evaluation system according to claim 1, characterized in that, The dielectric parameter measurement module includes: The measurement point selection unit is used to select the coordinates of the measurement point based on the partial discharge location list. The parameter acquisition unit is used to acquire dielectric constant and loss factor data at the measurement point using a dielectric spectrum analyzer. The frequency domain transformation unit is used to perform a fast Fourier transform on the acquired data and convert it into frequency domain features; The peak identification unit is used to identify peak points in the frequency domain characteristics and determine the trend of dielectric parameter variation. The curve generation unit is used to perform cubic spline interpolation calculations to generate continuously changing curves. The stability analysis unit is used to calculate the slope of the curve and the location of the inflection point to analyze the stability of dielectric properties. The curve sequence generation unit is used to generate a sequence of dielectric property change curves for each measurement point.

6. The busbar insulation performance evaluation system according to claim 1, characterized in that, The aging index calculation module includes: Peak marking unit, used to mark the point of maximum value in the change curve as the peak feature point; The slope calculation unit is used to calculate the slope value between adjacent peak points, forming a slope dataset. The aging trend calculation unit is used to extract the rate of change of the absolute value of the slope and calculate the aging trend index. The aging level determination unit is used to compare the aging trend index with a reference value to determine the aging level. The time series processing unit is used to perform time series smoothing on aging level data; The aging rate calculation unit is used to calculate the slope and peak distribution of the smoothed curve and obtain the aging rate value. The evaluation result generation unit is used to generate material aging degree evaluation results based on the aging rate value.

7. The busbar insulation performance evaluation system according to claim 1, characterized in that, The association model building module includes: The data acquisition unit is used to acquire historical data on material aging index, dielectric constant, and insulation strength. The outlier handling unit is used to remove outliers in the dataset that deviate from the mean by three times the standard deviation. The dimensionality reduction unit is used to perform principal component analysis dimensionality reduction and extract feature vectors. The regression model building unit is used to establish a linear regression model of dielectric constant change and insulation strength attenuation. The coefficient calculation unit is used to calculate regression coefficients and form a prediction coefficient matrix. The insulation strength calculation unit is used to calculate the insulation strength attenuation value based on the prediction coefficient matrix. The grading result generation unit is used to generate material aging grading results based on the insulation strength attenuation value.

8. The busbar insulation performance evaluation system according to claim 1, characterized in that, The evaluation and optimization module includes: The strength correction unit is used to obtain the standard insulation strength test value and apply the prediction coefficient matrix for correction; The standard comparison unit is used to compare the corrected insulation strength value with the operating standard value. The feature extraction unit is used to extract the statistical features of the corrected insulation strength value; The classification processing unit is used to perform K-means classification processing on the insulation strength characteristics; The parameter adjustment unit is used to adjust the insulation strength assessment parameters according to the classification results. The index calculation unit is used to calculate the adjusted insulation performance index value; The report generation unit is used to generate structured reports containing evaluation results.

9. The busbar insulation performance evaluation system according to claim 3, characterized in that, The gradient calculation unit is equipped with a gradient vector calculator, which calculates the electric field intensity gradient value according to the following formula: Where E represents the electric field strength scalar, This represents the partial derivative of the electric field intensity in the x-direction. This represents the partial derivative of the electric field intensity in the y-direction. denoted by , where x, y, and z represent the partial derivative of the electric field intensity in the z-direction, and x, y, and z represent the spatial coordinates.

10. The busbar insulation performance evaluation system according to claim 4, characterized in that: The clustering analysis unit includes: a distance calculation subunit, used to calculate the Euclidean distance between nodes in the electric field distortion region; a cluster initialization subunit, used to initialize K cluster centers; an iterative update subunit, used to iteratively assign nodes to the nearest center and update the center position; and a termination judgment subunit, used to complete clustering when the change in the center position is less than a specified value. The risk assessment unit includes: a feature construction subunit, used to construct node feature vectors; a hyperplane setting subunit, used to set hyperplane parameters; a distance calculation subunit, used to calculate the distance from the node to the hyperplane; and a risk determination subunit, used to determine the risk level of the node based on the distance.

Images