A method for on-site inspection of multi-point grounding of transformer core and clamps

Abstract

The present invention provides an on-site inspection method for multi-point grounding of a transformer core and clamps, which belongs to the technical field of transformer cores and clamps. The method comprises: using a handheld grounding resistance tester to preliminarily determine the basic positions of multiple grounding points; establishing an electromagnetic model of the transformer core and clamps to determine the approximate distribution area of ​​the multi-point grounding positions in the core and clamps, which is recorded as a preliminary position group; using a genetic algorithm to further optimize the electromagnetic model parameters to obtain a corrected position group; collecting an ultrasonic signal group to obtain a preprocessed ultrasonic signal group; extracting a feature vector to determine the source coordinates of each ultrasonic signal; performing correlation analysis on the source coordinates of each ultrasonic signal and the corrected position group to determine the specific positions of the multi-point grounding; and setting a local electromagnetic probe at the determined multi-point grounding position for actual detection to measure the local magnetic flux density and eddy current distribution of the core and clamps, and finally obtaining the determined position of the multi-point grounding fault.

Description

Technical Field

[0001] The present invention belongs to the technical field of transformer cores and clamps, and in particular relates to a method for on-site inspection of multi-point grounding of transformer cores and clamps. Background Art

[0002] Transformers are crucial components of power systems, performing key functions such as voltage conversion and power transmission. The operating condition of the transformer core and clamps, core components of the transformer, directly impacts the performance and reliability of the entire transformer. During actual transformer operation, the core and clamps may experience multi-point grounding faults. These faults can cause severe eddy current losses and localized overheating, leading to further damage such as insulation breakdown and mechanical deformation. Therefore, the effective detection and location of multi-point grounding faults in the transformer core and clamps has long been a key issue in transformer condition monitoring and fault diagnosis.

[0003] At present, the detection methods for multi-point grounding faults of transformer cores and clamps mainly include the following:

[0004] 1. Ground resistance measurement method: Use a handheld ground resistance tester to measure each suspected ground point one by one, looking for locations with lower ground resistance values ​​to preliminarily determine the locations of multiple ground points. This method is simple to operate, but can only provide approximate location information and is difficult to pinpoint accurately.

[0005] 2. Infrared thermal imaging: An infrared thermal imager is used to capture the temperature distribution on the transformer core and fixture surfaces. The location of multiple grounding points is determined by analyzing localized overheating areas. This method provides nondestructive testing, but due to the radiation characteristics of the core and fixture surfaces, accurate temperature data is difficult to obtain and positioning accuracy is limited.

[0006] 3. Electromagnetic field testing: Using magnetic field detection equipment, such as flux probes and eddy current probes, the local magnetic field and eddy current distribution within the core and fixtures are measured. The locations of multiple grounding points are inferred based on abnormal values ​​of these physical quantities. This method can obtain more information about the core and fixture interior, but it still requires prior knowledge of suspected locations and can be difficult to deploy.

[0007] 4. Acoustic Emission Detection: This method uses an acoustic emission sensor array to detect acoustic emission signals from the transformer core and fixture surfaces, and determines the locations of multiple grounding points through signal analysis. This method enables non-destructive, contactless detection, but it requires high signal processing and analysis, and its positioning accuracy is limited.

[0008] In summary, most existing transformer core and clamp multi-point grounding fault detection methods can only provide an approximate location area and cannot provide sufficiently detailed information. Summary of the Invention

[0009] In view of this, the present invention provides a method for on-site inspection of multi-point grounding of transformer cores and clamps, which can solve the technical problem of limited positioning accuracy in existing on-site inspection methods for multi-point grounding of transformer cores and clamps.

[0010] The present invention is achieved in that:

[0011] A first aspect of the present invention provides a method for on-site inspection of multi-point grounding of a transformer core and clamps, comprising the following steps:

[0012] S10. Use a handheld ground resistance tester to roughly test the suspected grounding points of the transformer core and clamps to preliminarily determine the basic positions of multiple grounding points;

[0013] S20. Using finite element analysis software, establish an electromagnetic model of the transformer core and the clamps to simulate and analyze the eddy current distribution and magnetic flux density distribution in the core and the clamps;

[0014] S30, based on the basic position, performing simulation analysis on the electromagnetic model of the transformer core and the clamp to determine the approximate distribution area of ​​the multi-point grounding positions in the core and the clamp, and recording the area as a preliminary position group;

[0015] S40, combining the preliminary position group, further optimizing the electromagnetic model parameters using a genetic algorithm to obtain a revised multi-point grounding position, which is recorded as a revised position group;

[0016] S50, using an ultrasonic sensor matrix to collect an ultrasonic signal group from the transformer core and the surface of the clamp, and preprocessing the collected ultrasonic signals to obtain a preprocessed ultrasonic signal group;

[0017] S60, extracting a feature vector from each ultrasonic signal in the preprocessed ultrasonic signal group, and determining the source coordinates of each ultrasonic signal based on the feature vector;

[0018] S70, performing correlation analysis using a machine learning algorithm on the source coordinates of each ultrasonic signal and the corrected position group to determine the specific locations of the multi-point grounding;

[0019] S80. At the determined multi-point grounding location, a local electromagnetic probe is set up to perform actual detection, and the local magnetic flux density and eddy current distribution of the core and the clamp are measured to finally determine the location of the multi-point grounding fault.

[0020] On the basis of the above technical solution, the on-site inspection method for multi-point grounding of a transformer core and a clamp of the present invention can also be improved as follows:

[0021] The steps of establishing the electromagnetic model of the transformer core and the clamp using finite element analysis software are as follows:

[0022] First, operators need to obtain the 3D geometric model data of the transformer core and clamps, including the dimensions and material parameters of the core and clamps.

[0023] Then, using finite element analysis software, an electromagnetic model of the transformer core and clamps is established;

[0024] Again, through simulation calculation, the eddy current distribution and magnetic flux density distribution cloud map inside the core and the clamp are obtained.

[0025] Furthermore, it also includes setting boundary conditions of the electromagnetic model, wherein the boundary conditions include insulation conditions of the iron core and the clamp surface and magnetic field conditions of the surrounding environment.

[0026] Wherein, the step S30 specifically includes:

[0027] First, the preliminary multiple grounding point position group obtained in step S10 is input into the electromagnetic model established in step S20;

[0028] Then, for each preliminarily determined multi-grounding point base position, the changes in eddy current distribution and magnetic flux density distribution inside the core and clamps are simulated when a ground fault occurs at the corresponding position;

[0029] Thirdly, by comparing and analyzing the changes in eddy current distribution and magnetic flux density distribution inside the core and the clamp, the approximate distribution area of ​​multiple grounding points in the core and the clamp is determined and recorded as the preliminary position group.

[0030] Furthermore, it also includes setting an eddy current density change rate threshold and a magnetic flux density change rate threshold. Only when the eddy current density or magnetic flux density change rate of a certain position exceeds these thresholds, the corresponding position is included in the preliminary position group.

[0031] Furthermore, the eddy current density change rate threshold is set to 10%, and the magnetic flux density change rate threshold is set to 5%.

[0032] The initial population of the genetic algorithm is the preliminary position group; the objective function of the genetic algorithm is to minimize the sum of the absolute values ​​of the differences between the grounding resistance value measured by the handheld grounding resistance tester and the grounding resistance value calculated by the electromagnetic model of the transformer core and clamps.

[0033] Wherein, the step S50 specifically includes:

[0034] First, an ultrasonic sensor array consisting of multiple ultrasonic sensors is arranged on the surface of the transformer core and the clamp to collect ultrasonic signals;

[0035] Secondly, during the acquisition process, it is necessary to ensure that the sensor array can fully cover the transformer core and the surface of the clamp to obtain a complete set of ultrasonic signals;

[0036] Once the acquisition is complete, the original ultrasonic signal needs to be preprocessed by filtering, amplifying, and normalizing to eliminate noise interference, enhance the effective signal, and obtain a preprocessed ultrasonic signal group.

[0037] The step S80 specifically includes:

[0038] First, based on the analysis results of step S70, a group of small electromagnetic probes, including flux probes and eddy current probes, are arranged at the determined multi-point grounding locations on the transformer core and the surface of the clamp;

[0039] Secondly, during actual testing, the specific locations of multi-point grounding are further confirmed by analyzing the magnetic flux density and eddy current distribution data collected by the probe.

[0040] Compared with the prior art, the method for on-site inspection of multi-point grounding of transformer core and clamps provided by the present invention has the following beneficial effects:

[0041] 1. Significantly Improved Positioning Accuracy: This method utilizes a combination of electromagnetic field analysis, ultrasonic testing, and machine learning algorithms to precisely locate the specific locations of multiple grounding points within the transformer core and fixtures, achieving centimeter-level accuracy. This significantly surpasses existing single-physical-quantity detection methods, such as ground resistance measurement and infrared thermal imaging.

[0042] 2. Comprehensively reflect the internal characteristics of the core and clamps: By establishing an electromagnetic field model of the transformer core and clamps, this method can simulate and analyze the eddy current distribution and magnetic flux density distribution inside the core and clamps, thereby better improving positioning accuracy.

[0043] In summary, the solution of the present invention solves the technical problem of limited positioning accuracy in the existing on-site inspection method for multi-point grounding of transformer cores and clamps. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0045] Figure 1 A flow chart of the method provided by the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0047] like Figure 1 FIG. 1 is a flow chart of a method for on-site inspection of multi-point grounding of a transformer core and clamps provided by the present invention. The method comprises the following steps:

[0048] S10. Use a handheld ground resistance tester to roughly test the suspected grounding points of the transformer core and clamps to preliminarily determine the basic positions of multiple grounding points;

[0049] S20. Using finite element analysis software, establish an electromagnetic model of the transformer core and the clamps to simulate and analyze the eddy current distribution and magnetic flux density distribution in the core and the clamps;

[0050] S30, based on the basic position, simulate and analyze the electromagnetic model of the transformer core and the clamp to determine the approximate distribution area of ​​the multi-point grounding positions in the core and the clamp, and record it as a preliminary position group;

[0051] S40, combining the preliminary position group, further optimizing the electromagnetic model parameters using a genetic algorithm to obtain a revised multi-point grounding position, which is recorded as a revised position group;

[0052] S50, using an ultrasonic sensor matrix to collect an ultrasonic signal group from the transformer core and the surface of the clamp, and preprocessing the collected ultrasonic signals to obtain a preprocessed ultrasonic signal group;

[0053] S60, extracting a feature vector from each ultrasonic signal in the preprocessed ultrasonic signal group, and determining the source coordinates of each ultrasonic signal based on the feature vector;

[0054] S70, performing correlation analysis using a machine learning algorithm on the source coordinates of each ultrasonic signal and the corrected position group to determine the specific locations of the multi-point grounding;

[0055] S80. At the determined multi-point grounding location, a local electromagnetic probe is set up to perform actual detection, and the local magnetic flux density and eddy current distribution of the core and the clamp are measured to finally determine the location of the multi-point grounding fault.

[0056] The specific implementation of the above steps is described in detail below:

[0057] Step S10: Use a handheld ground resistance tester to perform a rough test on the suspected grounding points of the transformer core and the clamps to preliminarily determine the basic positions of multiple grounding points.

[0058] The specific implementation is as follows: First, the operator needs to carefully inspect the transformer core and the surface of the clamps to identify locations where multiple grounding may occur. These locations are usually manifested by obvious rust, corrosion, or dents on the core and clamp surfaces.

[0059] Next, a handheld ground resistance tester is used to quickly test these suspected grounding points. Using the four-wire method, the tester measures the grounding resistance value at each suspected point. By analyzing the distribution of these grounding resistance values, the location of multiple grounding points within the core and clamps can be initially determined.

[0060] Generally speaking, locations with low ground resistance values ​​are considered to have multiple grounding points. A ground resistance threshold, such as 500 milliohms, can be set. Locations below this threshold are considered suspect grounding points. By testing all suspect points, a preliminary set of multiple grounding point locations can be obtained, laying the foundation for subsequent precise positioning.

[0061] Step S20: Using finite element analysis software, an electromagnetic model of the transformer core and the clamp is established to simulate and analyze the eddy current distribution and magnetic flux density distribution in the core and the clamp.

[0062] The specific implementation is as follows: First, the operator needs to obtain the three-dimensional geometric model data of the transformer core and clamps, including the dimensions, material parameters, etc. This data can be obtained through actual measurement or from the transformer's technical data.

[0063] Then, using finite element analysis software such as ANSYS Maxwell or COMSOL Multiphysics, an electromagnetic model of the transformer core and clamps is established. This model needs to consider the nonlinear magnetic properties of the core and clamp materials and set appropriate boundary conditions and excitation sources, such as the rated voltage and current parameters of the transformer.

[0064] After establishing the electromagnetic model, simulation calculations can be used to obtain eddy current distribution and magnetic flux density distribution cloud maps within the core and clamps. These results can intuitively reflect the electromagnetic field distribution characteristics in the core and clamps, providing an important basis for subsequent multi-point grounding location analysis.

[0065] Step S30: Based on the basic position, a simulation analysis is performed on the electromagnetic model of the transformer core and the clamp to determine the approximate distribution area of ​​the multi-point grounding positions in the core and the clamp, which is recorded as a preliminary position group.

[0066] The specific implementation is as follows: First, the preliminary set of multiple grounding point locations obtained in step S10 is input into the electromagnetic model established in step S20. Then, for each suspected grounding point, the eddy current distribution and magnetic flux density changes within the core and clamp are simulated when a ground fault occurs at that location.

[0067] By comparing and analyzing these simulation results, we can determine the approximate distribution of multiple grounding points within the core and fixtures. Generally speaking, eddy current density and magnetic flux density vary significantly near the grounding points, and these areas of variation can serve as a preliminary set of multiple grounding locations.

[0068] To improve analysis accuracy, you can set thresholds for the eddy current density change rate and the magnetic flux density change rate, such as 10% and 5%. Only when the eddy current density or magnetic flux density change rate at a location exceeds these thresholds will it be included in the preliminary location group. This can eliminate interference points that have little impact on the electromagnetic field.

[0069] Step S40: Based on the preliminary position group, the electromagnetic model parameters are further optimized using a genetic algorithm to obtain a revised multi-point grounding position, which is recorded as a revised position group.

[0070] The specific implementation is as follows: A genetic algorithm is an optimization algorithm based on natural selection and genetic mechanisms, and performs well in solving complex nonlinear problems. In this step, the genetic algorithm is used to optimize the parameters of the electromagnetic model established in step S20 to obtain more accurate multi-point grounding locations.

[0071] First, the preliminary position group obtained in step S30 is used as the initial population of the genetic algorithm. Then, by designing appropriate objective functions and genetic operators such as selection, crossover, and mutation, the simulation results of the electromagnetic model can be as close to the actual situation as possible.

[0072] The objective function can be designed to minimize the error between the ground resistance value measured in step S10 and the ground resistance value calculated by the model. During the iterative optimization process, the genetic algorithm continuously adjusts model parameters, such as the core and clamp material parameters, and the grounding point location, until the objective function reaches convergence conditions.

[0073] The final optimization result is the revised multi-point grounding position group, which is recorded as the revised position group. Compared with the preliminary result of step S30, this position group more accurately reflects the actual distribution of multiple grounding points in the core and the clamp.

[0074] Step S50: Using an ultrasonic sensor matrix, collect ultrasonic signal groups on the transformer core and the surface of the clamp, and preprocess the collected ultrasonic signals to obtain a preprocessed ultrasonic signal group.

[0075] The specific implementation is as follows: First, an ultrasonic sensor array is placed on the transformer core and the surface of the clamp. This array consists of multiple ultrasonic sensors, and the sensor spacing can be adjusted as needed. Each sensor can independently collect ultrasonic signals reflected from the core and the clamp surface.

[0076] During the acquisition process, it is necessary to ensure that the sensor array can fully cover the transformer core and the surface of the clamp to obtain a complete set of ultrasonic signals. To improve signal quality, the pulse echo method can be used for ultrasonic excitation and reception.

[0077] After acquisition, the raw ultrasonic signals need to be preprocessed, including filtering, amplification, and normalization, to eliminate noise interference and enhance the effective signal. The resulting ultrasonic signal group is recorded as the preprocessed ultrasonic signal group, preparing for subsequent feature extraction and position determination.

[0078] Step S60: extracting a feature vector from each ultrasonic signal in the pre-processed ultrasonic signal group, and determining the source coordinates of each ultrasonic signal based on the feature vector.

[0079] The specific implementation is as follows: First, for each preprocessed ultrasound signal, time-domain and frequency-domain feature extraction is performed. Time-domain features can include signal amplitude, envelope, and energy; frequency-domain features can include spectrum peaks and frequency centers. These features form a feature vector that describes the characteristics of each ultrasound signal.

[0080] Then, using machine learning algorithms, such as support vector machines (SVMs) or neural networks, a prediction model is built to map the ultrasonic signal's feature vectors to their spatial coordinates on the core and fixture surfaces. This model requires prior learning and training using training data with known coordinates.

[0081] In practice, by feeding each signal in the preprocessed ultrasonic signal group into a trained prediction model, the source coordinates of each signal can be obtained. These coordinates provide an important basis for subsequent multi-point grounding location correlation analysis.

[0082] Step S70: For each ultrasonic signal source coordinate and the correction position group, a machine learning algorithm is used to perform correlation analysis to determine the specific location of the multi-point grounding.

[0083] The specific implementation is as follows: First, the source coordinates of each ultrasound signal obtained in step S60 are paired and associated with the corrected position group obtained in step S40. This process can be implemented using a machine learning algorithm, such as a clustering algorithm or an anomaly detection algorithm.

[0084] The clustering algorithm can group coordinate points so that points near the same grounding point are clustered together. The anomaly detection algorithm can identify coordinate points that are significantly inconsistent with the corrected position group and identify them as possible other grounding points.

[0085] This correlation analysis can determine the specific locations of multiple grounding points corresponding to each cluster center or outlier. It can also calculate the confidence level or reliability of each grounding point, providing a basis for subsequent verification and testing.

[0086] In general, the purpose of this step is to use the spatial position information of the ultrasonic signal, combined with the optimization results of the electromagnetic model, to ultimately determine the specific locations of multi-point grounding in the transformer core and clamps.

[0087] Step S80: At the determined multi-point grounding location, a local electromagnetic probe is set up to perform actual detection to measure the local magnetic flux density and eddy current distribution of the core and the clamp, and finally determine the location of the multi-point grounding fault.

[0088] The specific implementation is as follows: First, based on the analysis results of step S70, a set of small electromagnetic probes are placed at multiple grounding locations on the transformer core and the surface of the clamp. These probes can be a combination of flux probes and eddy current probes to accurately measure the magnetic flux density and eddy current distribution at these locations.

[0089] During actual testing, the specific location of the multi-point ground fault can be further confirmed by analyzing the magnetic flux density and eddy current distribution data collected by the probe. If the measurement results match the analysis results of step S70, the location can be determined to be the specific location of the multi-point ground fault.

[0090] To improve detection accuracy, you can set some reference thresholds. For example, you can set the magnetic flux density change rate threshold to 15%, and the eddy current density change rate threshold to 20%. Only when the measured data at a certain location exceeds these thresholds is it determined as the final location of the multiple grounding point.

[0091] By following these steps, the locations of multiple ground faults in the transformer core and its clamps can be comprehensively and accurately determined. This method, which utilizes a combination of electromagnetic simulation, ultrasonic testing, and localized magnetic measurement, effectively addresses the inability of traditional detection methods to accurately locate faults.

[0092] In order to more clearly describe the solution of the present invention, the following is explained in conjunction with a specific formula:

[0093] Step S10: Use a handheld ground resistance tester to roughly test the suspected grounding points of the transformer core and clamps to preliminarily determine the basic positions of multiple grounding points. The specific implementation method is as follows:

[0094] First, for the transformer core and the surface of the clamp, a coordinate system (x, y, z) can be defined, where (x, y) represents the plane coordinates of the core and the clamp surface, and the z axis is perpendicular to the core and the clamp surface. In this coordinate system, the core and the clamp surface can be scanned to identify the location coordinates (x, y, z) where multiple grounding points may exist. i ,y i ), where i = 1, 2, ..., n, and n is the number of suspicious grounding points.

[0095] For each suspected grounding point (x i ,y i ), use a handheld ground resistance tester to measure and get the ground resistance R at that location i . Grounding resistance R i It can be expressed as follows:

[0096]

[0097] Among them, U i is the voltage at that location, I i is the current passing through this position. By measuring U i and I i , we can calculate R i .

[0098] In order to preliminarily determine the location of multiple grounding points, a grounding resistance threshold R can be set. th , for example R th = 500 milliohms. When the grounding resistance R i <R th When the grounding point is reached, it is classified as the base position of multiple grounding points and its coordinates (x i ,y i By performing this operation on all suspicious points, a preliminary multi-ground point location group can be obtained. Where i = 1, 2, ..., m, where m is the number of multiple grounding points initially determined.

[0099] Step S20: Using finite element analysis software, establish an electromagnetic model of the transformer core and the clamp to simulate and analyze the eddy current distribution and magnetic flux density distribution in the core and the clamp. The specific implementation is as follows:

[0100] First, we need to obtain the geometric parameters and material properties of the transformer core and clamps and build a three-dimensional finite element model. The geometric parameters of the core and clamps include length L, width W and thickness H, and the material properties include the specific conductivity σ and relative permeability μ of the core and clamp materials. r .

[0101] According to Maxwell's equations, the electromagnetic field model of the transformer core and the clamp can be established. The electromagnetic field equation can be expressed as:

[0102] Electric field equation:

[0103]

[0104] Magnetic field equation:

[0105]

[0106] Calculation relationship:

[0107]

[0108] in, is the electric field strength, is the magnetic induction intensity, is the magnetic field strength, is the current density, is the electric displacement, μ is the magnetic permeability, σ is the electrical conductivity, and ε is the dielectric constant.

[0109] To simulate the eddy current and magnetic flux density distribution in the core and fixtures, appropriate boundary conditions and excitation sources must be applied to the electromagnetic field equations. Boundary conditions can include insulation conditions on the core and fixture surfaces or the surrounding magnetic field. The excitation source can be the rated voltage or current of the transformer.

[0110] By solving the electromagnetic field model with finite element analysis software, the eddy current density distribution inside the core and the clamp can be obtained. and magnetic flux density distribution These results provide an important basis for subsequent multi-point grounding location analysis.

[0111] Step S30: Based on the basic position, simulate and analyze the electromagnetic model of the transformer core and the clamp to determine the approximate distribution area of ​​the multi-point grounding positions in the core and the clamp, which is recorded as a preliminary position group. The specific implementation method is as follows:

[0112] First, the preliminary multi-grounding point position group obtained in step S10 is Input into the electromagnetic model established in step S20. i ,y i ), applying a ground fault condition at this location to simulate the eddy current distribution inside the core and the clamp and magnetic flux density distribution

[0113] In order to determine the approximate distribution area of ​​multiple grounding points in the core and clamps, two indicators can be defined:

[0114] Eddy current density change rate Δ J and the rate of change of magnetic flux density Δ B , respectively:

[0115]

[0116] in, and They represent the eddy current distribution and magnetic flux density distribution under normal working conditions respectively.

[0117] Two thresholds Δ can be set J，th and Δ B，th , such as 10% and 5%. Only when the Δ J or Δ B When these thresholds are exceeded, they are classified into preliminary position groups. In this way, the approximate distribution area of ​​the multi-point grounding positions in the core and the clamps can be determined.

[0118] Step S40: Based on the preliminary position group, the electromagnetic model parameters are further optimized using a genetic algorithm to obtain a revised multi-point grounding position, which is recorded as a revised position group. The specific implementation is as follows:

[0119] The genetic algorithm is an optimization algorithm based on natural selection and heredity, and is highly effective at solving complex nonlinear problems. In this step, the genetic algorithm is used to optimize the parameters of the electromagnetic model established in step S20 to obtain more accurate multi-point grounding locations.

[0120] First, the preliminary position group obtained in step S30 is As the initial population of the genetic algorithm. Each individual (x i ,y i ) indicate a possible multiple grounding point location.

[0121] Then, we need to design appropriate objective functions and genetic operators. The objective function can be defined as:

[0122]

[0123] in, is the ground resistance value of the i-th suspicious point measured in step S10, The ground resistance value at this location is calculated for the electromagnetic model. We hope to minimize this objective function so that the model calculation results are as close as possible to the actual measured values.

[0124] Genetic operators include selection, crossover, and mutation. During the iterative optimization process, the genetic algorithm continuously adjusts individual genes, that is, model parameters such as core and clamp material parameters, grounding point locations, etc., until the objective function reaches convergence conditions.

[0125] The final optimization result is the corrected multi-point grounding position group Compared with the preliminary results of step S30 This position group It more accurately reflects the actual distribution of multiple grounding points in the core and clamps.

[0126] Step S50: Using an ultrasonic sensor matrix, collect ultrasonic signal groups from the transformer core and the surface of the clamp, and pre-process the collected ultrasonic signals to obtain a pre-processed ultrasonic signal group. The specific implementation is as follows:

[0127] An m×n ultrasonic sensor array is arranged on the surface of the transformer core and the clamp. The coordinates of each sensor are (x i ,y j ), where i = 1, 2, ..., m, j = 1, 2, ..., n. Each sensor can independently collect the ultrasonic signal s reflected from the core and the surface of the clamp. ij (t), where t represents time.

[0128] During the acquisition process, it is necessary to ensure that the sensor array can fully cover the transformer core and the surface of the clamp to obtain a complete set of ultrasonic signal In order to improve the signal quality, the pulse echo method can be used for ultrasonic excitation and reception.

[0129] After the acquisition is completed, the original ultrasonic signal s ij (t) Perform preprocessing, including filtering, amplification and normalization, to eliminate noise interference and enhance effective signals. The ultrasonic signal group obtained after preprocessing is recorded as Prepare for subsequent feature extraction and position judgment.

[0130] Step S60: Extract a feature vector from each ultrasonic signal in the preprocessed ultrasonic signal group, and determine the source coordinates of each ultrasonic signal based on the feature vector. Specific implementations are as follows:

[0131] For each preprocessed ultrasonic signal s i ′ j (t), extract time domain and frequency domain features to form a feature vector Time domain features can include signal amplitude, envelope, energy, etc.; frequency domain features can include spectrum peak, frequency center, etc. These features can be expressed by the following formula:

[0132] Time domain characteristics:

[0133]

[0134] Frequency domain characteristics:

[0135]

[0136] in, represents the Fourier transform. These features constitute the feature vector

[0137] Then, using supervised learning algorithms such as support vector machines (SVM) or neural networks, a prediction model is built The characteristic vector of the ultrasonic signal Mapped to its spatial coordinates (x i ,y j ). The model needs to be learned and trained in advance using training data with known coordinates.

[0138] In practical applications, the preprocessed ultrasound signal group The feature vector of each signal in is input into the trained prediction model The source coordinates (x i ,y j ). These coordinate information provides an important basis for the subsequent multi-point grounding location correlation analysis.

[0139] Step S70: For each ultrasonic signal source coordinate and the corrected position group, a machine learning algorithm is used to perform correlation analysis to determine the specific location of the multi-point grounding. The specific implementation is as follows:

[0140] First, the source coordinates (x i ,y j ), and the corrected position group obtained in step S40 Pairing and association are performed. This process can be achieved through clustering algorithms or anomaly detection algorithms.

[0141] The clustering algorithm can be used to cluster the coordinate points (x i ,y j ) to group the points near the same ground point so that they are clustered together. The k-means algorithm can be used, and its objective function is:

[0142]

[0143] in, is the lth cluster, is the lth cluster center. Through iterative optimization, k cluster centers can be obtained, which are the preliminary locations of multi-point grounding.

[0144] Another approach is anomaly detection algorithms, which identify those The coordinate points (x i ,y j ), identified as other possible grounding points. A one-class support vector machine (SVM) can be used as an anomaly detection model, and its objective function is:

[0145]

[0146] Among them, Φ(·) is the mapping function, v is the upper limit of the abnormal sample ratio, ξ i is a slack variable. By training this model, the coordinates of the outliers, i.e., the locations of other possible multi-ground points, can be obtained.

[0147] Through the above cluster analysis and anomaly detection, the specific location of the multi-point grounding corresponding to each cluster center or anomaly point can be determined At the same time, the confidence or credibility of each grounding point can also be calculated to provide a basis for subsequent verification and testing.

[0148] Step S80: At the determined multi-point grounding location, a local electromagnetic probe is set up for actual detection to measure the local magnetic flux density and eddy current distribution of the core and the clamp, ultimately determining the location of the multi-point grounding fault. Specific implementation methods are as follows:

[0149] According to the analysis results of step S70, multiple grounding positions on the transformer core and the surface of the clamp are A set of small electromagnetic probes are arranged on the surface of the magnet. These probes include flux probes and eddy current probes, which are used to accurately measure the magnetic flux density B at that location. l and eddy current density J l .

[0150] During actual testing, the rate of change of magnetic flux density ΔB at each probe position can be calculated l and the eddy current density change rate ΔJ l , defined as follows:

[0151]

[0152] Among them, B0 and J0 represent the magnetic flux density and eddy current density under normal operating conditions, respectively.

[0153] In order to improve the detection accuracy, some reference thresholds can be set, such as ΔB th =15% and ΔJ th = 20%. Only when ΔB at a certain position l or ΔJ l Only when these thresholds are exceeded are they determined as the final locations of the multiple grounding points.

[0154] By following these steps, the locations of multiple ground faults in the transformer core and its clamps can be comprehensively and accurately determined. This method, which utilizes a combination of electromagnetic analysis, ultrasonic testing, and localized magnetic measurement, effectively addresses the inability of traditional detection methods to accurately locate faults.

[0155] The following is an explanation of the variables involved in the above description:

[0156]

[0157]

[0158]

[0159] Specifically, the principle of the present invention is:

[0160] 1. Electromagnetic field analysis

[0161] Multiple grounding faults in the transformer core and its clamps can cause significant changes in local eddy current density and magnetic flux density. This paper first uses finite element analysis software to establish a three-dimensional electromagnetic field model of the transformer core and its clamps, enabling simulation and analysis of the electromagnetic field distribution characteristics within the core and clamps. By applying fault conditions at suspected grounding points and observing the changes in eddy current density and magnetic flux density, the approximate distribution of multiple grounding points within the core and clamps can be preliminarily determined.

[0162] 2. Genetic Algorithm Optimization

[0163] To further improve the positioning accuracy of multiple grounding points, this paper uses a genetic algorithm to optimize the parameters of the electromagnetic field model. A genetic algorithm, a heuristic optimization algorithm based on natural selection and genetic mechanisms, can adaptively adjust model parameters to ensure that simulation results are as close as possible to actual measured data. This iterative optimization method can produce a more accurate distribution of multiple grounding points.

[0164] 3. Ultrasonic detection and feature extraction

[0165] In addition to electromagnetic field analysis, the present invention also utilizes ultrasonic testing technology to scan the transformer core and fixture surfaces, acquiring a complete set of ultrasonic signals. By extracting the time-frequency features of these ultrasonic signals, a prediction model is developed that maps the signal features to the spatial coordinates of the core and fixture surfaces. This allows the source location of each ultrasonic signal to be determined, providing a basis for subsequent correlation analysis of multiple grounding points.

[0166] 4. Machine Learning Correlation Analysis

[0167] Finally, the present invention combines the distribution of multiple grounding points obtained by electromagnetic field model optimization with the spatial coordinate information obtained by ultrasonic signal detection, and uses a machine learning algorithm to perform correlation analysis. Through clustering or anomaly detection, these two types of information can be effectively integrated to determine the specific locations of multi-point grounding faults in the transformer core and clamps. This "multiple evidence" correlation analysis greatly improves the reliability of the detection results. In the above-mentioned embodiment, the core and clamp can be regarded as a whole or as two independent individuals, which does not affect the implementation of the steps of the present invention. That is, in the steps of the embodiment, the process of each step of the present invention is implemented on a separate core or a separate clamp.

[0168] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. A method for on-site inspection of multi-point grounding of transformer core and clamps, characterized in that: The following steps are involved: S10. Use a handheld ground resistance tester to roughly test the suspected grounding points of the transformer core and clamps to preliminarily determine the basic positions of multiple grounding points; S20. Using finite element analysis software, establish an electromagnetic model of the transformer core and the clamps to simulate and analyze the eddy current distribution and magnetic flux density distribution in the core and the clamps; S30, based on the basic position, performing simulation analysis on the electromagnetic model of the transformer core and the clamp to determine the distribution area of ​​the multi-point grounding positions in the core and the clamp, and recording it as a preliminary position group; S40, combining the preliminary position group, further optimizing the electromagnetic model parameters using a genetic algorithm to obtain a revised multi-point grounding position, which is recorded as a revised position group; S50, using an ultrasonic sensor matrix to collect an ultrasonic signal group from the transformer core and the surface of the clamp, and preprocessing the collected ultrasonic signals to obtain a preprocessed ultrasonic signal group; S60, extracting a feature vector from each ultrasonic signal in the preprocessed ultrasonic signal group, and determining the source coordinates of each ultrasonic signal based on the feature vector; S70, performing correlation analysis using a machine learning algorithm on the source coordinates of each ultrasonic signal and the corrected position group to determine the specific locations of the multi-point grounding; S80. At the determined multi-point grounding location, a local electromagnetic probe is set up to perform actual detection, and the local magnetic flux density and eddy current distribution of the core and the clamp are measured to finally determine the location of the multi-point grounding fault.

2. A method for on-site inspection of multi-point grounding of transformer core and clamps according to claim 1, characterized in that: The steps of establishing the electromagnetic model of the transformer core and the clamp using finite element analysis software are specifically: First, operators need to obtain the 3D geometric model data of the transformer core and clamps, including the dimensions and material parameters of the core and clamps. Then, using finite element analysis software, an electromagnetic model of the transformer core and clamps is established; Again, through simulation calculation, the eddy current distribution and magnetic flux density distribution cloud map inside the core and the clamp are obtained.

3. A method for on-site inspection of multi-point grounding of transformer core and clamps according to claim 2, characterized in that: The method also includes setting boundary conditions of the electromagnetic model, wherein the boundary conditions include insulation conditions of the iron core and the surface of the clamp and magnetic field conditions of the surrounding environment.

4. A method for on-site inspection of multi-point grounding of transformer core and clamps according to claim 1, characterized in that: The step S30 specifically includes: First, the preliminary multiple grounding point position group obtained in step S10 is input into the electromagnetic model established in step S20; Then, for each preliminarily determined multi-grounding point base position, the changes in eddy current distribution and magnetic flux density distribution inside the core and clamps are simulated when a ground fault occurs at the corresponding position; Thirdly, by comparing and analyzing the changes in eddy current distribution and magnetic flux density distribution inside the core and the clamp, the distribution area of ​​multiple grounding points in the core and the clamp is determined and recorded as the preliminary position group.

5. A method for on-site inspection of multi-point grounding of transformer core and clamps according to claim 4, characterized in that: It also includes setting an eddy current density change rate threshold and a magnetic flux density change rate threshold. Only when the eddy current density or magnetic flux density change rate of a certain position exceeds these thresholds, the corresponding position is classified into the preliminary position group.

6. A method for on-site inspection of multi-point grounding of transformer core and clamps according to claim 5, characterized in that: The eddy current density change rate threshold is set to 10%, and the magnetic flux density change rate threshold is set to 5%.

7. A method for on-site inspection of multi-point grounding of transformer core and clamps according to claim 1, characterized in that: The initial population of the genetic algorithm is the preliminary position group; the objective function of the genetic algorithm is to minimize the sum of the absolute values ​​of the differences between the grounding resistance value measured by the handheld grounding resistance tester and the grounding resistance value calculated by the electromagnetic model of the transformer core and the clamp.

8. The on-site inspection method for multi-point grounding of transformer core and clamps according to claim 1, characterized in that: The step S50 specifically includes: First, an ultrasonic sensor array consisting of multiple ultrasonic sensors is arranged on the surface of the transformer core and the clamp to collect ultrasonic signals; Secondly, during the acquisition process, it is necessary to ensure that the sensor array can fully cover the transformer core and the surface of the clamp to obtain a complete set of ultrasonic signals; Once the acquisition is complete, the original ultrasonic signal needs to be preprocessed by filtering, amplifying, and normalizing to eliminate noise interference, enhance the effective signal, and obtain a preprocessed ultrasonic signal group.

9. The on-site inspection method for multi-point grounding of transformer core and clamps according to claim 1, characterized in that: The step S80 specifically includes: First, based on the analysis results of step S70, a group of small electromagnetic probes, including flux probes and eddy current probes, are arranged at the determined multi-point grounding locations on the transformer core and the surface of the clamp; Secondly, during actual testing, the specific locations of multi-point grounding are further confirmed by analyzing the magnetic flux density and eddy current distribution data collected by the probe.

Images