Transformer winding fault sample generation method and device, equipment and medium

Abstract

The invention relates to the technical field of transformer monitoring, and discloses a transformer winding fault sample generation method and device, equipment and a medium. The method comprises the following steps: establishing a three-dimensional simulation model of the transformer; acquiring a vibration acceleration distribution cloud picture of the three-dimensional simulation model under the combined action of the electricity-magnetism-force-sound multi-physical field; selecting a region with the maximum vibration amplitude from the vibration acceleration distribution cloud picture as a vibration concentration region; obtaining an actual measurement vibration signal and a simulation vibration signal of the transformer vibration concentration area; optimizing the three-dimensional simulation model by taking the maximum similarity between the actually measured vibration signal and the simulation vibration signal as a target; and setting relative displacement between the high-voltage winding and the low-voltage winding of the transformer on the optimized three-dimensional simulation model to simulate different fault states of the transformer winding, and generating vibration data of the transformer in different fault states as fault samples of the transformer winding so as to realize expansion of the fault samples of the transformer winding.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer monitoring, and in particular to a method, device, equipment and medium for generating transformer winding fault samples. Background Art

[0002] As a key hub equipment in the power transmission and transformation link of the power grid, power transformers play multiple roles such as voltage conversion, power transmission and reactive power compensation. Their operating conditions are directly related to the safety and stability of the power system. Therefore, transformers must undergo systematic operation and maintenance during commissioning to ensure that they are continuously in the best operating state. With the continuous advancement and widespread application of smart grid and Internet of Things technologies, the maintenance mode of transformers has gradually transitioned from traditional regular power outage detection to non-stop online monitoring. For high-voltage transformers, the implementation of non-stop online monitoring can capture equipment operating status information in real time and accurately, thereby discovering and handling abnormal operating conditions and early faults of transformers in advance, significantly improving the power supply reliability of the power system.

[0003] At present, common transformer non-stop online monitoring technologies include oil chromatography, infrared temperature measurement, and partial discharge ultrasonic detection. These technologies collect physical parameters such as dissolved gas, temperature distribution, and ultrasonic waves in transformer oil to identify abnormal operating conditions and fault types of transformers. However, these traditional monitoring methods have limitations in quickly and effectively identifying mechanical faults such as loose transformer cores and winding deformation. If these mechanical anomalies cannot be detected and handled in a timely manner, the transformer may be in a poor operating state for a long time, which may induce more serious defects and even cause major power outages.

[0004] When a transformer fails, its operating sound is different. Therefore, the operating status of the transformer can be judged based on the sound of the transformer. Voiceprint monitoring technology uses voiceprint sensors to collect vibration or voiceprint data of normal operation and fault operation of the transformer. Classifying and identifying the transformer operation data based on the collected one-dimensional signal can accurately detect various types of transformer faults, and issue early warnings for faults, thereby improving equipment operation and maintenance efficiency. Specifically, during the operation of the transformer, the vibration of components such as the core and winding will generate mechanical waves. These waves are transmitted to the surface of the box through the oil medium and solid connections, and the mechanical vibration of the transformer is transmitted through the air to produce different sounds. Therefore, based on the mechanical vibration of the transformer casing or the operating sound of the transformer, the diagnosis of transformer mechanical faults can be achieved, among which the acquisition of transformer operation data (i.e., samples) used to achieve fault identification is particularly important.

[0005] However, in actual engineering applications, the probability of transformer winding failure is relatively low compared to normal operation, which makes the acquisition of winding fault samples particularly scarce. The imbalance between the number of fault samples and normal samples, as well as the lack of types and number of fault samples, lead to a decrease in the accuracy of fault vibration and sound signal fault identification. Summary of the invention

[0006] The object of the present invention is to provide a method, device, equipment and medium for generating transformer winding fault samples, which can expand transformer winding fault samples and thus improve the accuracy of transformer winding fault identification.

[0007] To solve the above technical problems, an embodiment of the present invention provides a method for generating a transformer winding fault sample, comprising the following steps:

[0008] According to the structure of the transformer, a three-dimensional simulation model of the transformer is established;

[0009] Obtaining a vibration acceleration distribution cloud map of the three-dimensional simulation model under the joint action of the electric-magnetic-force-acoustic multi-physical fields, wherein the vibration acceleration distribution cloud map is used to indicate the vibration amplitudes of different regions of the three-dimensional simulation model under the joint action of the electric-magnetic-force-acoustic multi-physical fields;

[0010] Selecting the area with the largest vibration amplitude from the vibration acceleration distribution cloud map of the three-dimensional simulation model as the vibration concentration area of ​​the three-dimensional simulation model, and obtaining the simulated vibration signal of the vibration concentration area of ​​the transformer during the simulation operation through the three-dimensional simulation model;

[0011] Obtain the measured vibration signal of the vibration concentration area of ​​the transformer during actual operation;

[0012] With the goal of maximizing the similarity between the measured vibration signal and the simulated vibration signal, the three-dimensional simulation model of the transformer is optimized. By setting the relative displacement between the high and low voltage windings of the transformer on the optimized three-dimensional simulation model, different fault states of the transformer windings are simulated to generate vibration data of the transformer under different fault states as fault samples of the transformer windings.

[0013] In some optional embodiments, obtaining a vibration acceleration distribution cloud map of the three-dimensional simulation model under the joint action of the electric-magnetic-mechanical-acoustic multi-physical fields includes:

[0014] Applying a preset three-phase AC voltage to the high voltage side of the three-dimensional simulation model and a preset current to the low voltage side, respectively, so as to apply a magnetic field to the three-dimensional simulation model;

[0015] Setting an external applied voltage and a load impedance for the three-dimensional simulation model according to the circuit wiring condition of the transformer to apply an electric field to the three-dimensional simulation model;

[0016] Connecting the electric field circuit to the magnetic field coil through an external I vs.U node to couple the magnetic field and electric field applied to the three-dimensional simulation model;

[0017] Applying a force field to the three-dimensional simulation model according to the forces borne by the three-dimensional simulation model in the magnetic field and the electric field to couple the magnetic field, the electric field and the force field applied to the three-dimensional simulation model;

[0018] According to the vibration acceleration of the three-dimensional simulation model under the action of the magnetic field, electric field and force field, an acoustic field is applied to the three-dimensional simulation model to couple the magnetic field, electric field, force field and acoustic field applied to the three-dimensional simulation model;

[0019] Obtain the vibration acceleration distribution cloud map of the 3D simulation model under the combined action of coupled magnetic field, electric field, force field and acoustic field.

[0020] In some optional embodiments, before obtaining the vibration acceleration distribution cloud map of the three-dimensional simulation model under the combined action of the coupled magnetic field, electric field, force field and acoustic field, the method further includes:

[0021] The bottom surface of the transformer winding in the three-dimensional simulation model is divided using free triangle mesh;

[0022] A five-layer swept mesh is used to divide the coils of the transformer winding in the three-dimensional simulation model;

[0023] The transformer core and the remaining parts except the transformer winding and the transformer core in the three-dimensional simulation model are divided by free tetrahedral mesh.

[0024] In some optional embodiments, selecting the area with the largest vibration amplitude from the vibration acceleration distribution cloud map of the three-dimensional simulation model as the vibration concentration area of ​​the three-dimensional simulation model includes:

[0025] Divide the front area of ​​the transformer oil tank into a number of measurement point units to be selected;

[0026] Apply the same vibration acceleration to each of the candidate measuring point units to obtain the vibration acceleration waveform of each of the candidate measuring point units within one cycle;

[0027] The vibration acceleration waveforms of the various candidate measuring point units are compared to select the area with the largest vibration amplitude as the vibration concentration area.

[0028] In some optional embodiments, the measured vibration signal is a vibration signal of the transformer under different excitation voltages and different load impedances, and the simulated vibration signal is a vibration signal of the three-dimensional simulation model under different excitation voltages and different load impedances;

[0029] The three-dimensional simulation model of the transformer is optimized with the goal of maximizing the similarity between the measured vibration signal and the simulated vibration signal, including:

[0030] Compare the measured vibration signal and the simulated vibration signal under the same excitation voltage and the same load impedance;

[0031] Obtain the excitation voltage and load impedance used by the measured vibration signal and the simulated vibration signal when their similarity is the greatest;

[0032] The three-dimensional simulation model is optimized by adjusting the excitation voltage and load impedance of the three-dimensional simulation model.

[0033] In some optional embodiments, when the similarity between the measured vibration signal and the simulated vibration signal is the greatest, the excitation voltage and load impedance used by the two signals include:

[0034] Dynamic time warping (DTW) distance is used to obtain the similarity between the measured vibration signal and the simulated vibration signal in the time domain, and spectral coherence is used to obtain the similarity between the measured vibration signal and the simulated vibration signal in the frequency domain.

[0035] The excitation voltage and load impedance used by the measured vibration signal and the simulated vibration signal are obtained when their similarities in the time domain and frequency domain are the largest.

[0036] In some optional embodiments, the core of the transformer adopts a three-phase three-column structure and is composed of stacked silicon steel sheets; the winding of the transformer is a solid hollow cylinder with multiple turns of coils evenly distributed.

[0037] An embodiment of the present invention further provides a device for generating a transformer winding fault sample, comprising:

[0038] A model building module is used to build a three-dimensional simulation model of the transformer according to the structure of the transformer;

[0039] A vibration analysis module, used to obtain a vibration acceleration distribution cloud map of a three-dimensional simulation model under the joint action of an electric-magnetic-mechanical-acoustic multi-physical field, wherein the vibration acceleration distribution cloud map is used to indicate the vibration amplitude of different regions of the three-dimensional simulation model under the joint action of an electric-magnetic-mechanical-acoustic multi-physical field;

[0040] The signal acquisition module is used to select the area with the largest vibration amplitude from the vibration acceleration distribution cloud map of the three-dimensional simulation model as the vibration concentration area of ​​the three-dimensional simulation model, and obtain the simulated vibration signal of the vibration concentration area of ​​the transformer during the simulation operation through the three-dimensional simulation model; and obtain the measured vibration signal of the vibration concentration area of ​​the transformer during the actual operation;

[0041] The sample generation module is used to optimize the three-dimensional simulation model of the transformer with the goal of maximizing the similarity between the measured vibration signal and the simulated vibration signal, and to simulate different fault states of the transformer winding by setting the relative displacement between the high and low voltage windings of the transformer on the optimized three-dimensional simulation model, so as to generate vibration data of the transformer under different fault states as fault samples of the transformer winding.

[0042] An embodiment of the present invention also provides a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned method for generating transformer winding fault samples.

[0043] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program, wherein the computer program implements the above-mentioned method for generating transformer winding fault samples when executed by a processor.

[0044] The method for generating transformer winding fault samples provided by the present invention has at least the following beneficial effects:

[0045] By establishing a three-dimensional simulation model of the transformer, setting the relative displacement between the high and low voltage windings of the transformer on the three-dimensional simulation model, different fault states of the transformer winding are simulated, and the vibration data of the transformer under different fault states are generated, which are used as fault samples of the transformer winding, so as to expand the fault samples of the transformer winding. Among them, the three-dimensional simulation model is obtained by optimizing the difference between the measured signal and the simulation signal in the transformer vibration concentration area after determining the transformer vibration concentration area, taking into account the actual vibration characteristics of the transformer, and being closer to the actual working conditions of the transformer, so that the simulated transformer winding fault state is more real and accurate, ensuring the accuracy of the generated transformer winding fault samples, thereby solving the problem of insufficient transformer winding fault samples and providing an accurate data source for subsequent transformer vibration and voiceprint recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] One or more embodiments are exemplarily described by the pictures in the corresponding drawings, and these exemplary descriptions do not constitute limitations on the embodiments.

[0047] Figure 1 is a flow chart of a method for generating a transformer winding fault sample according to an embodiment of the present invention;

[0048] FIG2 is a schematic diagram of a three-dimensional simulation model of a transformer provided according to an embodiment of the present invention; wherein (a) is a front view of the three-dimensional simulation model, and (b) is a side view of the three-dimensional simulation model;

[0049] Figure 3 is a schematic diagram of mesh division of a three-dimensional simulation model provided according to an embodiment of the present invention; wherein (a) is a schematic diagram of mesh division of a transformer winding, (b) is a schematic diagram of mesh division of a transformer core, and (c) is a schematic diagram of mesh division of the transformer as a whole. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. However, it will be appreciated by those skilled in the art that in the embodiments of the present invention, many technical details are proposed in order to enable the reader to better understand the present invention. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed in the present invention can be implemented. The division of the following embodiments is for the convenience of description and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined and referenced with each other without contradiction.

[0051] An embodiment of the present invention relates to a method for generating a transformer winding fault sample. The implementation details of the method for generating a transformer winding fault sample of this embodiment are specifically described below. The following content is only the implementation details provided for the convenience of understanding and is not necessary for implementing this solution.

[0052] The specific process of the method for generating transformer winding fault samples in this embodiment can be as follows: Figure 1 As shown, including:

[0053] Step 101: Establish a three-dimensional simulation model of the transformer according to the structure of the transformer.

[0054] Specifically, the iron core of the transformer in this embodiment adopts a three-phase three-column structure, which is composed of stacked silicon steel sheets. The winding of the transformer is a solid hollow cylinder with evenly distributed multi-turn coils, and the winding and the iron core are fixed with clamps. According to the structure of the above transformer, a finite element simulation model of the transformer is established using COMSOL software. As a three-dimensional simulation model, the model schematic diagram can be shown in Figure 2, where (a) is a front view of the three-dimensional simulation model, and (b) is a side view of the three-dimensional simulation model.

[0055] In the specific implementation, in order to improve the simulation efficiency without affecting the simulation results, this embodiment reasonably simplifies the transformer model and ignores some insignificant parameters and structures in the transformer, as described below: (1) Ignore the influence of the stacking of silicon steel sheets and the joint form of the transformer core, and regard the core as a whole; (2) The high and low voltage windings of the transformer are equivalent to hollow cylinders, and it is assumed that the high and low voltage windings are uniform multi-turn layered winding structures; (3) Ignore other structures such as the oil tank fan, and the oil tank is equivalent to a rectangular parallelepiped.

[0056] Step 102, obtaining a vibration acceleration distribution cloud map of the three-dimensional simulation model under the joint action of the electric-magnetic-mechanical-acoustic multi-physical fields, wherein the vibration acceleration distribution cloud map is used to indicate the vibration amplitudes of different regions of the three-dimensional simulation model under the joint action of the electric-magnetic-mechanical-acoustic multi-physical fields.

[0057] Specifically, after the three-dimensional simulation model of the transformer is established, the electric-magnetic-mechanical-acoustic multi-physical fields are added to the three-dimensional simulation model, and a transient solver is used to obtain the vibration acceleration distribution cloud map of the three-dimensional simulation model under the joint action of the electric-magnetic-mechanical-acoustic multi-physical fields.

[0058] Among them, adding electric-magnetic-force-acoustic multi-physical fields to the three-dimensional simulation model includes: applying a preset three-phase AC voltage on the high-voltage side of the three-dimensional simulation model and a preset current on the low-voltage side to apply a magnetic field to the three-dimensional simulation model; setting an external voltage and load impedance for the three-dimensional simulation model according to the circuit wiring of the transformer to apply an electric field to the three-dimensional simulation model; connecting the electric field circuit with the magnetic field coil through an external I vs.U node in the electric field to couple the magnetic field and electric field applied to the three-dimensional simulation model; applying a force field to the three-dimensional simulation model according to the force borne by the three-dimensional simulation model in the magnetic field and electric field to couple the magnetic field, electric field and force field applied to the three-dimensional simulation model; applying an acoustic field to the three-dimensional simulation model according to the vibration acceleration of the three-dimensional simulation model under the action of the magnetic field, electric field and force field to couple the magnetic field, electric field, force field and acoustic field applied to the three-dimensional simulation model.

[0059] In the specific implementation, first complete the magnetic field excitation setting for the three-dimensional simulation model. When the transformer is running under load conditions, the vibration of its winding is mainly affected by the rated current. The rated three-phase AC voltage is applied to the high-voltage side of the transformer, and the low-voltage side winding current is set to the rated current. The vibration of the transformer core can be ignored. Then, complete the field-circuit coupling of the circuit and the magnetic field for the three-dimensional simulation model. Set the external voltage and load impedance according to the actual transformer circuit wiring diagram, and connect the circuit to the magnetic field coil through the external I vs.U node in the circuit physics field to complete the field-circuit coupling. At this time, the coil boundary condition needs to be added to the magnetic field to connect with the external I vs.U node in the circuit physics field. Next, set the transformer winding boundary condition, add the body load boundary condition, select the Lorentz force as the load type, obtain the force condition of the coil, and complete the coupling of the electromagnetic field and solid mechanics through the magnetomechanics coupling interface in the multi-physics field node. Finally, the winding deformation is used as the vibration source of the sound field, and the transformer oil is used as the propagation medium. Based on the simulation results of the above windings, the pressure acoustics physics field is added, and the vibration acceleration of the winding is used as the sound source boundary condition. The coupling between solid mechanics and the acoustic field is completed through the acoustic-structure boundary coupling interface in the multi-physics field node.

[0060] Therefore, the above process couples the electric field and magnetic field together through field-path coupling, couples the magnetic field and solid mechanics together through the magnetomechanical coupling interface, and couples solid mechanics together through the acoustic-structure boundary coupling interface. Combining each step of coupling, all physical fields are coupled together, and finally the coupling of the multi-physical fields of electric-magnetic-force-acoustic is completed. Based on this, the vibration acceleration distribution cloud map of the three-dimensional simulation model under the combined action of the coupled magnetic field, electric field, force field and acoustic field can be obtained.

[0061] It should be noted that the simulation process is a simulation of the transformer in the time domain. The simulation steps in the frequency domain before acoustics are the same as those in the time domain. After the structural force field simulation is completed, a research step needs to be added to convert the time domain data into frequency domain data through Fast Fourier Transform (FFT).

[0062] In one example, before obtaining the vibration acceleration distribution cloud map of the three-dimensional simulation model under the combined action of the coupled magnetic field, electric field, force field and acoustic field, the three-dimensional simulation model of the transformer is first meshed using COMSOL software. The model division schematic diagram can be shown in Figure 3, where (a) is a schematic diagram of the mesh division of the transformer winding, (b) is a schematic diagram of the mesh division of the transformer core, and (c) is a schematic diagram of the mesh division of the transformer as a whole. In order to improve the calculation efficiency and reduce the calculation time, the bottom surface of the transformer winding in the three-dimensional simulation model is divided by a free triangle mesh, the coil of the transformer winding in the three-dimensional simulation model is divided by a five-layer swept mesh, and the transformer core in the three-dimensional simulation model and the remaining part except the transformer winding and the transformer core are divided by a free tetrahedral mesh, wherein the free tetrahedral mesh of the transformer core part is relatively refined, and the free tetrahedral mesh of the remaining part except the transformer winding and the transformer core is divided in a conventional manner. It can be understood that since acoustics requires the use of external field calculation boundary conditions, another layer of boundary layer mesh needs to be added outside the oil tank for external field calculation.

[0063] It should be noted that, in this embodiment, after adding the electric-magnetic-force-acoustic multi-physics field to the three-dimensional simulation model, it is also necessary to set the material properties of the transformer according to the applied electric-magnetic-force-acoustic multi-physics field, specifically including:

[0064] Add materials such as soft iron, copper and transformer oil to the built-in material library of COMSOL. The specific properties of the electrical conductivity, Young's modulus, Poisson's ratio, saturation magnetization, saturation magnetostriction coefficient and other specific properties required for each structural material of the transformer used in this example are shown in the following table. Table 1 describes the properties of the transformer core material, Table 2 describes the specific properties of the electrical conductivity, density, Young's modulus, Poisson's ratio and other specific properties required for the transformer winding material, and Table 3 describes the speed of sound propagation:

[0065] Table 1

[0066] Material Conductivity Young's modulus Poisson's ratio Saturation magnetization Saturation magnetostriction coefficient Iron Core 10[S / m] 60e9[Pa] 0.45 1.5e6*1e4[A / m] 200[ppm]

[0067] Table 2

[0068] Material Conductivity density Young's modulus Poisson's ratio Winding 5.998e7[S / m] 8960[kg / m^3] 110[GPa] 0.35

[0069] Table 3

[0070] Material Transformer oil Speed ​​of sound 1324(m / s)

[0071] When obtaining the vibration acceleration distribution cloud map of the three-dimensional simulation model under the combined action of the coupled magnetic field, electric field, force field and acoustic field, a transient study can be used. The transient study requires setting the study time and study step in the transient solver, and setting the full coupling solution method in the transient solver. The full coupling method will form a large set of equations to solve all unknown quantities (i.e. "fields") and include all couplings between unknown quantities (multi-physics field effects) at once in a single iteration.

[0072] Step 103, selecting an area with the largest vibration amplitude from the vibration acceleration distribution cloud map of the three-dimensional simulation model as the vibration concentration area of ​​the three-dimensional simulation model, and obtaining a simulated vibration signal of the vibration concentration area of ​​the transformer during the simulation operation through the three-dimensional simulation model.

[0073] Specifically, according to the vibration amplitude of different areas of the three-dimensional simulation model under the joint action of the electric-magnetic-mechanical-acoustic multi-physical fields indicated by the vibration acceleration distribution cloud map of the three-dimensional simulation model, it is found that the vibration of the transformer is mainly concentrated in the front and top areas of the transformer oil tank. Considering the actual structural characteristics of the transformer, the top area is located at a higher position and is equipped with components such as bushings and oil pillows, which is not convenient for actual installation and operation and poses certain safety risks. The back area is usually equipped with a cooling fan with loud noise, which interferes with the accuracy of the vibration signal. Therefore, the front of the transformer oil tank is selected as the vibration measurement point area.

[0074] At this time, the front area of ​​the transformer oil tank is divided into several candidate measuring point units, where each candidate measuring point unit has a specification of 0.3m*0.3m and a rib width of 0.07m. The same vibration acceleration is applied to each candidate measuring point unit to obtain the vibration acceleration waveform of each candidate measuring point unit within a cycle. Since the simulation data is a noise-free signal, the larger the vibration amplitude, the less the attenuation of the internal vibration of the transformer when it is transmitted to the transformer oil tank. The vibration acceleration waveforms of each candidate measuring point unit are compared to select the area with the largest vibration amplitude as the vibration concentration area. Then, the simulated vibration signal of the vibration concentration area of ​​the transformer during the simulation operation is obtained through the three-dimensional simulation model.

[0075] Step 104 , obtaining a measured vibration signal of a vibration concentration area of ​​the transformer during actual operation, and obtaining a simulated vibration signal of the vibration concentration area of ​​the transformer during simulated operation through a three-dimensional simulation model.

[0076] Specifically, a vibration signal sensor is placed in the vibration concentration area of ​​the transformer determined above to collect the actual vibration signal of the vibration concentration area of ​​the transformer during actual operation.

[0077] In a specific implementation, this embodiment actually obtains the above-mentioned measured vibration signal and simulated vibration signal under the parameter conditions of different excitation voltages and different load impedances, that is, the measured vibration signal of the transformer under different excitation voltages and different load impedances is obtained, and the simulated vibration signal of the three-dimensional simulation model under different excitation voltages and different load impedances is obtained.

[0078] Step 105, with the goal of maximizing the similarity between the measured vibration signal and the simulated vibration signal, the three-dimensional simulation model of the transformer is optimized, and by setting the relative displacement between the high and low voltage windings of the transformer on the optimized three-dimensional simulation model, different fault states of the transformer winding are simulated to generate vibration data of the transformer under different fault states as fault samples of the transformer winding.

[0079] Specifically, the vibration period, vibration amplitude and signal similarity measurement value of the measured vibration signal and the simulated vibration signal are used as standards, and the measured vibration signal and the simulated vibration signal are compared under the same excitation voltage and the same load impedance. Then, the excitation voltage and load impedance used by the measured vibration signal and the simulated vibration signal when the similarity is the largest are obtained, which are the model parameters closest to the actual working conditions. The excitation voltage and load impedance of the three-dimensional simulation model are adjusted so that the excitation voltage and load impedance of the three-dimensional simulation model are the excitation voltage and load impedance used by the measured vibration signal and the simulated vibration signal when the similarity is the largest, so as to achieve the optimization of the three-dimensional simulation model. At this time, the three-dimensional simulation model is the simulation model closest to the actual operating state of the transformer.

[0080] In a specific implementation, in order to describe the accuracy of the transformer vibration signal (i.e., the simulated vibration signal) obtained by simulation, the measured vibration signal and the simulated vibration signal can be used to perform similarity analysis in the time domain and frequency domain, respectively. For example, the dynamic time warping (DTW) distance and spectrum coherence are used to measure the similarity between the simulated signal and the measured signal in the time domain and the frequency domain, respectively, so as to evaluate the simulation accuracy of the vibration waveform. That is, the dynamic time warping DTW distance is used to obtain the similarity between the measured vibration signal and the simulated vibration signal in the time domain, and the spectrum coherence is used to obtain the similarity between the measured vibration signal and the simulated vibration signal in the frequency domain, and then the excitation voltage and load impedance used by the measured vibration signal and the simulated vibration signal are obtained when the similarity between the measured vibration signal and the simulated vibration signal in the time domain and the frequency domain is the maximum.

[0081] Among them, the DTW distance metric can normalize the time axis and effectively solve the problem of excessive errors caused by inconsistent phases of time series. The main goal of the DTW algorithm is to find an optimal normalized path, add the distances of all similar points as the normalized path distance, and use the normalized path distance to measure the similarity between two time series. The smaller the normalized path distance, the higher the similarity between the sequences.

[0082] Assume that the length of the vibration signal is k, and the measured vibration signal is represented by X = {x1, x2, ... x i ,…,x k}, the vibration signal obtained by simulation is represented by Y={y1,y2,…y i ,…,y k} indicates that the optimal regularization path determines the corresponding relationship between the data points of the two vibration signals. In the distance function dis(x i ,y j ) is determined, the goal of DTW is to minimize the sum of the distance values ​​of all corresponding data points in the two vibration data, and use it as the metric distance function dis(x) between the sequences. i ,y j ) and DTW are defined as follows:

[0083] DTW(X,Y)=∑dis(x i ,y j )

[0084] dis(x i ,y j )=||x i -y j || w

[0085] In the formula, w is usually taken as 2, dis is the Euclidean distance;

[0086] The DTW distance is calculated using the idea of ​​dynamic programming on the optimal regular path. The calculation formula is as follows:

[0087] D(i,j)=dis(x i ,x j )+min[D(i-1,j),D(i,j-1),D(i-1,j-1)]

[0088] Where D(i,j) reflects the distance difference between two time series of length i and j on the normalization path. Spectral coherence is a method used to describe the similarity of two vibration waveforms in the frequency domain. It can be obtained by calculating the cross power spectrum density and auto power spectrum density of the two waveforms. The cross power spectrum density describes the degree of interaction between the two signals at different frequencies, and the auto power spectrum density describes the power distribution of each signal at different frequencies. Assuming there are two signals x(t) and y(t), their spectral coherence can be expressed as γ xy (f), which is calculated as:

[0089]

[0090] In the formula, |S xy (f)| represents the modulus of the cross power spectral density of signals x(t) and y(t) at frequency f, S xx (f) and S yy (f) represents the auto-power spectral density of the two signals at frequency f. The cross power spectral density and auto-power spectral density can be calculated using Fourier transform, and the calculation formula is as follows:

[0091]

[0092] In the formula, represents the Fourier transform, Represents the conjugate complex numbers of y(t) and x(t), respectively. The value of spectral coherence usually ranges from 0 to 1. When there is a complete linear correlation between the two signals, γ xy (f) is equal to 1; when there is no correlation between them, γ xy When (f) is equal to 0, when there is high spectral coherence between two vibration signals, it means that they have similar vibration behaviors at the same frequency, which may mean that they come from the same equipment or the same working conditions.

[0093] By setting the relative displacement between the high and low voltage windings of the transformer on the optimized three-dimensional simulation model, different fault states of the transformer winding are simulated. At the same time, the relative displacement between the high and low voltage windings and the overall height of the transformer winding are used to simulate the fault degree of the transformer winding at 5%, 15% and 20%. The simulation waveform of the transformer vibration signal under different fault states is obtained by using COMSOL software. By comparing with the actual normal operating state data of the transformer, it can be used to analyze the simulated vibration characteristics of the transformer under different fault states. The fault state of the transformer is simulated by simulating the fault state of the simulated winding through the optimized three-dimensional simulation model of the transformer, and the vibration signal under different fault operating conditions of the transformer (i.e., the fault sample of the transformer winding) is obtained. Finally, it is fused with the measured normal operating condition data of the transformer to fill the missing fault signal in the transformer vibration signal database, and construct a virtual-real fusion vibration signal database.

[0094] The simulated vibration characteristics of the transformer under the above-mentioned different fault conditions include: in the time domain, compared with the normal operating state of the transformer, the axial displacement fault of the winding will not change the vibration period of the transformer, but the amplitude of the vibration waveform will change, and with the increase of the axial displacement degree, the amplitude of the transformer vibration acceleration will increase slightly. In the frequency domain, compared with the normal operation state of the transformer, the axial displacement fault of the winding significantly affects the distribution of the vibration spectrum, the proportion of high-frequency components will increase, and with the increase of the axial displacement degree of the transformer winding, the main frequency of the vibration signal will also change. When the axial displacement degree of the winding is 5%, the main frequency of the vibration signal is 100Hz, but the amplitude at the main frequency becomes larger, and a high-amplitude frequency component appears between 400Hz-800Hz. When the axial displacement degree of the winding is 15%, the main frequency of the vibration signal becomes 200Hz, and obvious frequency components are also observed at frequency points such as 100Hz, 300Hz, 500Hz, 600Hz and 700Hz. High-frequency components above 1000Hz begin to appear, but the amplitude is relatively small. When the axial displacement degree of the winding is 20%, the main frequency remains at 200Hz, but in the frequency bands of 500Hz, 600Hz, 700Hz and 800Hz, the frequency components are more prominent, and the high-frequency components above 1000Hz become more obvious, and the amplitude also increases.

[0095] In this embodiment, by establishing a three-dimensional simulation model of the transformer, setting the relative displacement between the high and low voltage windings of the transformer on the three-dimensional simulation model, different fault states of the transformer winding are simulated, and vibration data of the transformer under different fault states are generated, which are used as fault samples of the transformer winding, so as to expand the transformer winding fault samples. Among them, the three-dimensional simulation model is obtained by optimizing the difference between the measured signal and the simulation signal in the transformer vibration concentration area after determining the transformer vibration concentration area, taking into account the actual vibration characteristics of the transformer, and being closer to the actual working conditions of the transformer, so that the simulated transformer winding fault state is more real and accurate, ensuring the accuracy of the generated transformer winding fault samples, thereby solving the problem of insufficient transformer winding fault samples, and providing an accurate data source for subsequent transformer vibration and voiceprint recognition.

[0096] The step division of the various methods above is only for clear description. When implemented, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the protection scope of the present invention. Adding insignificant modifications or introducing insignificant designs to the algorithm or process without changing the core design of the algorithm and process are all within the protection scope of the invention.

[0097] Another embodiment of the present invention relates to a device for generating transformer winding fault samples. The implementation details of the device for generating transformer winding fault samples of this embodiment are described in detail below. The following content is only the implementation details provided for the convenience of understanding and is not necessary for the implementation of this solution. The device for generating transformer winding fault samples of this embodiment specifically includes: a model building module, a vibration analysis module, a model optimization module and a sample generation module.

[0098] Specifically, the model building module is used to build a three-dimensional simulation model of the transformer according to the structure of the transformer;

[0099] A vibration analysis module, used to obtain a vibration acceleration distribution cloud map of a three-dimensional simulation model under the joint action of an electric-magnetic-mechanical-acoustic multi-physical field, wherein the vibration acceleration distribution cloud map is used to indicate the vibration amplitude of different regions of the three-dimensional simulation model under the joint action of an electric-magnetic-mechanical-acoustic multi-physical field;

[0100] The signal acquisition module is used to select the area with the largest vibration amplitude from the vibration acceleration distribution cloud map of the three-dimensional simulation model as the vibration concentration area of ​​the three-dimensional simulation model, and obtain the simulated vibration signal of the vibration concentration area of ​​the transformer during the simulation operation through the three-dimensional simulation model; and obtain the measured vibration signal of the vibration concentration area of ​​the transformer during the actual operation;

[0101] The sample generation module is used to optimize the three-dimensional simulation model of the transformer with the goal of maximizing the similarity between the measured vibration signal and the simulated vibration signal, and to simulate different fault states of the transformer winding by setting the relative displacement between the high and low voltage windings of the transformer on the optimized three-dimensional simulation model, so as to generate vibration data of the transformer under different fault states as fault samples of the transformer winding.

[0102] It is not difficult to find that this embodiment is a device embodiment corresponding to the above method embodiment, and this embodiment can be implemented in conjunction with the above method embodiment. The relevant technical details and technical effects mentioned in the above embodiment are still valid in this embodiment, and in order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the above embodiment.

[0103] It is worth mentioning that all modules involved in this embodiment are logic modules. In practical applications, a logic unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, in order to highlight the innovative part of the present invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by the present invention, but this does not mean that there are no other units in this embodiment.

[0104] Another embodiment of the present invention relates to a computer device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for generating transformer winding fault samples in the above-mentioned embodiments.

[0105] Among them, the memory and the processor are connected in a bus manner, and the bus may include any number of interconnected buses and bridges, and the bus connects various circuits of one or more processors and memories together. The bus can also connect various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on a transmission medium. The data processed by the processor is transmitted on a wireless medium via an antenna, and further, the antenna also receives data and transmits the data to the processor.

[0106] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0107] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program, which implements the above method embodiment when executed by a processor.

[0108] That is, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including a number of instructions for a device (which can be a single-chip microcomputer, chip, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as: ROM), random access memory (Random Access Memory, referred to as: RAM), disk or optical disk and other media that can store program codes.

[0109] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present invention, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for generating transformer winding fault samples, characterized in that: include: According to the structure of the transformer, a three-dimensional simulation model of the transformer is established; Obtaining a vibration acceleration distribution cloud map of the three-dimensional simulation model under the joint action of the electric-magnetic-force-acoustic multi-physical fields, wherein the vibration acceleration distribution cloud map is used to indicate the vibration amplitudes of different regions of the three-dimensional simulation model under the joint action of the electric-magnetic-force-acoustic multi-physical fields; Selecting the area with the largest vibration amplitude from the vibration acceleration distribution cloud map of the three-dimensional simulation model as the vibration concentration area of ​​the three-dimensional simulation model, and obtaining the simulated vibration signal of the vibration concentration area of ​​the transformer during the simulation operation through the three-dimensional simulation model; Obtain the measured vibration signal of the vibration concentration area of ​​the transformer during actual operation; With the goal of maximizing the similarity between the measured vibration signal and the simulated vibration signal, the three-dimensional simulation model of the transformer is optimized. By setting the relative displacement between the high and low voltage windings of the transformer on the optimized three-dimensional simulation model, different fault states of the transformer windings are simulated to generate vibration data of the transformer under different fault states as fault samples of the transformer windings.

2. The method for generating transformer winding fault samples according to claim 1, characterized in that: The step of obtaining a vibration acceleration distribution cloud map of the three-dimensional simulation model under the joint action of the electric-magnetic-mechanical-acoustic multi-physical fields comprises: Applying a preset three-phase AC voltage to the high-voltage side of the three-dimensional simulation model and a preset current to the low-voltage side to apply a magnetic field to the three-dimensional simulation model; Setting an external applied voltage and a load impedance for the three-dimensional simulation model according to the circuit wiring condition of the transformer to apply an electric field to the three-dimensional simulation model; Connect the electric field circuit to the magnetic field coil through the external Ivs.U node to couple the magnetic field and electric field applied to the three-dimensional simulation model; Applying a force field to the three-dimensional simulation model according to the forces borne by the three-dimensional simulation model in the magnetic field and the electric field to couple the magnetic field, the electric field and the force field applied to the three-dimensional simulation model; According to the vibration acceleration of the three-dimensional simulation model under the action of the magnetic field, electric field and force field, an acoustic field is applied to the three-dimensional simulation model to couple the magnetic field, electric field, force field and acoustic field applied to the three-dimensional simulation model; Obtain the vibration acceleration distribution cloud map of the 3D simulation model under the combined action of coupled magnetic field, electric field, force field and acoustic field.

3. The method for generating transformer winding fault samples according to claim 2, characterized in that: Before obtaining the vibration acceleration distribution cloud map of the three-dimensional simulation model under the combined action of the coupled magnetic field, electric field, force field and acoustic field, the method further includes: The bottom surface of the transformer winding in the three-dimensional simulation model is divided using free triangle mesh; A five-layer swept mesh is used to divide the coils of the transformer winding in the three-dimensional simulation model; The transformer core and the remaining parts except the transformer winding and the transformer core in the three-dimensional simulation model are divided by free tetrahedral mesh.

4. The method for generating transformer winding fault samples according to claim 1, characterized in that: The step of selecting an area with the largest vibration amplitude from the vibration acceleration distribution cloud map of the three-dimensional simulation model as the vibration concentration area of ​​the three-dimensional simulation model includes: Divide the front area of ​​the transformer oil tank into a number of measurement point units to be selected; Apply the same vibration acceleration to each of the candidate measuring point units to obtain the vibration acceleration waveform of each of the candidate measuring point units within one cycle; The vibration acceleration waveforms of the various candidate measuring point units are compared to select the area with the largest vibration amplitude as the vibration concentration area.

5. The method for generating transformer winding fault samples according to claim 1, characterized in that: The measured vibration signal is a vibration signal of the transformer under different excitation voltages and different load impedances, and the simulated vibration signal is a vibration signal of the three-dimensional simulation model under different excitation voltages and different load impedances; The three-dimensional simulation model of the transformer is optimized with the goal of maximizing the similarity between the measured vibration signal and the simulated vibration signal, including: Compare the measured vibration signal and the simulated vibration signal under the same excitation voltage and the same load impedance; Obtain the excitation voltage and load impedance used by the measured vibration signal and the simulated vibration signal when their similarity is the greatest; The three-dimensional simulation model is optimized by adjusting the excitation voltage and load impedance of the three-dimensional simulation model.

6. The method for generating transformer winding fault samples according to claim 5, characterized in that: The excitation voltage and load impedance used by the measured vibration signal and the simulated vibration signal when the similarity between the measured vibration signal and the simulated vibration signal is the greatest include: Dynamic time warping (DTW) distance is used to obtain the similarity between the measured vibration signal and the simulated vibration signal in the time domain, and spectral coherence is used to obtain the similarity between the measured vibration signal and the simulated vibration signal in the frequency domain. The excitation voltage and load impedance used by the measured vibration signal and the simulated vibration signal are obtained when their similarities in the time domain and frequency domain are the largest.

7. The method for generating transformer winding fault samples according to claim 1, characterized in that: The iron core of the transformer adopts a three-phase three-column structure and is composed of stacked silicon steel sheets; the winding of the transformer is a solid hollow cylinder with evenly distributed multi-turn coils.

8. A device for generating transformer winding fault samples, characterized in that: include: A model building module is used to build a three-dimensional simulation model of the transformer according to the structure of the transformer; A vibration analysis module, used to obtain a vibration acceleration distribution cloud map of a three-dimensional simulation model under the joint action of an electric-magnetic-mechanical-acoustic multi-physical field, wherein the vibration acceleration distribution cloud map is used to indicate the vibration amplitude of different regions of the three-dimensional simulation model under the joint action of an electric-magnetic-mechanical-acoustic multi-physical field; The signal acquisition module is used to select the area with the largest vibration amplitude from the vibration acceleration distribution cloud map of the three-dimensional simulation model as the vibration concentration area of ​​the three-dimensional simulation model, and obtain the simulated vibration signal of the vibration concentration area of ​​the transformer during the simulation operation through the three-dimensional simulation model; and obtain the measured vibration signal of the vibration concentration area of ​​the transformer during the actual operation; The sample generation module is used to optimize the three-dimensional simulation model of the transformer with the goal of maximizing the similarity between the measured vibration signal and the simulated vibration signal, and to simulate different fault states of the transformer winding by setting the relative displacement between the high and low voltage windings of the transformer on the optimized three-dimensional simulation model, so as to generate vibration data of the transformer under different fault states as fault samples of the transformer winding.

9. A computer device, characterized in that: include: at least one processor; And, a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for generating transformer winding fault samples as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for generating transformer winding fault samples according to any one of claims 1 to 7 is implemented.

Images