Oil-immersed iron core reactor fault diagnosis method based on multi-physics field coupling and characteristic spectrum

Abstract

The invention discloses an oil-immersed iron core reactor fault diagnosis method based on multi-physics field coupling and a characteristic spectrum, and the method comprises the steps: constructing a multi-physics field coupling calculation model of an iron core reactor, and obtaining simulation results of an electromagnetic-structural force field and a sound field of the iron core reactor; establishing an iron core reactor simulation model under typical defects, and obtaining simulation results of an electromagnetic-structural force field and a sound field of the iron core reactor under different working conditions; analyzing the influence rule of different defects on the vibration and noise of the iron core reactor; according to a vibration and noise distribution rule, obtaining optimal vibration and noise measuring point distribution, and extracting vibration and noise signals of the iron core reactor under different working conditions; the extracted vibration and noise signals under different working conditions are converted into a Gramer angle and field characteristic spectrum; and inputting the Grubm angle and the field characteristic spectrum into AlexNet to carry out transfer learning so as to realize fault diagnosis. According to the oil-immersed iron core reactor fault diagnosis method, the fault of the iron core reactor can be identified more comprehensively and accurately, and the operation safety and reliability of the iron core reactor are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of core reactor fault diagnosis, and particularly to a fault diagnosis method for oil-immersed core reactors based on multi-physical field coupling and characteristic spectra. Background Art

[0002] As an important device in the power transmission and transformation system, the safe and reliable operation of the core reactor is crucial, and fault diagnosis of the core reactor should be carried out in a timely manner. Traditional fault diagnosis methods for oil-immersed core reactors include oil chromatography analysis, winding DC resistance detection, and partial discharge detection, etc. These methods usually rely on regular inspections and tests and cannot monitor the state of the winding in real time. For example, the oil chromatography analysis method infers whether a fault has occurred inside the core reactor by analyzing the gases dissolved in the oil of the core reactor. Although this method can reflect the fault conditions inside the core reactor, especially faults such as high temperature, discharge, and partial breakdown, and is easy to operate. However, this method can only reflect the indirect symptoms of the fault, cannot accurately locate the specific location of the winding fault, and is insensitive to small-scale and local winding insulation damage; the winding DC resistance detection method judges the health status of the winding by measuring the DC resistance of each winding of the core reactor, can only detect static electrical faults, cannot detect dynamic faults, and the change of the resistance may be slow, making it difficult to detect in the initial stage of the fault. Therefore, only relatively serious faults can be found, and slight damage or early insulation problems of the winding cannot be accurately diagnosed. The partial discharge detection method judges the insulation state of the winding by detecting partial discharge signals (such as ultrasonic waves, electrical pulses, etc.). For tiny and local discharge phenomena, this method may require high-precision instrument equipment, increasing the equipment and maintenance costs, and the discharge signal is easily interfered by external noise, which may lead to misdiagnosis or missed diagnosis. For relatively serious faults, the discharge may have occurred, making it difficult to detect in time.

[0003] Therefore, traditional core reactor fault diagnosis methods often cannot monitor the state of the core reactor in real time, comprehensively, and accurately, are difficult to detect tiny or complex faults at an early stage, and some methods can only detect a certain type of fault and need to be combined with other means for comprehensive diagnosis. Summary of the Invention

[0004] To solve the above technical problems, the present invention proposes a fault diagnosis method for oil-immersed core reactors based on multi-physical field coupling and characteristic spectra. Through technical means such as multi-physical field coupling modeling, feature extraction, data analysis, and fault prediction, the faults of the core reactor can be identified more comprehensively and accurately, improving its operation safety and reliability.

[0005] The technical solution adopted by the present invention is as follows:

[0006] A fault diagnosis method for oil-immersed core reactors based on multi-physical field coupling and characteristic spectra, comprising the following steps:

[0007] Step 1: Construct a multi-physics field coupling calculation model for the iron core reactor to obtain the simulation results of the electromagnetic-structural force field and acoustic field of the iron core reactor;

[0008] Step 2: Establish a simulation model of an iron core reactor under typical defects, including loose core, winding deformation, winding looseness, and inter-turn short circuit, and obtain simulation results of the electromagnetic-structural force field and acoustic field of the iron core reactor under different working conditions;

[0009] Step 3: Based on the simulation results of step 2, analyze the influence of different defects on the vibration and noise of the iron core reactor;

[0010] Step 4: According to the vibration and noise distribution rules, obtain the best vibration and noise measurement point distribution, and extract the vibration and noise signals of the iron core reactor under different working conditions;

[0011] Step 5: Convert the extracted vibration and noise signals under different working conditions into Gram angle and field characteristic spectra;

[0012] Step 6: Input the Gram angle and field feature maps into AlexNet for transfer learning to achieve fault diagnosis.

[0013] In step 1, a finite element simulation software is used to establish a three-dimensional simulation model of the iron core reactor to be tested, as shown in Figure 2(a). The reactor structure mainly includes an iron core, a winding and an oil tank. The three-dimensional simulation model of the iron core reactor is meshed, and material properties are set and boundary conditions and initial conditions are set. The mesh division is set to coarsening, and the division result is shown in Figure 2(b); the material property settings are shown in Table 1:

[0014] Table 1 Reactor material properties

[0015]

[0016] Where: T is the temperature of the reactor oil;

[0017] The boundary conditions of the reactor are set as electromagnetic force and magnetostrictive force in the core and winding as the loading conditions of the structural force field, and the initial sound pressure of pressure acoustics is set to 0Pa.

[0018] Add multi-physics fields, including magnetic field, structural force field and acoustic field, to obtain the vibration and acoustic field simulation results of the reactor.

[0019] The distribution of reactor vibration acceleration is shown in Figure 3(a). The maximum vibration acceleration is 12.6Gal, located at the side yoke of the core. The distribution of reactor sound pressure is shown in Figure 3(b). The maximum sound pressure is 96Pa, located at the upper yoke of the core.

[0020] Establish the magnetic-structure coupling, structure-acoustic field coupling, and magnetic-acoustic field coupling equations. The general expression form of the magnetic-structure-acoustic field coupling equation is:

[0021]

[0022] Equations (1) and (2) are magnetic field equations, where H represents the magnetic field intensity; B represents the magnetic induction intensity; J represents the current density; represents the Hamiltonian operator.

[0023] Equation (3) is the structural dynamics equation, where ρ represents the material density; u represents the structural displacement; σ represents the stress tensor; f represents the external load.

[0024] Equation (4) is the acoustic field equation, where ρ c represents the density of the acoustic medium; p represents the sound pressure; represents the sound pressure vector field; t represents time.

[0025] In step 2, set faults for the three-dimensional simulation model of the iron-core reactor in step 1:

[0026] a. Adjust the iron core clamping force to simulate the iron core loosening fault;

[0027] b. Set axial bulging for the primary winding to simulate the winding deformation fault;

[0028] c. Set the internal circuit of the winding through the simulation software and adjust the number of winding turns and the fault impedance to simulate the inter-turn short circuit fault of the iron-core reactor winding;

[0029] d. Simulate the winding clamping force loosening fault by adjusting the upper clamping force of the winding.

[0030] In step 3, according to the multi-physics field coupling calculation model and the fault settings, calculate the multi-physics field coupling calculation results of the iron-core reactor winding under different working conditions. The multi-physics field coupling simulation calculation results under normal working conditions of the reactor are shown in Figures 3(a) and 3(b); the simulation calculation results under the inter-turn short circuit working condition are shown in Figures 4(a) and 4(b); Figures 5(a) and 5(b) are the simulation calculation results under the winding clamping force loosening working condition of the reactor.

[0031] In step 4, reasonable observation points are selected according to the vibration distribution characteristics and noise distribution characteristics of the iron-core reactor. The sound field observation point is located at the upper yoke of the reactor core, as shown in Fig. 4(a); the vibration observation point is located on the upper wall surface of the reactor oil tank, as shown in Fig. 4(b). According to the multi-physical field coupling calculation results of the iron-core reactor under different working conditions, the vibration signals and noise signals at each observation point are extracted; taking the signals extracted under the normal working condition of the reactor as an example, Fig. 5(a) is the sound field signal extracted at the observation point; Fig. 6(a) is the vibration signal extracted.

[0032] In step 5, the principle of the Gramian Angular Summation Field (GASF) is as follows:

[0033] S5.1: Normalize the signal to the interval [-1, 1], as shown in Equation (6):

[0034]

[0035] In Equation (6): represents the normalized signal, n represents the data sequence number; x i is the time series signal; min(x) is the minimum sequence signal; max(x) is the maximum sequence signal;

[0036] S5.1: Map the normalized signal to the polar coordinate system, as shown in Equation (7):

[0037]

[0038] In Equation (7), r represents the polar radius in the polar coordinate system, and t i represents the polar angle in the polar coordinate system, and t i is the number of signal sampling points, and N is the length of the signal.

[0039] S5.3: Combining Equation (6) and Equation (7), define the Gramian Angular Summation Field (GASF) as shown in Equation (8):

[0040]

[0041] In Equation (8): Subscripts i and j respectively represent the sequence numbers of different sampling points;

[0042] Use the Gramian Angular Summation Field to convert the feature signals extracted in step 4 into two-dimensional feature pictures; taking the sound field and vibration signals extracted under the normal working condition of the reactor as an example, Fig. 5(b) is the two-dimensional feature map obtained by converting the sound field signal in Fig. 5(a); Fig. 6(b) is the two-dimensional feature map obtained by converting the vibration signal in Fig. 6.

[0043] In step 6, the basic structure of the AlexNet model includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer; among them, the role of the input layer is to receive image data and perform preprocessing; the role of the convolutional layer is to gradually extract local features of the image; the role of the pooling layer is to reduce the size of the feature map and enhance the robustness of the model; the role of the fully connected layer is to combine and map the extracted features to the category space; the output layer generates the final classification result. Among them, the fully connected layer uses the Dropout technique to prevent overfitting.

[0044] Step 6 includes the following steps:

[0045] S6.1: Use the Gram angle and field feature map set generated in step 5 as the target data set, and perform preprocessing on it, including resizing, normalizing, and data augmentation to improve the generalization ability of the model;

[0046] S6.2: Load the pre-trained AlexNet model. In transfer learning, freeze the convolutional layer of AlexNet to avoid retraining the learned features and accelerate the training process;

[0047] S6.3: Modify the fully connected layer of the AlexNet model according to the number of task categories;

[0048] S6.4: Set the loss function and optimizer;

[0049] S6.5: Create a data loader to process training and validation data according to the data set;

[0050] S6.6: Train the model and perform validation.

[0051] The fault diagnosis method of the oil-immersed iron-core reactor based on multi-physical field coupling and feature map of the present invention has the following technical effects:

[0052] 1) The present invention can accurately identify the fault categories of the iron-core reactor winding, thus ensuring the reliability and safety of the power system, reducing downtime losses, extending the service life of the equipment, optimizing the operation and maintenance decisions, and promoting the intelligentization and optimized design of the iron-core reactor equipment.

[0053] 2) The present invention uses the Gram angle and field to process the voiceprint and vibration signals, which helps to denoise the signals and compress the data, thereby improving the accuracy of fault diagnosis and reducing the time of fault diagnosis.

[0054] 3) The fault diagnosis method of the iron-core reactor winding based on the sound field-structural force field of the present invention combines the multi-physical field characteristics of acoustics and mechanics, and can realize early fault diagnosis and accurate fault location. By collecting and analyzing the vibration and acoustic signals during the operation of the iron-core reactor, the winding faults can be effectively identified. Description of the Drawings

[0055] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0056] Figure 1 It is a flowchart of a fault diagnosis method for an iron core reactor.

[0057] Figure 2(a) is a three-dimensional simulation model of an iron core reactor established using finite element simulation software;

[0058] Figure 2(b) is a schematic diagram of the mesh division of the three-dimensional simulation model of the iron core reactor.

[0059] Figure 3(a) is the simulation result of the sound field of the iron core reactor under normal operating conditions;

[0060] Figure 3(b) is the simulation result of the structural force field of the iron core reactor under normal operating conditions.

[0061] Figure 4(a) is the simulation result of the sound field of the iron core reactor under the condition of winding deformation;

[0062] Figure 4(b) is the simulation result of the structural force field of the iron core reactor under the condition of winding deformation.

[0063] Figure 5(a) is the simulation result of the sound field of the iron core reactor under the condition of inter-turn short circuit;

[0064] Figure 5(b) is the simulation result of the structural force field of the iron core reactor under the condition of inter-turn short circuit.

[0065] Figure 6(a) is the simulation result of the sound field of the iron core reactor under the condition of winding looseness;

[0066] Figure 6(b) is the simulation result of the structural force field of the iron core reactor under the condition of winding looseness.

[0067] Figure 7(a) is a schematic diagram of the position of the vibration signal observation point of the iron core reactor Figure 1 ;

[0068] Figure 7(b) is the second schematic diagram of the position of the sound field signal observation point of the iron core reactor.

[0069] Figure 8(a) is a schematic diagram of a random voiceprint signal collected by a simulated sensor by the simulation software.

[0070] Figure 8(b) is a two-dimensional feature spectrum formed by the voiceprint signal through the Gram angular field.

[0071] Figure 9(a) is a schematic diagram of a random vibration signal collected by a simulated sensor by the simulation software.

[0072] Figure 9(b) is a two-dimensional feature spectrum formed by the vibration signal through the Gram angular field.

[0073] Figure 10(a) shows the operation result (training set confusion matrix) after inputting the feature spectrum into the convolutional neural network;

[0074] Figure 10(b) shows the operation result (test set confusion matrix) after inputting the feature spectrum into the convolutional neural network. Specific implementation manner

[0075] As Figure 1 shown, the fault diagnosis process of the iron-core reactor based on multi-physical field coupling is as follows: build a finite element simulation model of the iron-core reactor, clarify the calculation methods of the electromagnetic field and the structural force field, and simulate the operating states of the iron-core reactor windings under different fault conditions; based on the coupled simulation of the sound field and the structural force field, analyze the performance of the iron-core reactor in normal and fault states; select the acoustic fingerprint and vibration signals as features, and convert the collected signals into a GASF image set; input the image set into the convolutional neural network for training; initialize the diagnostic model, input the training set into the model for transfer learning, and complete the training process; evaluate the diagnostic effect through the test set data to complete the fault diagnosis.

[0076] It includes the following steps:

[0077] Step S1: According to the parameters of the iron-core reactor, establish a physical model of the iron-core reactor, build a multi-physical field coupling calculation model, and obtain the simulation results of the electromagnetic-structural force field and the sound field of the reactor;

[0078] Step S2: Establish a simulation model of the iron-core reactor under typical defects, including: loose iron core, winding deformation, loose winding, and inter-turn short circuit, and obtain the simulation results of the electromagnetic-structural force field and the sound field of the reactor under different working conditions;

[0079] Step S3: According to the simulation results, analyze the influence law of the defects on vibration and noise;

[0080] Step S4: According to the vibration and noise distribution laws, obtain the optimal vibration and noise measurement point distributions, and extract the vibration and noise signals of the iron-core reactor under different working conditions;

[0081] Step S5: Convert the extracted vibration and noise signals under different working conditions into Gram angular field feature spectra;

[0082] Step S6: Input the Gram angular field feature spectra into AlexNet for transfer learning to achieve fault diagnosis.

[0083] In step S1, a three-dimensional simulation model of the iron-core reactor to be measured is established using finite element simulation software, and multiple physical fields are added, including magnetic field, structural force field, and sound field modules. Material parameter settings, mesh generation, and boundary and initial condition settings are performed on the model. Among them, in terms of material parameter settings, the hysteresis and magnetostriction characteristics of the iron-core material are obtained based on the data of silicon steel sheets measured in experiments, and the reactor oil and windings are set according to actual parameters; in terms of the mesh, a refined mesh is used at positions such as the iron core and windings, and a sparse mesh is used at positions far from the iron core and windings; in terms of boundary condition settings, since the bottom of the reactor is connected to the bottom of the oil tank, the bottom of the iron-core reactor is set as a fixed constraint, that is, zero displacement, which is a stationary wall surface.

[0084] In step S2, magnetic-structure coupling and structure-sound field coupling equations of the iron-core reactor are established. Among them, the magnetic field-structural force field coupling equation is as follows:

[0085]

[0086] Equations (1) and (2) are magnetic field equations, where H is the magnetic field strength; B is the magnetic induction intensity; J is the current density. Equation (3) is the structural dynamics equation, where ρ is the material density; u is the structural displacement; σ is the stress tensor; f is the external load.

[0087] Among them, the structural force field-sound field coupling equation is as follows:

[0088]

[0089] Equation (4) is the sound field equation, where ρ c is the density of the sound medium; p is the sound pressure.

[0090] In step S2, fault settings are made for the iron-core reactor model in step S1: The internal fault simulation of the oil-immersed iron-core reactor is mainly located in the iron core and windings. In terms of the iron core, the iron-core loosening fault is simulated by adjusting the iron-core pressing force. The normal working condition is set to 100% of the rated pressing force, and the pressing force is adjusted to 10%, 20%, 30%, and 40% of the rated value respectively to simulate the iron-core loosening fault; In terms of the windings, radial and axial bulges are set for the primary-side winding to simulate the winding deformation fault. Radial and axial bulges are set at different positions: Radial and axial bulges are set in the upper-middle, middle, and lower-middle parts of the winding respectively. By controlling the size of the bulges, the winding deformation faults under different fault degrees are simulated; The inter-turn short-circuit fault of the iron-core reactor winding is simulated by setting the internal circuit of the winding and adjusting the number of winding turns and fault impedance through simulation software. Inter-turn short-circuit faults are set in the upper-middle, middle, and lower-middle parts of the winding respectively. The different fault degrees are determined according to the proportion of the short-circuit turns in the number of winding turns under normal working conditions, such as 1%, 2%, 3%, and 4%; The winding pressing force loosening fault is simulated by adjusting the pressing force at the upper end of the winding. The normal working condition is set to 100% of the rated pressing force, and the pressing force is set to 10%, 20%, 30%, and 40% of the rated value to simulate the winding pressing force loosening fault.

[0091] In step S3, according to the above multi-physical-field coupling calculation model and fault settings, the multi-physical-field coupling calculation results of the iron-core reactor winding under different working conditions are calculated, and their vibration characteristics and noise distribution characteristics are compared and analyzed: According to the simulation results, the magnetic flux density distributions along different paths at the iron core and winding positions are extracted for comparative analysis; The stress distributions along different paths at the iron core and winding positions are extracted for comparative analysis; The vibration displacement distributions along different paths at the iron core and winding positions are extracted for comparative analysis; The noise distributions along different paths around the iron core and winding positions are extracted for comparative analysis.

[0092] In step S4, reasonable observation points are selected according to the vibration distribution characteristics and noise distribution characteristics of the iron-core reactor. The specific steps for optimizing the measuring points are as follows: The data with a time of 0.2 s are sequentially intercepted for each type of working condition sample at different measuring points for LMD decomposition, and multiple PF components are decomposed. The correlation between each component and the original data is calculated, and P components with large correlations are selected as effective components, and the information entropy is calculated to construct a feature vector matrix; Keeping the same working condition unchanged, the correlation analysis is carried out on the feature vector matrices X and Y at different measuring points respectively, and the correlations of each measuring point under each working condition are solved in turn; The number of correlation index quantities among each measuring point under each working condition is counted, and those without correlation are retained to achieve the purpose of optimizing the measuring points. According to the multi-physical-field coupling calculation results of the iron-core reactor under different working conditions, the vibration signals and noise signals at each observation point under each working condition are extracted.

[0093] In step S5, the Gram Angular Summation Field (GASF) is used to convert the feature signals extracted from different measurement points in step S4 into a two-dimensional feature map. The principle of the GASF is as follows:

[0094] A1: Normalize the signal to the interval [-1, 1] as shown in Equation (6):

[0095]

[0096] where represents the normalized signal and n represents the data sequence number.

[0097] A2: Map the normalized signal to the polar coordinate system as shown in Equation (7):

[0098]

[0099] where r represents the radius in the polar coordinate system, and t i represents the polar angle in the polar coordinate system, t i is the number of signal sampling points, and N is the length of the signal.

[0100] A3: Combining Equation (6) and Equation (7), define the Gram Angular Summation Field (GASF) as shown in Equation (8):

[0101]

[0102] where the subscripts i and j represent the sequence numbers of different sampling points respectively.

[0103] In step S6, the basic structure of the AlexNet model includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer. The Dropout technique is used in the fully connected layer to prevent overfitting. The specific fault classification steps are as follows:

[0104] B1: Use the GASF image set generated in step S5 as the target data set and preprocess it, including resizing, normalizing, and data augmentation to improve the generalization ability of the model.

[0105] B2: Load the pre-trained AlexNet model. In transfer learning, freeze the convolutional layer of AlexNet to avoid retraining the learned features and accelerate the training process.

[0106] B3: Modify the fully connected layer of AlexNet according to the number of task categories.

[0107] B4: Set the loss function and optimizer.

[0108] B5: Create a data loader to process the training and validation data according to the data set.

[0109] B6: Train the model and verify it.

[0110] Figures 2 and 3 are the models of oil-immersed transformers established by finite element simulation software and the simulation results after adding the acoustic field-structure force field, which is used to simulate the iron core reactor in actual operation. The noise and vibration acceleration distribution laws of the iron core reactor under different working conditions are compared and analyzed, and the optimal measuring point position is selected, as shown in Figure 7 (a) and Figure 7 (b). After the finite element simulation process is completed, the relationship curve between the soundprint amplitude and time and the relationship curve between the vibration acceleration and time under various working conditions at the selected measuring point are derived using a one-dimensional drawing group, as shown in Figures 8 (a) and 9 (a), and the transformer's soundprint signal and vibration signal are reflected in this way. Since the extracted soundprint and vibration signals usually have noise interfering with the useful part of the original signal, the present invention processes the extracted signal through the Gram angle field, converts the one-dimensional signal set into a two-dimensional feature spectrum, so as to achieve the purpose of noise reduction and signal compression, as shown in Figures 8 (b) and 9 (b).

[0111] As shown in Figure 10(a) and Figure 10(b), the accuracy of the above algorithm after operation is completed is compared with the actual situation. Figure 10(a) is the result of the training set, and Figure 10(b) is the result of the test set. It can be seen from the results Figure 10(a) and Figure 10(b) that the accuracy of both is higher than 94%, indicating that the algorithm operation result is accurate.

[0112] In actual operation, the specific steps are as follows:

[0113] Step 1: Use finite element simulation software to establish a three-dimensional simulation model of the iron core reactor to be tested and mesh the model, set material properties, boundary conditions and initial conditions. Add multiple physical fields, including magnetic field, structural force field and acoustic field. Establish magnetic-structural coupling, structural-acoustic field coupling and magnetic-acoustic field coupling equations.

[0114] Step 2: Set up faults for the iron core reactor model: adjust the core clamping force to simulate a loose core fault; set up an axial bulge on the primary winding to simulate a winding deformation fault; use simulation software to set up the internal circuit of the winding and adjust the number of winding turns and fault impedance to simulate an inter-turn short-circuit fault in the iron core reactor winding; and adjust the clamping force on the upper end of the winding to simulate a loose winding clamping force fault.

[0115] Step 3: According to the multi-physics field coupling calculation model and fault setting, the multi-physics field coupling calculation results of the iron core reactor winding under different working conditions are calculated, and the vibration characteristics and noise distribution characteristics are compared and analyzed.

[0116] Step 4: Select reasonable observation points according to the vibration distribution characteristics and noise distribution characteristics of the iron-core reactor. According to the multi-physics field coupling calculation results of the iron-core reactor under different working conditions, extract the magnetic characteristic signals, vibration signals and noise signals of each observation point under various working conditions. And convert the extracted characteristic signals into two-dimensional characteristic spectrograms using the Gramian angular summation field. Input the characteristic spectrograms into the improved AlexNet convolutional neural network for transfer learning to complete fault diagnosis.

Claims

1. A fault diagnosis method for oil-immersed iron core reactor based on multi-physical field coupling and characteristic spectrum, characterized in that The following steps are involved: Step 1: Construct a multi-physics field coupling calculation model for the iron core reactor to obtain the simulation results of the electromagnetic-structural force field and acoustic field of the iron core reactor; Step 2: Establish a simulation model of an iron core reactor under typical defects, including loose core, winding deformation, winding looseness, and inter-turn short circuit, and obtain simulation results of the electromagnetic-structural force field and acoustic field of the iron core reactor under different working conditions; Step 3: Based on the simulation results of step 2, analyze the influence of different defects on the vibration and noise of the iron core reactor; Step 4: According to the vibration and noise distribution rules, obtain the best vibration and noise measurement point distribution, and extract the vibration and noise signals of the iron core reactor under different working conditions; Step 5: Convert the extracted vibration and noise signals under different working conditions into Gram angle and field characteristic spectra; Step 6: Input the Gram angle and field feature maps into AlexNet for transfer learning to achieve fault diagnosis.

2. The oil-immersed iron core reactor fault diagnosis method based on multi-physical field coupling and characteristic spectrum according to claim 1 is characterized in that: In the step 1, a finite element simulation software is used to establish a three-dimensional simulation model of the iron core reactor to be tested, wherein the reactor structure includes an iron core, a winding and an oil tank; and the three-dimensional simulation model of the iron core reactor is meshed, and material properties, boundary conditions and initial conditions are set; the mesh segmentation is set to coarsening, and the boundary conditions of the reactor are set to the electromagnetic force and magnetostrictive force in the iron core and the winding as loading conditions of the structural force field; and multiple physical fields are added, including a magnetic field, a structural force field and an acoustic field, to obtain simulation results of the vibration and acoustic field of the reactor.

3. The fault diagnosis method for oil-immersed iron core reactor based on multi-physical field coupling and characteristic spectrum according to claim 2 is characterized in that: The magnetic-structure coupling, structure-acoustic field coupling and magnetic-acoustic field coupling equations are established. The magnetic-structure-acoustic field coupling equation is expressed as: Formula (1) and Formula (2) are magnetic field equations, where H represents the magnetic field intensity; B represents the magnetic induction intensity; J represents the current density; represents the Hamiltonian operator; Formula (3) is the structural dynamics equation, where ρ represents the material density; u represents the structural displacement; σ represents the stress tensor; f represents the external load; Formula (4) is the acoustic field equation, where ρ c represents the density of the sound medium; p represents the sound pressure; represents the sound pressure vector field; t represents time.

4. The fault diagnosis method for oil-immersed iron core reactor based on multi-physical field coupling and characteristic spectrum according to claim 1 is characterized in that: In step 2, a fault setting is performed on the three-dimensional simulation model of the iron core reactor in step 1: a. Adjust the core pressing force to simulate the core loosening fault; b. Perform axial bulging on the primary winding to simulate winding deformation fault; c. Use simulation software to set the internal circuit of the winding and adjust the number of winding turns and fault impedance to simulate the short-circuit fault between turns of the iron core reactor winding; d. Simulate the loose winding clamping force failure by adjusting the clamping force at the upper end of the winding.

5. The fault diagnosis method for oil-immersed iron core reactor based on multi-physical field coupling and characteristic spectrum according to claim 1 is characterized in that: In the step 4, a reasonable observation point is selected according to the vibration distribution characteristics and noise distribution characteristics of the iron core reactor; The acoustic field observation point is located at the upper yoke of the reactor core, and the vibration observation point is located at the upper wall of the reactor oil tank. According to the multi-physical field coupling calculation results of the iron core reactor under different working conditions, the vibration signal and noise signal of each observation point under each working condition are extracted.

6. The fault diagnosis method for oil-immersed iron core reactor based on multi-physical field coupling and characteristic spectrum according to claim 5 is characterized in that: In step 5, the principle of the Gram angle and field is: S5.1: Normalize the signal to the interval [-1,1], as shown in equation (6): In formula (6): represents the normalized signal, n represents the data sequence number; x i is a time series signal; min(x) is the minimum series signal; max(x) is the maximum series signal; S5.1: Map the normalized signal to the polar coordinate system, as shown in equation (7): In formula (7), r represents the polar diameter in the polar coordinate system, t i represents the polar angle in the polar coordinate system, t i is the number of sampling points of the signal, and N is the length of the signal; S5.3: Combining equations (6) and (7), define the Gram angle sum field (GASF) as shown in equation (8): In formula (8), the subscripts i and j represent the serial numbers of different sampling points respectively; The feature signal extracted in step 4 is converted into a two-dimensional feature image using the Gram angle and field.

7. The fault diagnosis method for oil-immersed iron core reactor based on multi-physical field coupling and characteristic spectrum according to claim 5 is characterized in that: In step 6, the basic structure of the AlexNet model includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an output layer; wherein the input layer is used to receive image data and perform preprocessing; the convolutional layer is used to gradually extract local features of the image; the pooling layer is used to reduce the size of the feature map and enhance the robustness of the model; the fully connected layer is used to combine and map the extracted features to the category space; the output layer generates the final classification result; wherein the fully connected layer uses the Dropout technology to prevent overfitting.

8. The fault diagnosis method for oil-immersed iron core reactor based on multi-physical field coupling and characteristic spectrum according to claim 7 is characterized in that: Step 6 includes the following steps: S6.1: Take the Gram angle and field feature atlas generated in step 5 as the target dataset and preprocess it, including resizing, normalizing and performing data augmentation to improve the generalization ability of the model; S6.2: Load the pre-trained AlexNet model and freeze the convolutional layers of AlexNet in transfer learning; S6.3: Modify the fully connected layer of the AlexNet model according to the number of task categories; S6.4: Set the loss function and optimizer; S6.5: Create a data loader to process training and validation data based on the dataset; S6.6: Train the model and perform validation.

Images