Transformer fault diagnosis method, system and equipment based on harmonic injection and storage medium

Abstract

The invention discloses a transformer fault diagnosis method, system and equipment based on harmonic injection and a storage medium, and the method comprises the steps: installing a vibration sensor on a transformer body, collecting steady-state and dynamic vibration signals, and carrying out the preprocessing of the vibration signals; time domain feature analysis is carried out on the collected vibration signals, frequency domain feature analysis is carried out on the vibration signals, and then spectrum complexity is analyzed; normal ranges of RMS values, peak factors and odd-even harmonic ratios are set through historical data, and if the data exceed the set ranges, the fan influence is further analyzed; analyzing in combination with a power system operation mode, and then performing fault judgment based on a machine learning classification algorithm; if the direct current magnetic bias is detected, an HVDC operation mode is adjusted; and if harmonic injection is detected, an active harmonic filter is adopted to suppress harmonic interference. The harmonic current is detected and compensated in real time, the voltage and current waveforms of the power grid are improved, the harmonic pollution of the power grid is reduced, and the electric energy quality is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power grid connection, and more specifically, particularly relates to a transformer fault diagnosis method and system based on harmonic injection. Background Technique

[0002] With the large-scale access of new energy (such as wind power and photovoltaic), the power quality problems of the power grid have become increasingly prominent. Among them, harmonic injection is one of the important factors affecting the stable operation of the power grid, mainly manifested as the enhancement of high-frequency odd harmonics, resulting in problems such as voltage waveform distortion, increased equipment losses, and misoperation of relay protection. In addition, non-linear loads such as HVDC transmission projects, high-speed railways, and subways may also cause DC bias problems in the power grid, affecting the normal operation of transformers.

[0003] At present, traditional harmonic suppression methods mainly include passive filters (LC filters) and active filters (APFs). However, passive filters have a resonance risk and a limited applicable range. Although active filters have better effects, they have a high cost and need to be coordinated and optimized with inverters. The adjustment of new energy inverter parameters is crucial for suppressing harmonics. For example, how to dynamically adjust the modulation mode of the inverter to optimize the current harmonic components, thereby reducing the impact of harmonics on the power grid. Therefore, the present invention proposes a harmonic suppression method based on an active harmonic filter, combined with a new energy inverter parameter optimization strategy, to improve the power quality of the power grid, reduce harmonic interference, optimize the grid connection characteristics of new energy, and improve the operation stability of the power system. Summary of the Invention

[0004] In view of the above or existing problems of the transformer fault diagnosis method and system based on harmonic injection, the present invention is proposed.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] An embodiment of the present invention provides a transformer fault diagnosis method based on harmonic injection, including: installing vibration sensors on the transformer body, collecting steady-state and dynamic vibration signals, and preprocessing the vibration signals;

[0007] Performing time-domain feature analysis on the collected vibration signals, then performing frequency-domain feature analysis on the vibration signals, and then analyzing the spectral complexity;

[0008] Setting the normal ranges of the RMS value, peak factor, and odd-even harmonic ratio through historical data. If the data exceeds the set range, further analyze the influence of the fan;

[0009] Analyze in combination with the operation mode of the power system, and then conduct fault discrimination based on the machine learning classification algorithm. If the low-frequency even harmonics increase and the spectral entropy decreases, it is DC bias magnetization; if the high-frequency odd harmonics increase and the spectral entropy increases, it is harmonic injection.

[0010] If DC bias magnetization is detected, adopt DC isolation or DC compensation technology to adjust the HVDC operation mode and reduce the impact on the AC system; if harmonic injection is detected, adopt an active harmonic filter to suppress harmonic interference,

[0011] Adjust the parameters of the new energy inverter and optimize the current harmonic components.

[0012] As a preferred scheme of the transformer fault diagnosis method based on harmonic injection described in the present invention, it includes: collecting steady-state and dynamic vibration signals, and preprocessing the vibration signals, including:

[0013] Set the sampling frequency f s = 10 kHz, collect data for more than 60 s for each condition of the steady-state signal, and collect data for more than 4 s for each condition of the dynamic signal;

[0014] Perform 5-layer decomposition using wavelet, and use soft threshold filtering for high-frequency noise: ,

[0015] where T is the threshold, σ is the noise standard deviation, and N is the signal length;

[0016] Then perform data smoothing and window segmentation, and use moving average filtering for processing: ,

[0017] where X′(n) is the filtered signal, X(n) is the original unprocessed signal, and N is the length of the moving window.

[0018] As a preferred scheme of the transformer fault diagnosis method based on harmonic injection described in the present invention, it includes: performing time-domain feature analysis on the collected vibration signals, then performing frequency-domain feature analysis on the vibration signals, and then analyzing the spectral complexity, including:

[0019] Calculate the spectrum of the vibration signal using Fourier transform:

[0020] ,

[0021] Obtain the energy distribution of different frequency components of the vibration signal by performing spectrum analysis on the vibration signal;

[0022] Define the odd harmonic energy ratio and the even harmonic energy ratio:

[0023] ,

[0024] Among them, Hodd is the proportion of odd - harmonic energy, Heven is the proportion of even - harmonic energy, and kf 0 is different harmonic frequencies;

[0025] Then, calculate the spectral complexity of the signal through spectral entropy: ,

[0026] Among them, P i is the normalized power of each frequency component. A higher spectral entropy indicates that the signal has more energy distribution at multiple frequencies.

[0027] As a preferred scheme of the transformer fault diagnosis method based on harmonic injection described in the present invention, wherein: set the normal ranges of the RMS value, peak factor, and odd - even harmonic ratio. If the data exceeds the set range, further analyze the influence of the fan, including:

[0028] For the collected vibration signal X(t), calculate the RMS value, peak factor, and odd - even harmonic ratio of the parameters within each time window T, and determine whether they exceed the threshold: ,

[0029] If any parameter exceeds the set range, enter the further analysis of the influence of the fan;

[0030] Analyze the interference of the fan influence, observe whether there is low - frequency modulation in the signal, and calculate the autocorrelation function: ,

[0031] If periodic peaks appear, they are caused by the influence of the fan;

[0032] Conduct FFT analysis and observe whether there are the fan characteristic frequency f f and its multiple frequencies:

[0033] ,

[0034] If f f has a high proportion in the vibration spectrum, the influence of the fan is greater.

[0035] As a preferred scheme of the transformer fault diagnosis method based on harmonic injection described in the present invention, wherein: conduct fault discrimination based on the machine - learning classification algorithm. If the low - frequency even harmonics are enhanced and the spectral entropy decreases, it is DC bias magnetization; if the high - frequency odd harmonics are enhanced and the spectral entropy increases, it is harmonic injection, including:

[0036] Select the support vector machine for fault discrimination. The extracted features are X = [X1, X2,..., Xn], where X1 is the low - frequency even - harmonic feature, X2 is the high - frequency odd - harmonic feature, X3 is the spectral entropy. Use the training data set to train the support vector machine model, and set appropriate penalty parameter C and kernel function parameter γ during training: ,

[0037] where α i is the Lagrange multiplier, y i is the label, x i is the training data, x is the new input data, and b is the bias term;

[0038] Calculate the spectral entropy of the signal and check if it changes. If the spectral entropy decreases and the low-frequency even harmonics increase, it is determined as DC bias magnetization; if the spectral entropy increases and the high-frequency odd harmonics increase, it is determined as harmonic injection;

[0039] If an increase in low-frequency even harmonics is detected, further exclude the influence of the fan; if an increase in high-frequency odd harmonics is detected, determine whether there is a harmonic source in combination with the operation mode of the power system.

[0040] As a preferred solution of the transformer fault diagnosis method based on harmonic injection according to the present invention, wherein: if DC bias magnetization is detected, adjust the HVDC operation mode using DC isolation or DC compensation technology to reduce the impact on the AC system, including:

[0041] In the HVDC converter station, by adjusting the DC current I dc of the converter to restore the balance of the transformer magnetic flux:

[0042] ,

[0043] where ΔHeven is the detected even harmonic increment and K is the compensation coefficient;

[0044] Install an active compensation device at the neutral point of the transformer to detect the DC current I dc in real time and inject a reverse DC current: ,

[0045] to make the total DC component in the system tend to zero;

[0046] If the DC component is isolated for a long time, use DC isolation; if it is dynamically adjusted, use DC compensation; check whether H even decreases and S increases to ensure that the DC bias magnetization is suppressed.

[0047] As a preferred solution of the transformer fault diagnosis method based on harmonic injection according to the present invention, wherein: if harmonic injection is detected, use an active harmonic filter to suppress harmonic interference and adjust the parameters of the new energy inverter to optimize the current harmonic components, including:

[0048] Install an active harmonic filter, collect the current signal in the power grid, detect the harmonic components, calculate the target compensation current, generate a reverse harmonic current in real time, and output a compensation signal through the power conversion circuit to cancel the harmonic current;

[0049] Adjust the pulse width modulation strategy, optimize the switching frequency, and reduce the high-frequency harmonic content; adopt the p-q theory to enhance the harmonic suppression ability, and use the phase-locked loop algorithm to improve the synchronization control accuracy and reduce the harmonic interference of the inverter to the power grid.

[0050] A transformer fault diagnosis system based on harmonic injection, comprising: an acquisition and preprocessing module, configured to install vibration sensors on the transformer body to acquire steady-state and dynamic vibration signals and preprocess the vibration signals;

[0051] A feature extraction module, configured to perform time-domain feature analysis on the acquired vibration signals, then perform frequency-domain feature analysis on the vibration signals, and then analyze the spectral complexity;

[0052] A setting analysis module, configured to set the normal ranges of the RMS value, peak factor, and odd-even harmonic ratio through historical data. If the data exceeds the set range, further analyze the influence of the fan;

[0053] An algorithm classification module, configured to analyze in combination with the operation mode of the power system, and then perform fault discrimination based on the machine learning classification algorithm. If the low-frequency even harmonics are enhanced and the spectral entropy decreases, it is DC bias magnetization; if the high-frequency odd harmonics are enhanced and the spectral entropy increases, it is harmonic injection;

[0054] A fault optimization module, configured to, if DC bias magnetization is detected, adjust the HVDC operation mode using DC isolation or DC compensation technology to reduce the influence on the AC system; if harmonic injection is detected, suppress harmonic interference using an active harmonic filter, adjust the parameters of the new energy inverter, and optimize the current harmonic components.

[0055] A computing device, the computing device comprising:

[0056] At least one processor, a memory, and an input / output unit;

[0057] Wherein, the memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the steps of the transformer fault diagnosis method based on harmonic injection.

[0058] A computer-readable storage medium, which includes instructions that, when running on a computer, cause the computer to execute the steps of the transformer fault diagnosis method based on harmonic injection.

[0059] The beneficial effects of the present invention are as follows: By detecting and compensating harmonic currents in real time, the present invention reduces the influence of high-frequency odd harmonics, effectively improves the voltage and current waveforms of the power grid, reduces power grid harmonic pollution, and improves power quality. By adjusting the pulse width modulation strategy, DC bus voltage, and filtering parameters of the new energy inverter, the harmonic components of the current are optimized, enabling the new energy power generation system to be connected to the grid more smoothly and reducing the impact on the power grid. The harmonic suppression method of the present invention can reduce these adverse effects, extend the service life of equipment, and improve the overall stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0061] Figure 1 It is a flowchart of a transformer fault diagnosis method based on harmonic injection provided by an embodiment of the present invention.

[0062] Figure 2 It is a schematic structural diagram of a transformer fault diagnosis system based on harmonic injection provided by an embodiment of the present invention.

[0063] Figure 3 It schematically shows a structural diagram of a medium according to an embodiment of the present invention.

[0064] Figure 4 It schematically shows a structural diagram of a computing device according to an embodiment of the present invention.

[0065] In the drawings, the same or corresponding reference numerals indicate the same or corresponding parts. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the drawings in the specification.

[0067] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0068] Second, as used herein, "one embodiment" or "an embodiment" refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other.

[0069] Embodiment

[0070] The following refers to Figure 1 , Figure 1 which is a flowchart of a transformer fault diagnosis method based on harmonic injection provided for an embodiment of the present invention. It should be noted that the implementation manner of the present invention can be applied to any applicable scenario.

[0071] Figure 1 The flow of the transformer fault diagnosis method based on harmonic injection provided for an embodiment of the present invention shown in the figure includes:

[0072] S1: Install vibration sensors on the transformer body, collect steady-state and dynamic vibration signals, and preprocess the vibration signals.

[0073] Preferably, set the sampling frequency f s = 10 kHz, collect data for more than 60 s for each condition of the steady-state signal, and collect data for more than 4 s for each condition of the dynamic signal;

[0074] Perform 5-layer decomposition using wavelet, and use soft threshold filtering for high-frequency noise: ,

[0075] where T is the threshold, σ is the noise standard deviation, and N is the signal length;

[0076] Then perform data smoothing and window segmentation, and use moving average filtering for processing: ,

[0077] where X′(n) is the filtered signal, X(n) is the original unprocessed signal, and N is the length of the moving window.

[0078] Furthermore, collect the vibration signals of a certain transformer under normal conditions and harmonic injection conditions. After wavelet decomposition, it is found that there are no obvious sharp pulses in the high-frequency detail signals under normal conditions, and the spectral energy is mainly concentrated in the low-frequency region; obvious sharp pulses appear in the high-frequency detail signals under harmonic injection conditions, the energy of odd harmonics is significantly enhanced, and the spectral complexity increases. In order to further remove the residual noise and improve the stability of feature extraction, moving average filtering is used for smoothing processing and window segmentation.

[0079] Set the window size, for example, a 1s window (10,000 sampling points), and perform sliding calculations. Adopt an overlapping window method (such as 50% overlap) to ensure data continuity and improve the accuracy of feature extraction. In a certain transformer, after applying the moving average filter, it is observed that the signal is stable under normal conditions, the RMS value has no obvious fluctuation, and the odd-to-even harmonic ratio is within the normal range; under the harmonic injection condition, the energy of the high-frequency signal increases, the odd harmonic ratio exceeds the set threshold, and the peak factor increases significantly.

[0080] S2: Based on the collected vibration signals, perform time-domain feature analysis on them, then perform frequency-domain feature analysis on the vibration signals, and then analyze the spectral complexity.

[0081] Preferably, use the Fourier transform to calculate the spectrum of the vibration signal:

[0082] ,

[0083] By performing spectrum analysis on the vibration signal, obtain the energy distribution of its different frequency components;

[0084] Define the odd harmonic energy ratio and the even harmonic energy ratio:

[0085] ,

[0086] where Hodd is the proportion of odd harmonic energy, Heven is the proportion of even harmonic energy, and kf 0 is different harmonic frequencies;

[0087] Then calculate the spectral complexity of the signal through spectral entropy: ,

[0088] where P i is the normalized power of each frequency component. A higher spectral entropy indicates that the signal has more energy distribution at multiple frequencies.

[0089] Furthermore, the sampling frequency fs = 10 kHz, sampling time: collect 60s of vibration signal data under steady-state conditions, dynamic conditions: collect 4s of short-time burst vibration data, and the number of signal channels uses a three-axis acceleration sensor,

[0090] Select wavelet for 5-layer decomposition: X(n)=A5+D5+D4+D3+D2+D1, A5 is the lowest-frequency signal, D5, D4,..., D1 are high-frequency detail components;

[0091] Calculate the noise standard deviation:

[0092] ,

[0093] Set the threshold ,

[0094] Process the high-frequency coefficients:

[0095] ,

[0096] Furthermore, calculate 1000 groups of data to obtain the statistical results under different working conditions:

[0097]

[0098] Use a support vector machine for classification, and the classification accuracy is 97.2%. If H even > 0.3 and SE < 1.0, then it is judged as DC bias; if H odd > 0.4 and SE > 1.5, then it is judged as harmonic injection.

[0099] S3: Set the normal ranges of the RMS value, peak factor, and odd and even harmonic ratios through historical data. If the data exceeds the set range, further analyze the influence of the fan.

[0100] Preferably, for the collected vibration signal X(t), calculate the RMS value, peak factor, and odd and even harmonic ratios of the parameters within each time window T, and judge whether they exceed the threshold: ,

[0101] If any parameter exceeds the set range, enter the further analysis of the influence of the fan;

[0102] Analyze the interference of the fan influence, observe whether there is low-frequency modulation in the signal, and calculate the autocorrelation function: ,

[0103] If periodic peaks appear, they are caused by the influence of the fan;

[0104] Perform FFT analysis to observe whether there is the fan characteristic frequency f f and its multiples:

[0105] ,

[0106] If f f has a high proportion in the vibration spectrum, the influence of the fan is greater.

[0107] Furthermore, set the empirical thresholds:

[0108] RMS > 0.4g → abnormal vibration

[0109] C > 3.5 → enhanced impact

[0110] H odd > 0.45 or Heven > 0.3 → abnormal harmonics

[0111] If any parameter exceeds the set range, it enters the fan interference analysis stage;

[0112] If the autocorrelation function shows periodic peaks and the FFT finds that the proportion of the fan characteristic frequency and its multiples increases, it can be determined that the fan has interfered with the transformer vibration signal; further calculate the energy proportion of the fan frequency f f in the vibration signal. If the proportion of the fan characteristic frequency in the total signal energy exceeds 20%, it is determined that the fan has a greater impact; if the fan has a greater impact, measures such as adjusting the fan operation mode, optimizing the transformer installation position, and adding a damping structure can be taken to reduce the vibration interference of the fan on the transformer;

[0113] Experimental analysis was carried out on 1000 groups of vibration signal data, and the following statistical results were obtained:

[0114]

[0115] In the FFT spectrum within 500Hz, the proportion of the fan characteristic frequency f f = 45Hz and its multiples (90Hz, 135Hz) increased significantly (>20%) under fan interference; under normal conditions, the fan frequency characteristics were weak (<5%).

[0116] S4: Analyze in combination with the power system operation mode, and then perform fault discrimination based on the machine learning classification algorithm. If the low-frequency even harmonics increase and the spectral entropy decreases, it is DC bias magnetization; if the high-frequency odd harmonics increase and the spectral entropy increases, it is harmonic injection.

[0117] Preferably, a support vector machine is selected for fault discrimination. The extracted features are X = [X1, X2,..., Xn], where X1 is the low-frequency even harmonic feature, X2 is the high-frequency odd harmonic feature, X3 is the spectral entropy, and the training data set is used to train the support vector machine model. When training, set appropriate penalty parameter C and kernel function parameter γ: ,

[0118] where, α i is the Lagrange multiplier, y i is the label, x i is the training data, x is the newly input data, and b is the bias term;

[0119] Calculate the spectral entropy of the signal and check whether it changes. If the spectral entropy decreases and the low-frequency even harmonics increase, it is determined to be DC bias magnetization; if the spectral entropy increases and the high-frequency odd harmonics increase, it is determined to be harmonic injection;

[0120] If it is detected that the low-frequency even harmonics increase, further exclude the influence of the fan; if the high-frequency odd harmonics increase, combine the operation mode of the power system to judge whether there is a harmonic source.

[0121] Further, input the collected characteristic data [X1, X2, X3] into the trained SVM classifier. The model will output a predicted label: 1 represents DC bias, and 2 represents harmonic injection. If the output label is 1, it is determined as DC bias, and the influence of the fan is further excluded, and DC isolation or compensation technology is executed to reduce the interference to the AC system. If the output label is 2, it is determined as harmonic injection, and it is analyzed in combination with the operation mode of the power system. If there is a harmonic source, interference suppression is performed through an active harmonic filter, and the parameter configuration of the inverter is optimized.

[0122] The data set contains vibration signals of different fault types, including normal conditions, DC bias, and harmonic injection. Under each condition, multiple data samples are collected. Three features are extracted from each sample: low-frequency even harmonics, high-frequency odd harmonics, and spectral entropy.

[0123]

[0124] 70% of the data is used for training, and 30% of the data is used for testing. The optimal penalty parameter and kernel function parameter are determined through cross-validation. The training accuracy is 98%, and the test accuracy is 95%. For the newly collected signal, after extracting the features [X1, X2, X3], it is input into the SVM model: the sample input is X1 = 0.52, X2 = 0.15, X3 = 0.71; the SVM outputs the label 1, indicating DC bias.

[0125] S5: If DC bias is detected, the DC isolation or DC compensation technology is adopted to adjust the HVDC operation mode to reduce the influence on the AC system; if harmonic injection is detected, the active harmonic filter is used to suppress harmonic interference,

[0126] Adjust the parameters of the new energy inverter to optimize the current harmonic components.

[0127] Preferably, in the HVDC converter station, by adjusting the DC current I of the converter dc to restore the balance of the transformer magnetic flux:

[0128] ,

[0129] where ΔHeven is the detected even harmonic increment, and K is the compensation coefficient;

[0130] An active compensation device is installed at the neutral point of the transformer to detect the DC current I in real time dc and inject a reverse DC current: ,

[0131] to make the total DC component in the system tend to zero;

[0132] For long-term isolation of the DC component, use DC isolation; for dynamic adjustment, use DC compensation; check whether H even decreases and whether S increases to ensure that DC bias is suppressed.

[0133] Preferably, install an active harmonic filter to collect current signals in the power grid, detect harmonic components, calculate the target compensation current, generate reverse harmonic current in real time, and output a compensation signal through a power conversion circuit to cancel the harmonic current; adjust the pulse width modulation strategy, optimize the switching frequency, and reduce the high-frequency harmonic content; adopt the p-q theory to improve the harmonic suppression ability, and adopt the phase-locked loop algorithm to improve the synchronous control accuracy and reduce the harmonic interference of the inverter to the power grid.

[0134] Furthermore, the collected DC current starts from 0 seconds and is 1000 A. There is also a spectrum analysis device installed in the system to detect the even harmonic increment in the vibration signal. The preliminary detection results show that the even harmonic increment measured by the system at the initial moment (0 seconds) is 0.15. At this time, the system needs to perform DC compensation to avoid the influence of DC bias on the flux balance of the transformer.

[0135] According to the detected even harmonic increment ΔH even =0.15, the system calculates the compensation coefficient K = 0.8. Use this coefficient to adjust the DC current so that the DC current is effectively compensated. The adjusted DC current is 1000.12 A. Therefore, at 0 seconds, the adjusted DC current becomes 1000.12 A.

[0136] During the implementation of dynamic compensation, the system continuously monitors the change of the even harmonic increment. After compensation, the even harmonic increment recalculated and detected by the system gradually decreases. According to the monitoring data, at 60 seconds, the even harmonic increment drops to 0.10. Then, at 120 seconds, after further adjustment, the even harmonic increment further decreases to 0.05. During this process, the DC current is also continuously adjusted to ensure that the DC bias problem is effectively suppressed. At this time, the adjusted DC current is 1000.08 A.

[0137] At 180 seconds, the effect of the compensation device is finally verified. At this time, the even harmonic increment drops to 0.02, and the DC current is finally adjusted to 1000.02 A. At this time, the S signal (i.e., the flux recovery signal of the transformer) increases to 1.7, indicating that the flux of the transformer has returned to balance. Through spectrum analysis, it can be confirmed that the even harmonic components in the system have decreased significantly, the DC bias has been effectively suppressed, and the S signal continues to increase, indicating that the DC compensation effect of the system is very significant.

[0138] After introducing the method of the exemplary embodiment of the present invention, next, refer to Figure 2An explanation is given to the harmonic injection-based transformer fault diagnosis system according to the exemplary embodiments of the present invention. The system includes:

[0139] An acquisition and preprocessing module, configured to install vibration sensors on the transformer body, acquire steady-state and dynamic vibration signals, and preprocess the vibration signals;

[0140] A feature extraction module, configured to perform time-domain feature analysis on the acquired vibration signals, then perform frequency-domain feature analysis on the vibration signals, and then analyze the spectral complexity;

[0141] A setting analysis module, configured to set the normal ranges of the RMS value, crest factor, and odd-even harmonic ratio through historical data. If the data exceeds the set ranges, the influence of the fan is further analyzed;

[0142] An algorithm classification module, configured to analyze in combination with the operation mode of the power system, and then perform fault discrimination based on the machine learning classification algorithm. If the low-frequency even harmonics are enhanced and the spectral entropy decreases, it is DC bias; if the high-frequency odd harmonics are enhanced and the spectral entropy increases, it is harmonic injection;

[0143] A fault optimization module, configured to, if DC bias is detected, adopt DC isolation or DC compensation technology to adjust the HVDC operation mode and reduce the influence on the AC system; if harmonic injection is detected, adopt an active harmonic filter to suppress harmonic interference, adjust the parameters of the new energy inverter, and optimize the current harmonic components.

[0144] After introducing the methods and devices according to the exemplary embodiments of the present invention, next, reference is made to Figure 3 An explanation is given to the computer-readable storage medium according to the exemplary embodiments of the present invention. Please refer to Figure 3, which shows that the computer-readable storage medium is an optical disc 30, on which a computer program (i.e., program product) is stored. When the computer program is run by a processor, it will implement the steps recorded in the above method embodiments. For example, vibration sensors are installed on the transformer body to collect steady-state and dynamic vibration signals, and the vibration signals are preprocessed; time-domain feature analysis is performed on the collected vibration signals, then frequency-domain feature analysis is performed on the vibration signals, and then the spectral complexity is analyzed; through historical data, the normal ranges of RMS value, peak factor, and odd-even harmonic ratio are set. If the data exceeds the set range, the influence of the fan is further analyzed; analysis is performed in combination with the operation mode of the power system, and then fault discrimination is performed based on the machine learning classification algorithm. If the low-frequency even harmonics are enhanced and the spectral entropy is reduced, it is DC bias; if the high-frequency odd harmonics are enhanced and the spectral entropy is increased, it is harmonic injection; if DC bias is detected, DC isolation or DC compensation technology is used to adjust the HVDC operation mode to reduce the influence on the AC system; if harmonic injection is detected, an active harmonic filter is used to suppress harmonic interference, the parameters of the new energy inverter are adjusted, and the current harmonic components are optimized; the specific implementation methods of each step will not be repeated here.

[0145] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other optical and magnetic storage media, which will not be elaborated here one by one.

[0146] After introducing the methods, devices, and media of the exemplary embodiments of the present invention, next, reference is made to Figure 4 a computing device for transformer fault diagnosis based on harmonic injection according to the exemplary embodiments of the present invention.

[0147] Figure 4 The block diagram of an exemplary computing device 40 suitable for implementing the embodiments of the present invention is shown. The computing device 40 may be a computer system or a server. Figure 4 The shown computing device 40 is merely an example and should not impose any limitation on the functions and usage scope of the embodiments of the present invention.

[0148] As Figure 4 shown, the components of the computing device 40 may include, but are not limited to: one or more processors or processing units 401, a system memory 402, and a bus 403 connecting different system components (including the system memory 402 and the processing unit 401).

[0149] Computing device 40 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by computing device 40, including volatile and non-volatile media, removable and non-removable media.

[0150] System memory 402 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 4021 and / or cache memory 4022. Computing device 40 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, ROM 4023 can be used to read and write to non-removable, non-volatile magnetic media ( Figure 4 not shown in the figure, commonly referred to as a "hard disk drive"). Although not shown in Figure 4 the figure, a disk drive can be provided for reading and writing to removable non-volatile disks (such as "floppy disks"), and an optical disk drive for reading and writing to removable non-volatile optical disks (such as CD-ROM, DVD-ROM or other optical media). In these cases, each drive can be connected to bus 403 through one or more data media interfaces. System memory 402 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the various embodiments of the present invention.

[0151] A program / utility 4025 having a set (at least one) of program modules 4024 can be stored, for example, in system memory 402, and such program modules 4024 include but are not limited to: an operating system, one or more application programs, other program modules, and program data, and the implementation of a network environment may be included in each or some combination of these examples. Program modules 4024 generally perform the functions and / or methods described in the embodiments of the present invention.

[0152] Computing device 40 can also communicate with one or more external devices 404 (such as a keyboard, pointing device, display, etc.). Such communication can be carried out through an input / output (I / O) interface 405. Moreover, computing device 40 can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN) and / or a public network, such as the Internet) through a network adapter 406. As Figure 4 shown, network adapter 406 communicates with other modules (such as processing unit 401, etc.) of computing device 40 through bus 403. It should be understood that although Figure 4 not shown in the figure, other hardware and / or software modules can be used in conjunction with computing device 40.

[0153] The processing unit 401 executes various functional applications and data processing by running the programs stored in the system memory 402. For example, vibration sensors are installed on the transformer body to collect steady-state and dynamic vibration signals, and the vibration signals are preprocessed; time-domain feature analysis is performed on the collected vibration signals, then frequency-domain feature analysis is performed on the vibration signals, and then the spectral complexity is analyzed; through historical data, the normal ranges of RMS value, peak factor, and odd-even harmonic ratio are set. If the data exceeds the set range, the influence of the fan is further analyzed; analysis is carried out in combination with the operation mode of the power system, and then fault discrimination is performed based on the machine learning classification algorithm. If the low-frequency even harmonics are enhanced and the spectral entropy decreases, it is DC bias magnetization; if the high-frequency odd harmonics are enhanced and the spectral entropy increases, it is harmonic injection; if DC bias magnetization is detected, DC isolation or DC compensation technology is used to adjust the HVDC operation mode to reduce the impact on the AC system; if harmonic injection is detected, an active harmonic filter is used to suppress harmonic interference, and the parameters of the new energy inverter are adjusted to optimize the current harmonic components.

[0154] The specific implementation methods of each step will not be repeated here. It should be noted that although several units / modules or sub-units / sub-modules of the synchronous escape wiring device based on multi-commodity flow are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to the embodiments of the present invention, the features and functions of the two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided and embodied by multiple units / modules.

[0155] In the description of the present invention, it should be noted that the terms "first", "second", and "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0156] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0157] In several embodiments provided by the present invention, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. Also, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be in electrical, mechanical, or other forms.

[0158] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed over multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0159] In addition, each functional unit in various embodiments of the present invention may be integrated in a processing unit, may exist physically separately for each unit, or two or more units may be integrated in one unit.

[0160] If the described function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0161] Finally, it should be noted that the above-described embodiments are only specific implementation manners of the present invention, used to illustrate the technical solutions of the present invention, and are not intended to limit it. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: any person skilled in the art within the technical scope disclosed by the present invention can still modify the technical solutions described in the foregoing embodiments, or can easily think of changes, or perform equivalent replacements for some of the technical features; and these modifications, changes, or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

[0162] In addition, although the operations of the method of the present invention are described in a specific order in the drawings, this does not require or imply that these operations must be performed in that specific order, or that all the shown operations must be performed to achieve the desired result. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step for execution, and / or one step may be decomposed into multiple steps for execution.

Claims

1. A transformer fault diagnosis method based on harmonic injection, characterized in that: include: Install vibration sensors on the transformer body to collect steady-state and dynamic vibration signals and pre-process the vibration signals; Based on the collected vibration signal, the time domain characteristics are analyzed, the frequency domain characteristics of the vibration signal are analyzed, and then the spectrum complexity is analyzed; Through historical data, set the normal range of RMS value, peak factor and odd-even harmonic ratio. If the data exceeds the set range, further analyze the impact of the fan; Combined with the power system operation mode, the fault is identified based on the machine learning classification algorithm. If the low-frequency even harmonics are enhanced and the spectral entropy is reduced, it is DC bias magnetism; if the high-frequency odd harmonics are enhanced and the spectral entropy is increased, it is harmonic injection. If DC bias is detected, DC isolation or DC compensation technology is used to adjust the HVDC operation mode to reduce the impact on the AC system; if harmonic injection is detected, active harmonic filters are used to suppress harmonic interference. Adjust the parameters of the new energy inverter and optimize the current harmonic components.

2. The transformer fault diagnosis method based on harmonic injection according to claim 1, characterized in that: The step of collecting steady-state and dynamic vibration signals and preprocessing the vibration signals includes: Set the sampling frequency f s =10kHz, steady-state signal data is collected for more than 60s in each working condition, and dynamic signal data is collected for more than 4s in each working condition; Use wavelet to perform 5-layer decomposition and use soft threshold filtering for high-frequency noise: , Where T is the threshold, σ is the noise standard deviation, and N is the signal length; Then the data is smoothed and windowed, and processed using sliding average filtering: , Where X′(n) is the filtered signal, X(n) is the original unprocessed signal, and N is the length of the sliding window.

3. The transformer fault diagnosis method based on harmonic injection according to claim 1, characterized in that: The method includes performing time domain feature analysis on the collected vibration signal, performing frequency domain feature analysis on the vibration signal, and then analyzing the spectrum complexity, including: Use Fourier transform to calculate the frequency spectrum of the vibration signal: ， By performing spectrum analysis on the vibration signal, the energy distribution of its different frequency components can be obtained; Define the odd harmonic energy ratio and even harmonic energy ratio: ， Among them, Hodd is the odd harmonic energy ratio, Heven is the even harmonic energy ratio, and kf0 is the different harmonic frequencies; Then the spectral complexity of the signal is calculated by spectral entropy: , Among them, P i is the normalized power of each frequency component. A higher spectral entropy indicates that the signal has more energy distributed over multiple frequencies.

4. The transformer fault diagnosis method based on harmonic injection according to claim 1, characterized in that: The normal range of the RMS value, peak factor and odd-even harmonic ratio is set. If the data exceeds the set range, further analysis of the fan impact is performed, including: For the collected vibration signal X(t), the parameter RMS value, peak factor, and odd-even harmonic ratio are calculated in each time window T to determine whether it exceeds the threshold: , If any parameter exceeds the set range, further analysis of the fan impact will be performed; Analyze the influence of the fan, observe whether the signal has low-frequency modulation, and calculate the autocorrelation function: , If periodic peaks occur, they are caused by the influence of the fan; Perform FFT analysis to see if there is a fan characteristic frequency f f And its frequency multiples: ， If f f If the proportion in the vibration spectrum is high, the fan has a greater impact.

5. The transformer fault diagnosis method based on harmonic injection according to claim 1, characterized in that: The fault identification is performed based on the machine learning classification algorithm. If the low-frequency even harmonics are enhanced and the spectral entropy is reduced, it is a DC bias magnetization; If the high-frequency odd harmonics are enhanced and the spectral entropy increases, it is harmonic injection, including: Select support vector machine for fault identification, and extract the feature X=[X1,X2,...,Xn], where X1 is the low-frequency even harmonic feature, X2 is the high-frequency odd harmonic feature, and X3 is the spectral entropy. Use the training data set to train the support vector machine model, and set the appropriate penalty parameter C and kernel function parameter γ during training: , Among them, α i is the Lagrange multiplier, y i is the label, x i is the training data, x is the new input data, and b is the bias term; Calculate the spectral entropy of the signal to see if it changes. If the spectral entropy decreases and the low-frequency even harmonics increase, it is determined to be DC bias magnetism; if the spectral entropy increases and the high-frequency odd harmonics increase, it is determined to be harmonic injection; If the low-frequency even-order harmonics are detected to be enhanced, the influence of the wind turbine is further eliminated; if the high-frequency odd-order harmonics are enhanced, it is determined whether there is a harmonic source in combination with the operating mode of the power system.

6. The transformer fault diagnosis method based on harmonic injection according to claim 1, characterized in that: If DC bias is detected, the HVDC operation mode is adjusted by using DC isolation or DC compensation technology to reduce the impact on the AC system, including: In HVDC converter stations, by adjusting the DC current I dc To restore transformer flux balance: ， Wherein, ΔHeven is the detected even harmonic increment, and K is the compensation coefficient; An active compensation device is installed at the neutral point of the transformer to detect the DC current I in real time. dc And inject reverse DC current: , Make the total DC component in the system approach zero; If the DC component is isolated for a long time, use DC isolation; if dynamic adjustment is required, use DC compensation; check H even Whether S decreases and whether S rises again ensures that the DC bias is suppressed.

7. The transformer fault diagnosis method based on harmonic injection according to claim 1, characterized in that: If harmonic injection is detected, an active harmonic filter is used to suppress harmonic interference, adjust the parameters of the new energy inverter, and optimize the current harmonic components, including: Install active harmonic filters, collect current signals in the power grid, detect harmonic components, calculate target compensation currents, generate reverse harmonic currents in real time, and output compensation signals through power conversion circuits to offset harmonic currents; Adjust the pulse width modulation strategy, optimize the switching frequency, and reduce the high-frequency harmonic content; adopt the pq theory to improve the harmonic suppression capability, adopt the phase-locked loop algorithm to improve the synchronization control accuracy, and reduce the harmonic interference of the inverter to the power grid.

8. A transformer fault diagnosis system based on harmonic injection, characterized in that: include: The acquisition and preprocessing module is used to install a vibration sensor on the transformer body, collect steady-state and dynamic vibration signals, and preprocess the vibration signals; The feature extraction module is used to perform time domain feature analysis on the collected vibration signal, then perform frequency domain feature analysis on the vibration signal, and then analyze the spectrum complexity; The setting analysis module is used to set the normal range of RMS value, peak factor and odd-even harmonic ratio through historical data. If the data exceeds the set range, the impact of the fan is further analyzed; The algorithm classification module is used to analyze the operation mode of the power system and then distinguish faults based on the machine learning classification algorithm. If the low-frequency even harmonics are enhanced and the spectral entropy is reduced, it is DC bias magnetism; if the high-frequency odd harmonics are enhanced and the spectral entropy is increased, it is harmonic injection; The fault optimization module is used to adjust the HVDC operation mode by using DC isolation or DC compensation technology to reduce the impact on the AC system if DC bias is detected; if harmonic injection is detected, an active harmonic filter is used to suppress harmonic interference, adjust the parameters of the new energy inverter, and optimize the current harmonic components.

9. A computing device, comprising: at least one processor, memory, and input-output unit; The memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the steps of the transformer fault diagnosis method based on harmonic injection as described in any one of claims 1 to 7.

10. A computer-readable storage medium comprising instructions, which, when executed on a computer, enables the computer to execute the steps of the transformer fault diagnosis method based on harmonic injection as claimed in any one of claims 1 to 7.

Images