Transformer fault diagnosis method and system based on vibration mechanisms of iron core and winding

Abstract

A transformer fault diagnosis method based on vibration mechanisms of an iron core and a winding. The method comprises: performing signal processing and feature extraction by means of calling simulation results of an iron core and a winding; and establishing a transformer fault diagnosis model. Thus, the accurate diagnosis of a transformer fault is realized. A conventional fault diagnosis method is improved, and chaotic mapping and an opposition-based learning strategy are introduced to optimize the diagnosis model, thereby improving the accuracy of diagnosis. Compared with the conventional method, a fault type and a fault degree of a transformer can be more accurately determined, so as to facilitate an engineering technician in maintenance, thereby improving the operation reliability of the transformer. In addition, a transformer fault diagnosis system based on vibration mechanisms of an iron core and a winding, and a computing device and a computer-readable storage medium are further provided.

Description

Method and system for diagnosing transformer faults based on vibration mechanism of core and winding Technical Field

[0001] The present invention relates to the field of electric power technology, and in particular to a method and system for diagnosing transformer faults based on the vibration mechanism of an iron core and a winding. Background Art

[0002] In recent years, with the construction of new power systems and the improvement of power grid intelligence, the safety and stability of transformers, which are important electrical equipment for electromagnetic conversion and voltage transformation in power systems, are particularly important. As new power systems develop towards a high proportion of new energy, the number of newly added transformers in operation has increased sharply. Due to the complex structure, long working cycle, poor operating environment and other reasons of transformers, the large-scale investment in transformers has brought huge safety hazards to the power grid. Ensuring the safe operation of transformers is one of the important tasks of power grid workers.

[0003] Establishing a transformer fault diagnosis model is the ultimate goal of exploring transformer vibration problems, and using vibration signals to effectively identify transformer internal mechanical faults. However, there are still many problems in applying the vibration method to actual detection work. The exploration of the internal vibration mechanism of the transformer is not deep enough, the exploration of various mechanical states is too little, and the understanding of the characteristics of the transformer vibration signal is insufficient. When using the vibration method to identify the transformer fault state, it still mainly relies on actual work experience, and the identification results are not ideal. In recent years, with the development of artificial intelligence technology, it has become a reality to use artificial intelligence to judge transformer faults without manual work, but artificial intelligence technology requires data support, and research institutions and industries have very little data on transformer vibration research. With the help of existing data, an artificial intelligence diagnostic model is established, which makes it even more difficult to judge faults without manual work. Based on the research on transformer core and winding vibration, and the obtained simulation data, artificial intelligence technology is used to realize transformer fault diagnosis.

[0004] Summary of the Invention

[0005] In view of the problems existing in the existing transformer fault diagnosis method and system based on the vibration mechanism of the iron core and winding, the present invention is proposed.

[0006] Therefore, the purpose of the present invention is to provide a method and system for diagnosing transformer faults based on the vibration mechanism of the iron core and winding. When diagnosing the transformer vibration mechanism, it is possible to achieve transformer fault diagnosis by obtaining simulation data and using artificial intelligence technology without manual fault judgment.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a method for diagnosing transformer faults based on the vibration mechanism of the core and winding, which includes calling finite element simulation results of the core and winding, selecting the core and winding vibration signals for combined superposition based on the proximity principle, and obtaining the synthetic vibration wave signals of the corresponding transformer core and winding through combined superposition; using the mathematical tool of the vibration eigenvalue method to process the vibration wave signal, extracting the eigenvalue of the processed signal based on the transformer vibration mechanism, and performing frequency domain transformation on the transformer based on the frequency domain information after Fourier transformation; establishing a transformer fault diagnosis model, introducing chaotic mapping and reverse learning strategy into the manta ray algorithm to optimize the MRFO initial population, using the optimized MRFO initial population as the parameter seeking method of the BP neural network, and establishing a transformer fault diagnosis model based on the improved manta ray algorithm to optimize the BP.

[0009] As a preferred solution of the method for diagnosing transformer faults based on the vibration mechanism of the iron core and windings described in the present invention, the combined superposition includes obtaining the time domain waveforms of the vibration signals of the transformer at different measuring points, the time domain waveforms of the vibration signals include vibration waveforms of mild, moderate, and severe iron core looseness and mild, moderate, and severe winding looseness, and the vibration wave signal includes arranging the vibration waveform measuring points in sequence on the transformer phase line cake, and synthesizing the vibration wave signal through signal superposition.

[0010] As a preferred solution of the method for diagnosing transformer faults based on the vibration mechanism of the core and windings of the present invention, the feature value extraction of the processed signal includes judging whether the transformer is faulty by comparing the feature value differences of each state. Multiple feature values ​​need to be extracted for judgment. The specific steps are as follows:

[0011] Extract the fundamental frequency component amplitude p1;

[0012] The extracted fundamental frequency component proportional coefficient p2 is:

[0013] in,

[0014] The proportion of the harmonic component vibration signal of the total harmonic distortion rate THD is:

[0015] Among them, A 100 is the 100Hz component amplitude, A i is the amplitude of the harmonic component of the multiple of the fundamental frequency of 100 Hz, and N is the harmonic order;

[0016] The weighted total harmonic distortion rate p3 is:

[0017] Among them, q i is the harmonic weight coefficient, which is q i =i, thus achieving high sensitivity to higher harmonics;

[0018] The extracted vibration information entropy p4 is:

[0019] Among them, B f is the proportion coefficient of 50Hz and 50Hz multiplier components, and p4 is the vibration information entropy.

[0020] As a preferred solution of the method for diagnosing transformer faults based on the vibration mechanism of the iron core and windings of the present invention, the manta ray algorithm includes the foraging strategies of the manta ray group, namely, the chain foraging strategy, the cyclone foraging strategy, and the somersault foraging strategy. The mathematical model of the foraging strategy of the manta ray group is as follows:

[0021] The chain foraging strategy is:

[0022] When i=1,

[0023] When i=2, 3...N,

[0024] in, is the position of the i-th manta ray in the d-dimensional space at the t-th iteration, is the optimal solution of the manta ray population in the d-dimensional space, α is the weight coefficient, and r is a random number;

[0025] The cyclone foraging strategy is:

[0026] When i=1,

[0027] When i=2,3...N,

[0028] Among them, β is the coefficient of the whirlwind foraging strategy, T, t are the maximum number of iterations and the current number of iterations respectively, r1, r are random numbers;

[0029] At this time, the cyclone foraging strategy uses cyclone characteristics to develop the area near the current optimal solution. If the current optimal solution is replaced by a random position in the parameter space, the mathematical model of the cyclone foraging as an exploration strategy in the parameter space is:

[0030] When i=1,

[0031] When i=2,3...N,

[0032] in, are the lower and upper bounds of the d-dimensional parameter space, is the randomly generated position in the parameter space at the tth iteration;

[0033] The somersault foraging strategy is:

[0034] Among them, S is the somersault range coefficient, and r is a random number.

[0035] As a preferred solution of the method for diagnosing transformer faults based on the vibration mechanism of the iron core and windings of the present invention, the method of introducing chaotic mapping and reverse learning strategy to optimize the MRFO initial population includes using Logistic chaotic mapping and combining reverse learning strategy to screen the initial population. The specific steps are as follows:

[0036] The mathematical expression of chaotic mapping using Logistic is:

[0037] in, μ∈[0, 4], when μ is 3.99, the Logistic mapping achieves a chaotic effect;

[0038] The method of combining the reverse learning strategy to select the initial population includes combining a multi-stage population initialization algorithm, and the specific steps are as follows:

[0039] Use Logistic mapping to generate N individuals to form the original population N1;

[0040] The reverse population N2 is formed by using the reverse solution:

[0041] in, For the initial individual The reverse individuals are merged into the original population and the reverse population, and the final initial population consisting of the first N individuals with the smallest fitness is calculated.

[0042] As a preferred solution of the method for diagnosing transformer faults based on the vibration mechanism of the iron core and windings described in the present invention, the parameter determination method includes training data and test data for normal and faulty transformers, and the training data includes assigning the initial population initial solution of the MRFO algorithm to the BP neural network to establish an MRFO-BP classification model.

[0043] As a preferred solution of the method for diagnosing transformer faults based on the vibration mechanism of the iron core and windings described in the present invention, wherein: the establishment of the transformer fault diagnosis model includes initializing the MRFO-BP classification model assigned by the population, calculating the model fitness value, and calculating the model fitness value. If the fitness value is satisfied, a fault diagnosis model is obtained. If not satisfied, the fault diagnosis model is returned to the MRFO algorithm for iterative solution. The iterative solution includes normalizing the transformer vibration characteristic value and adding a fitness function. The specific steps are as follows:

[0044] The normalization formula is:

[0045] Among them, P i is the vibration characteristic value, P max , P min are the maximum and minimum values ​​of the vibration eigenvalues, is the normalized eigenvalue;

[0046] The fitness function is:

[0047] Among them, y I is the state code corresponding to the real state of the transformer, y i Predict the type of fault for the model;

[0048] When the model is in a highly loose fault state, if it is misjudged as a normal state, the fitness function will rise, and the looseness of the fault will be determined according to the degree of increase;

[0049] When the model is in a serious loose fault, the fitness function rises to the peak position, triggering an alarm to prompt the operator to deal with the fault in time.

[0050] In a second aspect, an embodiment of the present invention provides a system for diagnosing transformer faults based on the vibration mechanism of the core and winding, which includes: a combination and superposition module, which calls the finite element simulation results of the core and winding, selects the core and winding vibration signals for combination and superposition based on the proximity principle, and obtains the synthetic vibration wave signal of the corresponding transformer core and winding through combination and superposition; an extraction module, which uses the mathematical tool of the vibration eigenvalue method to process the vibration wave signal, extracts the eigenvalue of the processed signal based on the transformer vibration mechanism, and performs frequency domain transformation on the transformer based on the frequency domain information after Fourier transformation; an optimization module, which establishes a transformer fault diagnosis model, introduces chaotic mapping and reverse learning strategy into the manta ray algorithm to optimize the MRFO initial population, uses the optimized MRFO initial population as the parameter seeking method of the BP neural network, and establishes a transformer fault diagnosis model based on the improved manta ray algorithm to optimize the BP.

[0051] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, it implements any step of the above-mentioned method for diagnosing transformer faults based on the vibration mechanism of the core and winding.

[0052] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the above-mentioned method for diagnosing transformer faults based on the vibration mechanism of the core and winding is implemented.

[0053] The beneficial effects of the present invention are as follows: the present invention calls the simulation results of the iron core and the winding, performs signal processing and feature extraction, establishes a transformer fault diagnosis model, and realizes accurate diagnosis of transformer faults. The traditional fault diagnosis method is improved, and chaos mapping and reverse learning strategies are introduced to optimize the diagnosis model, thereby improving the accuracy of diagnosis. Compared with the traditional method, the present solution can more accurately judge the fault type and fault degree of the transformer, facilitates maintenance by engineering and technical personnel, and improves the operational reliability of the transformer. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0055] FIG1 is a specific flow chart of a method and system for diagnosing transformer faults based on the vibration mechanism of the core and windings, provided by one embodiment of the present invention.

[0056] FIG2 is a Logistic map chaos effect diagram of a method and system for diagnosing transformer faults based on the vibration mechanism of the iron core and windings provided by an embodiment of the present invention.

[0057] FIG3 is an iterative diagram of a diagnostic model of a method and system for diagnosing transformer faults based on the vibration mechanism of the core and windings provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0059] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0060] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0061] The present invention is described in detail with reference to schematic diagrams. For ease of illustration, cross-sectional views of device structures may be partially enlarged and not to scale when describing embodiments of the present invention. Furthermore, the schematic diagrams are merely illustrative and should not limit the scope of the present invention. Furthermore, in actual production, the three-dimensional dimensions of length, width, and depth should be included.

[0062] In the description of the present invention, it should be noted that the terms "upper, lower, inner, and outer" and other references to orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0063] In this disclosure, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they may refer to fixed, removable, or integral connections. They may also refer to mechanical, electrical, or direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in this disclosure.

[0064] Example 1

[0065] 1 to 3 , which are the first embodiment of the present invention, provide a method for diagnosing transformer faults based on the vibration mechanism of the core and windings, including:

[0066] S1: Call the finite element simulation results of the core and winding, select the core and winding vibration signals based on the proximity principle for combined superposition, and obtain the synthetic vibration wave signals of the corresponding transformer core and winding through combined superposition.

[0067] Among them, the combined superposition includes obtaining the time domain waveforms of the vibration signals of the transformer at different measuring points. The time domain waveforms of the vibration signals include vibration waveforms of mild, moderate and severe core looseness and mild, moderate and severe winding looseness. The vibration wave signal includes arranging the vibration waveform measuring points in sequence on the transformer phase line cake and synthesizing the vibration wave signal through signal superposition.

[0068] Furthermore, when the transformer is operating normally at rated load, its vibration mainly comes from the core vibration and winding vibration. When superimposing the core and winding vibration signals, it is necessary to consider the selection of measurement points. The vibration signals of the winding and core with the smallest vibration attenuation index are selected for superposition based on the principle of proximity. The solution results obtained by the finite element simulation of the core and winding are called. The distribution of transformer vibration data is shown in Table 1 below:

[0069] Table 1 Transformer vibration data distribution table

[0070] In the actual vibration signal acquisition, due to measurement errors and various types of noise interference, in order to simulate the real signal on site, a Gaussian distribution of N(0, 0.6) is applied to the superimposed 22*7 transformer vibration data signals. 2 ) of white noise.

[0071] Furthermore, the transformer vibration wave signal is mainly distributed within 500Hz, the 100Hz fundamental frequency does not play the role of the main frequency, and 200Hz occupies a large level in all states. When the transformer core fails, its 500-1000Hz high frequency band will undergo more obvious changes. When the winding is highly loose, the 250Hz component will be greatly attenuated. In addition, it can be seen from the time domain waveform that after the noise signal is superimposed, the vibration signal of the core is greatly affected, and the waveform complexity is significantly increased, but the noise cannot be completely avoided.

[0072] S2: Use the mathematical tools of the vibration eigenvalue method to process the vibration wave signal, extract the eigenvalue of the processed signal based on the transformer vibration mechanism, and perform frequency domain transformation on the transformer based on the frequency domain information after Fourier transformation.

[0073] Furthermore, feature value extraction is performed on the processed signal, including comparing the feature value differences of each state to determine whether the transformer is faulty. Multiple feature values ​​need to be extracted for judgment. The specific steps are as follows:

[0074] Extract the fundamental frequency component amplitude p1;

[0075] The extracted fundamental frequency component proportional coefficient p2 is:

[0076] in,

[0077] The proportion of the harmonic component vibration signal of the total harmonic distortion rate THD is:

[0078] Among them, A 100 is the 100Hz component amplitude, A i is the amplitude of the harmonic component of the multiple of the fundamental frequency of 100 Hz, and N is the harmonic order;

[0079] The weighted total harmonic distortion rate p3 is:

[0080] Among them, q i is the harmonic weight coefficient, which is q i =i, thus achieving high sensitivity to higher harmonics;

[0081] The extracted vibration information entropy p4 is:

[0082] Among them, B f is the proportion coefficient of the 50Hz and 50Hz multiplier components, and p4 is the vibration information entropy, which is not less than 0. As can be seen from the expression, assuming the vibration spectrum contains only a 50Hz frequency component, its vibration information entropy is 0. When the frequency is dispersed, the smaller the amplitude of each frequency component, the smaller the proportion coefficient of each component, and the greater the vibration information entropy. Therefore, using vibration information entropy as a vibration characteristic value can reflect the overall dispersion characteristics of the spectrum.

[0083] S3: Establish a transformer fault diagnosis model, introduce chaotic mapping and reverse learning strategy into the manta ray algorithm to optimize the MRFO initial population, use the optimized MRFO initial population as the parameter method of the BP neural network, and establish a transformer fault diagnosis model based on the improved manta ray algorithm to optimize the BP.

[0084] The manta ray algorithm includes the foraging strategies of manta ray groups, namely the chain foraging strategy, the whirlwind foraging strategy, and the somersault foraging strategy. The mathematical model of the manta ray group's foraging strategy is as follows:

[0085] The chain foraging strategy is:

[0086] When i=1,

[0087] When i=2,3...N,

[0088] in, is the position of the i-th manta ray in the d-dimensional space at the t-th iteration, is the optimal solution of the manta ray population in the d-dimensional space, α is the weight coefficient, and r is a random number;

[0089] The cyclone foraging strategy is:

[0090] When i=1,

[0091] When i=2, 3...N,

[0092] Among them, β is the coefficient of the whirlwind foraging strategy, T, t are the maximum number of iterations and the current number of iterations respectively, r1, r are random numbers;

[0093] At this time, the cyclone foraging strategy uses cyclone characteristics to develop the area near the current optimal solution. If the current optimal solution is replaced by a random position in the parameter space, the mathematical model of the cyclone foraging as an exploration strategy in the parameter space is:

[0094] When i=1,

[0095] When i=2,3...N,

[0096] in, are the lower and upper bounds of the d-dimensional parameter space, is the randomly generated position in the parameter space at the tth iteration;

[0097] The somersault foraging strategy is:

[0098] Among them, S is the somersault range coefficient, and r is a random number.

[0099] S3.1: Introducing Chaotic Mapping and Reverse Learning Strategy to Optimize the MRFO Initial Population This includes using Logistic to perform chaotic mapping and combining reverse learning strategy to screen the initial population. The specific steps are as follows:

[0100] The mathematical expression of chaotic mapping using Logistic is:

[0101] in, μ∈[0, 4], when μ is 3.99, the Logistic mapping achieves a chaotic effect;

[0102] The combination of reverse learning strategy to select the initial population includes combining the multi-stage population initialization algorithm. The specific steps are as follows:

[0103] Use Logistic mapping to generate N individuals to form the original population N1;

[0104] The reverse population N2 is formed by using the reverse solution:

[0105] in, For the initial individual The reverse individuals are merged into the original population and the reverse population, and the final initial population consisting of the first N individuals with the smallest fitness is calculated.

[0106] S3.2: The parameter determination method includes training data and test data for normal and faulty transformers. The training data includes assigning the initial population solution of the MRFO algorithm to the BP neural network to establish the MRFO-BP classification model.

[0107] Furthermore, a multi-stage algorithm combining logistic mapping and backpropagation learning (OBL) was used to initialize the population. The generated initial population was used as the initial population for the MRFO algorithm. The initial solution of the initial population of the MRFO algorithm was assigned to the BP neural network as the weight and bias of the BP model, replacing the backpropagation algorithm of the BP neural network to establish the MRFO-BP classification model. The distribution of training data and test data is shown in Table 2 below:

[0108] Table 2 Distribution of training data and test data

[0109] Seven different looseness states of the transformer core and winding are shown. 17 training samples and 5 test samples are collected for each state to establish a transformer fault diagnosis model to realize the identification and diagnosis of different degrees of looseness faults of the core and winding.

[0110] S3.3: Establishing a transformer fault diagnosis model includes initializing the MRFO-BP classification model assigned by the population, calculating the model fitness value, and calculating the model fitness value. If the fitness value is satisfied, the fault diagnosis model is obtained. If not, the fault diagnosis model is returned to the MRFO algorithm for iterative solution. The iterative solution includes normalizing the transformer vibration eigenvalues ​​and adding the fitness function. The specific steps are as follows:

[0111] The normalization formula is:

[0112] Among them, P i is the vibration characteristic value, P max , P min are the maximum and minimum values ​​of the vibration eigenvalues, is the normalized eigenvalue;

[0113] The fitness function is:

[0114] Among them, y I is the state code corresponding to the real state of the transformer, y i Predict the type of fault for the model;

[0115] When the model is in a highly loose fault state, if it is misjudged as a normal state, the fitness function will rise, and the looseness of the fault will be determined according to the degree of increase;

[0116] When the model is in a serious loose fault, the fitness function rises to the peak position, triggering an alarm to prompt the operator to deal with the fault in time.

[0117] Furthermore, the normalized fundamental frequency component amplitude, fundamental frequency component proportional coefficient, weighted total harmonic distortion rate and vibration information entropy are used as the input of the model. In order to adapt to the digital output of the model, the normal and fault states of the transformer are digitally encoded. The fault coding is shown in Table 3 below:

[0118] Table 3 Fault code data table

[0119] Seven fault states of the transformer can be detected, including four types of loose core faults and three types of loose winding faults. A unique status code is set for each fault state to identify the fault state of the transformer in the diagnosis and monitoring system. The correspondence between the status code and the fault type can realize automatic identification of different faults.

[0120] The above-mentioned unit modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of the above-mentioned modules.

[0121] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0122] In summary, the present invention calls the simulation results of the iron core and winding, performs signal processing and feature extraction, establishes a transformer fault diagnosis model, and realizes accurate diagnosis of transformer faults. It improves the traditional fault diagnosis method and introduces chaos mapping and reverse learning strategy to optimize the diagnosis model, thereby improving the accuracy of diagnosis. Compared with the traditional method, this solution can more accurately judge the fault type and fault degree of the transformer, facilitates maintenance by engineering and technical personnel, and improves the operational reliability of the transformer.

[0123] Example 2

[0124] 1 to 3 , which are a second embodiment of the present invention, a multi-source power grid information fusion method based on the Internet of Things is provided. In order to verify the beneficial effects of the present invention, a scientific demonstration is conducted through simulation experiments.

[0125] Under the optimization of the improved manta ray algorithm, the model stabilized after 169 iterations, the model converged quickly, and achieved a small error. The fitness value reached a minimum of 0.2689, and the accuracy reached the target. The transformer fault diagnosis model training was completed. In order to verify the accuracy of the established transformer fault diagnosis model, 35 test data were input into the model for prediction.

[0126] Among the 35 test samples, the model made only 6 incorrect predictions, indicating good prediction results. Among them, all the moderate loose core faults were predicted correctly. However, there were misjudgments when predicting mild loose windings, misjudging mild looseness as normal and severe loose windings as moderate looseness. At the same time, the model also made misjudgments between normal and highly loose cores. This misjudgment was more serious than the previous one. It may be due to the similar vibration spectra caused by different measurement points selected for the data. Therefore, in related research, attention should be paid to the obvious impact of different measurement points on the signal. The actual data prediction results are shown in Table 4 below:

[0127] Table 4 Actual forecast data analysis table

[0128] The data used in the study is simulation data. To corroborate with actual cases and further illustrate the reliability of the algorithm, this article provides three sets of data: normal state, winding looseness, and core looseness data, and performs model fault prediction. That is, the model's prediction results are slightly different from the actual results.

[0129] The improved BP model achieved improved diagnostic results on the samples. The PSO optimized BP model failed to predict 9 times, and the BA optimized BP model failed to predict 8 times. In addition, the BA optimized BP model correctly predicted all the loose core heights. However, it also misjudged the loose winding height as loose core height. The diagnostic results of the corresponding fault diagnosis model are shown in Table 5 below:

[0130] The manta ray algorithm optimized BP model using Logistic mapping and reverse learning strategy has the highest accuracy, which is 20.01%, 8.58% and 5.72% higher than the traditional BP, particle swarm algorithm optimized BP and bat algorithm optimized BP respectively. It is an active and effective diagnostic model.

[0131] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for diagnosing transformer faults based on the vibration mechanism of the core and windings, characterized in that: include, The finite element simulation results of the core and winding are called, and the vibration signals of the core and winding are selected for combination and superposition based on the principle of proximity, so as to obtain the synthetic vibration wave signals of the corresponding transformer core and winding through combination and superposition; The vibration wave signal is processed using the mathematical tool of the vibration eigenvalue method, the eigenvalue of the processed signal is extracted based on the transformer vibration mechanism, and the transformer is transformed in the frequency domain based on the frequency domain information after Fourier transformation; A transformer fault diagnosis model is established. Chaotic mapping and reverse learning strategy are introduced into the manta ray algorithm to optimize the MRFO initial population. The optimized MRFO initial population is used as the parameter seeking method of the BP neural network, and a transformer fault diagnosis model based on the improved manta ray algorithm and optimized BP is established.

2. The method for diagnosing transformer faults based on the vibration mechanism of the core and windings according to claim 1, characterized in that: The combined superposition includes obtaining the time domain waveform of the vibration signal of the transformer at different measuring points, the time domain waveform of the vibration signal includes vibration waveforms of slight, moderate and severe core looseness and slight, moderate and severe winding looseness, and the vibration wave signal includes arranging the vibration waveform measuring points in sequence on the transformer phase line cake, and synthesizing the vibration wave signal by signal superposition.

3. The method for diagnosing transformer faults based on the vibration mechanism of the core and windings as claimed in claim 2, characterized in that: The feature value extraction of the processed signal includes judging whether the transformer is faulty by comparing the feature value differences of each state. It is necessary to extract multiple feature values ​​for judgment. The specific steps are as follows: Extract the fundamental frequency component amplitude p1; The extracted fundamental frequency component proportional coefficient p2 is: in, The proportion of the harmonic component vibration signal of the total harmonic distortion rate THD is: Among them, A 100 is the amplitude of the 100Hz component, Ai is the amplitude of the harmonic component of the multiple of the 100Hz fundamental frequency, and N is the harmonic order; The weighted total harmonic distortion rate p3 is: Among them, qi is the harmonic weight coefficient, and the value is qi=i here, so as to achieve high sensitivity of high-order harmonics; The extracted vibration information entropy p4 is: Among them, Bf is the proportion coefficient of 50Hz and 50Hz double frequency components, and p4 is the vibration information entropy.

4. The method for diagnosing transformer faults based on the vibration mechanism of the core and windings as claimed in claim 3, characterized in that: The manta ray algorithm includes the foraging strategies of the manta ray group, namely the chain foraging strategy, the whirlwind foraging strategy and the somersault foraging strategy. The mathematical model of the foraging strategy of the manta ray group is as follows: The chain foraging strategy is: in, is the position of the ith manta ray in the d-dimensional space at the tth iteration, is the optimal solution of the manta ray population in the d-dimensional space, α is the weight coefficient, and r is a random number; The cyclone foraging strategy is: Among them, β is the coefficient of the whirlwind foraging strategy, T, t are the maximum number of iterations and the current number of iterations respectively, r1, r are random numbers; At this time, the cyclone foraging strategy uses cyclone characteristics to develop the area near the current optimal solution. If the current optimal solution is replaced by a random position in the parameter space, the cyclone foraging is used as an exploration strategy in the parameter space. for: in, are the lower and upper bounds of the d-dimensional parameter space, is the randomly generated position in the parameter space at the tth iteration; The somersault foraging strategy is: Among them, S is the somersault range coefficient and r is a random number.

5. The method for diagnosing transformer faults based on the vibration mechanism of the core and windings as claimed in claim 4, characterized in that: The introduction of chaotic mapping and reverse learning strategy to optimize the MRFO initial population includes using Logistic to perform chaotic mapping and combining the reverse learning strategy to screen the initial population. The specific steps are as follows: The mathematical expression of chaotic mapping using Logistic is: in, μ∈[0, 4], when μ is 3.99, the Logistic mapping achieves a chaotic effect; The method of combining the reverse learning strategy to select the initial population includes combining a multi-stage population initialization algorithm, and the specific steps are as follows: Use Logistic mapping to generate N individuals to form the original population N1; The reverse population N2 is formed by using the reverse solution: in, For the initial individual The reverse individuals are combined into the original population and the reverse population, and the final initial population consisting of the first N individuals with the smallest fitness is calculated.

6. The method for diagnosing transformer faults based on the vibration mechanism of the core and windings as claimed in claim 5, characterized in that: The parameter seeking method includes training data and test data for normal and faulty transformers. The training data includes assigning the initial population initial solution of the MRFO algorithm to the BP neural network to establish a MRFO-BP classification model.

7. The method for diagnosing transformer faults based on the vibration mechanism of the core and windings according to claim 6, characterized in that: The establishment of the transformer fault diagnosis model includes initializing the MRFO-BP classification model assigned by the population, calculating the model fitness value, and calculating the model fitness value. If the fitness value is satisfied, the fault diagnosis model is obtained. If not satisfied, the fault diagnosis model is returned to the MRFO algorithm for iterative solution. The iterative solution includes normalizing the transformer vibration characteristic value and adding the fitness function. The specific steps are as follows: The normalization formula is: Among them, Pi is the vibration characteristic value, P max , P min are the maximum and minimum values ​​of the vibration eigenvalues, is the normalized eigenvalue; The fitness function is: Among them, yI is the state code corresponding to the real state of the transformer, and yi is the fault type predicted by the model; When the model is in a highly loose fault, if it is misjudged as a normal state, the fitness function will rise, and the looseness of the fault will be determined according to the degree of increase; When the model is in a serious loose fault, the fitness function rises to the peak position, triggering an alarm to prompt the operator to deal with the fault in time.

8. A system for diagnosing transformer faults based on the vibration mechanism of the iron core and the winding, based on the method for diagnosing transformer faults based on the vibration mechanism of the iron core and the winding according to any one of claims 1 to 7, characterized in that: include, A combined superposition module, which calls the finite element simulation results of the core and winding, selects the core and winding vibration signals for combined superposition based on the proximity principle, and obtains the synthetic vibration wave signals of the corresponding transformer core and winding through combined superposition; An extraction module, which uses the mathematical tool of the vibration eigenvalue method to process the vibration wave signal, extracts the eigenvalue of the processed signal based on the transformer vibration mechanism, and performs frequency domain transformation on the transformer based on the frequency domain information after Fourier transformation; The optimization module establishes a transformer fault diagnosis model, introduces chaotic mapping and reverse learning strategy into the manta ray algorithm to optimize the MRFO initial population, uses the optimized MRFO initial population as the parameter seeking method of the BP neural network, and establishes a transformer fault diagnosis model based on the improved manta ray algorithm to optimize the BP.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for diagnosing transformer faults based on the vibration mechanism of the core and the windings according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for diagnosing transformer faults based on the vibration mechanism of the core and the windings according to any one of claims 1 to 7 are implemented.

Images