Fault detection method and device of transformer and readable storage medium

Abstract

The invention discloses a transformer fault detection method and device and a readable storage medium, and relates to the technical field of equipment fault detection. The transformer fault detection method comprises the steps of obtaining first sound data of a transformer; performing feature extraction and data type conversion processing on the first sound data to obtain model training data; based on the model training data, establishing a fault detection model corresponding to the transformer; in the operation process of the transformer, the sound of the transformer is collected to obtain second sound data of the transformer; and inputting the second sound data into the fault detection model to obtain a fault detection result output by the fault detection model. According to the invention, the transformer fault detection accuracy is improved.

Description

Technical Field

[0001] This application relates to the field of equipment fault detection technology, and in particular to a fault detection method, apparatus and readable storage medium for a transformer. Background Technology

[0002] Currently, real-time inspections are necessary during transformer operation to ensure its normal operation. At present, various fault detection methods exist for transformers, such as noise signal detection and voltage signal detection. However, existing fault detection methods suffer from low accuracy. Summary of the Invention

[0003] This application provides a method, apparatus, and readable storage medium for detecting transformer faults, which addresses technical problems such as low detection accuracy in the prior art.

[0004] A first aspect of this application provides a method for detecting faults in a transformer, the method comprising: Acquire the first sound data of the transformer; Feature extraction and data type conversion are performed on the first sound data to obtain model training data; Based on the model training data, a fault detection model for transformers is established. During the operation of the transformer, the sound of the transformer is collected to obtain the transformer's second sound data; The second sound data is input into the fault detection model to obtain the fault detection result output by the fault detection model.

[0005] In some embodiments, feature extraction and data type conversion are performed on the first sound data to obtain model training data, including: Variational mode decomposition is performed on the first sound data to obtain the feature data in the first sound data; The feature data is processed by continuous wavelet transform to convert it into a time-frequency grayscale image; Image enhancement processing is performed on the time-frequency grayscale image to obtain model training data.

[0006] In some embodiments, variational mode decomposition is performed on the first sound data to obtain feature data in the first sound data, including: Determine the optimal constraint values ​​corresponding to the preset first variational mode model; Based on the optimal constraint values, the first variational mode model is updated to obtain the second variational mode model; The first sound data is input into the second variational mode model to obtain the feature data output by the second variational mode model.

[0007] In some embodiments, performing continuous wavelet transform processing on the feature data to convert the feature data into a time-frequency grayscale image includes: Based on the preset first transformation formula, a second transformation formula corresponding to the feature data is established. The feature data is input into the second transformation formula to obtain a time-frequency grayscale image.

[0008] In some embodiments, image enhancement processing is performed on the time-frequency grayscale image to obtain model training data, including: The time-frequency grayscale image is flipped to obtain the first image data; The pixels in the time-frequency grayscale image are transposed to obtain the second image data; The time-frequency grayscale image is smoothed and blurred to obtain the third image data; The first image data, the second image data, and the third image data are combined into model training data.

[0009] In some embodiments, a fault detection model for the transformer is established based on model training data, including: Input the model training data into the pre-defined convolutional neural network; Data training is performed on the convolutional neural network to obtain a fault detection model.

[0010] In some embodiments, acquiring first audio data of the transformer includes: The transformer's sound was collected under different operating conditions to obtain the transformer's third sound data; Based on the transformer's operating status, the third sound data is annotated to obtain the first sound data.

[0011] The transformer fault detection method in this embodiment improves the accuracy of model training data by performing feature extraction and data type conversion on the first sound data of the transformer. Then, a fault detection model is established using the model training data, which improves the accuracy of the fault detection model. The fault detection model improves the accuracy of transformer fault detection.

[0012] A second aspect of this application provides a transformer fault detection device, comprising: The acquisition unit is used to acquire the first sound data of the transformer; The processing unit is used to perform feature extraction and data type conversion on the first sound data to obtain model training data; The processing unit is also used to establish a fault detection model for the transformer based on the model training data; The acquisition unit is also used to collect the sound of the transformer during its operation to obtain the transformer's second sound data. The processing unit is also used to input the second sound data into the fault detection model to obtain the fault detection result output by the fault detection model.

[0013] The transformer fault detection device in this embodiment improves the accuracy of the model training data by performing feature extraction and data type conversion on the first sound data of the transformer. Then, it establishes a fault detection model using the model training data, thereby improving the accuracy of the fault detection model and enhancing the accuracy of transformer fault detection.

[0014] A third aspect of this application provides another transformer fault detection device, including a processor and a memory. The memory stores a computer program, which, when executed by the processor, implements the steps of the transformer fault detection method as described in any of the above embodiments. Therefore, this transformer fault detection device possesses all the beneficial effects of the transformer fault detection method in any of the above embodiments, and will not be elaborated further here.

[0015] A fourth aspect of this application provides a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the transformer fault detection method as described in any of the above embodiments. Therefore, this readable storage medium possesses all the beneficial effects of the transformer fault detection method in any of the above embodiments, which will not be elaborated further here. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating a transformer fault detection method provided in an embodiment of this application; Figure 2 A functional block diagram of a transformer fault detection device provided in an embodiment of this application; Figure 3 This is a structural block diagram of a transformer fault detection device provided in an embodiment of this application; Detailed Implementation To better understand the technical solutions provided in the embodiments of this specification, the technical solutions of the embodiments of this specification will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of this specification and the specific features in the embodiments are detailed descriptions of the technical solutions of the embodiments of this specification, rather than limitations on the technical solutions of this specification. In the absence of conflict, the embodiments of this specification and the technical features in the embodiments can be combined with each other.

[0018] In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, without necessarily requiring or implying any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The term "two or more" includes two or more cases.

[0019] In some embodiments, such as Figure 1 As shown, an embodiment of this application provides a transformer fault detection method, including: Step S101: Obtain the first sound data of the transformer; Step S102: Perform feature extraction and data type conversion on the first sound data to obtain model training data; Step S103: Based on the model training data, establish a fault detection model corresponding to the transformer; Step S104: During the operation of the transformer, the sound of the transformer is collected to obtain the second sound data of the transformer; Step S105: Input the second sound data into the fault detection model to obtain the fault detection result output by the fault detection model.

[0020] In this embodiment, a fault detection method for transformers is proposed, which is used to detect the fault type of the transformer during operation, and then to carry out fault repair on the transformer in order to maintain the normal operation of the transformer.

[0021] For example, the transformer may specifically be a DC-DC transformer.

[0022] For example, the transformer may specifically be an AC-AC transformer.

[0023] The first sound data of the transformer is collected, which is the historical sound data of the transformer.

[0024] For example, the first sound data may include the sound data of the transformer operating under overload conditions.

[0025] For example, the first sound data may include the operating sound data of the transformer in a partial discharge state.

[0026] For example, the first audio data may include the operating audio data of the transformer under overvoltage conditions.

[0027] For example, the first sound data may include the operating sound data of the transformer when the core is loose.

[0028] For example, the first sound data may include the operating sound data of the transformer in normal operation.

[0029] The first sound data undergoes feature extraction and data type conversion to be transformed into model training data, whereby the model training data is data that can be used for model training.

[0030] For example, the model training data may include image data converted from the first sound data.

[0031] For example, the model training data may include text data converted from the first audio data.

[0032] The model training process is carried out based on the model training data to establish a fault detection model for the transformer. The fault detection model is a deep learning model for detecting transformer faults.

[0033] For example, the fault detection model can determine whether a transformer has a fault, and if a transformer has a fault, the fault detection model can also determine the type of fault in the transformer.

[0034] During the operation of the transformer, the second sound data of the transformer is collected, which is the real-time sound data of the transformer.

[0035] The second sound data is used as input data and fed into the fault detection model to obtain the fault detection result output by the fault detection model. The fault detection result is used to represent the fault type of the transformer.

[0036] For example, the fault detection result can be specifically that the transformer is operating normally and there is no fault in the transformer.

[0037] For example, the fault detection result can specifically indicate that the transformer has an overload fault.

[0038] For example, the fault detection result can specifically be that the transformer has a partial discharge fault.

[0039] For example, the fault detection result can specifically indicate that the transformer has an overvoltage fault.

[0040] For example, the fault detection result can be specifically that the transformer has a loose core fault.

[0041] It should be noted that in this embodiment, feature extraction and data type conversion are performed on the first sound data of the transformer to obtain model training data, which improves the accuracy of the model training data. By establishing a fault detection model with accurate model training data, the recognition accuracy of the fault detection model is greatly improved, thereby ensuring the accuracy of the fault detection results output by the fault detection model.

[0042] The transformer fault detection method in this embodiment improves the accuracy of model training data by performing feature extraction and data type conversion on the first sound data of the transformer. Then, a fault detection model is established using the model training data, which improves the accuracy of the fault detection model. The fault detection model improves the accuracy of transformer fault detection.

[0043] In some embodiments, this application provides a transformer fault detection method, which performs feature extraction and data type conversion processing on first sound data to obtain model training data, including: Step S201: Perform variational mode decomposition on the first sound data to obtain the feature data in the first sound data; Step S202: Perform continuous wavelet transform on the feature data to convert the feature data into a time-frequency grayscale image; Step S203: Perform image enhancement processing on the time-frequency grayscale image to obtain model training data.

[0044] In this embodiment, feature data is extracted from the first sound data based on the variational mode decomposition algorithm, wherein the feature data is data representing the sound features in the first sound data.

[0045] For example, the first sound data can be input into a variational mode decomposition program to extract features from the first sound data and obtain feature data.

[0046] For example, the feature data may specifically include the frequency features, peak features, and mean squared error features of the first sound data.

[0047] The feature data is processed by continuous wavelet transform to convert the feature data into a time-frequency grayscale image, which is a grayscale image combining the time domain and the frequency domain.

[0048] For example, a time-frequency grayscale image can be specifically a grayscale image where the horizontal axis is time and the vertical axis is frequency.

[0049] For example, feature extraction is performed on the first sound data, the feature data is used as sample set data, and then continuous wavelet transform is performed on the sample set data to generate a time-frequency grayscale image.

[0050] Image enhancement processing is performed on the time-frequency grayscale images to increase the number of images in the time-frequency grayscale images. Then, the supplemented time-frequency grayscale images are packaged into model training data.

[0051] For example, the time-frequency grayscale image is processed by image mirroring, image transposition, image 180-degree transformation, and adding noise to the image to increase the amount of training data and obtain model training data.

[0052] In some embodiments, this application provides a transformer fault detection method, which performs variational mode decomposition on first sound data to obtain feature data in the first sound data, including: Step S301: Determine the optimal constraint value corresponding to the preset first variational mode model; Step S302: Based on the optimal constraint values, update the first variational mode model to obtain the second variational mode model; Step S303: Input the first sound data into the second variational mode model to obtain the feature data output by the second variational mode model.

[0053] In this embodiment, the variational modal model corresponding to the first sound data is determined, wherein the variational modal model is a model for feature extraction processing.

[0054] Determine the optimal constraint values ​​for the variational modal model, where the optimal constraint values ​​are the conditional values ​​that constrain the output data of the variational modal model.

[0055] For example, the first variational mode model is a mathematical model that can implement the variational mode decomposition algorithm and can perform adaptive signal decomposition processing on the first sound data.

[0056] In the first variational mode model, the eigenmode functions of the first variational mode model are represented by amplitude-modulated (AM) and frequency-modulated (FM) signals. The expressions for the eigenmode functions are as follows: u k (t)=A k (t)×cos( k (t)); Among them, u k (t) represents the k-th modal component, A k (t) represents the instantaneous amplitude of the k modal components. k (t) represents the instantaneous phase of the k modal components, where t represents time. In the intrinsic mode functions, the signal is analyzed using the Hilbert transform to obtain their respective one-sided spectra. Based on the Fourier transform principle, each mode is multiplied by an exponential signal, the center frequency is adjusted, and the modes are demodulated to their corresponding fundamental frequency bands. Gaussian smoothing is then used to estimate the bandwidth of the demodulated signal. Assuming f represents the original signal and K represents the number of modal components after decomposition, the formula is as follows: ; ; Among them, u k (t) represents the k-th modal component, ω k Let t be the center frequency corresponding to the k-th modal component, K be the number of decomposition layers, and δ(t) be the impulse function.

[0057] Introducing α and λ(t) into the above equation, where α is the quadratic penalty factor and λ(t) is the Lagrange multiplier operator, gives it good convergence properties and strict constraint enforcement, and transforms it into an unconstrained problem. The extended Lagrange expression is as follows: ; The first variational modal model employs the alternating direction multiplier algorithm to obtain the optimal solution to the above equations, thereby acquiring the optimal constraint values. The algorithm for obtaining the optimal solution is as follows: Step (1), when n=0, initialize {u k 1}、{ω k 1}、λ; Step (2): Let k=0, k=k+1, and update the modal component u in the above formula. k and center frequency ω k The following formula is obtained: ; ； Step (3), update the Lagrange multiplier λ: ; Step (4): Repeat steps (2) and (3) above. Stop iterating once the formula satisfies the constraints. ; Where ɛ is the previously set convergence error.

[0058] For example, the number of parameter modes K and the penalty factor α in the first variational modal model are optimized using an optimization algorithm. The specific steps are as follows: Step 1, initialize the population, which mainly involves initializing various parameters, including population N, scaling factor F, and crossover probability CR.

[0059] Step 2, mutation operation. Mutation refers to the change of an element. Randomly select two different objects in the population, scale the difference vector between the two, and then sum it with the mutated individual. The formula is as follows. U i (g+1)=X t1 (g)+F×(X t2 (g)-X t3 (g)); Where t1, t2, t3 and i are all distinct, i∈[0,N], i takes a random positive integer, F represents the scaling factor of the vector, also called the mutation operator, g represents the evolutionary generation, and X t1 (g) represents the t1th individual in the g-th generation population.

[0060] Step 3, crossover operation, for the g-th generation population X i (g) and the intermediate population U of the variation i (g+1) performs a crossover operation, as shown in the following formula; ; Where CR is the crossover factor, also known as the crossover probability, and the values ​​of rand and CR range from [0,1]. rand Let X be a random integer. ij (g) and U ij (g+1) represent X respectively i (g) and U i The j-th component in (g+1).

[0061] Step 4: Select the operation and take the experimental individuals U obtained from the crossover operation. i (g+1) and parent individual X i Compare (g+1) and select the individual in the population with the best fitness function value: ; Where, f(U) i (g+1)) is the fitness function value corresponding to the g+1th generation.

[0062] Step 5: Determine the termination condition to identify the number of parameter modes K and the penalty factor α. The evolutionary operation terminates if the predetermined number of iterations or accuracy is reached.

[0063] Based on the optimal constraint values, the first variational modal model is updated to obtain the second variational modal model. Then, the first sound data is input into the second variational modal model to obtain the feature data output by the second variational modal model.

[0064] For example, the second variational mode model is the variational mode model after the optimal constraint values ​​have been determined.

[0065] For example, according to optimization analysis, setting the parameters of variational mode decomposition to α=2000 and K=6 decomposes the sound of the operating transformer into 6 layers of modal components with different frequency bands, without mode aliasing or over-decomposition. The steps for energy feature extraction from variational mode decomposition are as follows: Step 1: After performing variational mode decomposition on the brake pad impact signal, K modal components U are obtained. i (i=1,2,3...K).

[0066] Step 2: Use the summation formula to calculate the energy of each modal component after decomposition. The expression for the energy is as follows: ; Where E represents energy, U represents modal components, and i represents the i-th modal component.

[0067] Step 3, normalization process, let T i =E i / E (i=1,2,3...K).

[0068] Step 4: Obtain the feature vector T=[T1,T2,T3...T] that can characterize the internal defects of the brake pads. K The feature vector of the defect is the feature data.

[0069] In some embodiments, this application provides a transformer fault detection method, which performs continuous wavelet transform processing on feature data to convert the feature data into a time-frequency grayscale image, including: Step S401: Based on the preset first transformation formula, establish the second transformation formula corresponding to the feature data; Step S402: Input the feature data into the second transformation formula to obtain a time-frequency grayscale image.

[0070] In this embodiment, a preset first transformation formula is obtained, and then a second transformation formula corresponding to the feature data is established based on the first transformation formula. The first transformation formula is a preset data transformation formula, and the second transformation formula is an updated data transformation formula.

[0071] The feature data is input into the second transformation formula to obtain a time-frequency grayscale image.

[0072] For example, a time-frequency grayscale image is generated by performing a continuous wavelet transform on the sample set data from which features have been extracted. A wavelet is a small, continuous waveform with peaks and troughs. Let the first transform formula be ψ(t)∈L.1 (R)∩L 2 (R), such that ψ(0) = 0, the integral of the function ψ(0) is 0. ψ(0) is called a mother wavelet or the basic wavelet. When the basic wavelet ψ(0) is further subjected to scaling and translation transformations, a new formula is obtained, as follows: ; Here, letter 'a' represents the scale of the function, called the scale factor, and letter 'b' indicates the translation position of the wavelet function on the time axis. Generally, the energy of the mother wavelet ψ(t) is most concentrated at the origin, while point 'b' is the energy concentration point of the wavelet function. ψ(t) becomes ψ(t) after scaling and transformation. a,b (t), this function is defined as a continuous wavelet function, and its formula after continuous wavelet transform or integral wavelet transform is: ; In the formula, the letter 'a' is not equal to 0, b and t are continuous variables, and the complex conjugate function of ψ(t) is ψ*(t). Continuous wavelet transform also possesses two important properties: linearity and translation invariance. The rectangular method constitutes the basic approach to wavelet transform. Because 'a' is greater than 0 in the formula, ψ(t) is always a real-function wavelet. s Let the sampling interval be n times and k times the sampling interval in the signal function f(t) and wavelet function ψ(t), respectively. The wavelet transform formula obtained by rectangular numerical integration is as follows: ; It can also be simplified to the second transformation formula: ; Here, f(n) and ψ(n) are the sampling sequences of f(t) and ψ(t), respectively. A set of wavelet coefficients is obtained by varying k values ​​based on the value of 'a'. 'a' is a discrete value, typically taken as a=2. j .

[0073] In some embodiments, this application provides a transformer fault detection method, which performs image enhancement processing on a time-frequency grayscale image to obtain model training data, including: Step S501: Perform image flipping processing on the time-frequency grayscale image to obtain the first image data; Step S502: Transpose the pixels in the time-frequency grayscale image to obtain the second image data; Step S503: Perform smoothing and blurring processing on the time-frequency grayscale image to obtain the third image data; Step S504: Combine the first image data, the second image data, and the third image data into model training data.

[0074] In this embodiment, the time-frequency grayscale image is subjected to image flipping processing so that the time-frequency grayscale image is updated to first image data, wherein the first image data is the image data after the time-frequency grayscale image is flipped.

[0075] For example, the first image data can be obtained by mirroring and flipping a time-frequency grayscale image.

[0076] For example, rotating a time-frequency grayscale image by 180 degrees can yield the first image data.

[0077] The pixels in the time-frequency grayscale image are transposed to obtain the second image data, wherein the second image data is the image data after the time-frequency grayscale image is transposed.

[0078] The time-frequency grayscale image is smoothed and blurred to obtain the third image data, which is the image data after blurring the time-frequency grayscale image.

[0079] For example, Gaussian blurring can be applied to a time-frequency grayscale image to obtain third image data.

[0080] The first image data, the second image data, and the third image data are combined into model training data.

[0081] For example, performing image flipping, transposition, and smoothing blurring on time-frequency grayscale images can increase the number of time-frequency grayscale images, thereby ensuring the total amount of data for model training.

[0082] In some embodiments, this application provides a transformer fault detection method, which establishes a fault detection model corresponding to the transformer based on model training data, including: Step S601: Input the model training data into the preset convolutional neural network; Step S602: Perform data training processing on the convolutional neural network to obtain a fault detection model.

[0083] In this embodiment, a preset convolutional neural network is obtained, and model training data is input into the preset convolutional neural network. The convolutional neural network is a type of feedforward neural network that includes convolution calculations and has a deep structure.

[0084] Data training is performed on the convolutional neural network to obtain a fault detection model.

[0085] For example, a convolutional neural network (CNN) is constructed. A CNN is a type of feedforward neural network where artificial neurons process complex images by responding to surrounding units. It consists of four basic structures: an input layer, convolutional layers, pooling layers, and fully connected layers (FC). One convolutional layer and one pooling layer form the image feature extractor. In a neural network, there is a function between the output of an upper-layer neuron and the input of a lower-layer neuron; this function is called the activation function. In this case, the activation function is the Sigmoid function. The Sigmoid function is a non-linear function commonly used in binary classification problems; it is also known as the Logistic function. As shown in the diagram, the function has an S-shaped shape, with 0.5 as its center of symmetry. Its domain is from negative infinity to positive infinity, and its range is from 0 to 1. The output center is not 0. We prefer functions where the input value is 0 and the output value is also 0; therefore, the Sigmoid function is generally not used in deep neural networks. The mathematical form of the Sigmoid function is: h(x) = 1 / (1+e^(-x))^(-x ... -x ), where x is the input data.

[0086] For example, this embodiment comprehensively considers factors such as image pixel size, convolutional kernel, and stride, and selects an 11-layer neural network structure, including 5 convolutional layers, 5 pooling layers, and 1 fully connected layer. The learning rate and batch size parameters in the model are set. The learning rate determines the speed of gradient descent; an improper setting can lead to local optima or failure to converge, neither of which yields the optimal value. Choosing an appropriate learning rate can optimize the performance of the network model. In this embodiment, the initial value of the learning rate is set to 0.01. The batch size refers to the number of images required for one training iteration of the model. An improper setting can cause system crashes or reduced running speed due to insufficient memory. The training cycle and accuracy determine the batch size; in this embodiment, the batch size is set to 50. Finally, the model is determined, and the sample dataset is input for learning and training. Later, the sound of transformer operation is collected, analyzed, and classified using a microphone to achieve the purpose of detecting internal faults in the transformer.

[0087] In some embodiments, this application provides a transformer fault detection method, which acquires first sound data of the transformer, including: Step S701: Under different operating conditions of the transformer, collect the sound of the transformer to obtain the third sound data of the transformer; Step S702: Based on the operating status of the transformer, perform data annotation processing on the third sound data to obtain the first sound data.

[0088] In this embodiment, the sound of the transformer is collected under different operating conditions to obtain the third sound data of the transformer, wherein the third sound data is the original sound data of the transformer.

[0089] For example, a sound signal acquisition system is used to collect the sounds of the operating transformer under overload, partial discharge, overvoltage, core loosening and normal operation conditions, and then summarize them into third sound data.

[0090] Based on the transformer's operating status, the third sound data is annotated to obtain the first sound data.

[0091] For example, five categories of labels are identified, namely overload, partial discharge, overvoltage, loose iron core, and normal operation, and the third sound data is processed by data labeling to obtain the first sound data.

[0092] In some embodiments, such as Figure 2 As shown, an embodiment of this application provides a transformer fault detection device 800, comprising: Acquisition unit 802 is used to acquire the first sound data of the transformer; The processing unit 804 is used to perform feature extraction and data type conversion on the first sound data to obtain model training data; The processing unit 804 is also used to establish a fault detection model for the transformer based on the model training data. The acquisition unit 802 is also used to collect the sound of the transformer during the operation of the transformer in order to obtain the second sound data of the transformer; The processing unit 804 is also used to input the second sound data into the fault detection model to obtain the fault detection result output by the fault detection model.

[0093] In this embodiment, a transformer fault detection device 800 is proposed, which is used to detect the fault type of the transformer during operation, and then to carry out fault repair on the transformer in order to maintain the normal operation of the transformer.

[0094] For example, the transformer may specifically be a DC-DC transformer.

[0095] For example, the transformer may specifically be an AC-AC transformer.

[0096] The first sound data of the transformer is collected, which is the historical sound data of the transformer.

[0097] For example, the first sound data may include the sound data of the transformer operating under overload conditions.

[0098] For example, the first sound data may include the operating sound data of the transformer in a partial discharge state.

[0099] For example, the first audio data may include the operating audio data of the transformer under overvoltage conditions.

[0100] For example, the first sound data may include the operating sound data of the transformer when the core is loose.

[0101] For example, the first sound data may include the operating sound data of the transformer in normal operation.

[0102] The first sound data undergoes feature extraction and data type conversion to be transformed into model training data, whereby the model training data is data that can be used for model training.

[0103] For example, the model training data may include image data converted from the first sound data.

[0104] For example, the model training data may include text data converted from the first audio data.

[0105] The model training process is carried out based on the model training data to establish a fault detection model for the transformer. The fault detection model is a deep learning model for detecting transformer faults.

[0106] For example, the fault detection model can determine whether a transformer has a fault, and if a transformer has a fault, the fault detection model can also determine the type of fault in the transformer.

[0107] During the operation of the transformer, the second sound data of the transformer is collected, which is the real-time sound data of the transformer.

[0108] The second sound data is used as input data and fed into the fault detection model to obtain the fault detection result output by the fault detection model. The fault detection result is used to represent the fault type of the transformer.

[0109] For example, the fault detection result can be specifically that the transformer is operating normally and there is no fault in the transformer.

[0110] For example, the fault detection result can specifically indicate that the transformer has an overload fault.

[0111] For example, the fault detection result can specifically be that the transformer has a partial discharge fault.

[0112] For example, the fault detection result can specifically indicate that the transformer has an overvoltage fault.

[0113] For example, the fault detection result can be specifically that the transformer has a loose core fault.

[0114] It should be noted that in this embodiment, feature extraction and data type conversion are performed on the first sound data of the transformer to obtain model training data, which improves the accuracy of the model training data. By establishing a fault detection model with accurate model training data, the recognition accuracy of the fault detection model is greatly improved, thereby ensuring the accuracy of the fault detection results output by the fault detection model.

[0115] The transformer fault detection device 800 in this embodiment improves the accuracy of the model training data by performing feature extraction and data type conversion on the first sound data of the transformer. Then, it establishes a fault detection model using the model training data, thereby improving the accuracy of the fault detection model and enhancing the accuracy of transformer fault detection.

[0116] In some embodiments, this application provides a transformer fault detection device 800, comprising: The processing unit 804 is also used to perform variational mode decomposition on the first sound data to obtain feature data in the first sound data; The processing unit 804 is also used to perform continuous wavelet transform processing on the feature data to convert the feature data into a time-frequency grayscale image; The processing unit 804 is also used to perform image enhancement processing on the time-frequency grayscale image to obtain model training data.

[0117] In some embodiments, this application provides a transformer fault detection device 800, comprising: The processing unit 804 is also used to determine the optimal constraint value corresponding to the preset first variational mode model; The processing unit 804 is also used to update the first variational mode model based on the optimal constraint values ​​to obtain the second variational mode model; The processing unit 804 is also used to input the first sound data into the second variational mode model to obtain the feature data output by the second variational mode model.

[0118] In some embodiments, this application provides a transformer fault detection device 800, comprising: The processing unit 804 is also used to establish a second transformation formula corresponding to the feature data based on a preset first transformation formula; The processing unit 804 is also used to input feature data into the second transformation formula to obtain a time-frequency grayscale image.

[0119] In some embodiments, this application provides a transformer fault detection device 800, comprising: The processing unit 804 is also used to perform image flipping processing on the time-frequency grayscale image to obtain the first image data; The processing unit 804 is also used to transpose the pixels in the time-frequency grayscale image to obtain the second image data; The processing unit 804 is also used to perform smoothing and blurring processing on the time-frequency grayscale image to obtain third image data; The processing unit 804 is also used to combine the first image data, the second image data and the third image data into model training data.

[0120] In some embodiments, this application provides a transformer fault detection device 800, comprising: The processing unit 804 is also used to input model training data into a preset convolutional neural network; The processing unit 804 is also used to perform data training on the convolutional neural network to obtain a fault detection model.

[0121] In some embodiments, this application provides a transformer fault detection device 800, comprising: The processing unit 804 is also used to collect the sound of the transformer when the transformer is in different operating states, so as to obtain the third sound data of the transformer; The processing unit 804 is also used to perform data annotation processing on the third sound data based on the operating status of the transformer to obtain the first sound data.

[0122] In some embodiments, such as Figure 3 As shown, a transformer fault detection device 900 is proposed. The transformer fault detection device 900 includes a processor 902 and a memory 904. The memory 904 stores a computer program, which, when executed by the processor 902, implements the steps of the transformer fault detection method as described in any of the above embodiments. Therefore, the transformer fault detection device 900 possesses all the beneficial effects of the transformer fault detection method in any of the above embodiments, which will not be elaborated further here.

[0123] In some embodiments, a readable storage medium is provided having a program stored thereon, which, when executed by a processor, implements the steps of the transformer fault detection method as described in any of the above embodiments, and thus has all the beneficial technical effects of the transformer fault detection method in any of the above embodiments.

[0124] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0125] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-readable program code.

[0126] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0127] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0128] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0129] This application also provides a computer program product, which includes computer software instructions that, when executed on a processing device, cause the processing device to execute a process for a transformer fault detection method.

[0130] A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0131] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0132] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms.

[0133] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0134] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0135] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0136] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions 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 this application.

[0137] Although preferred embodiments have been described in this specification, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this specification.

[0138] Obviously, those skilled in the art can make various modifications and variations to this specification without departing from its spirit and scope. Therefore, if such modifications and variations fall within the scope of the claims and their equivalents, this specification is also intended to include such modifications and variations.

Claims

1. A method for detecting faults in a transformer, characterized in that, The method includes: Acquire the first sound data of the transformer; The first sound data is subjected to feature extraction and data type conversion to obtain model training data; Based on the model training data, a fault detection model corresponding to the transformer is established; During the operation of the transformer, the sound of the transformer is collected to obtain the second sound data of the transformer; The second sound data is input into the fault detection model to obtain the fault detection result output by the fault detection model.

2. The method according to claim 1, characterized in that, The step of performing feature extraction and data type conversion on the first sound data to obtain model training data includes: The first sound data is subjected to variational mode decomposition to obtain the feature data in the first sound data; The feature data is subjected to continuous wavelet transform processing to convert the feature data into a time-frequency grayscale image; The time-frequency grayscale image is subjected to image enhancement processing to obtain the model training data.

3. The method according to claim 2, characterized in that, The step of performing variational mode decomposition on the first sound data to obtain feature data from the first sound data includes: Determine the optimal constraint values ​​corresponding to the preset first variational mode model; Based on the optimal constraint values, the first variational mode model is updated to obtain the second variational mode model; The first sound data is input into the second variational mode model to obtain the feature data output by the second variational mode model.

4. The method according to claim 2, characterized in that, The step of performing continuous wavelet transform processing on the feature data to convert the feature data into a time-frequency grayscale image includes: Based on the preset first transformation formula, a second transformation formula corresponding to the feature data is established; The feature data is input into the second transformation formula to obtain the time-frequency grayscale image.

5. The method according to claim 2, characterized in that, The image enhancement processing of the time-frequency grayscale image to obtain the model training data includes: The time-frequency grayscale image is subjected to image flipping processing to obtain the first image data; The pixels in the time-frequency grayscale image are transposed to obtain the second image data; The time-frequency grayscale image is smoothed and blurred to obtain third image data; The first image data, the second image data, and the third image data are combined to form the model training data.

6. The method according to any one of claims 1 to 5, characterized in that, The step of establishing a fault detection model for the transformer based on the model training data includes: The training data of the model is input into a preset convolutional neural network; The convolutional neural network is trained with data to obtain the fault detection model.

7. The method according to any one of claims 1 to 5, characterized in that, The acquisition of the first sound data of the transformer includes: The transformer is subjected to different operating conditions. The sound of the transformer is collected to obtain the third sound data of the transformer. Based on the operating status of the transformer, the third sound data is labeled to obtain the first sound data.

8. A fault detection device for a transformer, characterized in that, The device includes: Acquisition unit, used to acquire the first sound data of the transformer; The processing unit is used to perform feature extraction and data type conversion on the first sound data to obtain model training data; The processing unit is also used to establish a fault detection model corresponding to the transformer based on the model training data; The acquisition unit is also used to collect the sound of the transformer during the operation of the transformer to obtain the second sound data of the transformer; The processing unit is further configured to input the second sound data into the fault detection model to obtain the fault detection result output by the fault detection model.

9. A fault detection device for a transformer, characterized in that, include: processor; A memory, which stores programs or instructions, wherein a processor, when executing the programs or instructions in the memory, implements the steps of the transformer fault detection method as described in any one of claims 1 to 7.

10. A readable storage medium, characterized in that, A program or instructions are stored on a readable storage medium, which, when executed by a processor, implement the steps of the transformer fault detection method as described in any one of claims 1 to 7.

Images