Insulation partial discharge mode identification method, device and equipment and storage medium

Abstract

The invention discloses an insulation partial discharge mode identification method, apparatus and device, and a storage medium. The method comprises the steps of collecting a partial discharge signal of a to-be-identified insulation defect sample; extracting phase distribution information of the partial discharge signal; generating a time-frequency graph of the insulation defect sample to be identified based on the time-frequency domain feature of the phase distribution information and a weight coefficient corresponding to the time-frequency domain feature; and inputting the time-frequency graph into a pre-trained discharge mode recognition model based on a deep learning network, and outputting the defect type of the insulation defect sample to be recognized. By using the method, under the condition that effective samples of different types of partial discharge are very limited, the problem that a classifier is very easy to generate an underfitting phenomenon due to insufficient discharge characteristics is solved, and the identification accuracy is improved.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of power grid technology, and in particular to a method, device, equipment, and storage medium for identifying an insulation partial discharge pattern. Background Art

[0002] Insulation degradation in electrical equipment is a major cause of equipment failure. Depending on the factors causing the insulation defect, the defects can manifest in various forms and cause varying degrees of damage. Effective identification of partial discharge (PD), a manifestation of insulation degradation, can provide a reference for assessing equipment status. Early detection of PD and timely elimination of insulation defects, thus ensuring the safe and stable operation of electrical equipment, is of great theoretical and practical value.

[0003] Traditional methods for identifying partial discharge signals primarily rely on two-dimensional phase-resolved partial discharge (PRPD) statistical maps and consist of two steps: feature extraction and classifier recognition. Feature extraction typically employs statistical, fractal, moment, and texture feature methods. Commonly used classifiers include support vector machines, BP neural networks, extreme learning machines, and convolutional neural network models. In engineering practice, the probability of insulation failure in electrical equipment is low, and valid samples of different types of partial discharge are very limited. Insufficient discharge features can easily lead to underfitting of the classifier, affecting the final recognition results. Summary of the Invention

[0004] The embodiments of the present invention provide a method, apparatus, device and storage medium for identifying partial discharge patterns of insulation. When valid samples of different types of partial discharges are very limited, the method solves the problem that insufficient discharge features easily lead to underfitting of the classifier, thereby improving the recognition accuracy.

[0005] In a first aspect, an embodiment of the present invention provides a method for identifying a partial discharge pattern of an insulation, comprising:

[0006] Collecting partial discharge signals of insulation defect samples to be identified;

[0007] extracting phase distribution information of the partial discharge signal;

[0008] Generating a time-frequency diagram of the insulation defect sample to be identified based on the time-frequency domain characteristics of the phase distribution information and the weight coefficients corresponding to the time-frequency domain characteristics;

[0009] The time-frequency graph is input into a pre-trained discharge pattern recognition model based on a deep learning network, and the defect type of the insulation defect sample to be identified is output.

[0010] In a second aspect, an embodiment of the present invention further provides an insulation partial discharge pattern recognition device, the device comprising:

[0011] A discharge signal acquisition module is used to collect partial discharge signals of insulation defect samples to be identified;

[0012] An information extraction module, configured to extract phase distribution information of the partial discharge signal;

[0013] a time-frequency diagram generating module, configured to generate a time-frequency diagram of the insulation defect sample to be identified based on the time-frequency domain features of the phase distribution information and weight coefficients corresponding to the time-frequency domain features;

[0014] A defect type output module is used to input the time-frequency diagram into a pre-trained discharge pattern recognition model based on a deep learning network, and output the defect type of the insulation defect sample to be identified.

[0015] In a third aspect, an embodiment of the present disclosure further provides an electronic device, comprising:

[0016] one or more processors;

[0017] a storage device for storing one or more programs,

[0018] When the one or more programs are executed by the one or more processors, the one or more processors implement the insulation partial discharge pattern recognition method provided by the embodiment of the present disclosure.

[0019] In a fourth aspect, the embodiments of the present disclosure further provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to implement the insulation partial discharge pattern recognition method provided by the embodiments of the present disclosure.

[0020] In a fifth aspect, an embodiment of the present disclosure further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the insulation partial discharge pattern recognition method provided by the embodiment of the present disclosure.

[0021] The present invention discloses a method, apparatus, device, and storage medium for identifying insulation partial discharge patterns. The method comprises: collecting partial discharge signals of insulation defect samples to be identified; extracting phase distribution information from the partial discharge signals; generating a time-frequency graph of the insulation defect samples to be identified based on the time-frequency domain characteristics of the phase distribution information and weight coefficients corresponding to the time-frequency domain characteristics; and inputting the time-frequency graph into a pre-trained discharge pattern recognition model based on a deep learning network to output the defect type of the insulation defect samples to be identified. This method solves the problem of underfitting of the classifier due to insufficient discharge features, which can easily lead to underfitting of the classifier when valid samples of different types of partial discharge are very limited, thereby improving recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that the originals and elements are not necessarily drawn to scale.

[0023] Figure 1 A flowchart of a method for identifying an insulation partial discharge pattern provided by an embodiment of the present disclosure;

[0024] Figure 2 An example diagram of a test circuit system provided by an embodiment of the present disclosure;

[0025] Figure 3 This is an example diagram of a discharge pattern recognition model based on a deep learning network provided in an embodiment of the present disclosure;

[0026] Figure 4 This is an example diagram of a defect model for a needle-plate defect provided by an embodiment of the present disclosure;

[0027] Figure 5 This is an example diagram of a defect model for a column-plate defect provided by an embodiment of the present disclosure;

[0028] Figure 6 An example diagram of a defect model for a ball-plate defect provided by an embodiment of the present disclosure;

[0029] Figure 7 This is an example diagram of partial discharge identification results under a small sample size provided by an embodiment of the present disclosure;

[0030] Figure 8 A schematic structural diagram of an insulation partial discharge pattern recognition device provided by an embodiment of the present disclosure;

[0031] Figure 9 A schematic structural diagram of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0032] The following describes embodiments of the present disclosure in more detail with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of protection of the present disclosure.

[0033] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0034] As used herein, the term "including" and its variations are open-ended, i.e., "including but not limited to." The term "based on" means "based, at least in part, on." The term "one embodiment" means "at least one embodiment," the term "another embodiment" means "at least one additional embodiment," and the term "some embodiments" means "at least some embodiments." Other terms are defined in the following description.

[0035] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0036] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, they should be understood as "one or more".

[0037] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0038] It is understandable that before using the technical solutions disclosed in the various embodiments of this disclosure, the type, scope of use, usage scenarios, etc. of the personal information involved in this disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.

[0039] For example, in response to a user's active request, a prompt message is sent to the user to clearly inform the user that the operation requested will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the electronic device, application, server, storage medium, or other software or hardware that performs the operations of the disclosed technical solution based on the prompt message.

[0040] As an optional but non-limiting implementation, in response to receiving a user's active request, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. Furthermore, the pop-up window may also contain a selection control for the user to select "agree" or "disagree" to provide personal information to the electronic device.

[0041] It is understandable that the above notification and user authorization process are merely illustrative and do not limit the implementation of the present disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0042] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) must comply with the requirements of relevant laws, regulations and relevant provisions.

[0043] Figure 1 This is a flow chart of an insulation partial discharge pattern recognition method provided by an embodiment of the present disclosure. The embodiment of the present disclosure is suitable for providing a solution to the problem that when the valid samples of different types of partial discharges are very limited, insufficient discharge characteristics can easily lead to underfitting of the classifier. The method can be executed by an insulation partial discharge pattern recognition method device, which can be implemented in the form of software and / or hardware. Optionally, it can be implemented by an electronic device, which can be a mobile terminal, PC or server, etc.

[0044] like Figure 1 As shown, an insulation partial discharge pattern recognition method provided by an embodiment of the present disclosure may specifically include the following steps:

[0045] S110, collecting partial discharge signals of the insulation defect sample to be identified. In this embodiment, the test circuit system is used to obtain partial discharge signals generated after partial discharge of the insulation defect sample to be identified.

[0046] For example, Figure 2 This is an example diagram of the test circuit system provided by the embodiment of the present disclosure. Figure 2 As shown, the partial discharge signal of the insulation defect sample to be identified is collected; during the collection, a power frequency voltage with a voltage amplitude of 1.6 times the initial discharge voltage is applied, and 500 power frequency discharge cycle signals are collected.

[0047] On the basis of the above embodiment, after collecting the local partial discharge signal of the insulation defect sample to be identified, the method further includes:

[0048] Perform denoising on the partial discharge signal.

[0049] Specifically, the sample signals are denoised using a wavelet denoising method. The wavelet base can be db2 wavelet, with 7 decomposition layers, and an adaptive threshold and soft threshold method for denoising.

[0050] S120: Extract phase distribution information of the partial discharge signal.

[0051] Phase distribution information includes: maximum discharge capacity, average discharge capacity and discharge times.

[0052] Specifically, the partial discharge signal of each cycle is divided into multiple data according to the phase, and the maximum discharge amount, average discharge amount and discharge number of each data are determined. All the maximum discharge amount, average discharge amount and discharge number are used as phase distribution information.

[0053] Based on the above embodiment, extracting the phase distribution information of the partial discharge signal includes the following steps:

[0054] a1) Dividing the partial discharge signal of each period into a first number of data portions according to phases.

[0055] b1) Determine the maximum discharge capacity, average discharge capacity, and number of discharges for each first number of pieces of data.

[0056] c1) All maximum discharge amounts, average discharge amounts, and discharge times are taken as phase distribution information.

[0057] In this embodiment, the first number of portions may be preset and set according to actual conditions, and is not specifically limited in this embodiment. The maximum discharge amount may be the maximum discharge amount among multiple discharges, the average discharge amount may be the average of the discharge amounts among the multiple discharges, and the number of discharges may be the number of discharges. The cycle may be a power frequency cycle.

[0058] Specifically, the partial discharge signal of each cycle is divided into a first number of data according to the phase, the maximum discharge amount, average discharge amount and discharge number of each first number of data are determined, and all the maximum discharge amounts, average discharge amounts and discharge numbers are used as phase distribution information.

[0059] For example, the m sampling point data of the insulation defect sample under each power frequency cycle are divided into 360 parts according to the phase, the maximum discharge amount, average discharge amount and number of discharges in each part of the data are counted, and the discharge data of the insulation defect sample under 500 power frequency cycles are used as the phase distribution information of the local discharge signal.

[0060] S130 : generating a time-frequency diagram of the insulation defect sample to be identified based on the time-frequency domain features of the phase distribution information and the weight coefficients corresponding to the time-frequency domain features.

[0061] In this embodiment, the time-frequency plot provides information about the joint distribution of insulation defect samples in both the time and frequency domains, clearly depicting the relationship between signal frequency and time. The time-frequency domain features can be information about features in both the time and frequency domains. The weight coefficients can be weights assigned to the corresponding colors when transforming the color matrix of the time-frequency plot.

[0062] Specifically, the phase distribution information is transformed according to the S-transformation formula to obtain the time-frequency domain feature matrix of the phase distribution information, the time-frequency domain feature matrix is ​​multiplied by the weight coefficient to generate a color matrix, and the color matrix is ​​used as the time-frequency diagram.

[0063] On the basis of the above embodiment, generating a time-frequency diagram of the insulation defect sample to be identified based on the time-frequency domain features of the phase distribution information and the weight coefficients corresponding to the time-frequency domain features includes the following steps:

[0064] a2) transforming the phase distribution information according to an S-transformation formula to obtain a time-frequency domain feature matrix of the phase distribution information, wherein the behavioral frequencies of the time-frequency domain feature matrix are listed as phases.

[0065] b2) Multiplying the time-frequency domain feature matrix by the weight coefficient to generate a color matrix, and using the color matrix as the time-frequency graph.

[0066] For example, the phase distribution information is subjected to S transformation to obtain the time-frequency matrix H, U, O, which is used as the three-primary color matrix to determine the color matrix C of the S-transformed time-frequency diagram of the power frequency periodic discharge. i,j , use the spectral color matrix as the spectral map.

[0067]

[0068] Among them, R i,j , G i,j 、B i,j are red, green, and blue components respectively; α, β, and γ are the weights of the corresponding colors; |H i,j ||U i,j ||O i,j | is the modulus of the element in the i-th row and j-th column of the matrix; |H| MAX |U| MAX |O| MAX is the maximum value of the modulus of all elements in the matrix.

[0069] Based on the above embodiment, the process of setting the weight coefficient includes the following steps:

[0070] a3) Obtaining the set initial weight coefficient;

[0071] b3) Iteratively optimizing the initial weight coefficients using a particle swarm algorithm to obtain multiple optimized weight coefficients;

[0072] c3) The time-frequency graph corresponding to each optimized weight coefficient is input into the pre-trained convolutional neural network recognition model for recognition, and the optimized weight coefficient with the highest recognition accuracy is used as the weight coefficient.

[0073] Specifically, the weight coefficients are iteratively optimized through the particle swarm optimization algorithm, and the time-frequency graphs corresponding to each optimized weight coefficient are input into the pre-trained convolutional neural network recognition model for recognition, and the optimized weight coefficient with the highest recognition accuracy is used as the weight coefficient.

[0074] S140: Input the time-frequency graph into a pre-trained discharge pattern recognition model based on a deep learning network, and output the defect type of the insulation defect sample to be identified.

[0075] Specifically, this step is used to input the time-frequency graph into a pre-trained discharge pattern recognition model based on a deep learning network, and output the defect type of the insulation defect sample to be identified.

[0076] Figure 3 This is an example diagram of a discharge pattern recognition model based on a deep learning network provided by an embodiment of the present disclosure. Figure 3 As shown, the discharge pattern recognition model based on a deep learning network consists of an input layer, a convolutional layer, a pooling layer, a bidirectional gated recurrent network, and three fully connected layers. The convolutional layers have a kernel size of 3×3, 16 kernels, a stride of 1, and use the ReLU function as the activation function. The pooling layers have a kernel size of 2×2, a stride of 2, and a bidirectional gated recurrent network with 32 units. There are three fully connected layers, with output sizes of 32, 16, and n, respectively. A dropout layer with a probability of 10% is used between each fully connected layer. The input to fully connected layer 1 consists of the flattened vector of the pooling result and the output vector of the BIGRU layer. The insulation defect samples to be identified include pin-plate, column-plate, and ball-plate defects. For each type of insulation defect, 10% of the S-transform time-frequency graphs of power-frequency periodic discharges are selected as the model training set, and the remaining portion is used as the test set. The training set is input into the CNN-BIGRU model, and the network parameters are optimized using the Adam algorithm. Set the maximum number of training times to 100, the initial learning rate to 0.01, and the learning rate reduction factor to 0.1.

[0077] The present invention discloses a method for identifying insulation partial discharge patterns. The method comprises: collecting partial discharge signals from insulation defect samples to be identified; extracting phase distribution information from the partial discharge signals; generating a time-frequency graph of the insulation defect samples to be identified based on the time-frequency domain features of the phase distribution information and weight coefficients corresponding to the time-frequency domain features; and inputting the time-frequency graph into a pre-trained discharge pattern recognition model based on a deep learning network to output the defect type of the insulation defect samples to be identified. This method solves the problem of underfitting of the classifier due to insufficient discharge features when valid samples of different types of partial discharge are very limited, thereby improving recognition accuracy.

[0078] In an exemplary embodiment, the insulation partial discharge mode of the insulation defect sample under a small sample can be identified. The method may include the following steps:

[0079] Step 1: Perform a partial discharge test on n types of defect samples of insulation defects to be identified, collect the power frequency periodic discharge signal generated in the test as a sample signal, and perform wavelet threshold denoising on the collected sample signal.

[0080] During partial discharge testing, 50 power-frequency cycles of partial discharge signals were collected for each of the n types of insulation defects to be identified. These signals were then denoised using wavelet denoising. The DB2 wavelet was used as the wavelet base, with a decomposition layer of seven. Adaptive thresholding and soft thresholding were used for denoising.

[0081] Step 2: Draw the normalized maximum discharge phase distribution spectrum of each defect sample according to each sample signal Average discharge phase distribution spectrum and discharge number phase distribution spectrum

[0082] The m sampling point data of each defective sample under each power frequency cycle are divided into 360 parts according to the phase. The maximum discharge amount, average discharge amount and discharge times of the data in each part are counted. Based on the discharge data of each defective sample under 500 power frequency cycles, the phase distribution spectrum of the maximum discharge amount is drawn. Average discharge phase distribution spectrum and discharge number phase distribution spectrum

[0083] Step 3: Phase distribution spectrum of the maximum discharge amount of each defective sample Average discharge phase distribution spectrum and discharge number phase distribution spectrum Perform S-transformation and use it as the three-primary color matrix of the S-transformation time-frequency diagram of power frequency periodic discharge;

[0084] Phase distribution spectrum of the maximum discharge amount of each defect sample Average discharge phase distribution spectrum and discharge number phase distribution spectrum Spectral data Q max (n), Q qave (n) and N(n) are transformed by S, and Q max (n) as an example, the transformation formula is as follows:

[0085]

[0086] Where j, m, n = 0, 1, 2, ..., N-1; It's Q max Fourier spectrum of (n).

[0087] For the sample graph data Q max (n), Q qave After discrete S transformation of (n) and N(n), we get the matrix complex time-frequency matrix H, U, O, which is used as the three primary color matrix to determine the color matrix C of the power frequency periodic discharge S transform time-frequency diagram i,j .

[0088] For the sample graph data Q max (n), Q qave After discrete S transformation of (n) and N(n), we get the matrix complex time-frequency matrix H, U, O, which is used as the three primary color matrix to determine the color matrix C of the power frequency periodic discharge S transform time-frequency diagram i,j .

[0089]

[0090] Among them, R i,j , G i,j 、B i,j are the red, green, and blue components; α, β, and γ are the corresponding color weights; |H i,j | is the modulus of the element in the i-th row and j-th column of the matrix H; |H| max is the maximum value of the modulus of all elements in the matrix H.

[0091] Step 4: Optimize the color matrix C of the power frequency periodic discharge S transform time-frequency diagram using the particle swarm algorithm i,j The weight of is used to adjust the color ratio of the S-transform time-frequency diagram of power frequency periodic discharge.

[0092] The color weights α, β, and γ are iteratively optimized through the optimization algorithm. The three primary color matrices of the sample data under different color weights are input into the simple convolutional neural network (CNN) recognition model for training and recognition. The optimal color weight with the highest accuracy is iteratively optimized, and the color matrix C of the time-frequency diagram is transformed according to the power frequency period discharge of the optimal color weight. i,j Draw the S-transformation time-frequency graph of power-frequency periodic discharge. To reduce the impact of excessive image size on computing performance, the bilinear interpolation method can be used to reduce the size of each power-frequency periodic discharge S-transformation time-frequency graph to 360×360, and set a digital label for each power-frequency periodic discharge S-transformation time-frequency graph according to the corresponding data type. The data type can be such as maximum discharge amount, average discharge amount, and number of discharges. Furthermore, the color matrix C of the power-frequency periodic discharge S-transformation time-frequency graph is i,j The optimization algorithm for optimizing the weights may also include other methods, such as genetic algorithm, differential evolution algorithm, ant colony optimization algorithm, etc.

[0093] Step 5: Build a deep learning network model with strong generalization ability and initialize the model parameters, and train the model using the gradient descent method.

[0094] For example, you can construct Figure 3 The deep learning network model shown in Figure 1 can be composed of an input layer, a convolutional pooling layer, a bidirectional gated recurrent network, and three fully connected layers. The convolutional layer (conv) has a convolution kernel size of 3×3, 16 kernels, and a stride of 1, and uses a Relu function to activate the output. The pooling layer (maxpool) has a pooling kernel size of 2×2 and a stride of 2. The bidirectional gated recurrent network (BIGRU) has 32 units, and its input is the flattened vector of the pooling result. There are three fully connected layers, with output sizes of 32, 16, and n, respectively. The dropout layer probability between each fully connected layer is set to 10%. The input of fully connected layer 1 (fc1) is the flattened vector of the pooling result and the output vector of the BIGRU layer. Finally, the softmax function is used to predict the probability of each pattern. It is understood that the above deep learning network model is only an example and is not limiting. Deep learning network models of other structures or types, such as ResNet deep learning network models and graph convolutional neural network models, can also be used.

[0095] In an embodiment of the present invention, for each insulation defect type, 10% of the power-frequency periodic discharge S-transform time-frequency graphs are selected to construct a model training set, and the remaining power-frequency periodic discharge S-transform time-frequency graphs serve as a test set. In some embodiments, interference images may be added to the model training set to enhance the deep learning network model's ability to process power-frequency periodic discharge S-transform time-frequency graphs. The training set is input into the deep learning network model, and the network parameters are optimized using the Adam algorithm. The maximum number of training attempts, initial learning rate, and learning rate reduction factor are set. The specific values ​​can be set according to actual needs.

[0096] Step 6: Input the test set into the trained deep learning model for recognition verification.

[0097] The S-transformed time-frequency graphs of power frequency periodic discharges of different samples in the test set were input into the model for recognition to verify the recognition effect of the above method. Figure 4 、 Figure 5 as well as Figure 6 Three classic insulation defect models, namely needle-plate defect, pillar-plate defect and ball-plate defect, are established respectively. Figure 2 In the partial discharge test system shown, the partial discharge signal of each type of defect is collected; during the test, each test sample is applied with an industrial frequency voltage with a voltage amplitude of 1.6 times the initial discharge voltage. Four samples are tested for each defect, and 500 industrial frequency discharge cycle signals are collected for each sample. Secondly, the sample signals are denoised using the wavelet denoising method. The db2 wavelet is selected as the wavelet base, the number of decomposition layers is 7, and an adaptive threshold is adopted to perform denoising using the soft threshold method. Then, each sample data is converted into a time-frequency matrix through S transformation, and the corresponding three-primary color matrix is ​​formed. After optimizing the color weights, the industrial frequency periodic discharge S transform time-frequency graph is drawn, and the image size is reduced to 360×360×3. Finally, a deep learning network model is constructed, and 10% of the industrial frequency periodic discharge S transform time-frequency graphs in each sample defect are input into the model for training. After the loss function converges, the remaining images are input for recognition verification. The pattern recognition results are shown as follows. Figure 7 , the accuracy rate reached 98.67%.

[0098] Figure 8 The present invention also provides a schematic diagram of a device structure for identifying a partial discharge pattern of an insulation, as shown in FIG. Figure 8 , a discharge signal acquisition module 210, an information extraction module 220, a time-frequency diagram generation module 230, and a defect type output module 240.

[0099] The discharge signal acquisition module 210 is used to collect the partial discharge signal of the insulation defect sample to be identified;

[0100] An information extraction module 220 is configured to extract phase distribution information of the partial discharge signal;

[0101] A time-frequency diagram generating module 230 is configured to generate a time-frequency diagram of the insulation defect sample to be identified based on the time-frequency domain features of the phase distribution information and weight coefficients corresponding to the time-frequency domain features;

[0102] The defect type output module 240 is used to input the time-frequency diagram into a pre-trained discharge pattern recognition model based on a deep learning network, and output the defect type of the insulation defect sample to be identified.

[0103] The technical solution provided by the embodiment of the present disclosure utilizes this method to solve the problem that insufficient discharge features easily lead to underfitting of the classifier when valid samples of different types of partial discharges are very limited, thereby improving the recognition accuracy.

[0104] Furthermore, the information extraction module 220 may be used to:

[0105] The phase distribution information includes: maximum discharge amount, average discharge amount and discharge times;

[0106] Dividing the partial discharge signal of each cycle into a first number of data portions according to phase;

[0107] determining the maximum discharge capacity, the average discharge capacity, and the number of discharges for each of the first number of pieces of data;

[0108] All of the maximum discharge amounts, the average discharge amounts, and the number of discharges are used as the phase distribution information.

[0109] Furthermore, the time-frequency diagram generating module 230 may also be used to:

[0110] The phase distribution information is transformed according to an S-transformation formula to obtain a time-frequency domain characteristic matrix of the phase distribution information, wherein the behavior frequency of the time-frequency domain characteristic matrix is ​​a column of phase;

[0111] The time-frequency domain feature matrix is ​​multiplied by the weight coefficient to generate a color matrix, and the color matrix is ​​used as the time-frequency graph.

[0112] Furthermore, the device can also be used for:

[0113] Get the set initial weight coefficient;

[0114] Iteratively optimizing the initial weight coefficients by using a particle swarm algorithm to obtain multiple optimized weight coefficients;

[0115] The time-frequency graph corresponding to each of the optimized weight coefficients is input into a pre-trained convolutional neural network recognition model for recognition, and the optimized weight coefficient with the highest recognition accuracy is used as the weight coefficient.

[0116] Furthermore, the device can also be used for:

[0117] The discharge pattern recognition model based on a deep learning network includes: an input layer, a convolutional layer, a pooling layer, a bidirectional gated recurrent network, and three fully connected layers; wherein the convolution kernel in the convolutional layer has a size of 3×3, a number of 16, a step size of 1, and a ReLU function as an activation function; the pooling kernel size of the pooling layer has a size of 2×2, and a step size of 2; and the bidirectional gated recurrent network includes 32 units.

[0118] Furthermore, the device can also be used for:

[0119] The defect types of the insulation defect samples to be identified include: pin-plate defect, pillar-plate defect and ball-plate defect.

[0120] Furthermore, the device can also be used for:

[0121] After collecting the partial discharge signal of the insulation defect sample to be identified, the method further includes:

[0122] De-noising is performed on the partial discharge signal.

[0123] The above device can execute the methods provided by all the above embodiments of the present invention, and has the corresponding functional modules and beneficial effects of executing the above methods. For technical details not fully described in this embodiment, please refer to the methods provided by all the above embodiments of the present invention.

[0124] Figure 9 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is provided. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0125] like Figure 9As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0126] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0127] The processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any other suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the insulation partial discharge pattern recognition method.

[0128] In some embodiments, the insulation partial discharge pattern recognition method can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the insulation partial discharge pattern recognition method described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to execute the insulation partial discharge pattern recognition method in any other suitable manner (e.g., via firmware).

[0129] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0130] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0131] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0132] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0133] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0134] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0135] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0136] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for identifying insulation partial discharge patterns, characterized in that: include: Collecting partial discharge signals of insulation defect samples to be identified; extracting phase distribution information of the partial discharge signal; Generating a time-frequency diagram of the insulation defect sample to be identified based on the time-frequency domain characteristics of the phase distribution information and the weight coefficients corresponding to the time-frequency domain characteristics; The time-frequency graph is input into a pre-trained discharge pattern recognition model based on a deep learning network, and the defect type of the insulation defect sample to be identified is output.

2. The method according to claim 1, characterized in that The phase distribution information includes: maximum discharge amount, average discharge amount and number of discharges; correspondingly, the phase distribution information of the partial discharge signal is extracted, including: Dividing the partial discharge signal of each cycle into a first number of data portions according to phase; determining the maximum discharge capacity, the average discharge capacity, and the number of discharges for each of the first number of pieces of data; All of the maximum discharge amounts, the average discharge amounts, and the number of discharges are used as the phase distribution information.

3. The method according to claim 1, characterized in that The generating of the time-frequency diagram of the insulation defect sample to be identified based on the time-frequency domain features of the phase distribution information and the weight coefficients corresponding to the time-frequency domain features includes: The phase distribution information is transformed according to an S-transformation formula to obtain a time-frequency domain characteristic matrix of the phase distribution information, wherein the behavior frequency of the time-frequency domain characteristic matrix is ​​a column of phase; The time-frequency domain feature matrix is ​​multiplied by the weight coefficient to generate a color matrix, and the color matrix is ​​used as the time-frequency graph.

4. The method according to claim 1, wherein The process of setting the weight coefficient includes: Get the set initial weight coefficient; Iteratively optimizing the initial weight coefficients by using a particle swarm algorithm to obtain multiple optimized weight coefficients; The time-frequency graph corresponding to each of the optimized weight coefficients is input into a pre-trained convolutional neural network recognition model for recognition, and the optimized weight coefficient with the highest recognition accuracy is used as the weight coefficient.

5. The method according to claim 1, characterized in that: The discharge pattern recognition model based on a deep learning network includes: an input layer, a convolutional layer, a pooling layer, a bidirectional gated recurrent network, and three fully connected layers; wherein the convolution kernel in the convolutional layer has a size of 3×3, a number of 16, a step size of 1, and a ReLU function as an activation function; the pooling kernel size of the pooling layer has a size of 2×2, and a step size of 2; and the bidirectional gated recurrent network includes 32 units.

6. The method according to claim 1, characterized in that The defect types of the insulation defect samples to be identified include: pin-plate defect, pillar-plate defect and ball-plate defect.

7. The method according to claim 1, characterized in that After collecting the partial discharge signal of the insulation defect sample to be identified, the method further includes: De-noising is performed on the partial discharge signal.

8. An insulation partial discharge pattern recognition device, characterized in that: include: A discharge signal acquisition module is used to collect partial discharge signals of insulation defect samples to be identified; An information extraction module, configured to extract phase distribution information of the partial discharge signal; a time-frequency diagram generating module, configured to generate a time-frequency diagram of the insulation defect sample to be identified based on the time-frequency domain features of the phase distribution information and weight coefficients corresponding to the time-frequency domain features; A defect type output module is used to input the time-frequency diagram into a pre-trained discharge pattern recognition model based on a deep learning network, and output the defect type of the insulation defect sample to be identified.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the insulation partial discharge pattern recognition method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the insulation partial discharge pattern recognition method according to any one of claims 1 to 7 when executed.

Images