Self-adaptive data processing and self-diagnosis method and system for partial discharge characteristics of KYN equipment
By using adaptive data processing and convolutional neural network technology in KYN equipment to build a feature map, the problem of insufficient efficiency and accuracy of local discharge diagnosis of KYN cable accessories in the existing technology is solved, and efficient fault identification and evaluation is achieved .
Patent Information
- Application Number
- CN202410807569.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-21
- Publication Date
- 2024-11-15
AI Technical Summary
The existing KYN cable accessories partial discharge diagnosis methods are difficult to fully reflect the complex PD phenomenon, the data acquisition efficiency of a single sensor is low, and the traditional feature extraction algorithm cannot dig deep into the data, resulting in insufficient diagnostic accuracy and efficiency.
The modular KYN device is equipped with a built-in TEV sensor and data acquisition device to collect data in real time, and data cleaning and dimensionality reduction are performed through adaptive data processing methods. The convolutional neural network is used to extract features and build feature maps, and automatic diagnosis is performed in combination with attention mechanism.
It realizes a comprehensive characterization of KYN equipment localized characteristics, improves the accuracy and efficiency of diagnosis, reduces calculation costs, and ensures the reliability and practical application capabilities of the diagnostic system.
Smart Images

Figure CN118965207A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of partial discharge diagnosis of KYN cable accessories, and in particular to a method and system for adaptive data processing and self-diagnosis of partial discharge characteristics of KYN equipment. Background Art
[0002] In the current field of partial discharge (PD) diagnosis of KYN cable accessories, traditional methods have many limitations. First, it is difficult for a single sensor data collection to fully reflect the complexity and diversity of PD phenomena, and it is impossible to fully mine the correlation information between different sensors. Secondly, the existing feature extraction algorithms are often limited to simple statistical features or frequency domain features, and cannot fully mine the deep information in the data. In addition, the traditional methods have low processing efficiency for large-scale data, and the accuracy and robustness of PD pattern recognition need to be improved. Therefore, in view of the limitations of existing technologies, a more advanced and efficient data processing and diagnosis method is needed to improve the accuracy, efficiency and reliability of PD diagnosis.
[0003] Traditional PD diagnosis methods rely only on partial discharge data collected by a single sensor, which cannot fully reflect the complex PD phenomenon in cable accessories. Traditional methods are limited by simple statistical features and frequency domain features, and it is difficult to fully mine the deep information in the data, resulting in low diagnostic accuracy. Traditional methods are inefficient when processing large-scale data and cannot meet the requirements for data processing efficiency in engineering practice. The existing algorithms have high complexity in identifying PD feature maps, resulting in high computational costs and are not conducive to practical applications. Traditional methods cannot achieve real-time diagnosis of PD feature maps, which limits their scope of use in practical applications. Summary of the invention
[0004] In view of the above existing problems, the present invention is proposed.
[0005] Therefore, the present invention provides an adaptive data processing and self-diagnosis method for partial discharge characteristics of KYN equipment, which can collect, preprocess and extract characteristics of partial discharge signals of KYN equipment, and then construct a partial discharge characteristic map, so that the partial discharge characteristics of the equipment can be comprehensively and systematically characterized.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for adaptive data processing and self-diagnosis of partial discharge characteristics of KYN equipment, comprising: using the TEV sensor and data acquisition device built into the modular KYN equipment to collect various data generated during the operation of the equipment in real time; using an adaptive data processing method to clean, denoise and reduce the dimension of the collected raw data; using an adaptive convolutional neural network to extract key features from the preprocessed data and construct a feature map; based on the constructed feature map, using an adaptive convolutional neural network combined with an attention mechanism to automatically diagnose and analyze the operating status of the equipment.
[0007] As a preferred solution of the method for adaptive data processing and self-diagnosis of partial discharge characteristics of KYN equipment described in the present invention, the data collection includes arranging TEV sensors at key parts of the equipment to collect data of various physical quantities, and the TEV sensors are arranged at various important parts inside the KYN equipment;
[0008] The ARM+FPGA dual-core system is used to complete the sampling and simple processing of discharge data. The FPGA is the main control, responsible for controlling the sampling speed of the AD chip and caching the AD sampling data into the internal RAM memory; the ARM is responsible for reading the data in the FPGA internal RAM and uploading it to the PC for further processing; in terms of hardware selection, the Cyclone IV series FPGA is selected, and the RAM in the FPGA is used as an external memory. The values stored in the two RAMs are read and instructions are transmitted through the FSMC, and the data read from the RAM is uploaded to the PC through the UART;
[0009] Transient ground voltage frequency upper limit f h =60MHz, lower cutoff f l =3MHz, characteristic frequency f0=32MHz, passband ripple δ=0.05dB, stopband attenuation 40dB, matching impedance 50Ω.
[0010] As a preferred solution of the KYN equipment partial discharge feature adaptive data processing and self-diagnosis method described in the present invention, wherein: the adaptive data processing method includes data preprocessing, feature extraction and selection, and feature representation and encoding;
[0011] First, data cleaning is performed to remove outliers and noise interference factors, and a Chebyshev I filter is used to design a bandpass filter circuit:
[0012]
[0013] Where C(x) represents the cleaned signal, μ k , σ krepresents the mean and standard deviation of the kth Gaussian distribution, K represents the number of Gaussian distributions, Tn(x) is the nth order Chebyshev polynomial, ε is the maximum value of the passband fluctuation, and x0 represents the normalized angular frequency;
[0014] The wavelet transform method is used for time-frequency analysis. The signal is decomposed into wavelet coefficients of different scales and frequencies. The short-time Fourier transform and Wigner-Ville distribution are used to deeply explore the time-frequency characteristics of the signal. The adaptive filtering algorithm is introduced to perform adaptive mean filtering and adaptive median filtering. The wavelet analysis and wavelet packet analysis are used to perform time-frequency analysis and feature extraction on the signal to obtain representative time-frequency features:
[0015]
[0016] Where T(x) represents the result of short-time Fourier transform, V(x) represents the result of Wigner-Ville distribution, A(x) represents the signal after adaptive filtering, O(x) represents the final output signal, and c j,k represents the wavelet packet coefficient of the kth node in the jth layer, and E represents the signal energy.
[0017] As a preferred solution of the KYN equipment partial discharge feature adaptive data processing and self-diagnosis method described in the present invention, wherein: the feature extraction and selection includes extracting effective features that can reflect the equipment status and fault characteristics from the pre-processed data, and extracting the features of the discharge data as the judgment conditions for defect classification and identification;
[0018] Characterizes the average level of discharge intensity for different defect types:
[0019]
[0020] Among them, x i is the amplitude in the data, and n is the number;
[0021] Characterize the distribution dispersion of different defect discharges:
[0022]
[0023] The average amount of information after excluding redundant information:
[0024]
[0025] Among them, H(X) is the information entropy, X is the set of events, n is the number of event sets, p(x i ) is the probability of the number in the entire set, and the calculation method is:
[0026]
[0027] As a preferred solution of the KYN equipment partial discharge feature adaptive data processing and self-diagnosis method described in the present invention, the feature representation and encoding includes converting the original data into a numerical vector form, realizing it through vectorization processing, using principal component analysis and linear discriminant analysis to extract the most representative features, reducing data complexity, combining features, and generating new features through addition, subtraction, multiplication, division, and polynomial expansion; after the feature representation is completed, the discrete features are encoded using the one-hot encoding method for learning and prediction by the machine learning model.
[0028] As a preferred solution of the KYN equipment partial discharge feature adaptive data processing and self-diagnosis method of the present invention, wherein: the feature map and construction include, based on the extracted features, further screening and selection, determining the most representative and discriminating feature subset for the construction of the feature map;
[0029] Correlation analysis, variance analysis, and principal component analysis are used for feature selection. For high-dimensional feature space, feature dimension reduction methods are used to map the feature space to low-dimensional space. Principal component analysis, linear discriminant analysis, and t-distribution neighborhood embedding are used to retain the main structure and features of the data while reducing the dimension of the features.
[0030] In the process of constructing the feature map, the features are clustered and grouped according to the similarity and correlation between the features to form representative feature groups. The local discharge data in bin format collected by the detection terminal is processed by Python. The collected data is read in real-time loop with a period of 2s. The data is processed with a period of 2s. 64 cycles of discharge data are collected in each cycle at a working frequency of 50Hz. The discharge peak in the first cycle is set to 90 phases, and the phase is corrected every 64 cycles.
[0031] As a preferred solution of the KYN equipment partial discharge feature adaptive data processing and self-diagnosis method described in the present invention, the automatic diagnosis includes: using PyTorch's `ImageFolder` to load an image dataset, the dataset is organized into a folder structure, each folder represents a category, which contains images of this category, and pre-processing the image through a predefined transformation `transform`, resizing the image to 224x224 pixels, converting it into a tensor, and normalizing the image to scale the pixel value from the [0,255] range to the [-1,1] range;
[0032] Build an image classification model `FourClassNet` based on VGG neural network. The model includes three convolutional layers, pooling layers and two fully connected layers. Use ReLU activation function for nonlinear transformation and output probability distribution after log_softmax. Use cross entropy loss function to measure the difference between model output and true label. Use stochastic gradient descent as optimizer to update model parameters. Set learning rate to 0.001 and momentum to 0.9. Train the model through multiple epoch iterations. Perform multiple mini-batch training on the training data set in each epoch. Calculate loss and accuracy.
[0033] During the training process, the training data set is used to adjust the model parameters and minimize the loss function. The performance of the model is evaluated on the test set. The classification effect of the model is measured by calculating the accuracy and confusion matrix of the test set. The confusion matrix shows the classification of the model in each category, including true positive examples, false positive examples, true negative examples, and false negative examples, which can intuitively evaluate the classification accuracy and error rate of the model in different categories.
[0034] Based on the prediction results of the model on the test set, automatic diagnosis is performed to identify possible abnormal conditions, fault characteristics or warning signals of the equipment. The evaluation results of the model on the test set are saved as Excel files, and a visualization diagram of the confusion matrix is drawn. Further analysis and improvement are carried out based on the evaluation results.
[0035] Another object of the present invention is to provide a KYN equipment partial discharge feature adaptive data processing and self-diagnosis system, which can use deep learning algorithms and frequency domain analysis methods to extract rich and efficient features from partial discharge feature spectrum data, fully tap the potential information in the data, and improve the performance and robustness of the diagnostic model. By training and optimizing the model, the model's recognition accuracy and stability for different defect types can be improved, ensuring the reliability and effectiveness of the diagnostic system in practical applications.
[0036] As a preferred solution of the KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis system described in the present invention, it includes: a data acquisition module, a data preprocessing module, a feature extraction module, a feature spectrum construction module, an automatic diagnosis module, a real-time monitoring and early warning module, a user interface and a report generation module;
[0037] The data acquisition module collects the partial discharge signal of the KYN equipment, and the signal includes an acoustic emission signal, an ultrasonic signal, and a transient ground voltage signal;
[0038] The data preprocessing module performs denoising, filtering, and normalization preprocessing operations on the collected signals to improve data quality;
[0039] The feature extraction module extracts key features from the preprocessed signal, including time domain features, frequency domain features, and time-frequency features;
[0040] The feature map construction module displays the extracted features in the form of a map to facilitate visual analysis and subsequent processing;
[0041] The automatic diagnosis module uses a deep learning algorithm to identify the device status and diagnose faults based on the feature map;
[0042] The real-time monitoring and early warning module monitors the equipment status in real time and issues early warnings in time when abnormalities are found to ensure safe operation of the equipment;
[0043] The user interface and report generation module provides a user operation interface, displays diagnostic results and warning information, and generates reports for reference by maintenance and management personnel.
[0044] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the KYN device partial discharge characteristic adaptive data processing and self-diagnosis method are implemented.
[0045] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of a method for adaptive data processing and self-diagnosis of partial discharge characteristics of a KYN device are implemented.
[0046] Beneficial effects of the present invention: This method can automatically adapt to different types and sizes of partial discharge characteristic spectrum data through a modular data processing process, thereby realizing flexible processing and efficient use of data. A partial discharge characteristic spectrum diagnostic system based on an advanced model has been established, which can accurately and quickly diagnose and evaluate the defects of KYN equipment, thereby improving the efficiency of fault detection and troubleshooting. The diagnostic results are presented to users in the form of charts, reports, etc., so that users can intuitively understand the health status of cable accessories, take necessary maintenance and repair measures in a timely manner, and improve the reliability and safety of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:
[0048] Figure 1 A schematic flow chart of a method for adaptive data processing and self-diagnosis of partial discharge characteristics of a KYN device provided in accordance with an embodiment of the present invention.
[0049] Figure 2 A schematic diagram of data sampling and processing of a method for adaptive data processing and self-diagnosis of partial discharge characteristics of a KYN device provided in one embodiment of the present invention.
[0050] Figure 3 A PRPD spectrum of a KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method provided by an embodiment of the present invention.
[0051] Figure 4 This is a diagram of confusion matrix recognition results of the KYN equipment partial discharge feature adaptive data processing and self-diagnosis method provided by an embodiment of the present invention.
[0052] Figure 5 A schematic diagram of the working modules of a KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis system provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0053] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.
[0054] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0055] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0056] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.
[0057] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.
[0058] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0059] Example 1
[0060] Reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method, including:
[0061] S1: Use the TEV sensor and data acquisition device built into the modular KYN equipment to collect various data generated during the operation of the equipment in real time.
[0062] Furthermore, the data collection includes arranging TEV sensors at key parts of the equipment to collect data of various physical quantities. The TEV sensors are arranged at various important parts inside the KYN equipment;
[0063] The ARM+FPGA dual-core system is used to complete the sampling and simple processing of discharge data. The FPGA is the main control, responsible for controlling the sampling speed of the AD chip and caching the AD sampling data into the internal RAM memory; the ARM is responsible for reading the data in the FPGA internal RAM and uploading it to the PC for further processing; in terms of hardware selection, the Cyclone IV series FPGA is selected, and the RAM in the FPGA is used as an external memory. The values stored in the two RAMs are read and instructions are transmitted through the FSMC, and the data read from the RAM is uploaded to the PC through the UART;
[0064] Transient ground voltage frequency upper limit f h =60MHz, lower cutoff f l=3MHz, characteristic frequency f0=32MHz, passband ripple δ=0.05dB, stopband attenuation 40dB, matching impedance 50Ω. Perform quality control on the collected data, including data verification, outlier detection, data cleaning, etc. Through data quality control, noise and interference that may be introduced during the collection process can be eliminated to ensure that the collected data is of high quality and credibility.
[0065] It should be noted that the TEV sensor and data acquisition device built into the modular KYN equipment are used to collect various data generated during the operation of the equipment in real time, including parameters such as voltage, current, and temperature. The collected data covers all aspects of the equipment's operating status, such as electrical characteristics, thermal characteristics, etc., and has rich information. The acquisition frequency is set to 3MHz~60MHz to meet the real-time and accuracy requirements in different application scenarios.
[0066] S2: Adopt adaptive data processing methods to clean, denoise and reduce the dimension of the collected raw data.
[0067] Furthermore, the adaptive data processing method includes data preprocessing, feature extraction and selection, and feature representation and encoding;
[0068] First, data cleaning is performed to remove outliers and noise interference factors, and a Chebyshev I filter is used to design a bandpass filter circuit:
[0069]
[0070] Where C(x) represents the cleaned signal, μ k , σ k represents the mean and standard deviation of the kth Gaussian distribution, K represents the number of Gaussian distributions, Tn(x) is the nth order Chebyshev polynomial, ε is the maximum value of the passband fluctuation, and x0 represents the normalized angular frequency;
[0071] Apply wavelet transform method for time-frequency analysis, decompose the signal into wavelet coefficients of different scales and frequencies, and use short-time Fourier transform and Wigner-Ville distribution to deeply explore the time-frequency characteristics of the signal:
[0072]
[0073] Adaptive filtering algorithm is introduced to perform adaptive mean filtering and adaptive median filtering. Wavelet analysis and wavelet packet analysis are used to perform time-frequency analysis and feature extraction on the signal to obtain representative time-frequency features:
[0074]
[0075] Where T(x) represents the result of short-time Fourier transform, V(x) represents the result of Wigner-Ville distribution, A(x) represents the signal after adaptive filtering, O(x) represents the final output signal, and c j,k represents the wavelet packet coefficient of the kth node in the jth layer, and E represents the signal energy.
[0076] It should be noted that the feature extraction and selection includes extracting effective features that can reflect the equipment status and fault characteristics from the preprocessed data, and extracting the features of the discharge data as the judgment conditions for defect classification and identification;
[0077] Characterizes the average level of discharge intensity for different defect types:
[0078]
[0079] Among them, x i is the amplitude in the data, and n is the number;
[0080] Characterize the distribution dispersion of different defect discharges:
[0081]
[0082] The average amount of information after excluding redundant information:
[0083]
[0084] Among them, H(X) is the information entropy, X is the set of events, n is the number of event sets, p(x i ) is the probability of the number in the entire set, and the calculation method is:
[0085]
[0086] Furthermore, the feature representation and encoding includes converting the original data into a numerical vector form through vectorization processing, using principal component analysis and linear discriminant analysis to extract the most representative features, reduce data complexity, combine features, and generate new features through addition, subtraction, multiplication, division, and polynomial expansion. After the feature representation is completed, the discrete features are encoded using a one-hot encoding method to facilitate learning and prediction by the machine learning model.
[0087] S3: Use an adaptive convolutional neural network to extract key features from the preprocessed data and construct a feature map.
[0088] Furthermore, the feature map and construction includes, based on the extracted features, further screening and selecting to determine the most representative and discriminative feature subset for the construction of the feature map;
[0089] Correlation analysis, variance analysis, and principal component analysis are used for feature selection. For high-dimensional feature space, feature dimension reduction methods are used to map the feature space to low-dimensional space. Principal component analysis, linear discriminant analysis, and t-distribution neighborhood embedding are used to retain the main structure and features of the data while reducing the dimension of the features.
[0090] In the process of constructing the feature map, the features are clustered and grouped according to the similarity and correlation between the features to form representative feature groups. The local discharge data in bin format collected by the detection terminal is processed by Python. The collected data is read in real-time loop with a period of 2s. The data is processed with a period of 2s. 64 cycles of discharge data are collected in each cycle at a working frequency of 50Hz. The discharge peak in the first cycle is set to 90 phases, and the phase is corrected every 64 cycles.
[0091] S4: Based on the constructed feature graph, an adaptive convolutional neural network combined with an attention mechanism is used to automatically diagnose and analyze the operating status of the device.
[0092] Furthermore, the automatic diagnosis includes loading an image dataset using PyTorch's `ImageFolder`, where the dataset is organized into a folder structure, each folder represents a category, and contains images of this category, performing a preprocessing operation on the image through a predefined transformation `transform`, resizing the image to 224x224 pixels, converting it to a tensor, and normalizing the image to scale the pixel value from the [0,255] range to the [-1,1] range;
[0093] Build an image classification model `FourClassNet` based on VGG neural network. The model includes three convolutional layers, pooling layers and two fully connected layers. Use ReLU activation function for nonlinear transformation and output probability distribution after log_softmax. Use cross entropy loss function to measure the difference between model output and true label. Use stochastic gradient descent as optimizer to update model parameters. Set learning rate to 0.001 and momentum to 0.9. Train the model through multiple epoch iterations. Perform multiple mini-batch training on the training data set in each epoch. Calculate loss and accuracy.
[0094] During the training process, the training data set is used to adjust the model parameters and minimize the loss function. The performance of the model is evaluated on the test set. The classification effect of the model is measured by calculating the accuracy and confusion matrix of the test set. The confusion matrix shows the classification of the model in each category, including true positive examples, false positive examples, true negative examples, and false negative examples, which can intuitively evaluate the classification accuracy and error rate of the model in different categories.
[0095] Based on the prediction results of the model on the test set, automatic diagnosis is performed to identify possible abnormal conditions, fault characteristics or warning signals of the equipment. The evaluation results of the model on the test set are saved as Excel files, and a visualization diagram of the confusion matrix is drawn. Further analysis and improvement are carried out based on the evaluation results.
[0096] Example 2
[0097] Reference Figure 3 and Figure 4 , which is an embodiment of the present invention, provides a method for adaptive data processing and self-diagnosis of partial discharge characteristics of KYN equipment. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.
[0098] This design uses Python to process the local discharge data in bin format collected by the detection terminal. The collected data is read in real time with a cycle of 2s, so the data is also processed with a cycle of 2s. At a 50Hz power frequency, 64 cycles of discharge data can be collected in each cycle. The discharge peak in the first cycle is set to 90 degrees, and the phase is corrected every 64 cycles. This ensures that the phase does not shift in each cycle of data collection. The PRPD spectrum is as follows: Figure 3 Shown
[0099] Build an image classification model `FourClassNet` based on VGG neural network. The model includes three convolutional layers, pooling layers and two fully connected layers. The ReLU activation function is used for nonlinear transformation, and the final output is the probability distribution after log_softmax. The cross entropy loss function is used to measure the difference between the model output and the true label, and the stochastic gradient descent (SGD) is used as the optimizer to update the model parameters. The learning rate is set to 0.001 and the momentum is set to 0.9. The training model is iterated through multiple epochs. In each epoch, the training data set is trained multiple times with mini-batch training, and the loss and accuracy are calculated.
[0100] During the training process, the training dataset is used to adjust the model parameters to minimize the loss function, so that the model can better fit the training data and improve the classification performance.
[0101] Evaluate the performance of the model on the test set, and measure the classification effect of the model by calculating the accuracy and confusion matrix of the test set. The confusion matrix shows the classification of the model in each category, including true positive examples, false positive examples, true negative examples, and false negative examples, so that the classification accuracy and error rate of the model in different categories can be intuitively evaluated. Based on the prediction results of the model on the test set, automatic diagnosis is performed to identify possible abnormal conditions, fault characteristics or early warning signals of the equipment, thereby improving the safety, stability and reliability of the equipment. Save the evaluation results of the model on the test set as an Excel file, and draw a visualization of the confusion matrix so that engineers and technicians can intuitively understand the classification effect of the model and conduct further analysis and improvement based on the evaluation results. The confusion matrix recognition results are as follows: Figure 4 shown.
[0102] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
[0103] Example 3
[0104] The third embodiment of the present invention is different from the first two embodiments in that:
[0105] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.
[0106] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.
[0107] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.
[0108] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0109] Example 4
[0110] Reference Figure 5 , which is an embodiment of the present invention, provides a KYN equipment partial discharge feature adaptive data processing and self-diagnosis system, which is characterized by: including a data acquisition module, a data preprocessing module, a feature extraction module, a feature spectrum construction module, an automatic diagnosis module, a real-time monitoring and early warning module, a user interface and a report generation module.
[0111] The data acquisition module collects the partial discharge signals of the KYN equipment, which include acoustic emission signals, ultrasonic signals, and transient ground voltage signals.
[0112] The data preprocessing module performs denoising, filtering, and normalization preprocessing operations on the collected signals to improve data quality.
[0113] The feature extraction module extracts key features from the preprocessed signal, including time domain features, frequency domain features, and time-frequency features.
[0114] The feature map construction module displays the extracted features in the form of a map to facilitate visual analysis and subsequent processing.
[0115] The automatic diagnosis module uses deep learning algorithms to identify device status and diagnose faults based on feature maps.
[0116] The real-time monitoring and early warning module monitors the equipment status in real time and issues early warnings when abnormalities are found to ensure the safe operation of the equipment.
[0117] The user interface and report generation module provides a user operation interface, displays diagnostic results and warning information, and generates reports for reference by maintenance and management personnel.
[0118] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method, characterized by: include, Use the TEV sensors and data acquisition devices built into the modular KYN equipment to collect equipment operation data in real time; Adopt adaptive data processing methods to clean, denoise and reduce the dimension of the collected raw data of equipment operation; Use adaptive convolutional neural networks to extract key features from preprocessed data and construct feature maps; Based on the constructed feature graph, an adaptive convolutional neural network combined with an attention mechanism is used to automatically diagnose and analyze the operating status of the equipment.
2. The KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method according to claim 1, characterized in that: The real-time acquisition of equipment operation data includes using an ARM+FPGA dual-core system to complete the sampling and processing of discharge data. The FPGA is used as the main control to control the sampling speed of the AD chip and cache the AD sampling data into the internal RAM memory; ARM is responsible for reading the data in the FPGA's internal RAM and uploading it to the PC for processing; The values stored in the two RAMs are read and instructions are transmitted through FSMC, and the data read from the RAMs are uploaded to the PC through UART. Transient ground voltage frequency upper limit f h =60MHz, lower cutoff f l =3MHz, characteristic frequency f0=32MHz, passband ripple δ=0.05dB, stopband attenuation 40dB, matching impedance 50Ω.
3. The KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method according to claim 2, characterized in that: The adaptive data processing method includes data preprocessing, feature extraction and selection, and feature representation and encoding; The data is cleaned to remove outliers and noise interference factors, and a Chebyshev I type filter is used to design a bandpass filter circuit: Where C(x) represents the cleaned signal, μ k , σ k represents the mean and standard deviation of the kth Gaussian distribution, K represents the number of Gaussian distributions, Tn(x) is the nth order Chebyshev polynomial, ε is the maximum value of the passband fluctuation, x a represents the normalized angular frequency; Apply the wavelet transform method to perform time-frequency analysis, decompose the signal into wavelet coefficients of different scales and frequencies, introduce the adaptive filtering algorithm, perform adaptive mean filtering and adaptive median filtering, use wavelet analysis and wavelet packet analysis to perform time-frequency analysis and feature extraction on the signal, and obtain representative time-frequency features: Where T(x) represents the result of short-time Fourier transform, V(x) represents the result of Wigner-Ville distribution, A(x) represents the signal after adaptive filtering, O(x) represents the final output signal, and c j,k represents the wavelet packet coefficient of the kth node in the jth layer, and E represents the signal energy.
4. The KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method according to claim 3, characterized in that: The feature extraction and selection includes extracting effective features that can reflect the equipment status and fault characteristics from the preprocessed data, and extracting the features of the discharge data as the judgment conditions for defect classification and identification; Characterizes the average level of discharge intensity for different defect types: Among them, x i is the amplitude in the data, and n is the number; Characterize the distribution dispersion of different defect discharges: The average amount of information after excluding redundant information: Among them, H(X) is the information entropy, X is the set of events, n is the number of event sets, p(x i ) is the probability of the number in the entire set, and the calculation method is:
5. The KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method according to claim 4, characterized in that: The feature representation and encoding includes converting the original data into a numerical vector form, implementing it through vectorization processing, using principal component analysis and linear discriminant analysis to extract the most representative features, reduce data complexity, and perform feature combination to generate new features; After the feature representation is completed, the discrete features are encoded using the one-hot encoding method.
6. The KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method according to claim 5, characterized in that: The feature map and construction includes, based on the extracted features, further screening and selecting to determine the most representative and discriminative feature subset for the construction of the feature map; After feature selection, for high-dimensional feature space, feature dimensionality reduction method is used to map the feature space to low-dimensional space and construct feature map; The features are clustered and grouped according to the similarity and correlation between the features to form feature groups. The local discharge data in bin format collected by the detection terminal is processed by Python. The collected data is read in real-time loop with a period of 2s. The data is processed with a period of 2s. 64 cycles of discharge data are collected in each cycle at a working frequency of 50Hz. The discharge peak in the first cycle is set to 90 phases, and the phase is corrected every 64 cycles.
7. The KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method according to claim 6, characterized in that: The automatic diagnosis includes loading an image dataset using PyTorch's `ImageFolder`, organizing the dataset into a folder structure, where each folder represents a category and contains images of the current category, performing a preprocessing operation on the image through a predefined transformation `transform`, resizing the image to 224x224 pixels, converting it to a tensor, and normalizing the image to scale the pixel value from the range of [0,255] to the range of [-1,1]; Build an image classification model `FourClassNet` based on the VGG neural network. The model includes three convolutional layers, a pooling layer, and two fully connected layers. The ReLU activation function is used for nonlinear transformation, and the output is a probability distribution after log_softmax. The cross entropy loss function is used to measure the difference between the model output and the true label. Stochastic gradient descent is used as the optimizer to update the model parameters. The learning rate is set to 0.001 and the momentum is set to 0.
9. The training model is iterated through multiple epochs. Multiple mini-batch training is performed on the training data set in each epoch. The loss and accuracy are calculated; During the training process, the training data set is used to adjust the model parameters and minimize the loss function. The performance of the model is evaluated on the test set. The classification effect of the model is measured by calculating the accuracy and confusion matrix of the test set. The confusion matrix shows the classification of the model in each category, including true positive examples, false positive examples, true negative examples, and false negative examples, which can intuitively evaluate the classification accuracy and error rate of the model in different categories. Based on the prediction results of the model on the test set, automatic diagnosis is performed to identify abnormal conditions, fault characteristics and warning signals of the equipment. The evaluation results of the model on the test set are saved as Excel files, and a visualization diagram of the confusion matrix is drawn. Further analysis and improvement are carried out based on the evaluation results.
8. A system using the KYN equipment partial discharge characteristic adaptive data processing and self-diagnosis method as claimed in any one of claims 1 to 7, characterized in that: It includes data acquisition module, data preprocessing module, feature extraction module, feature map construction module, automatic diagnosis module, real-time monitoring and early warning module, user interface and report generation module; The data acquisition module collects the partial discharge signal of the KYN equipment, and the signal includes an acoustic emission signal, an ultrasonic signal, and a transient ground voltage signal; The data preprocessing module performs denoising, filtering, and normalization preprocessing operations on the collected signals to improve data quality; The feature extraction module extracts key features from the preprocessed signal, including time domain features, frequency domain features, and time-frequency features; The feature map construction module displays the extracted features in the form of a map to facilitate visual analysis and subsequent processing; The automatic diagnosis module uses a deep learning algorithm to identify the device status and diagnose faults based on the feature map; The real-time monitoring and early warning module monitors the equipment status in real time and issues early warnings in time when abnormalities are found to ensure safe operation of the equipment; The user interface and report generation module provides a user operation interface, displays diagnostic results and warning information, and generates reports for reference by maintenance and management personnel.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
Citation Information
Cited By
Reactor partial discharge diagnosis method based on multi-source feature fusion and deep network
CN120011863A
Power equipment insulation state detection method and device, equipment and storage medium
CN120352743A
Method, device, equipment and storage medium for detecting insulation status of power equipment
CN120352743B
