Electrical equipment leakage identification method and identification device based on voiceprint signal

Abstract

The application provides an electrical equipment leakage identification method and device based on a voiceprint signal. The method comprises: processing a collected sound signal by using a variational mode decomposition algorithm to obtain a plurality of intrinsic mode functions; selecting three key intrinsic mode functions from the plurality of intrinsic mode functions as gray values of red, green and blue three color channels of an image respectively to generate a voiceprint image of the sound signal; extracting texture features of the voiceprint image through a gray level co-occurrence matrix, performing deep feature extraction and classification identification on the texture features by using a CNN-SVM model to obtain a leakage detection result; and when the leakage detection result indicates that a leakage fault occurs, determining a position of a fault source by calculating time delay differences of the sound signal propagating to a plurality of sound signal sensors. The scheme solves the problem that the prior art cannot provide accurate leakage fault source positioning, thereby prolonging the time for maintenance and fault elimination.

Description

Technical Field

[0001] This application relates to the field of power system automation technology, and more specifically, to a method, identification device, computer-readable storage medium, and electronic device for identifying leakage current in electrical equipment based on voiceprint signals. Background Technology

[0002] Leakage current detection in electrical equipment of power distribution networks is crucial for the safe operation of power systems. With the widespread application of AC / DC hybrid power supply networks, the diversification of electrical equipment types, and the increasing complexity of systems, traditional leakage current detection methods have gradually revealed their limitations. Traditional detection methods mainly rely on current difference, leakage current detection, or voltage imbalance analysis. These methods are easily affected by noise interference and nonlinear loads when dealing with complex electrical environments, resulting in low detection accuracy for leakage faults, especially in the detection of minor leakage and early-stage faults, where false positives or false negatives are prone to occur. Furthermore, with the continuous expansion of power distribution systems and the wide distribution of electrical equipment, the rapid location of leakage faults also faces significant challenges. Traditional methods often fail to provide accurate fault source location, leading to prolonged maintenance and troubleshooting times, affecting the stability and safety of power grid operation. Therefore, there is an urgent need for a detection method that can quickly locate fault sources to ensure the efficient operation of power distribution networks and the safe and stable operation of electrical equipment. Summary of the Invention

[0003] The main objective of this application is to provide a method, device, computer-readable storage medium, and electronic device for identifying electrical equipment leakage current based on voiceprint signals, so as to at least solve the problem that the prior art cannot provide accurate location of leakage current fault sources, resulting in extended maintenance and troubleshooting time.

[0004] To achieve the above objectives, according to one aspect of this application, a method for identifying leakage current in electrical equipment based on voiceprint signals is provided. The electrical equipment is AC / DC hybrid power distribution network equipment. The method includes: processing a collected acoustic signal using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions (IMFs), and selecting three key IMFs from the multiple IMFs; using the three key IMFs as grayscale values ​​for the red, green, and blue color channels of an image, respectively, to generate a voiceprint image of the acoustic signal; extracting texture features from the voiceprint image using a grayscale co-occurrence matrix, the texture features including contrast features, energy features, and entropy features; performing deep feature extraction and classification on the texture features using a CNN-SVM model to obtain a leakage current detection result; and when the leakage current detection result indicates a leakage current fault, calculating the time delay difference of the acoustic signal propagating to multiple acoustic signal sensors, and using the time delay difference to determine the location of the fault source.

[0005] Optionally, the acquired acoustic signal is processed using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions (EMFs), and three key EMFs are selected from these EMFs, including: applying Formula 1. The original acoustic signal is decomposed into a set of modal components, where K is the number of modal components, {u k} represents a set of modal components, {ω k} represents the center frequency corresponding to the modal component, u k (t) is the k-th modal component, ω k It is the center frequency of the kth modal component. The expression represents the derivative operation of the signal, j represents the imaginary unit, t is the time variable, * represents the convolution operation, and δ(t) is the Dirac function; for the first formula, the Lagrange multiplier λ and the penalty term α are introduced to construct the Lagrange function. The iterative optimization algorithm is used to solve u k ω k And λ(t), until the change of the modal component is less than a preset threshold, where λ(t) is the Lagrange multiplier function and f(t) is the original acoustic signal; analyze the characteristics of all the modal components, and select three modal components containing key signal information as the three key intrinsic mode functions.

[0006] Optionally, the three key intrinsic mode functions are used as grayscale values ​​of the red, green, and blue color channels of the image, respectively, to generate the voiceprint image of the sound signal, including: normalizing each key intrinsic mode function; creating a framework for an RGB image, the RGB image being composed of red, green, and blue color channels, each color channel representing information of different colors in the RGB image; and using formula P... n The voiceprint image is constructed using (i,j)=B((n-1)M×N / 2+M(i-1)+j), where P(.) represents the pixel grayscale value at the corresponding position in the RGB image, and B(.) represents the normalized amplitude of the key intrinsic mode function. The size of the RGB image is M×N, i=1:M, j=1:N, and the normalized amplitudes of the three key intrinsic mode functions are respectively used as the grayscale values ​​of the red (R), green (G), and blue (B) color channels of the RGB image.

[0007] Optionally, the process of constructing the gray-level co-occurrence matrix includes: selecting a reference point (x, y) and a co-occurrence point (x+a, y+b) on the voiceprint image, such that the gray values ​​of the reference point and the co-occurrence point form a gray-level value combination (g1, g2); letting the gray-level level of the voiceprint image be k, transforming the coordinates of the reference point and the co-occurrence point to obtain various different gray-level value combinations (g1, g2), wherein the number of combinations of gray-level value combinations (g1, g2) is k. 2 The process involves iterating through all reference points in the voiceprint image, counting the frequency of each grayscale value combination (g1, g2), and obtaining frequency statistics. These frequency statistics are then organized into a square matrix, which is normalized to convert into a probability matrix P(g1, g2). This probability matrix is ​​an initial grayscale co-occurrence matrix, where the rows and columns of the initial grayscale co-occurrence matrix correspond to the grayscale values ​​g1 and g2 in the voiceprint image, respectively. The formula is then used... The initial gray-level co-occurrence matrix is ​​normalized, where p'(g1,g2,a,b) is the normalized gray-level co-occurrence matrix, a and b are the distance difference values ​​of gray values ​​g1 and g2 in the voiceprint image, N' is the number of gray levels, and R is the normalization factor.

[0008] Optionally, the texture features of the voiceprint image are extracted using the gray-level co-occurrence matrix, including: using the formula The contrast feature quantity is extracted from the gray-level co-occurrence matrix using the formula. The energy feature quantity is extracted from the gray-level co-occurrence matrix using the formula. The entropy feature is extracted from the gray-level co-occurrence matrix, where i and j traverse the rows and columns of the gray-level co-occurrence matrix, respectively representing a gray value in the gray-level co-occurrence matrix, and p(i,j) is an element in the gray-level co-occurrence matrix, representing the probability of a pixel pair with gray values ​​i and j appearing at a given distance difference value (a, b).

[0009] Optionally, the construction process of the CNN-SVM model includes: constructing a CNN model, which includes an input layer, a convolutional layer, an activation layer, a pooling layer, and a fully connected layer. The input layer takes the voiceprint image as input; the convolutional layer automatically learns the texture features of the voiceprint image through convolutional kernels; the activation layer increases the non-linear expressive power of the network; the pooling layer reduces the size of the voiceprint image, extracts high-level features of the voiceprint image, reduces computational complexity, and prevents overfitting; the fully connected layer uses the ReLU activation function to convert the two-dimensional feature map output by the convolutional layer into a one-dimensional vector, obtaining a feature vector; and the feature vector is input into the SVM model for training to obtain the CNN-SVM model.

[0010] Optionally, determining the location of the fault source using the time delay difference includes: using formula d i =v×(t) i -t0) calculates the distance d between the fault source and the i-th sensor. i Where t0 is the time when the leakage fault occurs, t i Let v be the time when the i-th sensor receives the sound signal, and v be the speed of sound; solve for the objective function. Calculate the coordinates of the fault source in space, and use these coordinates as the location of the fault source, where P is the location of the fault source. i Let be the position of the i-th sensor, and D be the number of acoustic signal sensors.

[0011] To achieve the above objectives, according to another aspect of this application, an electrical equipment leakage current identification device based on voiceprint signals is provided, comprising:

[0012] The processing unit is used to process the acquired acoustic signal using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions, and select three key intrinsic mode functions from the multiple intrinsic mode functions.

[0013] The generation unit is used to generate the voiceprint image of the sound signal by taking the three key intrinsic mode functions as the gray values ​​of the red, green and blue color channels of the image, respectively.

[0014] The extraction unit is used to extract the texture features of the voiceprint image through the gray-level co-occurrence matrix, wherein the texture features include contrast features, energy features and entropy features;

[0015] The classification and recognition unit is used to perform deep feature extraction and classification and recognition of the texture features using a CNN-SVM model to obtain leakage detection results.

[0016] The calculation unit is used to calculate the time delay difference of the sound signal propagating to multiple sound signal sensors when the leakage detection result indicates that a leakage fault has occurred, and to determine the location of the fault source using the time delay difference.

[0017] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform any of the described electrical equipment leakage current identification methods based on voiceprint signals.

[0018] According to another aspect of this application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing any of the described methods for identifying electrical equipment leakage current based on voiceprint signals.

[0019] The technical solution of this application firstly processes the acquired acoustic signal using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions (EMFs), and then selects three key EMFs from these EMFs. These three key EMFs are then used as the grayscale values ​​of the red, green, and blue color channels of the image to generate a voiceprint image of the acoustic signal. Next, texture features of the voiceprint image are extracted using a gray-level co-occurrence matrix (GLCM), including contrast features, energy features, and entropy features. A CNN-SVM model is then used to perform deep feature extraction and classification of the texture features to obtain the leakage current detection result. Finally, when the leakage current detection result indicates a leakage current fault, the time delay difference between the acoustic signal propagation to multiple acoustic signal sensors is calculated, and this time delay difference is used to determine the location of the fault source. This solution solves the problem in existing technologies where accurate location of leakage current fault sources cannot be provided, leading to prolonged repair and troubleshooting time. Attached Figure Description

[0020] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0021] Figure 1 A hardware structure block diagram of a mobile terminal for implementing an electrical equipment leakage current identification method based on voiceprint signals, according to an embodiment of this application, is shown.

[0022] Figure 2 A schematic flowchart of an electrical equipment leakage current identification method based on voiceprint signals according to an embodiment of this application is shown.

[0023] Figure 3 A CNN-SVM recognition algorithm structure diagram is shown for an electrical equipment leakage current identification method based on voiceprint signals according to an embodiment of this application;

[0024] Figure 4 A structural block diagram of an electrical equipment leakage current identification device based on voiceprint signals according to an embodiment of this application is shown.

[0025] The attached figures are labeled as follows:

[0026] 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device. Detailed Implementation

[0027] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0028] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0030] As described in the background section, existing technologies are easily affected by factors such as noise interference and nonlinear loads when dealing with complex electrical environments, resulting in low detection accuracy of leakage faults. In particular, in the detection of minute leakage and early faults, it is easy to miss or falsely detect. In order to solve the problem of not being able to provide accurate fault source location, which leads to extended maintenance and troubleshooting time, the embodiments of this application provide an electrical equipment leakage identification method, identification device, computer-readable storage medium and electronic device based on voiceprint signals.

[0031] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0032] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for an electrical equipment leakage current identification method based on voiceprint signals, according to an embodiment of the present invention. Figure 1As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0033] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the device information display method in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0034] This embodiment provides a method for identifying electrical equipment leakage current based on voiceprint signals, which runs on a mobile terminal, computer terminal, or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Also, although the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0035] Figure 2This is a flowchart illustrating a method for identifying leakage current in electrical equipment based on voiceprint signals, according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0036] Step S201: The acquired acoustic signal is processed using the variational mode decomposition algorithm to obtain multiple intrinsic mode functions, and three key intrinsic mode functions are selected from the multiple intrinsic mode functions.

[0037] Specifically, Variational Mode Decomposition (VMD) is a signal processing technique used to decompose complex signals into a set of Intrinsic Mode Functions (IMFs), which can better reflect the intrinsic characteristics of the signal. In the embodiments of this application, the variational mode decomposition algorithm is used to preprocess the acquired acoustic signal, with the aim of separating and extracting acoustic features related to leakage faults.

[0038] Among the multiple intrinsic mode functions (IMFs) obtained by variational mode decomposition (VMD) algorithms, not all IMFs are directly related to leakage faults. Some IMFs may primarily reflect ambient noise, the acoustic patterns of normal equipment operation, or other interference signals. Therefore, it is crucial to select the three key IMFs most relevant to leakage faults from all the decomposed IMFs.

[0039] Step S202: The three key intrinsic mode functions mentioned above are used as the gray values ​​of the red, green and blue color channels of the image, respectively, to generate the voiceprint image of the sound signal.

[0040] Specifically, the amplitudes of these three key intrinsic mode functions need to be converted into grayscale values ​​in the image. Grayscale values ​​are typically between 0 and 255, where 0 represents the darkest (black) and 255 represents the brightest (white). The mapping process normalizes the amplitude range of the key intrinsic mode functions to the interval [0, 255], thus converting the signal intensity information in each mode component into the brightness information of the color channels in the image.

[0041] When generating a speakerprint image, the first key intrinsic mode function (IEM) is mapped to the red (R) channel of the image, the second to the green (G) channel, and the third to the blue (B) channel. The grayscale value of each channel represents the intensity information of the corresponding IEM. This allocation of color channels allows for a visual representation of the signal's spectrum and intensity characteristics within the image. Once the amplitudes of the key IEMs are converted to grayscale values ​​in the RGB channels, the next step is to construct the speakerprint image. A speakerprint image is a textured image that visually represents the characteristics of a sound signal, including its spectral distribution, intensity, and temporal characteristics. In the RGB space, different color combinations can create rich visual effects, highlighting specific patterns or features in the sound signal, which facilitates subsequent feature extraction and pattern recognition.

[0042] Step S203: Extract the texture features of the above voiceprint image through the gray-level co-occurrence matrix. The texture features include contrast features, energy features and entropy features.

[0043] Specifically, the gray-level co-occurrence matrix (GLCM) is a method for quantifying image texture. It statistically analyzes the frequency of gray-level value combinations between all pixels in an image and their neighboring pixels. For a given image and a specific offset vector (i.e., the relative positions between pixels), each element in the GLCM represents the number of times the gray-level values ​​of two pixels occur simultaneously. Since speaker images encode the characteristics of acoustic signals in gray-level values, the GLCM can reveal the distribution patterns of these acoustic characteristics in the image. By calculating texture features such as contrast, energy, and entropy, key information in speaker images can be further extracted, providing richer feature inputs for subsequent CNN-SVM models to identify and classify different acoustic patterns, including normal states and leakage fault states.

[0044] Step S204: Use the CNN-SVM model to perform deep feature extraction and classification on the above texture features to obtain leakage current detection results.

[0045] Specifically, a Convolutional Neural Network (CNN) is a type of neural network specifically designed to process input data (such as images) with a grid structure. In this embodiment, a CNN is used to extract deeper features from a voiceprint image. Through a multi-layered structure including convolutional layers, activation layers, and pooling layers, the CNN can automatically learn and extract local and global features of the image. For voiceprint images, the CNN can identify and extract texture patterns related to leakage faults, such as edges, shapes, and distributions.

[0046] Support Vector Machine (SVM) is a supervised learning model used for classification and regression analysis. In this embodiment, SVM is used to classify the feature vectors extracted by CNN to determine whether the voiceprint image contains features of a leakage fault. SVM distinguishes samples of different categories by finding an optimal hyperplane that maximizes the margin between different categories, thereby achieving accurate classification of new samples.

[0047] The CNN-SVM model combines the automatic feature extraction capabilities of deep learning with the classification and recognition capabilities of SVM. CNN can automatically learn highly abstract features from raw image data, while SVM can make decisions based on these features to determine whether a leakage fault exists. This combination improves the model's generalization ability and robustness, enabling it to accurately detect leakage faults even in complex and variable electrical environments.

[0048] Step S205: When the leakage detection result indicates that a leakage fault has occurred, calculate the time delay difference of the sound signal propagating to multiple sound signal sensors, and use the time delay difference to determine the location of the fault source.

[0049] Specifically, when a fault source generates an acoustic signal, this signal propagates through the medium of the power distribution network (such as air or insulating material). Because different sensors are at different distances from the fault source, the time it takes for the acoustic signal to reach each sensor also varies; this time difference is called time delay. By recording the precise arrival time of the acoustic signal through the sensors, the time delay difference between each pair of sensors can be calculated. Once the time delay differences between different sensors are calculated, these differences can be used to determine the location of the fault source.

[0050] In the above embodiments, firstly, the acquired acoustic signal is processed using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions (IMFs), and three key IMFs are selected from these IMFs. These three key IMFs are then used as the grayscale values ​​of the red, green, and blue color channels of the image, respectively, to generate a voiceprint image of the acoustic signal. Next, texture features of the voiceprint image are extracted using a gray-level co-occurrence matrix (GLCM). These texture features include contrast features, energy features, and entropy features. A CNN-SVM model is then used to perform deep feature extraction and classification of the texture features to obtain the leakage current detection result. Finally, when the leakage current detection result indicates a leakage current fault, the time delay difference between the acoustic signal propagation to multiple acoustic signal sensors is calculated, and the location of the fault source is determined using this time delay difference. This solves the problem in existing technologies where accurate location of leakage current fault sources cannot be provided, leading to prolonged repair and troubleshooting times.

[0051] In the specific implementation process, the acquired acoustic signal is processed using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions (EMFs). From these EMFs, three key EMFs are selected, including those obtained by applying formula one. The original acoustic signal is decomposed into a set of modal components, where K is the number of modal components, {u k} represents a set of the above modal components, {ω k} represents the center frequency corresponding to the above modal components, u k (t) is the k-th modal component, ω k It is the center frequency of the kth modal component mentioned above. The expression represents the derivative operation of the signal, j represents the imaginary unit, * represents the convolution operation, and δ(t) is the Dirac function; for the above formula one, the Lagrange multiplier λ and the penalty term α are introduced to construct the Lagrange function. The iterative optimization algorithm is used to solve u k ω k And λ(t), until the change of the above modal components is less than a preset threshold, where λ(t) is the Lagrange multiplier function and f(t) is the original above acoustic signal; analyze the characteristics of all the above modal components, and select three of the above modal components containing key information of the signal as the three above key intrinsic mode functions.

[0052] Specifically, the basic idea of ​​the variational mode decomposition algorithm is to decompose the signal into a set of modal components by constructing a variational model, where the center frequency ω of each modal component is... k The algorithm adaptively estimates frequencies through an optimization process. The core idea is to modulate the signal so that the spectrum of each mode is concentrated near zero frequency. Then, a band-limited filter is used to extract the frequency components of each mode. By minimizing the bandwidth of each mode component, the modal components and their corresponding center frequencies ω are solved. k .

[0053] In Formula 1 The expression here represents the objective of the variational mode decomposition algorithm: minimizing the spectral bandwidth of each modal component while satisfying constraints that ensure the sum of all modal components equals the original acoustic signal. To solve this constrained optimization problem, the variational mode decomposition algorithm introduces Lagrange multipliers and penalty parameters to construct the Lagrange function. The Lagrange function combines the objective function and constraints, and through iterative optimization algorithms, such as the Alternating Direction Method of Multipliers (ADMM), it solves for the modal components, center frequencies, and Lagrange multipliers until the variation of the modal components is less than a preset threshold. This process ensures the stability and convergence of the decomposition.

[0054] After obtaining all modal components, the characteristics of each modal component are analyzed, and three modal components containing key signal information are selected as key intrinsic mode functions (IMFs). This is typically based on the center frequency, energy distribution, and temporal characteristics of the modal components to determine which modal components best reflect signal anomalies, such as leakage faults. The selected key IMFs will be used for subsequent speakerprint image construction and feature extraction to achieve more accurate fault detection and localization.

[0055] In the specific implementation process, the three key intrinsic mode functions mentioned above are used as the grayscale values ​​of the red, green, and blue color channels of the image, respectively, to generate the voiceprint image of the sound signal. This includes: normalizing each of the key intrinsic mode functions; creating an RGB image framework, wherein the RGB image is composed of red, green, and blue color channels, and each color channel represents information of different colors in the RGB image; and using formula P... n The above voiceprint image is constructed by (i,j)=B((n-1)M×N / 2+M(i-1)+j), where P(.) represents the pixel grayscale value at the corresponding position in the above RGB image, B(.) represents the normalized amplitude of the above key intrinsic mode function, the size of the above RGB image is M×N, i=1:M, j=1:N, and the normalized amplitudes of the three above key intrinsic mode functions are respectively used as the grayscale values ​​of the red (R), green (G), and blue (B) color channels of the above RGB image.

[0056] Specifically, firstly, the three key intrinsic mode functions extracted from the acoustic signal need to be normalized. Normalization maps the amplitudes of these key intrinsic mode functions to the range of image grayscale values, typically between 0 and 255. This ensures that the information of each mode component is displayed in the image with appropriate brightness, preventing certain signal features from being ignored or misread due to large amplitude differences.

[0057] Creating an RGB image frame of size M×N means that the image will have M rows and N columns of pixels, and each pixel contains information from three color channels: red, green, and blue. Each channel of the RGB image uses 8 bits (256 levels of grayscale) to represent color information; therefore, the grayscale value of each channel ranges from 0 to 255.

[0058] Using P nThe formula (i,j) = B((n-1)M×N / 2 + M(i-1) + j) constructs the voiceprint image. This formula maps the normalized amplitudes of the key intrinsic mode functions (IEMs) to the pixel grayscale values ​​of the RGB image. This formula means that for each pixel in the image, its RGB color value will be determined by the normalized amplitudes of the three key IEMs at the corresponding location. In this way, information about different frequencies and intensities in the acoustic signal is encoded in the image's colors, forming the voiceprint image.

[0059] In the specific implementation process, the construction process of the aforementioned gray-level co-occurrence matrix includes: selecting a reference point (x, y) and a co-occurrence point (x+a, y+b) on the aforementioned voiceprint image, such that the gray values ​​of the reference point and the co-occurrence point form a gray-level value combination (g1, g2); letting the gray-level level of the aforementioned voiceprint image be k, transforming the coordinates of the reference point and the co-occurrence point to obtain various different gray-level value combinations (g1, g2), and the number of combinations of the gray-level value combinations (g1, g2) is k. 2 The process involves iterating through all the reference points in the aforementioned voiceprint image, counting the frequency of each grayscale value combination (g1, g2), and obtaining frequency statistics. These frequency statistics are then organized into a square matrix, which is normalized to convert into a probability matrix P(g1, g2). This probability matrix is ​​the initial grayscale co-occurrence matrix, where the rows and columns of the initial grayscale co-occurrence matrix correspond to the grayscale values ​​g1 and g2 in the aforementioned voiceprint image, respectively. The formula is then used... The initial gray-level co-occurrence matrix is ​​normalized, where p'(g1,g2,a,b) is the normalized gray-level co-occurrence matrix, a and b are the distance difference values ​​of the gray values ​​g1 and g2 in the voiceprint image, N' is the number of gray levels, and R is the normalization factor.

[0060] Specifically, a gray-level co-occurrence matrix is ​​constructed to quantify the distribution and spatial relationships of gray levels in an image, thereby extracting statistics describing the image's texture features. In the voiceprint image, a reference point (x, y) is first selected, followed by a co-occurrence point (x+a, y+b) located a certain distance from the reference point in a specific direction. Here, a and b define the spatial relationship between the two points, which can be a horizontal, vertical, or diagonal offset. The gray values ​​of the reference point and the co-occurrence point constitute a gray-level combination (g1, g2), where g1 is the gray value of the reference point and g2 is the gray value of the co-occurrence point. Since the voiceprint image has k gray levels, there are k possible gray-level combinations. 2 The process involves iterating through the entire voiceprint image and, for each pair of reference points and co-occurring points, recording the number of occurrences of their gray value combinations (g1, g2). This process needs to be performed on every possible reference point in the image, ultimately yielding a result that statistically analyzes the frequency of all gray value combinations.

[0061] The frequency statistics above are organized into a square matrix, where rows correspond to grayscale values ​​g1 and columns correspond to grayscale values ​​g2. Each element in the matrix represents the frequency of occurrence of the (g1, g2) combination. This matrix is ​​then normalized by dividing it by the total number of pixels in the image, converting it into a probability matrix P(g1, g2). The elements of the normalized matrix represent the probability of the grayscale value combination (g1, g2) appearing in the image. The formula is used... The probability matrix P(g1,g2) is further normalized to ensure that the sum of its elements equals 1. Normalization is performed by dividing each element of the matrix by a total amount or factor (R), ensuring that the sum of the matrix elements statistically represents a probability distribution.

[0062] After completing the above steps, the normalized gray-level co-occurrence matrix p'(g1,g2,a,b) is a statistical matrix that quantitatively describes the relative frequency and spatial relationship of gray values ​​in the image. This matrix can be used to calculate various texture features, such as contrast, energy, and entropy, further revealing the intrinsic structure and characteristics of the voiceprint image and providing valuable feature input for leakage detection of electrical equipment based on voiceprint signals. In practical applications, since the construction of the gray-level co-occurrence matrix depends on the selection of reference points and co-occurrence points, different selection strategies and distance differences a and b will lead to different matrices. Therefore, multiple gray-level co-occurrence matrices are usually constructed in different directions and offsets, and then the average or maximum value features are extracted from these matrices to comprehensively reflect the texture features of the image.

[0063] In the specific implementation process, the texture features of the aforementioned voiceprint image are extracted using the gray-level co-occurrence matrix, including: using the formula The contrast features are extracted from the gray-level co-occurrence matrix using the formula. The energy features are extracted from the gray-level co-occurrence matrix using the formula. Extract the entropy feature from the gray-level co-occurrence matrix, where i and j traverse the rows and columns of the gray-level co-occurrence matrix, respectively representing a gray value in the gray-level co-occurrence matrix, and p(i,j) is an element in the gray-level co-occurrence matrix, representing the probability of a pixel pair with gray values ​​i and j appearing at a given distance difference value (a, b).

[0064] Specifically, texture features of voiceprint images are extracted using the gray-level co-occurrence matrix (GLCM), including contrast features, energy features, and entropy features. These features quantify the texture attributes of the image and provide feature input for subsequent image recognition and classification. Contrast features, also known as contrast ratios, directly reflect the brightness contrast of image pixel values. They are directly proportional to the depth and clarity of texture grooves; the higher the value, the deeper the grooves and the clearer the image. Energy features measure the stability of gray-level changes in a texture image, reflecting the fineness of the texture and its gray-level distribution. The energy value is equal to the sum of the squares of the element values ​​in the matrix; the higher the energy value, the more stable the regular changes in the texture. Entropy features measure the amount of information in an image, reflecting the complexity of the texture. Their value is directly proportional to the image complexity. When pixel values ​​in the GLCM exhibit high randomness or all their values ​​are identical, the entropy value is at its maximum, and the image is more complex.

[0065] By calculating the contrast, energy, and entropy features of the gray-level co-occurrence matrix, the texture structure of the voiceprint image can be quantified, providing crucial feature inputs for image-based leakage detection methods. These features not only reveal the distribution and contrast of gray values ​​in the image but also reflect the image's complexity and information content, playing a vital role in distinguishing normal and faulty images. In the embodiments of this application, these features are used to train and optimize the CNN-SVM model to improve the accuracy and robustness of leakage detection.

[0066] In the specific implementation process, the construction process of the above-mentioned CNN-SVM model includes: constructing a CNN model, which includes an input layer, a convolutional layer, an activation layer, a pooling layer, and a fully connected layer. The input layer is used to take the voiceprint image as input. The convolutional layer is used to automatically learn the texture features of the voiceprint image through the convolutional kernel. The activation layer is used to increase the non-linear expressive power of the network. The pooling layer is used to reduce the size of the voiceprint image, extract the high-level features of the voiceprint image, reduce computational complexity, and prevent overfitting. The fully connected layer uses the ReLU activation function to convert the two-dimensional feature map output by the convolutional layer into a one-dimensional vector to obtain a feature vector. The feature vector is then input into the SVM model for training to obtain the above-mentioned CNN-SVM model.

[0067] Specifically, the process of building a CNN-SVM model combines two machine learning models: Convolutional Neural Network (CNN) and Support Vector Machine (SVM), to efficiently extract features from voiceprint images and perform classification. A CNN is a feedforward neural network containing convolutional computations and has a deep structure. CNN networks not only support multi-layer neural networks but can also learn feature parameters and classifiers simultaneously. The structure of a CNN network is similar to a multilayer perceptron, featuring a feedforward structure and characteristics such as weight sharing and local perception in image analysis. Its basic network structure includes five parts: input layer, convolutional layer, pooling layer, fully connected layer, and output layer.

[0068] Convolutional layers are the core layers in building convolutional neural networks. They process input image data through convolution operations to extract local features. The convolution kernel is essentially a weight matrix. Besides the weight matrix itself, the number of kernels and the stride are also important parameters. The appropriate number of kernels is selected based on the different dimensions of the extracted feature matrix. The stride proportionally reduces the input size; that is, convolutional layers can change the image size and the number of channels, as shown in the formula. As shown, w1 is the width of the input image, h1 is the height, c1 and m represent the number of channels, k is the kernel size, s is the stride, and p is the zero-padding size. Then w2, h2, and c2 are the width, height, and number of channels of the output image.

[0069] Pooling layers downsample the input image from the previous layer, reducing the size of the matrix and thus decreasing the number of nodes in the fully connected layer. This reduces network parameters and avoids overfitting. Common pooling methods include average pooling and max pooling. Pooling layers only change the image size, do not affect the depth of the 3D matrix, and preserve the invariance of local features. Its mathematical model is shown in the formula... As shown, w1 is the width of the input image, h1 is the height, c1 represents the number of channels, k is the kernel size, s is the stride, and p is the zero-padding size. Then w2, h2, and c2 are the width, height, and number of channels of the output image.

[0070] A fully connected layer is a traditional feedforward neural network structure where each neuron is fully connected to all neurons in the previous layer. Its main function is to summarize the information from the previous layer's input and act as a "classifier" within the neural network, classifying the features extracted from the feature maps of the previous layer. Fully connected layers typically use the ReLU activation function and are usually located in the last few layers of the network, converting the two-dimensional feature maps output by the convolutional layers into one-dimensional vectors, thus achieving an end-to-end learning process.

[0071] The output layer is the last layer of a neural network. Its function is to pass the output of the previous layer to the output function, typically the logistic regression function (i.e., softmax regression). The softmax function converts the output of the fully connected layer into a value between 0 and 1 and outputs the result as a probability. It is commonly used in classification tasks in convolutional neural networks. The softmax function is defined as follows: As shown in the figure, pi is the probability of the category output, the numerator is the exponential transformation of the corresponding category output, and the denominator is the sum of the exponential transformations of the corresponding category output, and n represents the number of categories.

[0072] VGG16 is a relatively mature convolutional neural network among current CNN networks. Using the VGG16 architecture as the CNN part of the network model is shown in the formula: input → [(conv×2 → pool)×2] → [(conv×3 → pool)×3] → FC×3. VGG16 uses the softmax function as the final output of its architecture. However, since the softmax classifier only performs probability transformation on the previous layer's input and lacks autonomous learning capabilities, it cannot effectively improve the generalization ability of the image classification and recognition model. Therefore, SVM is used to replace the softmax function to recalculate the values ​​in the final output layer.

[0073] The output feature vector of the CNN model is fed into the SVM model for training. SVM is a supervised learning model used for classification and regression tasks. Its core principle is to find a hyperplane that maximizes the class margin to distinguish samples from different classes. In this embodiment, SVM is used to learn the decision boundary between leakage and normal states based on the feature vectors extracted by the CNN. The training process of the SVM model involves finding the optimal classification hyperplane that enables the model to accurately distinguish between voiceprint images with and without leakage. Through training, the SVM model learns key features describing leakage faults, which will be used for subsequent leakage detection and fault source localization. Once the SVM model is trained, it can predict whether there is a leakage phenomenon based on the feature vectors of the new voiceprint images.

[0074] In summary, the CNN-SVM model is constructed by automatically learning and extracting texture features from voiceprint images using CNN, and then mapping these feature vectors into the decision space using SVM to classify the leakage status of electrical equipment. This method, which combines deep learning and traditional machine learning, can not only automatically extract key features from images but also leverage the strong classification capabilities of SVM for accurate classification tasks, thus improving the accuracy and efficiency of leakage detection.

[0075] In the specific implementation process, the location of the fault source is determined by utilizing the aforementioned time delay difference, including: using formula d i =v×(t)i -t0) calculates the distance d between the above fault source and the i-th sensor. i Where t0 is the time when the aforementioned leakage fault occurred, t i Let v be the time when the i-th sensor receives the aforementioned acoustic signal, and v be the speed of sound; solve for the objective function. Calculate the coordinates of the aforementioned fault source in space, and use these coordinates as the location of the fault source, where P is the location of the fault source. i Let represent the position of the i-th sensor, and D represent the number of acoustic signal sensors.

[0076] Specifically, the location of the sound source is determined based on the time difference of arrival of the sound signal at different sensors. The distance is calculated using the speed of sound and the time difference. Then, the sound source location is solved using triangulation or an overdetermined system of equations based on the time difference data from multiple sensors. First, the relative distance between the fault source and each sensor is estimated by detecting the time difference of the sound signal arriving at each sensor; this is the basis for location using time difference. Next, an objective function is used... The objective function for determining the exact coordinates of the fault source in space is to minimize the difference between the actual distance and the estimated distance. By solving this optimization problem, a precise location estimate of the fault source can be obtained. Generally, gradient descent, least squares, or more complex nonlinear optimization algorithms can be used to solve this type of problem.

[0077] The CNN-SVM recognition algorithm structure diagram is as follows: Figure 3 As shown, for voiceprint images, key feature information is first extracted using convolutional layers, and then pooling layers are used to "purify" the feature information. This process is repeated multiple times, ultimately yielding a multidimensional vector describing the internal features of the image. Since these multidimensional feature vectors cannot be directly used as input to the SVM, data transformation is required. Therefore, a fully connected neural network is added to the model to transform the final pooling result into a one-dimensional feature vector, which is then input into the SVM for model training.

[0078] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the electrical equipment leakage current identification method based on voiceprint signal of this application will be described in detail below with reference to specific embodiments.

[0079] This application also provides an electrical equipment leakage current identification device based on voiceprint signals. It should be noted that the electrical equipment leakage current identification device based on voiceprint signals of this application can be used to execute the electrical equipment leakage current identification method based on voiceprint signals provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0080] The following describes the electrical equipment leakage current identification device based on voiceprint signal provided in the embodiments of this application.

[0081] Figure 4 This is a structural block diagram of an electrical equipment leakage current identification device based on voiceprint signals according to an embodiment of this application.

[0082] like Figure 4 As shown, the device includes:

[0083] The processing unit 10 is used to process the acquired acoustic signal using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions, and select three key intrinsic mode functions from the multiple intrinsic mode functions.

[0084] Specifically, Variational Mode Decomposition (VMD) is a signal processing technique used to decompose complex signals into a set of Intrinsic Mode Functions (IMFs), which can better reflect the intrinsic characteristics of the signal. In the embodiments of this application, the variational mode decomposition algorithm is used to preprocess the acquired acoustic signal, with the aim of separating and extracting acoustic features related to leakage faults.

[0085] Among the multiple intrinsic mode functions (IMFs) obtained by variational mode decomposition (VMD) algorithms, not all IMFs are directly related to leakage faults. Some IMFs may primarily reflect ambient noise, the acoustic patterns of normal equipment operation, or other interference signals. Therefore, it is crucial to select the three key IMFs most relevant to leakage faults from all the decomposed IMFs.

[0086] The generation unit 20 is used to generate the voiceprint image of the above-mentioned sound signal by taking the three key intrinsic mode functions mentioned above as grayscale values ​​of the red, green and blue color channels of the image respectively.

[0087] Specifically, the amplitudes of these three key intrinsic mode functions need to be converted into grayscale values ​​in the image. Grayscale values ​​are typically between 0 and 255, where 0 represents the darkest (black) and 255 represents the brightest (white). The mapping process normalizes the amplitude range of the key intrinsic mode functions to the interval [0, 255], thus converting the signal intensity information in each mode component into the brightness information of the color channels in the image.

[0088] When generating a speakerprint image, the first key intrinsic mode function (IEM) is mapped to the red (R) channel of the image, the second to the green (G) channel, and the third to the blue (B) channel. The grayscale value of each channel represents the intensity information of the corresponding IEM. This allocation of color channels allows for a visual representation of the signal's spectrum and intensity characteristics within the image. Once the amplitudes of the key IEMs are converted to grayscale values ​​in the RGB channels, the next step is to construct the speakerprint image. A speakerprint image is a textured image that visually represents the characteristics of a sound signal, including its spectral distribution, intensity, and temporal characteristics. In the RGB space, different color combinations can create rich visual effects, highlighting specific patterns or features in the sound signal, which facilitates subsequent feature extraction and pattern recognition.

[0089] Extraction unit 30 is used to extract the texture features of the aforementioned voiceprint image through a gray-level co-occurrence matrix. The texture features include contrast features, energy features, and entropy features.

[0090] Specifically, the gray-level co-occurrence matrix (GLCM) is a method for quantifying image texture. It statistically analyzes the frequency of gray-level value combinations between all pixels in an image and their neighboring pixels. For a given image and a specific offset vector (i.e., the relative positions between pixels), each element in the GLCM represents the number of times the gray-level values ​​of two pixels occur simultaneously. Since speaker images encode the characteristics of acoustic signals in gray-level values, the GLCM can reveal the distribution patterns of these acoustic characteristics in the image. By calculating texture features such as contrast, energy, and entropy, key information in speaker images can be further extracted, providing richer feature inputs for subsequent CNN-SVM models to identify and classify different acoustic patterns, including normal states and leakage fault states.

[0091] The classification and recognition unit 40 is used to perform deep feature extraction and classification and recognition of the above texture features using a CNN-SVM model to obtain leakage current detection results.

[0092] Specifically, a Convolutional Neural Network (CNN) is a type of neural network specifically designed to process input data (such as images) with a grid structure. In this embodiment, a CNN is used to extract deeper features from a voiceprint image. Through a multi-layered structure including convolutional layers, activation layers, and pooling layers, the CNN can automatically learn and extract local and global features of the image. For voiceprint images, the CNN can identify and extract texture patterns related to leakage faults, such as edges, shapes, and distributions.

[0093] Support Vector Machine (SVM) is a supervised learning model used for classification and regression analysis. In this embodiment, SVM is used to classify the feature vectors extracted by CNN to determine whether the voiceprint image contains features of a leakage fault. SVM distinguishes samples of different categories by finding an optimal hyperplane that maximizes the margin between different categories, thereby achieving accurate classification of new samples.

[0094] The CNN-SVM model combines the automatic feature extraction capabilities of deep learning with the classification and recognition capabilities of SVM. CNN can automatically learn highly abstract features from raw image data, while SVM can make decisions based on these features to determine whether a leakage fault exists. This combination improves the model's generalization ability and robustness, enabling it to accurately detect leakage faults even in complex and variable electrical environments.

[0095] The calculation unit 50 is used to calculate the time delay difference of the sound signal propagating to multiple sound signal sensors when the above leakage detection result indicates that a leakage fault has occurred, and to determine the location of the fault source using the above time delay difference.

[0096] Specifically, when a fault source generates an acoustic signal, this signal propagates through the medium of the power distribution network (such as air or insulating material). Because different sensors are at different distances from the fault source, the time it takes for the acoustic signal to reach each sensor also varies; this time difference is called time delay. By recording the precise arrival time of the acoustic signal through the sensors, the time delay difference between each pair of sensors can be calculated. Once the time delay differences between different sensors are calculated, these differences can be used to determine the location of the fault source.

[0097] The above processing unit includes:

[0098] The decomposition module is used to apply Formula 1. The original acoustic signal is decomposed into a set of modal components, where K is the number of modal components, {u k} represents a set of the above modal components, {ω k} represents the center frequency corresponding to the above modal components, u k (t) is the k-th modal component, ω k It is the center frequency of the kth modal component mentioned above. The derivative operation of the signal is represented by j, the imaginary unit is represented by *, the convolution operation is represented by δ(t), and the Dirac function is represented by δ(t).

[0099] The solver module is used to construct the Lagrange function by introducing the Lagrange multiplier λ and the penalty term α for Equation 1 above. The iterative optimization algorithm is used to solve u k ω k And λ(t), until the change of the above modal components is less than a preset threshold, where λ(t) is the Lagrange multiplier function and f(t) is the original above acoustic signal;

[0100] The selection module is used to analyze the characteristics of all the above modal components and select three of the above modal components containing key signal information as the three above key intrinsic mode functions.

[0101] Specifically, the basic idea of ​​the variational mode decomposition algorithm is to decompose the signal into a set of modal components by constructing a variational model, where the center frequency ω of each modal component is... k The algorithm adaptively estimates frequencies through an optimization process. The core idea is to modulate the signal so that the spectrum of each mode is concentrated near zero frequency. Then, a band-limited filter is used to extract the frequency components of each mode. By minimizing the bandwidth of each mode component, the modal components and their corresponding center frequencies ω are solved. k .

[0102] In Formula 1 The expression here represents the objective of the variational mode decomposition algorithm: minimizing the spectral bandwidth of each modal component while satisfying constraints that ensure the sum of all modal components equals the original acoustic signal. To solve this constrained optimization problem, the variational mode decomposition algorithm introduces Lagrange multipliers and penalty parameters to construct the Lagrange function. The Lagrange function combines the objective function and constraints, and through iterative optimization algorithms, such as the Alternating Direction Method of Multipliers (ADMM), it solves for the modal components, center frequencies, and Lagrange multipliers until the variation of the modal components is less than a preset threshold. This process ensures the stability and convergence of the decomposition.

[0103] After obtaining all modal components, the characteristics of each modal component are analyzed, and three modal components containing key signal information are selected as key intrinsic mode functions (IMFs). This is typically based on the center frequency, energy distribution, and temporal characteristics of the modal components to determine which modal components best reflect signal anomalies, such as leakage faults. The selected key IMFs will be used for subsequent speakerprint image construction and feature extraction to achieve more accurate fault detection and localization.

[0104] The above-mentioned generation unit includes:

[0105] The processing module is used to normalize each of the aforementioned key intrinsic mode functions;

[0106] The creation module is used to create a framework for an RGB image. The RGB image consists of three color channels: red, green, and blue. Each of the color channels represents information about different colors in the RGB image.

[0107] Build a module for using formula P n The above voiceprint image is constructed by (i,j)=B((n-1)M×N / 2+M(i-1)+j), where P(.) represents the pixel grayscale value at the corresponding position in the above RGB image, B(.) represents the normalized amplitude of the above key intrinsic mode function, the size of the above RGB image is M×N, i=1:M, j=1:N, and the normalized amplitudes of the three above key intrinsic mode functions are respectively used as the grayscale values ​​of the red (R), green (G), and blue (B) color channels of the above RGB image.

[0108] Specifically, firstly, the three key intrinsic mode functions extracted from the acoustic signal need to be normalized. Normalization maps the amplitudes of these key intrinsic mode functions to the range of image grayscale values, typically between 0 and 255. This ensures that the information of each mode component is displayed in the image with appropriate brightness, preventing certain signal features from being ignored or misread due to large amplitude differences.

[0109] Creating an RGB image frame of size M×N means that the image will have M rows and N columns of pixels, and each pixel contains information from three color channels: red, green, and blue. Each channel of the RGB image uses 8 bits (256 levels of grayscale) to represent color information; therefore, the grayscale value of each channel ranges from 0 to 255.

[0110] Using P nThe formula (i,j) = B((n-1)M×N / 2 + M(i-1) + j) constructs the voiceprint image. This formula maps the normalized amplitudes of the key intrinsic mode functions (IEMs) to the pixel grayscale values ​​of the RGB image. This formula means that for each pixel in the image, its RGB color value will be determined by the normalized amplitudes of the three key IEMs at the corresponding location. In this way, information about different frequencies and intensities in the acoustic signal is encoded in the image's colors, forming the voiceprint image.

[0111] In the above extraction unit, the process of constructing the gray-level co-occurrence matrix includes: selecting a reference point (x, y) and a co-occurrence point (x+a, y+b) on the above voiceprint image, such that the gray values ​​of the reference point and the co-occurrence point form a gray-level value combination (g1, g2); letting the gray-level level of the above voiceprint image be k, transforming the coordinates of the reference point and the co-occurrence point to obtain various different gray-level value combinations (g1, g2), and the number of combinations of gray-level value combinations (g1, g2) is k. 2 The process involves iterating through all the reference points in the aforementioned voiceprint image, counting the frequency of each grayscale value combination (g1, g2), and obtaining frequency statistics. These frequency statistics are then organized into a square matrix, which is normalized to convert into a probability matrix P(g1, g2). This probability matrix is ​​the initial grayscale co-occurrence matrix, where the rows and columns of the initial grayscale co-occurrence matrix correspond to the grayscale values ​​g1 and g2 in the aforementioned voiceprint image, respectively. The formula is then used... The initial gray-level co-occurrence matrix is ​​normalized, where p'(g1,g2,a,b) is the normalized gray-level co-occurrence matrix, a and b are the distance difference values ​​of the gray values ​​g1 and g2 in the voiceprint image, N' is the number of gray levels, and R is the normalization factor.

[0112] Specifically, a gray-level co-occurrence matrix is ​​constructed to quantify the distribution and spatial relationships of gray levels in an image, thereby extracting statistics describing the image's texture features. In the voiceprint image, a reference point (x, y) is first selected, followed by a co-occurrence point (x+a, y+b) located a certain distance from the reference point in a specific direction. Here, a and b define the spatial relationship between the two points, which can be a horizontal, vertical, or diagonal offset. The gray values ​​of the reference point and the co-occurrence point constitute a gray-level combination (g1, g2), where g1 is the gray value of the reference point and g2 is the gray value of the co-occurrence point. Since the voiceprint image has k gray levels, there are k possible gray-level combinations. 2 The process involves iterating through the entire voiceprint image and, for each pair of reference points and co-occurring points, recording the number of occurrences of their gray value combinations (g1, g2). This process needs to be performed on every possible reference point in the image, ultimately yielding a result that statistically analyzes the frequency of all gray value combinations.

[0113] The frequency statistics above are organized into a square matrix, where rows correspond to grayscale values ​​g1 and columns correspond to grayscale values ​​g2. Each element in the matrix represents the frequency of occurrence of the (g1, g2) combination. This matrix is ​​then normalized by dividing it by the total number of pixels in the image, converting it into a probability matrix P(g1, g2). The elements of the normalized matrix represent the probability of the grayscale value combination (g1, g2) appearing in the image. The formula is used... The probability matrix P(g1,g2) is further normalized to ensure that the sum of its elements equals 1. Normalization is performed by dividing each element of the matrix by a total amount or factor (R), ensuring that the sum of the matrix elements statistically represents a probability distribution.

[0114] After completing the above steps, the normalized gray-level co-occurrence matrix p'(g1,g2,a,b) is a statistical matrix that quantitatively describes the relative frequency and spatial relationship of gray values ​​in the image. This matrix can be used to calculate various texture features, such as contrast, energy, and entropy, further revealing the intrinsic structure and characteristics of the voiceprint image and providing valuable feature input for leakage detection of electrical equipment based on voiceprint signals. In practical applications, since the construction of the gray-level co-occurrence matrix depends on the selection of reference points and co-occurrence points, different selection strategies and distance differences a and b will lead to different matrices. Therefore, multiple gray-level co-occurrence matrices are usually constructed in different directions and offsets, and then the average or maximum value features are extracted from these matrices to comprehensively reflect the texture features of the image.

[0115] The above extraction unit includes:

[0116] Extraction module, used to extract formulas The contrast features are extracted from the gray-level co-occurrence matrix using the formula. The energy characteristics are extracted from the gray-level co-occurrence matrix using the formula. Extract the entropy feature from the gray-level co-occurrence matrix, where i and j traverse the rows and columns of the gray-level co-occurrence matrix, respectively representing a gray value in the gray-level co-occurrence matrix, and p(i,j) is an element in the gray-level co-occurrence matrix, representing the probability of a pixel pair with gray values ​​i and j appearing at a given distance difference value (a, b).

[0117] Specifically, texture features of voiceprint images are extracted using the gray-level co-occurrence matrix (GLCM), including contrast features, energy features, and entropy features. These features quantify the texture attributes of the image and provide feature input for subsequent image recognition and classification. Contrast features, also known as contrast ratios, directly reflect the brightness contrast of image pixel values. They are directly proportional to the depth and clarity of texture grooves; the higher the value, the deeper the grooves and the clearer the image. Energy features measure the stability of gray-level changes in a texture image, reflecting the fineness of the texture and its gray-level distribution. The energy value is equal to the sum of the squares of the element values ​​in the matrix; the higher the energy value, the more stable the regular changes in the texture. Entropy features measure the amount of information in an image, reflecting the complexity of the texture. Their value is directly proportional to the image complexity. When pixel values ​​in the GLCM exhibit high randomness or all their values ​​are identical, the entropy value is at its maximum, and the image is more complex.

[0118] By calculating the contrast, energy, and entropy features of the gray-level co-occurrence matrix, the texture structure of the voiceprint image can be quantified, providing crucial feature inputs for image-based leakage detection methods. These features not only reveal the distribution and contrast of gray values ​​in the image but also reflect the image's complexity and information content, playing a vital role in distinguishing normal and faulty images. In the embodiments of this application, these features are used to train and optimize the CNN-SVM model to improve the accuracy and robustness of leakage detection.

[0119] In the aforementioned classification and recognition unit, the construction process of the aforementioned CNN-SVM model includes: constructing a CNN model, which comprises an input layer, a convolutional layer, an activation layer, a pooling layer, and a fully connected layer. The input layer takes the aforementioned voiceprint image as input; the convolutional layer automatically learns the texture features of the aforementioned voiceprint image through convolutional kernels; the activation layer increases the non-linear expressive power of the network; the pooling layer reduces the size of the aforementioned voiceprint image, extracts high-level features of the aforementioned voiceprint image, and simultaneously reduces computational complexity and prevents overfitting; the fully connected layer uses the ReLU activation function to convert the two-dimensional feature map output by the aforementioned convolutional layer into a one-dimensional vector, obtaining a feature vector; and the aforementioned feature vector is input into the SVM model for training to obtain the aforementioned CNN-SVM model.

[0120] Specifically, the process of building a CNN-SVM model combines two machine learning models: Convolutional Neural Network (CNN) and Support Vector Machine (SVM), to efficiently extract features from voiceprint images and perform classification. A CNN is a feedforward neural network containing convolutional computations and has a deep structure. CNN networks not only support multi-layer neural networks but can also learn feature parameters and classifiers simultaneously. The structure of a CNN network is similar to a multilayer perceptron, featuring a feedforward structure and characteristics such as weight sharing and local perception in image analysis. Its basic network structure includes five parts: input layer, convolutional layer, pooling layer, fully connected layer, and output layer.

[0121] Convolutional layers are the core layers in building convolutional neural networks. They process input image data through convolution operations to extract local features. The convolution kernel is essentially a weight matrix. Besides the weight matrix itself, the number of kernels and the stride are also important parameters. The appropriate number of kernels is selected based on the different dimensions of the extracted feature matrix. The stride proportionally reduces the input size; that is, convolutional layers can change the image size and the number of channels, as shown in the formula. As shown, w1 is the width of the input image, h1 is the height, c1 and m represent the number of channels, k is the kernel size, s is the stride, and p is the zero-padding size. Then w2, h2, and c2 are the width, height, and number of channels of the output image.

[0122] Pooling layers downsample the input image from the previous layer, reducing the size of the matrix and thus decreasing the number of nodes in the fully connected layer. This reduces network parameters and avoids overfitting. Common pooling methods include average pooling and max pooling. Pooling layers only change the image size, do not affect the depth of the 3D matrix, and preserve the invariance of local features. Its mathematical model is shown in the formula... As shown, w1 is the width of the input image, h1 is the height, c1 represents the number of channels, k is the kernel size, s is the stride, and p is the zero-padding size. Then w2, h2, and c2 are the width, height, and number of channels of the output image.

[0123] A fully connected layer is a traditional feedforward neural network structure where each neuron is fully connected to all neurons in the previous layer. Its main function is to summarize the information from the previous layer's input and act as a "classifier" within the neural network, classifying the features extracted from the feature maps of the previous layer. Fully connected layers typically use the ReLU activation function and are usually located in the last few layers of the network, converting the two-dimensional feature maps output by the convolutional layers into one-dimensional vectors, thus achieving an end-to-end learning process.

[0124] The output layer is the last layer of a neural network. Its function is to pass the output of the previous layer to the output function, typically the logistic regression function (i.e., softmax regression). The softmax function converts the output of the fully connected layer into a value between 0 and 1 and outputs the result as a probability. It is commonly used in classification tasks in convolutional neural networks. The softmax function is defined as follows: As shown in the figure, pi is the probability of the category output, the numerator is the exponential transformation of the corresponding category output, and the denominator is the sum of the exponential transformations of the corresponding category output, and n represents the number of categories.

[0125] VGG16 is a relatively mature convolutional neural network among current CNN networks. Using the VGG16 architecture as the CNN part of the network model is shown in the formula: input → [(conv×2 → pool)×2] → [(conv×3 → pool)×3] → FC×3. VGG16 uses the softmax function as the final output of its architecture. However, since the softmax classifier only performs probability transformation on the previous layer's input and lacks autonomous learning capabilities, it cannot effectively improve the generalization ability of the image classification and recognition model. Therefore, SVM is used to replace the softmax function to recalculate the values ​​in the final output layer.

[0126] The output feature vector of the CNN model is fed into the SVM model for training. SVM is a supervised learning model used for classification and regression tasks. Its core principle is to find a hyperplane that maximizes the class margin to distinguish samples from different classes. In this embodiment, SVM is used to learn the decision boundary between leakage and normal states based on the feature vectors extracted by the CNN. The training process of the SVM model involves finding the optimal classification hyperplane that enables the model to accurately distinguish between voiceprint images with and without leakage. Through training, the SVM model learns key features describing leakage faults, which will be used for subsequent leakage detection and fault source localization. Once the SVM model is trained, it can predict whether there is a leakage phenomenon based on the feature vectors of the new voiceprint images.

[0127] In summary, the CNN-SVM model is constructed by automatically learning and extracting texture features from voiceprint images using CNN, and then mapping these feature vectors into the decision space using SVM to classify the leakage status of electrical equipment. This method, which combines deep learning and traditional machine learning, can not only automatically extract key features from images but also leverage the strong classification capabilities of SVM for accurate classification tasks, thus improving the accuracy and efficiency of leakage detection.

[0128] In the aforementioned calculation unit, determining the location of the fault source using the aforementioned time delay difference includes: using formula d i =v×(t)i -t0) calculates the distance d between the above fault source and the i-th sensor. i Where t0 is the time when the aforementioned leakage fault occurred, t i Let v be the time when the i-th sensor receives the aforementioned acoustic signal, and v be the speed of sound; solve for the objective function. Calculate the coordinates of the aforementioned fault source in space, and use these coordinates as the location of the fault source, where P is the location of the fault source. i Let represent the position of the i-th sensor, and D represent the number of acoustic signal sensors.

[0129] Specifically, the location of the sound source is determined based on the time difference of arrival of the sound signal at different sensors. The distance is calculated using the speed of sound and the time difference. Then, the sound source location is solved using triangulation or an overdetermined system of equations based on the time difference data from multiple sensors. First, the relative distance between the fault source and each sensor is estimated by detecting the time difference of the sound signal arriving at each sensor; this is the basis for location using time difference. Next, an objective function is used... The objective function for determining the exact coordinates of the fault source in space is to minimize the difference between the actual distance and the estimated distance. By solving this optimization problem, a precise location estimate of the fault source can be obtained. Generally, gradient descent, least squares, or more complex nonlinear optimization algorithms can be used to solve this type of problem.

[0130] The aforementioned electrical equipment leakage current identification device based on voiceprint signals includes a processor and a memory. The processing unit, generation unit, extraction unit, classification and identification unit, and calculation unit are all stored as program units in the memory. The processor executes these program units stored in the memory to achieve the corresponding functions. All of the above modules are located in the same processor; alternatively, the modules may be located in different processors in any combination.

[0131] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0132] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the aforementioned electrical equipment leakage current identification method based on voiceprint signals.

[0133] This invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps:

[0134] Step S201: The acquired acoustic signal is processed using the variational mode decomposition algorithm to obtain multiple intrinsic mode functions, and three key intrinsic mode functions are selected from the multiple intrinsic mode functions.

[0135] Step S202: The three key intrinsic mode functions mentioned above are used as the gray values ​​of the red, green and blue color channels of the image, respectively, to generate the voiceprint image of the sound signal.

[0136] Step S203: Extract the texture features of the above voiceprint image through the gray-level co-occurrence matrix. The texture features include contrast features, energy features and entropy features.

[0137] Step S204: Use the CNN-SVM model to perform deep feature extraction and classification on the above texture features to obtain leakage current detection results;

[0138] Step S205: When the leakage detection result indicates that a leakage fault has occurred, calculate the time delay difference of the sound signal propagating to multiple sound signal sensors, and use the time delay difference to determine the location of the fault source.

[0139] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps:

[0140] Step S201: The acquired acoustic signal is processed using the variational mode decomposition algorithm to obtain multiple intrinsic mode functions, and three key intrinsic mode functions are selected from the multiple intrinsic mode functions.

[0141] Step S202: The three key intrinsic mode functions mentioned above are used as the gray values ​​of the red, green and blue color channels of the image, respectively, to generate the voiceprint image of the sound signal.

[0142] Step S203: Extract the texture features of the above voiceprint image through the gray-level co-occurrence matrix. The texture features include contrast features, energy features and entropy features.

[0143] Step S204: Use the CNN-SVM model to perform deep feature extraction and classification on the above texture features to obtain leakage current detection results;

[0144] Step S205: When the leakage detection result indicates that a leakage fault has occurred, calculate the time delay difference of the sound signal propagating to multiple sound signal sensors, and use the time delay difference to determine the location of the fault source.

[0145] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0146] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0147] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0148] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0149] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0150] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0151] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0152] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0153] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0154] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for identifying leakage current in electrical equipment based on voiceprint signals, characterized in that, The electrical equipment is AC / DC hybrid power distribution network electrical equipment, including: The acquired acoustic signal is processed using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions (IMFs), and three key IMFs are selected from these IMFs. The three key intrinsic mode functions are used as the grayscale values ​​of the red, green, and blue color channels of the image, respectively, to generate the voiceprint image of the sound signal; Texture features of the voiceprint image are extracted using a gray-level co-occurrence matrix, and the texture features include contrast features, energy features, and entropy features. The CNN-SVM model is used to perform deep feature extraction and classification of the texture features to obtain leakage current detection results. When the leakage detection result indicates that a leakage fault has occurred, the time delay difference of the sound signal propagating to multiple sound signal sensors is calculated, and the location of the fault source is determined using the time delay difference; Using the three key intrinsic mode functions as the grayscale values ​​of the red, green, and blue color channels of the image, respectively, to generate the voiceprint image of the acoustic signal, the method includes: Normalize each of the key intrinsic mode functions; Create a framework for an RGB image, which consists of three color channels: red, green, and blue. Each color channel represents information about different colors in the RGB image. Using formula The voiceprint image is constructed, where P(.) represents the pixel grayscale value at the corresponding position in the RGB image, B(.) represents the normalized amplitude of the key intrinsic mode function, the size of the RGB image is M×N, i=1:M, j=1:N, and the normalized amplitudes of the three key intrinsic mode functions are respectively used as the grayscale values ​​of the red (R), green (G), and blue (B) color channels of the RGB image.

2. The method according to claim 1, characterized in that, The acquired acoustic signal is processed using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions (EMFs). From these EMFs, three key EMFs are selected, including: Applying Formula 1 The original acoustic signal is decomposed into a set of modal components, where K is the number of modal components. For a set of modal components, The center frequency corresponding to the modal component. It is the k-th modal component. It is the center frequency of the kth modal component. t This represents the derivative operation of a signal, where j represents the imaginary unit and t is the time variable. This represents the convolution operation. It is the Dirac function; For Formula 1, we introduce Lagrange multipliers. and penalty items Construct the Lagrange function The solution is obtained by using an iterative optimization algorithm. , and Until the change in the modal component is less than a preset threshold, wherein, Let Lagrange multipliers be the functions of the Lagrange multipliers. The original acoustic signal; Analyze the characteristics of all the modal components, and select three modal components containing key signal information as the three key intrinsic mode functions.

3. The method according to claim 1, characterized in that, The construction process of the gray-level co-occurrence matrix includes: Select a reference point (x, y) and a co-occurring point (x+a, y+b) on the voiceprint image, such that the gray values ​​of the reference point and the co-occurring point form a gray value combination (g1, g2). Let the grayscale level of the voiceprint image be k. By transforming the coordinates of the reference point and the co-occurring point, various combinations of grayscale values ​​(g1, g2) are obtained. The number of combinations of grayscale values ​​(g1, g2) is k. 2 kind; Traverse all the reference points in the voiceprint image, count the frequency of occurrence of each gray value combination (g1, g2), and obtain the frequency statistics result; The frequency statistics results are organized into a square matrix, and the square matrix is ​​normalized to convert it into a probability matrix P(g1, g2). The probability matrix is ​​an initial gray-level co-occurrence matrix, wherein the rows and columns of the initial gray-level co-occurrence matrix correspond to the gray values ​​g1 and g2 in the voiceprint image, respectively. Using formula The initial gray-level co-occurrence matrix is ​​normalized, wherein, The gray-level co-occurrence matrix is ​​the normalized form, and a and b are the distance difference values ​​of the gray values ​​g1 and g2 in the voiceprint image. R represents the number of gray levels, and R is the normalization factor.

4. The method according to claim 1, characterized in that, Texture features of the voiceprint image are extracted using the gray-level co-occurrence matrix, including: Using formula The contrast feature quantity is extracted from the gray-level co-occurrence matrix using the formula. The energy feature quantity is extracted from the gray-level co-occurrence matrix using the formula. Extract the entropy feature from the gray-level co-occurrence matrix. Where i and j traverse the rows and columns of the gray-level co-occurrence matrix, respectively representing a gray value in the gray-level co-occurrence matrix. The element in the gray-level co-occurrence matrix represents the probability of a pixel pair with gray values ​​i and j appearing at a given distance difference value (a, b).

5. The method according to claim 1, characterized in that, The construction process of the CNN-SVM model includes: A CNN model is constructed, comprising an input layer, a convolutional layer, an activation layer, a pooling layer, and a fully connected layer. The input layer takes the voiceprint image as input. The convolutional layer automatically learns the texture features of the voiceprint image through convolutional kernels. The activation layer increases the non-linear expressive power of the network. The pooling layer reduces the size of the voiceprint image, extracts high-level features of the voiceprint image, reduces computational complexity, and prevents overfitting. The fully connected layer uses the ReLU activation function to convert the two-dimensional feature map output by the convolutional layer into a one-dimensional vector, thus obtaining a feature vector. The feature vectors are input into the SVM model for training to obtain the CNN-SVM model.

6. The method according to claim 1, characterized in that, Determining the location of the fault source using the time delay difference includes: Using formula Calculate the distance d between the fault source and the i-th sensor. i Where t0 is the time when the leakage fault occurs, t i The time when the i-th sensor receives the sound signal is v, where v is the speed of sound; Solve the objective function The coordinates of the fault source in space are calculated, and these coordinates are taken as the location of the fault source, where P is the location of the fault source. Let be the position of the i-th sensor, and D be the number of acoustic signal sensors.

7. An electrical equipment leakage current identification device based on voiceprint signals, characterized in that, include: The processing unit is used to process the acquired acoustic signal using a variational mode decomposition algorithm to obtain multiple intrinsic mode functions, and select three key intrinsic mode functions from the multiple intrinsic mode functions. The generation unit is used to generate the voiceprint image of the sound signal by taking the three key intrinsic mode functions as the gray values ​​of the red, green and blue color channels of the image, respectively. The extraction unit is used to extract the texture features of the voiceprint image through the gray-level co-occurrence matrix, wherein the texture features include contrast features, energy features and entropy features; The classification and recognition unit is used to perform deep feature extraction and classification and recognition of the texture features using a CNN-SVM model to obtain leakage detection results. The calculation unit is used to calculate the time delay difference of the sound signal propagating to multiple sound signal sensors when the leakage detection result indicates that a leakage fault has occurred, and to use the time delay difference to determine the location of the fault source. The generation unit includes: The processing module is used to normalize each of the key intrinsic mode functions; A creation module is used to create a framework for an RGB image, which consists of three color channels: red, green, and blue. Each color channel represents information about different colors in the RGB image. Build modules for using formulas The voiceprint image is constructed, where P(.) represents the pixel grayscale value at the corresponding position in the RGB image, B(.) represents the normalized amplitude of the key intrinsic mode function, the size of the RGB image is M×N, i=1:M, j=1:N, and the normalized amplitudes of the three key intrinsic mode functions are respectively used as the grayscale values ​​of the red (R), green (G), and blue (B) color channels of the RGB image.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the computer-readable storage medium to perform the electrical equipment leakage current identification method based on voiceprint signals as described in any one of claims 1 to 6.

9. An electronic device, characterized in that, include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing the electrical equipment leakage current identification method based on voiceprint signals as described in any one of claims 1 to 6.

Images