GIS equipment acoustic fault diagnosis and monitoring method and system based on edge calculation
By using edge computing-based wavelet packet decomposition and MFCC feature extraction technology to process voiceprint information, the problem that traditional monitoring methods are difficult to capture highly concealed faults is solved, and more accurate fault diagnosis is achieved.
Patent Information
- Application Number
- CN202510094161.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-21
- Publication Date
- 2025-04-25
AI Technical Summary
Traditional monitoring methods are difficult to effectively capture and analyze hidden and difficult to detect faults in GIS equipment, such as slight looseness of mechanical components and instantaneous fluctuations in electrical parameters, especially the acoustic characteristics of faults.
The wavelet packet decomposition module and the MFCC feature extraction module based on edge computing are used to process the voiceprint information of GIS equipment, extract the wavelet packet characteristics and MFCC characteristics, and obtain the prediction results through the classification module, build a voiceprint recognition model, and deploy it on the edge computing device to achieve fault diagnosis.
By combining wavelet packet decomposition and MFCC feature extraction technology, the features in the voiceprint information of GIS equipment can be extracted more comprehensively, thereby obtaining more accurate fault diagnosis results, and improving the ability to capture and analyze highly concealed faults.
Smart Images
Figure CN119889360A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of audio data processing technology, and in particular to a method and system for GIS equipment acoustic fault diagnosis and monitoring based on edge computing. Background Art
[0002] GIS (Gas Insulated Switchgear) equipment, as an indispensable switchgear in modern power systems, is of great importance. They are like the heart valves of the power system, and their precise opening and closing are directly related to the stability and safety of the entire power grid. However, just as even the strongest machinery has its weaknesses, GIS equipment will inevitably encounter various challenges and failures in its day-to-day and year-to-year operation.
[0003] In the operation of the power system, the opening and closing operation of GIS equipment is a crucial link. The success or failure of this action not only affects the normal transmission of electric energy, but also affects the safety and stability of the entire power grid. However, due to the complexity of the internal structure of GIS equipment and the particularity of the operating environment, various faults may be encountered during the opening and closing process. Among them, the most common is the problem of inadequate opening and closing due to mechanical structure wear, poor lubrication or improper electrical parameter settings. In addition, as a key component for driving the opening and closing of GIS equipment, the voltage stability of the energy storage motor also directly affects the normal operation of the equipment. Once the voltage is abnormal, it may cause the opening and closing action to be slow or inadequate, or even cause more serious faults.
[0004] In the face of these challenges, traditional monitoring methods seem to be powerless. Although traditional mechanical sensors can monitor some of the operating conditions of the equipment, they are often powerless to detect some special faults that are highly hidden and difficult to detect, such as slight looseness of mechanical parts and instantaneous fluctuations in electrical parameters. Similarly, although electrical detection devices can capture changes in the electrical signals of the equipment, it is difficult to effectively capture and analyze the acoustic characteristics of the fault - those weak sounds caused by mechanical friction, electrical discharge, etc. Summary of the invention
[0005] Based on this, it is necessary to provide an edge computing-based GIS equipment acoustic fault diagnosis and monitoring method and system that can improve fault diagnosis in response to the above technical problems.
[0006] In the first aspect, the present application provides a method for GIS equipment acoustic fault diagnosis and monitoring based on edge computing. The method comprises:
[0007] The wavelet packet decomposition module and the MFCC feature extraction module are used to process the voiceprint information of the GIS equipment, and the wavelet packet features and the MFCC features are extracted respectively; the classification module is used to obtain the first prediction result and the second prediction result based on the wavelet packet features and the MFCC features; the first prediction result and the second prediction result are combined to obtain the target prediction result;
[0008] Construct a voiceprint recognition model including wavelet packet decomposition module, MFCC feature extraction module and classification module;
[0009] Train the voiceprint recognition model and deploy the trained voiceprint recognition model on the edge computing device to obtain fault diagnosis results based on the voiceprint information of the GIS equipment to be identified.
[0010] In one embodiment, the extraction of wavelet packet features includes:
[0011] The voiceprint information of GIS equipment is processed by multi-layer wavelet packet decomposition technology to obtain the interval energy in each frequency band and extract the wavelet packet features composed of interval energy characteristics.
[0012] In one embodiment, the extraction of MFCC features includes:
[0013] Perform short-time Fourier transform on the voiceprint information of GIS equipment to obtain the spectrum;
[0014] Filter the spectrum using a Mel filter bank to obtain the filtered spectrum energy;
[0015] Take the logarithm of the spectrum energy and perform discrete cosine transform to obtain MFCC features.
[0016] In one embodiment, obtaining the first prediction result and the second prediction result based on the wavelet packet feature and the MFCC feature by using the classification module includes:
[0017] Construct the discriminant function of the classification module; the discriminant function includes Lagrange multipliers, bias terms and kernel functions;
[0018] Training the voiceprint recognition model includes:
[0019] The Lagrange multiplier and bias term are determined by solving the optimal problem to obtain the trained voiceprint recognition model.
[0020] In one embodiment, the method further comprises:
[0021] The attention mechanism is introduced to perform weighted operations on wavelet packet features and MFCC features respectively to obtain the wavelet packet attention matrix and MFCC attention matrix;
[0022] Using the classification module based on the wavelet packet feature and the MFCC feature, obtaining the first prediction result and the second prediction result includes:
[0023] Input the wavelet packet attention matrix into the classification module to obtain the first prediction result;
[0024] The MFCC attention matrix is input into the classification module to obtain the second prediction result.
[0025] In one embodiment, combining the first prediction result and the second prediction result to obtain the target prediction result includes:
[0026] A genetic algorithm is used to obtain weights of the first prediction result and the second prediction result, and a weighted sum is performed on the first prediction result and the second prediction result based on the weights to obtain a target prediction result.
[0027] In the second aspect, the present application also provides a GIS equipment acoustic fault diagnosis and monitoring system based on edge computing, the system comprising:
[0028] The model preparation module is used to extract wavelet packet features and MFCC features; obtain the first prediction result and the second prediction result based on the wavelet packet features and the MFCC features using the classification module; and obtain the target prediction result by combining the first prediction result and the second prediction result;
[0029] A model building module is used to build a voiceprint recognition model including a wavelet packet decomposition module, an MFCC feature extraction module and a classification module;
[0030] The model deployment module is used to train the voiceprint recognition model and deploy the trained voiceprint recognition model on the edge computing device to obtain the fault diagnosis results based on the voiceprint information of the GIS equipment to be identified.
[0031] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps in the above-mentioned edge computing-based GIS equipment acoustic fault diagnosis and monitoring method when executing the computer program.
[0032] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps in the above-mentioned edge computing-based GIS equipment acoustic fault diagnosis and monitoring method when executed by a processor.
[0033] In a fifth aspect, the present application also provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned GIS equipment acoustic fault diagnosis and monitoring method based on edge computing are implemented.
[0034] The above-mentioned GIS equipment acoustic fault diagnosis and monitoring method and system based on edge computing uses a wavelet packet decomposition module and an MFCC feature extraction module to process the voiceprint information of the GIS equipment, and extracts wavelet packet features and MFCC features respectively; uses a classification module to obtain a first prediction result and a second prediction result based on wavelet packet features and MFCC features; combines the first prediction result and the second prediction result to obtain a target prediction result; constructs a voiceprint recognition model including a wavelet packet decomposition module, an MFCC feature extraction module, and a classification module; trains the voiceprint recognition model, and deploys the trained voiceprint recognition model on an edge computing device to obtain a fault diagnosis result based on the voiceprint information of the GIS equipment to be identified. This application combines the wavelet packet decomposition technology and the MFCC feature extraction technology to make the two feature extraction technologies complementary, and more comprehensively extracts the features contained in the voiceprint information of the GIS equipment, thereby obtaining a more accurate fault diagnosis result. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 The figure is a flow chart of a method for GIS equipment acoustic fault diagnosis and monitoring based on edge computing in one embodiment;
[0036] Figure 2 The present invention is a structural block diagram of an edge computing-based GIS equipment acoustic fault diagnosis and monitoring system in one embodiment. DETAILED DESCRIPTION
[0037] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0038] The present application embodiment provides a GIS equipment acoustic fault diagnosis and monitoring method based on edge computing, such as Figure 1 As shown, the following steps are included:
[0039] Step 102, using a wavelet packet decomposition module and an MFCC feature extraction module to process the GIS equipment voiceprint information, extracting wavelet packet features and MFCC features respectively; using a classification module to obtain a first prediction result and a second prediction result based on the wavelet packet features and the MFCC features; and combining the first prediction result and the second prediction result to obtain a target prediction result.
[0040] In this embodiment, the wavelet packet feature has good time-frequency localization capability, can analyze the signal in both the time domain and the frequency domain, has a better processing effect on non-stationary signals, can adaptively select the best wavelet basis function according to the characteristics of the signal, divide the time-frequency plane more finely, and has a relatively high resolution for the high-frequency part of the signal. The MFCC (Mel-scale Frequency Cepstral Coefficients) feature is based on the Mel frequency scale in the frequency domain, which is close to the frequency perceived by the human ear, but the calculation accuracy in the high-frequency part is relatively low. The MFCC feature mainly focuses on the description of the spectral envelope of the speech signal, and loses the time domain information to a certain extent.
[0041] In addition, wavelet packet decomposition can decompose the signal in different frequency bands. By analyzing and processing each frequency band, it can extract features that are insensitive to noise to a certain extent and have a certain ability to suppress noise, but the performance may decrease in a strong noise environment. The MFCC feature itself has a certain robustness to noise, but its performance will be affected when the noise is large, especially in the high-frequency part, which may lead to a decrease in the accuracy of the feature due to the decrease in calculation accuracy.
[0042] The wavelet packet decomposition technology is encapsulated into a wavelet packet decomposition module, and the MFCC feature extraction technology is encapsulated into an MFCC feature extraction module. The collected GIS equipment voiceprint information is input into the wavelet packet decomposition module and the MFCC feature extraction module to obtain the wavelet packet features and the MFCC features. The wavelet packet features and the MFCC features are input into the same or different classification models to obtain the first prediction result corresponding to the wavelet packet features and the second prediction result corresponding to the MFCC features. The first prediction result and the second prediction result are comprehensively analyzed to obtain the target prediction result of the GIS equipment voiceprint information. The target prediction result can determine the fault status of the GIS equipment and the fault type.
[0043] Step 104, constructing a voiceprint recognition model including a wavelet packet decomposition module, an MFCC feature extraction module and a classification module.
[0044] The wavelet packet decomposition module, MFCC feature extraction module and classification module are encapsulated into one to form a voiceprint recognition model.
[0045] Step 106: train the voiceprint recognition model and deploy the trained voiceprint recognition model on the edge computing device to obtain the fault diagnosis result according to the voiceprint information of the GIS device to be identified.
[0046] Collect voiceprint information of abnormal opening and closing of GIS equipment during actual operation. Specifically, collect abnormal voiceprint information of opening and closing of GIS equipment during actual operation, and through laboratory simulation tests, collect voiceprint data of various abnormal states such as energy storage mechanism and operating mechanism, and build a sample library of GIS equipment fault voiceprints as training data for training voiceprint recognition models.
[0047] The trained voiceprint recognition model is deployed on the edge computing device. The edge computing device of this embodiment can be a smart terminal, mobile device, etc., which cooperates with microphones and other basic devices to collect voiceprint information of GIS equipment in real time, performs local computing tasks in real time, and quickly makes fault identification decisions, reducing dependence on cloud services, improving data processing real-time and fault response speed, thereby reducing production losses.
[0048] In one embodiment, the extraction of wavelet packet features includes: processing GIS equipment voiceprint information by multi-layer wavelet packet decomposition technology, obtaining interval energy in each frequency band, and extracting wavelet packet features composed of interval energy features.
[0049] Wavelet Packet Decomposition is a multi-resolution analysis method for non-stationary signal processing, which can subdivide audio signals at multiple scales and frequencies. The decomposition formula is as follows:
[0050]
[0051] Among them, W j，k (t) is the kth wavelet packet coefficient of the jth layer, x(n) is the original GIS equipment voiceprint information, and h(n) is the wavelet packet function.
[0052] Then, the interval energy E in each frequency band is calculated by multi-layer wavelet packet decomposition. j，k , the formula is:
[0053]
[0054] Finally, the interval energy features are extracted to form feature vectors, namely wavelet packet features, whose formula is expressed as:
[0055] F=[E j，1 , E j，2 , ..., E j，N ].
[0056] The characteristic coefficients of the audio signal extracted by wavelet packet decomposition are used to obtain the interval energy characteristics of multi-level frequency bands. The characteristic data is represented as a characteristic vector F as the input data of the classification module.
[0057] In one embodiment, the extraction of MFCC features includes: performing short-time Fourier transform on the voiceprint information of the GIS device to obtain a spectrum; filtering the spectrum using a Mel filter group to obtain the filtered spectrum energy; taking the logarithm of the spectrum energy and performing discrete cosine transform to obtain MFCC features.
[0058] The feature extraction of audio signals is performed based on MFCC. Its main steps include short-time Fourier transform (STFT), filter bank filtering, logarithmic value extraction and discrete cosine transform (DCT).
[0059] Specifically, the GIS device voiceprint information x(n) is first subjected to STFT transformation to obtain the spectrum X(k), which is expressed as follows:
[0060]
[0061] Then, the Mel filter bank Hm(k) filters the spectrum and calculates the energy of the filtered spectrum, which is expressed as:
[0062]
[0063] Finally, the logarithm of the spectrum energy is taken and DCT transformation is performed to obtain the characteristic coefficient Cm of the mth audio signal, which is expressed as follows:
[0064]
[0065] Here, M refers to the number of Mel filters.
[0066] MFCC features are extracted using MFCC feature extraction technology as input data for the classification module.
[0067] Based on the extracted acoustic features of wavelet packet features and MFCC features, an intelligent voiceprint recognition model is constructed by combining the characteristics of the two features, and machine learning and deep learning algorithms are used to identify and classify the fault voiceprints of GIS equipment.
[0068] In one embodiment, in step 102, using the classification module based on wavelet packet features and MFCC features to obtain the first prediction result and the second prediction result includes: constructing a discriminant function of the classification module; the discriminant function includes Lagrange multipliers, bias terms and kernel functions. In step 106, training the voiceprint recognition model includes: determining the Lagrange multipliers and bias terms by solving the optimal problem to obtain a trained voiceprint recognition model.
[0069] In the process of training the voiceprint recognition model, N training samples are given, each sample is represented by (x i,y i ), where x i is the acoustic feature, y i is the corresponding classification label (such as normal, fault type 1, or fault type 2, etc.).
[0070] During the training of the voiceprint recognition model, the Lagrange multiplier α is determined by solving the optimization problem i And bias term b, so as to construct the optimal classification hyperplane.
[0071] The discriminant function of the specific classification model is as follows:
[0072]
[0073] Among them, α i is the Lagrange multiplier, which represents the influence weight of the sample point on the classification hyperplane, y i is the classification label of each sample, with a value of +1 or -1, K(x i , x) is the kernel function, which is used to calculate the difference between input data x and training sample x i The similarity.
[0074] When the value of f(x) is greater than 0, the input data x is classified as a positive class (such as a normal state); when the value of f(x) is less than 0, the input data is classified as a negative class (such as a fault state).
[0075] By choosing a suitable kernel function K(x i , x) and parameter α i , the feature data can be mapped into a high-dimensional feature space, so as to find a better classification hyperplane in the space.
[0076] In practical applications, the extracted wavelet packet features and MFCC features of the GIS equipment voiceprint can be directly input into the corresponding same classification module SVM, and the discriminant function f(x) can be used to calculate the category of each input sample, and the fault state of the equipment can be judged based on the results. The wavelet packet features and MFCC features can also be input into the corresponding different classification modules, for example, the wavelet packet features can be input into the SVM, and the MFCC features can be input into the CNN model.
[0077] If the voiceprint recognition model detects that the input voiceprint is highly matched with the fault category, a corresponding fault alarm prompt will be issued.
[0078] In one embodiment, the method further includes: introducing an attention mechanism to perform weighted operations on the wavelet packet features and the MFCC features respectively, and obtaining a wavelet packet attention matrix and an MFCC attention matrix. In step 102, using a classification module to obtain a first prediction result and a second prediction result based on the wavelet packet features and the MFCC features includes: inputting the wavelet packet attention matrix into the classification module to obtain a first prediction result; inputting the MFCC attention matrix into the classification module to obtain a second prediction result.
[0079] This embodiment introduces an attention mechanism to assign a weight to each element of the input sequence, which represents the importance of the element in the current processing step. Then, the features are weighted and summed by these weights to obtain a context vector, which integrates the information of each element in the input sequence, but focuses more on the important parts. Therefore, the attention mechanism allows the model to dynamically focus on different input parts when processing each position of the input sequence, thereby improving the model's ability to capture important information.
[0080] In one embodiment, in step 102, combining the first prediction result and the second prediction result to obtain a target prediction result includes: using a genetic algorithm to obtain weights of the first prediction result and the second prediction result, and performing a weighted summation of the first prediction result and the second prediction result based on the weights to obtain the target prediction result.
[0081] Genetic Algorithm (GA) is an optimization method based on natural selection and genetics principles, and is widely used in optimization and search problems. In machine learning, genetic algorithms can be used to combine the results of multiple classifiers so that the combined classification results have the highest accuracy or other evaluation indicators to improve classification performance.
[0082] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.
[0083] Based on the same inventive concept, the embodiment of the present application also provides an edge computing-based GIS equipment acoustic fault diagnosis and monitoring system for implementing the above-mentioned edge computing-based GIS equipment acoustic fault diagnosis and monitoring method. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more edge computing-based GIS equipment acoustic fault diagnosis and monitoring system embodiments provided below can refer to the limitations of the edge computing-based GIS equipment acoustic fault diagnosis and monitoring method above, and will not be repeated here.
[0084] In one embodiment, Figure 2 As shown, a GIS equipment acoustic fault diagnosis and monitoring system based on edge computing is provided, including:
[0085] The model preparation module 202 is used to process the voiceprint information of the GIS equipment using the wavelet packet decomposition module and the MFCC feature extraction module to extract the wavelet packet features and the MFCC features respectively; obtain the first prediction result and the second prediction result based on the wavelet packet features and the MFCC features using the classification module; and obtain the target prediction result by combining the first prediction result and the second prediction result;
[0086] A model building module 204 is used to build a voiceprint recognition model including a wavelet packet decomposition module, an MFCC feature extraction module and a classification module;
[0087] The model deployment module 206 is used to train the voiceprint recognition model and deploy the trained voiceprint recognition model on the edge computing device to obtain the fault diagnosis result according to the voiceprint information of the GIS device to be identified.
[0088] Aiming at the abnormal situations that may occur during the opening and closing operation of GIS equipment, the present invention proposes a monitoring system based on voiceprint feature extraction and intelligent voiceprint recognition model. The system collects and analyzes the abnormal voiceprint information of GIS equipment, establishes a voiceprint feature sample library, and realizes real-time monitoring and fault identification of the opening and closing status of the equipment based on machine learning and pattern recognition algorithms. The overall technical solution is divided into four parts: voiceprint data collection, feature extraction, intelligent voiceprint recognition model construction, and non-contact acoustic monitoring platform development.
[0089] Each module in the above-mentioned edge computing-based GIS equipment acoustic fault diagnosis and monitoring system can be implemented in whole or in part through software, hardware, and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.
[0090] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in all the above method embodiments when executing the computer program.
[0091] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in all the above method embodiments are implemented.
[0092] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in all the above method embodiments when executed by a processor.
[0093] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.
[0094] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited thereto. The processors involved in each embodiment provided in this application may be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, data processing logic devices based on quantum computing, etc., but are not limited thereto.
[0095] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0096] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.
Claims
1. A GIS equipment acoustic fault diagnosis and monitoring method based on edge computing, characterized in that: The method comprises: The wavelet packet decomposition module and the MFCC feature extraction module are used to process the voiceprint information of the GIS equipment, and the wavelet packet features and the MFCC features are extracted respectively; the classification module is used to obtain the first prediction result and the second prediction result based on the wavelet packet features and the MFCC features; the first prediction result and the second prediction result are combined to obtain the target prediction result; Constructing a voiceprint recognition model including the wavelet packet decomposition module, the MFCC feature extraction module and the classification module; The voiceprint recognition model is trained, and the trained voiceprint recognition model is deployed on an edge computing device to obtain a fault diagnosis result based on the voiceprint information of the GIS device to be identified.
2. The method according to claim 1, characterized in that: The extraction of the wavelet packet features includes: The GIS equipment voiceprint information is processed by multi-layer wavelet packet decomposition technology to obtain the interval energy in each frequency band, and the wavelet packet features composed of interval energy features are extracted.
3. The method according to claim 1, characterized in that The extraction of the MFCC features includes: Performing short-time Fourier transform on the voiceprint information of the GIS device to obtain a frequency spectrum; Filtering the spectrum using a Mel filter bank to obtain filtered spectrum energy; The logarithm of the spectrum energy is taken, and discrete cosine transform is performed to obtain the MFCC feature.
4. The method according to claim 1, characterized in that The obtaining of the first prediction result and the second prediction result by using the classification module based on the wavelet packet feature and the MFCC feature includes: Constructing a discriminant function of the classification module; the discriminant function includes a Lagrange multiplier, a bias term and a kernel function; The training of the voiceprint recognition model comprises: The Lagrange multiplier and the bias term are determined by solving the optimal problem to obtain the trained voiceprint recognition model.
5. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: An attention mechanism is introduced to perform weighted operations on the wavelet packet features and the MFCC features respectively to obtain a wavelet packet attention matrix and an MFCC attention matrix; The obtaining of the first prediction result and the second prediction result by using the classification module based on the wavelet packet feature and the MFCC feature includes: Inputting the wavelet packet attention matrix into the classification module to obtain the first prediction result; The MFCC attention matrix is input into the classification module to obtain the second prediction result.
6. The method according to claim 1, characterized in that The step of combining the first prediction result and the second prediction result to obtain a target prediction result includes: A genetic algorithm is used to obtain weights of the first prediction result and the second prediction result, and a weighted sum is performed on the first prediction result and the second prediction result based on the weights to obtain the target prediction result.
7. A GIS equipment acoustic fault diagnosis and monitoring system based on edge computing, characterized in that: The system comprises: A model preparation module is used to process the voiceprint information of GIS equipment using a wavelet packet decomposition module and an MFCC feature extraction module to extract wavelet packet features and MFCC features respectively; obtain a first prediction result and a second prediction result based on the wavelet packet features and the MFCC features using a classification module; and obtain a target prediction result by combining the first prediction result and the second prediction result; A model building module, used to build a voiceprint recognition model including the wavelet packet decomposition module, the MFCC feature extraction module and the classification module; The model deployment module is used to train the voiceprint recognition model and deploy the trained voiceprint recognition model on the edge computing device to obtain the fault diagnosis result according to the voiceprint information of the GIS device to be identified.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.