Switch discharge detection method and system based on multi-source information fusion voiceprint signal

Abstract

The invention discloses a switch discharge detection method and system based on multi-source information fusion voiceprint signals. The method comprises the following steps: acquiring multi-source data of an open switch; preprocessing the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and performing feature extraction on the acoustic vibration reconstruction signal according to an optimized MCKD algorithm to obtain an acoustic vibration reconstruction signal feature; and constructing a CNN-LSTM fault identification model, introducing an attention mechanism, inputting the acoustic vibration reconstruction signal features in combination with the voltage data, the current data and the temperature data into the CNN-LSTM fault identification model for feature classification, and obtaining a fault type by using a softmax activation function according to the classified features. By combining the respective characteristics of the convolutional neural network and the long and short base memory network, the defect information existing in the voiceprint signal is effectively extracted, and the accurate diagnosis of the open switch state is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power detection equipment, and in particular relates to a switch discharge detection method and system based on multi-source information fusion voiceprint signals. Background Art

[0002] Open switches play an important role in power systems. They can effectively isolate circuits and electrical equipment, ensuring that relevant parts can be safely isolated during equipment maintenance, overhaul or failure, thereby preventing accidental electric shock accidents. With the continuous development of power systems and the increasing requirements for equipment performance, it is particularly necessary to develop open switch detection technology. Accurate detection technology can prevent potential failures of open switches, detect and deal with possible problems in a timely manner, and thus improve the safety of power systems. By monitoring the status of open switches in real time, the operation of power systems can be optimized, unnecessary shutdowns or switching operations can be reduced, and the reliability and stability of power supply can be improved. In addition, detection technology can also reduce operating costs, extend the service life of open switches, and reduce maintenance costs by timely detecting and repairing potential problems. In the context of smart grids, the requirements for monitoring and control of power system equipment are getting higher and higher. The development of open switch detection technology can adapt to this trend and realize remote monitoring and intelligent control of power system equipment. At the same time, with the growth of power load, the power system has higher and higher requirements for the reliability and safety of equipment. Faults such as damage caused by heat due to increased resistance of the selector switch contacts, improper coordination between the selector switch and the switching switch, stuck gear boxes, loose fastening bolts of the horizontal connecting rods, and worn contacts can be divided into electrical faults and mechanical faults. Many electrical faults are caused by poor contact due to degradation and deformation of mechanical parts. Therefore, research and development of open switch detection technology can ensure that the power system can still operate stably when facing high loads and meet the growing demand for electricity. The detection technology based on voiceprint signals has the characteristics of non-contact and non-power outage. The present invention uses voiceprint signals combined with other operating condition information to comprehensively diagnose the status of open switches, which has higher feasibility than a single information source. Summary of the invention

[0003] The present invention provides a switch discharge detection method and system based on multi-source information fusion voiceprint signals, which are used to solve the technical problem of electrical failure caused by poor contact due to degradation and deformation of mechanical parts.

[0004] In a first aspect, the present invention provides a switch discharge detection method based on multi-source information fusion voiceprint signal, comprising: Acquire multi-source data of the open switch, wherein the multi-source data includes voltage data, current data, temperature data, and acoustic vibration signals; Preprocessing the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and extracting features of the acoustic vibration reconstruction signal according to an optimized MCKD algorithm to obtain features of the acoustic vibration reconstruction signal; A CNN-LSTM fault identification model is constructed, and an attention mechanism is introduced. The acoustic vibration reconstruction signal features are combined with the voltage data, the current data, and the temperature data to be input into the CNN-LSTM fault identification model for feature classification. A softmax activation function is used to obtain the fault type according to the classified features.

[0005] In a second aspect, the present invention provides a switch discharge detection system based on multi-source information fusion voiceprint signal, comprising: An acquisition module configured to acquire multi-source data of the open switch, wherein the multi-source data includes voltage data, current data, temperature data, and acoustic vibration signals; An extraction module is configured to preprocess the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and perform feature extraction on the acoustic vibration reconstruction signal according to an optimized MCKD algorithm to obtain features of the acoustic vibration reconstruction signal; The classification module is configured to build a CNN-LSTM fault identification model and introduce an attention mechanism to input the acoustic vibration reconstruction signal features into the CNN-LSTM fault identification model in combination with the voltage data, the current data and the temperature data for feature classification, and use a softmax activation function to obtain the fault type according to the classified features.

[0006] According to a third aspect, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the steps of the switch discharge detection method based on multi-source information fusion voiceprint signals of any embodiment of the present invention.

[0007] In a fourth aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the program instructions are executed by a processor, the processor executes the steps of a switch discharge detection method based on multi-source information fusion voiceprint signals according to any embodiment of the present invention.

[0008] The switch discharge detection method and system based on multi-source information fusion voiceprint signals of the present application can more comprehensively understand the operating status of the equipment by combining the rich frequency information contained in the voiceprint signal and the characteristics of the vibration signal that more directly reflects the mechanical state of the equipment, and integrating the complementarity between other operating parameters; by optimizing the parameters of the MCKD algorithm using the improved zebra optimization algorithm, the sensitivity of the MCKD algorithm to defect signals can be effectively improved; by combining the respective characteristics of the convolutional neural network and the long-short base memory network, the defect information existing in the voiceprint signal can be effectively extracted to achieve accurate diagnosis of the open switch state. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 A flow chart of a switch discharge detection method based on multi-source information fusion voiceprint signal provided by an embodiment of the present invention; Figure 2 A structural block diagram of a switch discharge detection system based on multi-source information fusion voiceprint signal provided by an embodiment of the present invention; Figure 3 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0012] See also Figure 1 , which shows a flow chart of a switch discharge detection method based on multi-source information fusion voiceprint signal of the present application.

[0013] like Figure 1 As shown, the switch discharge detection method based on multi-source information fusion voiceprint signal specifically includes the following steps: Step S101 : acquiring multi-source data of an open switch, wherein the multi-source data includes voltage data, current data, temperature data, and acoustic vibration signals.

[0014] Step S102 , preprocessing the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and extracting features of the acoustic vibration reconstruction signal according to the optimized MCKD algorithm to obtain features of the acoustic vibration reconstruction signal.

[0015] In this step, multi-source data of the open switch including voltage, current, temperature, and acoustic vibration signals (vibration signals and voiceprint signals) are collected; in view of the complex acoustic vibration signal collection environment of the open switch in the substation environment, the acoustic vibration signals (vibration signals and voiceprint signals) are enhanced, that is, the wavelet packet algorithm is used to decompose and reduce the noise of the collected acoustic vibration signals to obtain a pure acoustic vibration signal, that is, an acoustic vibration reconstruction signal.

[0016] Specifically, a 4-layer Morse wavelet basis function is used to perform wavelet packet decomposition on the acoustic vibration signal; The wavelet packet coefficients obtained by wavelet packet decomposition are processed by using an improved threshold processing method. Perform noise reduction, where the noise reduction expression is: , , In the formula, is the wavelet packet coefficient The result obtained after threshold processing is is the threshold parameter, is the symbolic function, is the signal length of the noise, is the noise standard deviation of the acoustic vibration signal, level is the number of layers of wavelet packet coefficients, is an exponential function; When the fault signal When filtering is performed until the correlation kurtosis reaches the preset maximum value, the inverse FIR filter meets the requirements, and the inverse FIR filter length is calculated to obtain the inverse FIR filter output result, which is expressed as: Set the actual output fault signal for: , In the formula, is the original input fault signal, It is the sound signal of switch action. for noise; Calculation fault signal The kurtosis of , the expression is: , , In the formula, is the fault signal sampling length, is the displacement number, is the number of random displacements, is the fault signal period, To find the derivative symbol, are the inverse FIR filter coefficients, is the inverse FIR filter length; When the fault signal When filtering is performed until the correlation kurtosis reaches the preset maximum value, the inverse FIR filter meets the requirements, and the length of the inverse FIR filter is calculated to obtain the characteristic result of the inverse FIR filter output, which is expressed as: , , , In the formula, is the gain coefficient of the filter, The signal matrix Lines contain Start to m+1 consecutive but spaced of value, For the Fault signal value, , For the Fault signal value, is the signal matrix, is the transpose of the signal matrix, The displacement is , the period is The signal matrix, For the The elements are Multiply by Start to m+1 The product of .

[0017] In a specific embodiment, before extracting features of the vibroacoustic reconstruction signal according to the optimized MCKD algorithm to obtain features of the vibroacoustic reconstruction signal, the method further includes: The improved zebra optimization algorithm is used to globally optimize the parameters of the MCKD algorithm to obtain the optimal parameters of the MCKD algorithm. The improved zebra optimization algorithm is specifically as follows: Phase 1, foraging behavior: The zebra population members are updated based on the simulation of the members’ behavior when searching for feed; The optimal zebra population member is set as the pioneer zebra, and the other members of the zebra population except the pioneer zebra are led to the position of the pioneer zebra in the search space, and the position of the zebra population members in the foraging stage is updated. The expression is: , , In the formula, The latest in the first phase The only member of the zebra race is ranked first Dimension values, For the The only member of the zebra race is ranked first Dimension values, is a random number between 0 and 1. is the position of the pioneer zebra in the zebra population, is a random value in the set {1, 2}, Update for the first phase The location of the zebra race members, For the first stage The function evaluation value of only the zebra race members, is the evaluation function, For the A member of the zebra race; Use the Levy flight distribution function for initialization, the expression is: , , In the formula, For the The component of the jth dimension of the position of each zebra population member after the update in the first stage, For the The component of the jth dimension on the position of each zebra population member, is a random number generated in the range of 0 to 2. is the Levy flight distribution function, , They are all random numbers generated in the range of 0 to 1. is a fixed constant, is the gamma function; Phase 2: Defense strategies against predators: Assume that the zebra population members choose the escape strategy or the attack strategy with the same probability, update the position of the zebra population members in the search space, and gradually converge the population to the current optimal solution according to the adaptive "attraction-disturbance" strategy. The expression of the "attraction-disturbance" strategy is: , , , In the formula, The latest in the second phase The first member of the zebra race Dimension values, is a constant of 0.01, is the number of iterations, is the maximum number of iterations, is the switching probability between escape strategy and attack strategy, is the status of the attacked member in the zebra population, For the second stage The function evaluation value of only the zebra race members, The latest in the second phase The location of the zebra race members, For the The component of the j-th dimension of the position of each zebra population member after the update in the second stage.

[0018] Specifically, the improved zebra optimization algorithm is used to globally optimize the parameters of the MCKD algorithm, and the optimal parameters of the MCKD algorithm include: Initialize the zebra population and combine the influencing parameters of the MCKD algorithm As a member of the zebra population, each zebra population member corresponds to a pair of influencing parameters, and a certain number of influencing parameter combinations are randomly generated as the initial positions of the zebra population members, and the moving speed of each zebra population member is randomly initialized, where: is the FIR filter length, is the period of the FIR filter; Based on the position of the pioneer zebra in the zebra population, the MCKD algorithm is used to operate the acoustic vibration reconstruction signal and calculate the fitness function corresponding to each zebra population member; Compare the fitness values ​​of each zebra population member to update the local extreme value and global extreme value of each zebra population member in the zebra population, as well as the speed and position of each zebra population member; When the number of iterations reaches the maximum preset value, the best fitness value and the positions of zebra population members are output, and the optimal parameter combination of the MCKD algorithm is obtained.

[0019] Step S103, constructing a CNN-LSTM fault identification model, and introducing an attention mechanism, inputting the acoustic vibration reconstruction signal features into the CNN-LSTM fault identification model in combination with the voltage data, the current data and the temperature data for feature classification, and using a softmax activation function to obtain the fault type according to the classified features.

[0020] In this step, appropriate convolutional layers, pooling layers and activation functions are used to construct a convolutional neural network CNN, and the size of the convolution kernel in the convolutional layer is determined. A long short-term memory network LSTM is constructed, and the number of long short-term memory network LSTM, the number of hidden units and hyperparameters are defined. The constructed convolutional neural network CNN and the long short-term memory network LSTM are fused and connected to form a CNN-LSTM fault identification model; the convolutional neural network CNN in the CNN-LSTM fault identification model is used to extract the spatial features of multi-source data features, and the long short-term memory network LSTM is used to perform time series modeling on the extracted spatial features of multi-source data.

[0021] It should be noted that the current data, voltage data and temperature data are defined as , that is, the one-dimensional channel output characteristics, the acoustic vibration data is defined as , that is, the two-dimensional channel output feature; Calculate separately and The attention mechanism is and , the expression is: , , In the formula, for The output of the attention mechanism is for Query matrix, for The key matrix of for The value matrix of for The output of the attention mechanism is for The query matrix, for The key matrix of is the feature dimension of one-dimensional convolution, is the feature dimension of the two-dimensional convolution, for The value matrix of in, , , , In the formula, for or , Used to generate the query matrix, For generating the key matrix, Used to generate value matrices, collectively referred to as weight matrices; According to the weight parameters and Perform attention allocation and obtain attention parameters , the expression is: , In the formula, is the self-attention weight, is the cross attention weight.

[0022] In summary, the method of the present application, by combining the rich frequency information contained in the voiceprint signal and the characteristics of the vibration signal that more directly reflects the mechanical state of the equipment, and integrating the complementarity between other operating parameters, can more comprehensively understand the operating state of the equipment; by optimizing the parameters of MCKD using the zebra optimization algorithm, the sensitivity of MCKD to defect signals can be effectively improved; by combining the respective characteristics of the convolutional neural network and the long-short base memory network, the defect information existing in the voiceprint signal can be effectively extracted to achieve accurate diagnosis of the open switch state.

[0023] See also Figure 2 , which shows a structural block diagram of a switch discharge detection system based on multi-source information fusion voiceprint signal of the present application.

[0024] like Figure 2 As shown, the switch discharge detection system 200 includes an acquisition module 210 , an extraction module 220 and a classification module 230 .

[0025] Among them, the acquisition module 210 is configured to acquire multi-source data of the open switch, and the multi-source data includes voltage data, current data, temperature data and acoustic vibration signal; the extraction module 220 is configured to pre-process the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and perform feature extraction on the acoustic vibration reconstruction signal according to the optimized MCKD algorithm to obtain the acoustic vibration reconstruction signal feature; the classification module 230 is configured to construct a CNN-LSTM fault identification model, and introduce an attention mechanism, and input the acoustic vibration reconstruction signal feature into the CNN-LSTM fault identification model in combination with the voltage data, the current data and the temperature data for feature classification, and use the softmax activation function to obtain the fault type according to the classified features.

[0026] It should be understood that Figure 2 Modules and references documented in Figure 1 Therefore, the operations and features described above for the method and the corresponding technical effects are also applicable to Figure 2 The modules in it will not be described in detail here.

[0027] In some other embodiments, the present invention also provides a computer-readable storage medium on which a computer program is stored. When the program instructions are executed by a processor, the processor executes the switch discharge detection method based on multi-source information fusion voiceprint signal in any of the above method embodiments; As an implementation mode, the computer-readable storage medium of the present invention stores computer-executable instructions, and the computer-executable instructions are configured as follows: Acquire multi-source data of the open switch, wherein the multi-source data includes voltage data, current data, temperature data, and acoustic vibration signals; Preprocessing the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and extracting features of the acoustic vibration reconstruction signal according to an optimized MCKD algorithm to obtain features of the acoustic vibration reconstruction signal; A CNN-LSTM fault identification model is constructed, and an attention mechanism is introduced. The acoustic vibration reconstruction signal features are combined with the voltage data, the current data, and the temperature data to be input into the CNN-LSTM fault identification model for feature classification. A softmax activation function is used to obtain the fault type according to the classified features.

[0028] The computer-readable storage medium may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required by at least one function; the data storage area may store data created according to the use of the switch discharge detection system based on multi-source information fusion voiceprint signals, etc. In addition, the computer-readable storage medium may include a high-speed random access memory, and may also include a memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the switch discharge detection system based on multi-source information fusion voiceprint signals via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0029] Figure 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 3 As shown, the device includes: a processor 310 and a memory 320. The electronic device may also include: an input device 330 and an output device 340. The processor 310, the memory 320, the input device 330 and the output device 340 may be connected via a bus or other means. Figure 3In the example, the bus connection is used. The memory 320 is the above-mentioned computer-readable storage medium. The processor 310 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions and modules stored in the memory 320, that is, the switch discharge detection method based on multi-source information fusion voiceprint signal of the above-mentioned method embodiment is realized. The input device 330 can receive input digital or character information, and generate key signal input related to user settings and function control of the switch discharge detection system based on multi-source information fusion voiceprint signal. The output device 340 may include display devices such as display screens.

[0030] The electronic device can execute the method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the method provided by the embodiment of the present invention.

[0031] As an implementation mode, the electronic device is applied to a switch discharge detection system based on multi-source information fusion voiceprint signals, and is used for a client, and includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: Acquire multi-source data of the open switch, wherein the multi-source data includes voltage data, current data, temperature data, and acoustic vibration signals; Preprocessing the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and extracting features of the acoustic vibration reconstruction signal according to an optimized MCKD algorithm to obtain features of the acoustic vibration reconstruction signal; A CNN-LSTM fault identification model is constructed, and an attention mechanism is introduced. The acoustic vibration reconstruction signal features are combined with the voltage data, the current data, and the temperature data to be input into the CNN-LSTM fault identification model for feature classification. A softmax activation function is used to obtain the fault type according to the classified features.

[0032] Through the description of the above implementation modes, those skilled in the art can clearly understand that each implementation mode can be implemented by means of software plus a necessary general hardware platform, or of course by hardware. Based on such an understanding, the above technical solution can essentially or in other words be embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0033] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A switch discharge detection method based on multi-source information fusion voiceprint signal, characterized in that: include: Acquire multi-source data of the open switch, wherein the multi-source data includes voltage data, current data, temperature data, and acoustic vibration signals; Preprocessing the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and extracting features of the acoustic vibration reconstruction signal according to an optimized MCKD algorithm to obtain features of the acoustic vibration reconstruction signal; A CNN-LSTM fault identification model is constructed, and an attention mechanism is introduced. The acoustic vibration reconstruction signal features are combined with the voltage data, the current data, and the temperature data to be input into the CNN-LSTM fault identification model for feature classification. A softmax activation function is used to obtain the fault type according to the classified features.

2. According to claim 1, a switch discharge detection method based on multi-source information fusion voiceprint signal is characterized in that: The step of combining the acoustic vibration reconstruction signal feature with the voltage data, the current data and the temperature data and inputting the feature into the CNN-LSTM fault identification model for feature classification includes: Define current data, voltage data and temperature data as , that is, the one-dimensional channel output characteristics, the acoustic vibration data is defined as , that is, the two-dimensional channel output feature; Calculate separately and The attention mechanism is and , the expression is: ， ， In the formula, for The output of the attention mechanism is for Query matrix, for The key matrix of for The value matrix of for The output of the attention mechanism is for The query matrix, for The key matrix of is the feature dimension of one-dimensional convolution, is the feature dimension of the two-dimensional convolution, for The value matrix of in, , ， ， In the formula, for or , Used to generate the query matrix, For generating the key matrix, Used to generate value matrices, collectively referred to as weight matrices; According to the weight parameters and Perform attention allocation and obtain attention parameters , the expression is: ， In the formula, is the self-attention weight, is the cross attention weight.

3. The switch discharge detection method based on multi-source information fusion voiceprint signal according to claim 1 is characterized in that: The preprocessing of the acoustic vibration signal to obtain an acoustic vibration reconstruction signal comprises: Selecting a 4-layer Morse wavelet basis function to perform wavelet packet decomposition on the acoustic vibration signal; The wavelet packet coefficients obtained by wavelet packet decomposition are processed by using an improved threshold processing method. Perform noise reduction, where the noise reduction expression is: ， ， In the formula, is the wavelet packet coefficient The result obtained after threshold processing is is the threshold parameter, is the symbolic function, is the signal length of the noise, is the noise standard deviation of the acoustic vibration signal, level is the number of layers of wavelet packet coefficients, is an exponential function; The wavelet packet coefficients after threshold processing Reconstruction is performed to obtain an acoustic vibration reconstruction signal.

4. The switch discharge detection method based on multi-source information fusion voiceprint signal according to claim 1 is characterized in that: The feature extraction of the acoustic vibration reconstruction signal according to the optimized MCKD algorithm to obtain the acoustic vibration reconstruction signal features includes: Set the actual output fault signal for: ， In the formula, is the original input fault signal, It is the sound signal of switch action. for noise; Calculation fault signal The kurtosis of , the expression is: ， ， In the formula, is the fault signal sampling length, is the displacement number, To represent a specific data point in the signal, To represent a specific period sequence number in a periodic signal, is the number of random displacements, is the fault signal period, To find the derivative symbol, are the inverse FIR filter coefficients, is the inverse FIR filter length; When the fault signal When filtering is performed until the correlation kurtosis reaches the preset maximum value, the inverse FIR filter meets the requirements, and the inverse FIR filter length is calculated to obtain the inverse FIR filter output result, which is expressed as: ， ， ， In the formula, is the gain coefficient of the filter, The signal matrix Lines contain Start to m+1 consecutive but spaced of value, For the Fault signal value, , For the Fault signal value, is the signal matrix, is the transpose of the signal matrix, The displacement is , the period is The signal matrix, For the The elements are Multiply by Start, to m+1 The product of .

5. The switch discharge detection method based on multi-source information fusion voiceprint signal according to claim 1 is characterized in that: Before extracting features of the vibroacoustic reconstruction signal according to the optimized MCKD algorithm to obtain features of the vibroacoustic reconstruction signal, the method further includes: The improved zebra optimization algorithm is used to globally optimize the parameters of the MCKD algorithm to obtain the optimal parameters of the MCKD algorithm. The improved zebra optimization algorithm is specifically as follows: Phase 1, foraging behavior: The zebra population members are updated based on the simulation of the members’ behavior when searching for feed; The optimal zebra population member is set as the pioneer zebra, and the other members of the zebra population except the pioneer zebra are led to the position of the pioneer zebra in the search space, and the position of the zebra population members in the foraging stage is updated. The expression is: ， ， In the formula, The latest in the first phase The only member of the zebra race is ranked first Dimension values, For the The only member of the zebra race is ranked first Dimension values, is a random number between 0 and 1. is the position of the pioneer zebra in the zebra population, is a random value in the set {1, 2}, Update for the first phase The location of the zebra race members, For the first stage The function evaluation value of only the zebra race members, is the evaluation function, For the A member of the zebra race; Use the Levy flight distribution function for initialization, the expression is: ， ， In the formula, For the The component of the jth dimension of the position of each zebra population member after the update in the first stage, For the The component of the jth dimension on the position of each zebra population member, is a random number generated in the range of 0 to 2. is the Levy flight distribution function, , They are all random numbers generated in the range of 0 to 1. is a fixed constant, is the gamma function; Phase 2: Defense strategies against predators: Assume that the zebra population members choose the escape strategy or the attack strategy with the same probability, update the position of the zebra population members in the search space, and gradually converge the population to the current optimal solution according to the adaptive "attraction-disturbance" strategy. The expression of the "attraction-disturbance" strategy is: ， ， ， In the formula, The latest in the second phase The first member of the zebra race Dimension values, is a constant of 0.01, is the number of iterations, is the maximum number of iterations, is the switching probability between escape strategy and attack strategy, is the status of the attacked member in the zebra population, For the second stage The function evaluation value of only the zebra race members, The latest in the second phase The location of the zebra race members, For the The component of the j-th dimension of the position of each zebra population member after the update in the second stage.

6. A switch discharge detection method based on multi-source information fusion voiceprint signal according to claim 5, characterized in that: The improved zebra optimization algorithm is used to globally optimize the parameters of the MCKD algorithm, and the optimal parameters of the MCKD algorithm include: Initialize the zebra population and combine the influencing parameters of the MCKD algorithm As a member of the zebra population, each zebra population member corresponds to a pair of influencing parameters, and a certain number of influencing parameter combinations are randomly generated as the initial positions of the zebra population members, and the moving speed of each zebra population member is randomly initialized, where: is the FIR filter length, is the period of the FIR filter; Based on the position of the pioneer zebra in the zebra population, the MCKD algorithm is used to operate the acoustic vibration reconstruction signal and calculate the fitness function corresponding to each zebra population member; Compare the fitness values ​​of each zebra population member to update the local extreme value and global extreme value of each zebra population member in the zebra population, as well as the speed and position of each zebra population member; When the number of iterations reaches the maximum preset value, the best fitness value and the positions of zebra population members are output, and the optimal parameter combination of the MCKD algorithm is obtained.

7. A switch discharge detection system based on multi-source information fusion voiceprint signal, characterized in that: include: An acquisition module configured to acquire multi-source data of the open switch, wherein the multi-source data includes voltage data, current data, temperature data, and acoustic vibration signals; An extraction module is configured to preprocess the acoustic vibration signal to obtain an acoustic vibration reconstruction signal, and perform feature extraction on the acoustic vibration reconstruction signal according to an optimized MCKD algorithm to obtain features of the acoustic vibration reconstruction signal; The classification module is configured to build a CNN-LSTM fault identification model and introduce an attention mechanism to input the acoustic vibration reconstruction signal features into the CNN-LSTM fault identification model in combination with the voltage data, the current data and the temperature data for feature classification, and use a softmax activation function to obtain the fault type according to the classified features.

8. An electronic device, characterized in that: include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images