Power transmission line defect identification method and system based on voiceprint identification

Abstract

The invention discloses a voiceprint recognition-based power transmission line defect recognition method and system, and the method comprises the steps: firstly installing a sound collection system on a power transmission line, collecting sound signals transmitted by the power transmission line and equipment thereof, and then carrying out the preprocessing and feature extraction of the sound signals; and finally, classifying the collected sound signals by adopting a Bayesian classifier to achieve the purpose of identifying different types of defects. According to the invention, an effective technical means can be provided for determining the defect type of the power transmission line, early warning when the power transmission line has defects and saving the cost of power transmission line inspection.

Description

Technical Field

[0001] The present invention belongs to the technical field of power line inspection, and more specifically relates to a method and system for identifying transmission line defects based on voiceprint recognition. Background Art

[0002] With the rapid development of China's economy and the increasing demand for electric power, the construction scale of China's power grid has been continuously expanding, and more and more attention has been paid to the safe and stable operation of electric energy. Today, with the continuous development of China's economy, the demand for electric power resources in China is also increasing continuously, and the uses of electric energy are very extensive. Whether it is industry and agriculture, or scientific research and production, all walks of life are closely related to electric energy.

[0003] The power system includes links such as power generation, transformation, transmission, distribution, and power consumption. Among them, the transmission line, as an important part of the power system, plays a crucial role in the long-distance transmission of electric energy. However, most transmission lines are erected in the wild, which is not only affected by complex and changeable external environments, but also affected by factors such as its own structural design and human damage. Therefore, it is very easy to have faults. If not detected and maintained in time, major accidents are extremely likely to occur, resulting in inestimable economic losses. Therefore, it becomes particularly important to regularly inspect the transmission lines, reduce the occurrence of transmission line accidents, and ensure the safe and stable operation of the transmission lines and their equipment.

[0004] Traditional manual inspection requires power maintenance personnel to conduct on-site inspections, climb high towers, and use the naked eye or sensors to determine whether there are potential safety hazards and defects. This inspection method is not only costly and risky, but also has high requirements for the physical strength and energy of the inspection personnel, resulting in the inability to maximize the inspection efficiency. Therefore, traditional manual inspection can no longer meet the monitoring and maintenance requirements of transmission lines.

[0005] The prior art 1 (CN 107014827 A) provides a method, device, and system for analyzing transmission line defects based on image processing, including: setting the flight instructions of the unmanned aerial vehicle according to the position information of the transmission line input by the user; sending the flight instructions to the unmanned aerial vehicle so that the unmanned aerial vehicle flies according to the flight instructions; monitoring the unmanned aerial vehicle, and when it is detected that the distance between the unmanned aerial vehicle and the transmission line is less than the set threshold, sending an instruction to start shooting to the unmanned aerial vehicle so that the unmanned aerial vehicle shoots the image data of the transmission line; during the shooting process of the unmanned aerial vehicle, sending adjustment instructions to the unmanned aerial vehicle multiple times to adjust the operating state of the unmanned aerial vehicle multiple times so that the unmanned aerial vehicle shoots the transmission line from multiple angles and directions; obtaining the image data shot by the unmanned aerial vehicle; and determining the defects of the transmission line according to the image data. In this method, the collected images have many influencing factors, including the influence of light, different weather conditions, light environments, etc., which will all cause differences in the image formation of the pictures, resulting in errors in the identification of transmission line defects and low identification rates.

[0006] Prior art document 2 (CN 111340787 A) discloses a method, device and computer equipment for detecting and identifying conductor defects of power transmission lines. The method uses image enhancement and rendering to process conductor defect images to obtain a defect image sample library, and then uses the RetinaNet deep neural network to extract defect data, train and test defect data, and train defect recognition on the images in the defect image sample library to obtain a defect recognition neural network model. The image to be tested on the collected transmission line is input into the defect recognition neural network model for detection, and the conductor defects in the image to be tested are identified. This method also processes the collected transmission line image. Although the recognition rate is improved, the factors affecting the transmission line defect recognition based on the image still exist and cannot be completely removed. Summary of the invention

[0007] In order to solve the deficiencies in the prior art, the present invention provides a method and system for identifying defects in a power transmission line based on voiceprint recognition, which uses a sound signal collection device installed on the transmission line to extract the sound signal of the transmission line, extracts MFCC feature parameters and RCC feature parameters after processing the extracted sound signal through framing and Hamming window function, and then performs fusion and dimensionality reduction, identifies defects through a trained Bayesian classifier, realizes efficient defect identification of the transmission line, and transmits the identified defects to the background for early warning, thereby improving the inspection efficiency of the transmission line and reducing the inspection cost of the transmission line.

[0008] The present invention adopts the following technical solution.

[0009] A first aspect of the present invention provides a method for identifying defects in a power transmission line based on voiceprint recognition, which specifically comprises the following steps:

[0010] Step 1: non-contactly install a sound signal collection device on the transmission line to collect sound signals when different defects occur in the transmission line in real time;

[0011] Step 2, transmitting the sound signal to the sound signal processing module, processing the collected sound signal, extracting MFCC feature parameters and RCC feature parameters, and performing feature parameter fusion and dimensionality reduction;

[0012] Step 3: The fusion feature parameters after dimension reduction and the transmission line sound signal are transmitted to the Bayesian classifier in the background terminal for training until the accuracy of the recognition result of the Bayesian classifier is greater than the set threshold. The trained Bayesian classifier is used to identify transmission line defects on the transmission line sound signal collected in real time.

[0013] Preferably, in step 1, for the non-contact installation, the acoustic signal collection device is fixed at an arbitrary position within the range of 1 meter above and below the midpoint of the transmission line tower using a clip.

[0014] Preferably, in step 2, the processing of the transmission line sound signal includes pre-emphasis, framing, and applying a Hamming window.

[0015] Preferably, in step 2, the pre-emphasis of the audio of the transmission line defect is achieved through a first-order high-pass filter;

[0016] The pre-emphasized signal is framed and a Hamming window is applied, as shown in the following formula:

[0017]

[0018] S(n) = s(n) * h(n)

[0019] Where: s(n) is the transmission line sound signal before framing, N is the size of each frame after framing, h(n) is the Hamming window function, a is the overlap rate, S(n) is the transmission line sound signal after framing, and n is the nth discrete point of the transmission line sound signal;

[0020] The spectrum of each frame signal after preprocessing is shown in the following formula:

[0021]

[0022] Where: S(k) is the spectrum of the processed transmission line sound signal, k is the selected number of frames, and K is the maximum number of frames in the spectrum of the transmission line sound signal;

[0023] The power spectrum of the signal is shown in the following formula:

[0024]

[0025] Where: P(k) is the power spectrum of the signal, and N is the total number of discrete points of the transmission line sound signal;

[0026] The relationship between the Mel frequency, i.e., the Mel frequency and the linear frequency, is shown in the following formula:

[0027]

[0028] Where: Mel(f) is the Mel frequency of the signal, and f is the linear frequency of the signal;

[0029] The logarithmic energy of the Mel triangular filter bank is shown in the following formula:

[0030]

[0031] Where: S(m) is the logarithmic energy of the Mel triangular filter bank, H m (k) is the filter bank, m is the selected m-th filter, and M is the number of triangular filters;

[0032] The MFCC parameters of the sound signal of the preprocessed transmission line are shown by the following formula:

[0033]

[0034] Where: C(n) are the MFCC parameters of the sound signal of the preprocessed transmission line.

[0035] Extract the RCC feature parameters of the processed signal, as shown by the following formula:

[0036]

[0037] Where: R(f) are the RCC feature parameters.

[0038] Preferably, after obtaining the above two types of feature parameters, set the weight coefficient of the RCC feature parameters as α = {α1, α2,..., α s1} and the weight coefficient of the MFCC feature parameters as β = {β1, β2,..., β s2}. By fusing the two, the weight matrix η of the fused feature parameters can be obtained as η = {α1, α2,..., α s1 , β1, β2,..., β s2};

[0039] The dimensionality reduction processing of the fused feature parameters adopts the following formula:

[0040]

[0041] Among them, σ between represents the mean variance between different feature parameters; σ within represents the mean variance between different samples of the same feature parameter; r f is the weight value of different feature parameters after dimensionality reduction;

[0042]

[0043]

[0044] Where: the total number of samples is W, represents the mean of the u-th component in the i-th type of sample, m u represents the mean of the u-th component of all types of samples, n i represents the number of samples in the i-th type of sample, represents the u-th dimensional feature parameter of the i-th type of sample.

[0045] Preferably, in step 3, the Bayesian classifier is trained using the power transmission line sound signal and the dimension-reduced fusion feature parameters. The posterior probability and the prior probability of the Bayesian classifier are as shown in the following formula:

[0046]

[0047] In the formula: P(A d |Y) is the posterior probability, P(A d ) is the probability of a certain defect occurring, P(x|A d ) is the prior probability, and P(x) is the probability of a certain defect feature occurring;

[0048] The joint probability of the classifier is as shown in the following formula:

[0049]

[0050] In the formula: P(A d |x1,x2,...,x L ) is the joint probability, P(x i |A d ) is the prior probability of the i-th classification, and L is the number of types of power transmission line defects;

[0051] The sound signal and the dimension-reduced fusion feature parameters of the power transmission line collected in real time pass through the trained Bayesian classifier to obtain the joint probability of various types of power transmission line defects, and the discrimination and types of power transmission line defects are obtained based on the joint probability.

[0052] The second aspect of the present invention provides a power transmission line defect recognition system based on voiceprint recognition, which implements and runs the above-mentioned power transmission line defect recognition method based on voiceprint recognition, including:

[0053] A sound signal acquisition module, a sound signal processing module, and a background terminal;

[0054] The sound signal acquisition module and the sound signal processing module are arranged on-site. The sound signal acquisition module is installed on the transmission tower, and the sound signal emitted during the operation of the power transmission line is collected in real time through non-contact measurement, and the collected sound signal data is transmitted to the sound signal processing module;

[0055] The sound signal processing module processes the received sound signal data, extracts the feature parameters, and transmits the extracted feature parameters to the background terminal;

[0056] After receiving the sound signal feature parameters, the background terminal uses the Bayesian classifier to classify the data to obtain the power transmission line defect category result.

[0057] The third aspect of the present invention provides an electronic device for a method of identifying transmission line defects based on voiceprint recognition, which runs the method of predicting the movement trajectory of cold vortex affecting new energy output, including:

[0058] A processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete mutual communication through the communication bus;

[0059] The memory is used for storing computer programs;

[0060] The processor is used for executing the programs stored on the memory to implement the method of identifying transmission line defects based on voiceprint recognition.

[0061] The fourth aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the method of identifying transmission line defects based on voiceprint recognition.

[0062] The beneficial effects of the present invention are as follows. Compared with the prior art, the present invention provides a method and system for identifying transmission line defects based on voiceprint recognition. The adopted sound signal collection device supports non-contact installation, has a small device volume and is easy to use. At the same time, since the sound signal does not generate an electromagnetic field, it will not affect the normal operation of the transmission line and the devices therein; it will not be affected by the complex and changeable external environment and is not prone to failure; a sound signal processing system is provided to preprocess the collected sound signals, extract characteristic parameters, obtain the MFCC characteristic parameters of the transmission line sound signals, and send them to the background system; the present invention properly processes the collected sound signals of the transmission line, and then uses a Bayesian classifier to classify and identify more clearly and accurately. The trained Bayesian classifier in the background system can quickly classify the characteristic parameters, obtain the types of defects, and perform real-time early warning, alarm, etc., meeting the current efficiency requirements and alarm speed requirements for identifying transmission line defects, and can realize real-time monitoring of the state of the transmission line and its devices; avoiding the traditional manual inspection method not only reduces the risks of staff, but also improves the inspection efficiency of the transmission line and reduces the inspection cost of the transmission line. Description of the Drawings

[0063] Figure 1 It is a schematic diagram of a method for identifying transmission line defects based on voiceprint recognition of the present invention;

[0064] Figure 2 It is a flowchart of MFCC characteristic parameter extraction of the present invention;

[0065] Figure 3 It is a schematic diagram of a system for identifying transmission line defects based on voiceprint recognition of the present invention;

[0066] Figure 4 This is a practical layout relationship diagram of a transmission line defect recognition system based on voiceprint recognition according to the present invention. Specific implementation manners

[0067] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. The embodiments described in this application are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the spirit of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0068] The present invention provides a method for recognizing transmission line defects based on voiceprint recognition, as Figure 1 shown, including the following steps:

[0069] Step 1, non-contact installation of a sound signal collection device on the transmission line to collect sound signals in real time when different defects occur on the transmission line.

[0070] In a preferred but non-limiting implementation manner, in Step 1, the non-contact installation means fixing the sound signal collection device at any position within a range of 1 meter above and below the midpoint of the transmission line tower using a clip to achieve the collection of sound signals of various devices on the transmission line.

[0071] Step 2, transmitting the sound signal to a sound signal processing module, processing the collected sound signal, extracting MFCC feature parameters and RCC feature parameters, and performing fusion and dimensionality reduction.

[0072] In a preferred but non-limiting implementation manner, in Step 2, the sound signal is transmitted to the MCU module of the sound signal processing system, as Figure 2 shown, the processing includes pre-emphasis, framing, and adding a Hamming window;

[0073] Feature extraction is performed on the processed audio of the transmission line defect line, including performing a fast Fourier transform on the framed and windowed signal to obtain the frequency domain features of each frame, and taking the modulus square of the obtained frequency domain features to obtain the power spectrum, passing the power spectrum through a Mel filter, i.e., a Mel filter, and calculating its logarithmic energy, and then bringing the logarithmic energy into a discrete cosine transform to obtain MFCC feature parameters, i.e., Mel frequency cepstral coefficient feature parameters.

[0074] Furthermore, pre-emphasis of the transmission line defect audio is achieved through a first-order high-pass filter to amplify the high-frequency components of the input sound signal and compensate for the high-frequency components of the signal;

[0075] Frame the pre-emphasized signal, and multiply each frame by the Hamming window function to increase the continuity at both ends of the frame. The amplitude-frequency characteristic of the Hamming window is that the side lobe attenuation is relatively large, and the attenuation between the main lobe peak and the first side lobe peak can reach 40 dB, as shown in the following formula:

[0076]

[0077] S(n) = s(n) * h(n)

[0078] Where: s(n) is the power line sound signal before framing, N is the size of each frame after framing, h(n) is the Hamming window function, a is the overlap rate, S(n) is the power line sound signal after framing, and n is the nth discrete point of the power line sound signal;

[0079] Perform Fourier transform on each frame signal after preprocessing to obtain the spectrum of each frame, as shown in the following formula:

[0080]

[0081] Where: S(k) is the spectrum of the processed power line sound signal, k is the selected number of frames, and K is the maximum number of frames in the power line sound signal spectrum;

[0082] Take the modulus square of the spectrum of the signal to obtain the power spectrum of the signal, as shown in the following formula:

[0083]

[0084] Where: P(k) is the power spectrum of the signal, and N is the total number of discrete points of the power line sound signal;

[0085] In the Mel triangular filter, the span of each triangular filter corresponds to the Mel scale. The relationship between the Mel frequency, that is, the Mel frequency and the linear frequency, is shown in the following formula:

[0086]

[0087] Where: Mel(f) is the Mel frequency of the signal, and f is the linear frequency of the signal;

[0088] Multiply each filter bank by the power spectrum, add the coefficients and take the logarithm to obtain the logarithmic energy of the filter bank, as shown in the following formula:

[0089]

[0090] Where: S(m) is the logarithmic energy of the Mel triangular filter bank, H m (k) is the filter bank, m is the selected mth filter, and M is the number of triangular filters;

[0091] Finally, the obtained logarithmic energy is brought into the discrete cosine transform to calculate the MFCC parameters, as shown in the following formula:

[0092]

[0093] Where: C(n) is the MFCC parameter of the sound signal of the power transmission line after preprocessing.

[0094] Furthermore, in the voiceprint recognition technology, the MFCC feature parameters are easily affected by environmental noise, resulting in the loss of some high-frequency information and being unable to accurately represent the feature parameters. To solve this problem, the present invention provides an RCC feature parameter, which can concentrate the voice information in the high-frequency region, and a method of fusing the RCC feature parameter and the MFCC feature parameter to form a new feature parameter for classification; extracting the RCC feature parameter of the processed signal, as shown in the following formula:

[0095]

[0096] Where: R(f) is the RCC feature parameter.

[0097] Step 3: Transmit the dimension-reduced fusion feature parameter and the sound signal of the power transmission line to the Bayesian classifier of the background terminal for training until the correct rate of the recognition result of the Bayesian classifier is greater than the set threshold, and use the trained Bayesian classifier to identify the defects of the power transmission line for the sound signal of the power transmission line collected in real time.

[0098] After obtaining the above two feature parameters, set the coefficient of the RCC feature parameter as α = {α1, α2,..., α s1}, the coefficient of the MFCC feature parameter as β = {β1, β2,..., β s2}, and the two can be fused to obtain the fusion feature parameter η = {α1, α2,..., α s1 , β1, β2,..., β s2};

[0099] The dimension of the fused feature parameter increases greatly, which will increase the difficulty and efficiency of subsequent classification training. To eliminate redundant information, it is necessary to perform dimensionality reduction on it. The dimensionality reduction process uses the following formula:

[0100]

[0101] Among them, σ between represents the mean variance between different feature parameters; σ within represents the mean variance between different samples of the same feature parameter; r f is the weight value of different feature parameters after dimensionality reduction; the weight value represents the influence of the measured feature on the sound signal of the power transmission line.

[0102]

[0103]

[0104] Where: the total number of samples is W, represents the mean of the u-th dimensional component in the i-th type of samples, m u represents the mean of the u-th dimensional component of all types of samples, n i represents the number of samples in the i-th type of samples, represents the u-th dimensional feature parameter of the i-th type of samples.;

[0105] According to the above method, the weight values of all dimensions of the fused feature parameters are calculated, and the contribution degrees are compared. Finally, the dimensions with the largest set proportion of contribution degrees are left as the fused feature parameters after dimensionality reduction.

[0106] In a preferred but non-limiting implementation manner, in step 3, the above-mentioned collected and processed sound signals and the fused feature parameters after dimensionality reduction are used to train the Bayesian classifier. In order to reduce the computational complexity, the Bayesian classifier needs to calculate the posterior probability additionally, which is also the biggest difference between it and ordinary classifiers. It uses the original probability of the hypothesis that does not include redundant information to determine the posterior probability, that is, the probability of correcting an event after obtaining additional information, as shown in the following formula:

[0107]

[0108] Where: P(A d |Y) is the posterior probability, P(A d ) is the probability of the occurrence of a certain defect, P(x|A d ) is the prior probability, and P(x) is the probability of the occurrence of a certain defect feature;

[0109] Its joint probability is shown in the following formula:

[0110]

[0111] Where: P(A d |x1,x2,...,x L ) is the joint probability, P(x i |A d ) is the prior probability of the i-th type of classification, and L is the number of types of transmission line defects;

[0112] The sound signals of the transmission line collected in real time and the fused feature parameters after dimensionality reduction pass through the trained Bayesian classifier, and the joint probabilities of various types of transmission line defects are obtained. Based on the joint probabilities, the discrimination and types of transmission line defects are obtained;

[0113] The category of transmission line defects is represented by d in the formula. The categories include insulator defects, clamp defects, arrester defects, etc. If the machine is to automatically identify various defects, it is necessary to record the sound signals of different types of defects and train them with the processed data set. Among them, different sound signals correspond to different transmission line defects. The defect data set can be divided into multiple categories, such as insulator defects, clamp defects, arrester defects, etc. Finally, after each data set is trained, a classifier can be formed, and an algorithm can be formed based on this. The advantage of the Bayesian classifier is that it can achieve correct classification with fewer training sets.

[0114] The present invention provides a transmission line defect identification system based on voiceprint recognition, and implements the operation of a transmission line defect identification method based on voiceprint recognition, such as Figure 3 、 4 shown, including:

[0115] A sound signal acquisition module, a sound signal processing module, and a background terminal;

[0116] The sound signal acquisition module and the sound signal processing module are arranged on-site. The sound signal acquisition module is installed on the transmission tower, and the sound signals emitted during the operation of the transmission line are collected in real time through non-contact measurement, and the collected sound signal data is transmitted to the sound signal processing module;

[0117] The sound signal processing module processes the received sound signal data, extracts feature parameters, and transmits the extracted feature parameters to the background terminal;

[0118] After the background terminal receives the sound signal feature parameters, it uses a Bayesian classifier to classify the data to obtain the result of the transmission line defect category.

[0119] The present invention provides an electronic device for a transmission line defect identification method based on voiceprint recognition, and runs a cold vortex movement trajectory prediction method that affects new energy output, including:

[0120] A processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus;

[0121] The memory is used to store computer programs;

[0122] The processor is used to execute the programs stored on the memory to implement the transmission line defect identification method based on voiceprint recognition.

[0123] The present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the described method for identifying transmission line defects based on voiceprint recognition is implemented.

[0124] This disclosure can be a system and / or a method.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific implementation manners of the present invention, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. A method for identifying transmission line defects based on voiceprint recognition, characterized in that, The method includes the following steps: Step 1, non - contact install a sound signal collection device on the transmission line to collect sound signals in real - time when different defects occur on the transmission line; Step 2, transmit the sound signal to the sound signal processing module. After processing the collected sound signal, extract MFCC feature parameters and RCC feature parameters, and perform feature parameter fusion and dimensionality reduction; Step 3, train the reduced - dimension fused feature parameters and the Bayesian classifier in the transmission line sound signal transmission background terminal until the correct rate of the recognition result of the Bayesian classifier is greater than the set threshold, and use the trained Bayesian classifier to identify transmission line defects for the sound signals of the transmission line collected in real - time.

2. A method for identifying transmission line defects based on voiceprint recognition according to claim 1, characterized in that: In step 1, the non - contact installation means fixing the sound signal collection device at any position within the range of 1 meter above and below the mid - point of the transmission line tower using a clip.

3. A method for identifying transmission line defects based on voiceprint recognition according to claim 1, characterized in that: In step 2, the processing of the transmission line sound signal includes pre - emphasis, framing, and applying a Hamming window.

4. A method for identifying transmission line defects based on voiceprint recognition according to claim 3, characterized in that: In step 2, the pre - emphasis of the transmission line defect audio is achieved through a first - order high - pass filter; The pre - emphasized signal is framed and a Hamming window is applied, as shown in the following formula: S(n) = s(n) * h(n) Where: s(n) is the transmission line sound signal before framing, N is the size of each frame after framing, h(n) is the Hamming window function, a is the overlap rate, S(n) is the transmission line sound signal after framing, and n is the nth discrete point of the transmission line sound signal; The spectrum of each frame signal after pre - processing is as shown in the following formula: Where: S(k) is the spectrum of the processed transmission line sound signal, k is the selected number of frames, and K is the maximum number of frames in the transmission line sound signal spectrum; The power spectrum of the signal is as shown in the following formula: Where: P(k) is the power spectrum of the signal, and N is the total number of discrete points of the transmission line sound signal; The relationship between Mel frequency (i.e., Mel frequency) and linear frequency is as shown in the following formula: Where: Mel(f) is the Mel frequency of the signal, and f is the linear frequency of the signal; The logarithmic energy of the Mel - scale filter bank is as shown in the following formula: Where: S(m) is the logarithmic energy of the Mel triangular filter bank, H m (k) is the filter bank, m is the selected m-th filter, and M is the number of triangular filters.

5. A method for identifying transmission line defects based on voiceprint recognition according to claim 4, characterized in that: The MFCC parameters of the pre - processed transmission line sound signal are as shown in the following formula: Where: C(n) are the MFCC parameters of the pre - processed transmission line sound signal.

6. A method for identifying transmission line defects based on voiceprint recognition according to claim 1 or 5, characterized in that: Extract the RCC feature parameters of the processed signal, as shown in the following formula: Where: R(f) are the RCC feature parameters.

7. The method for identifying transmission line defects based on voiceprint recognition according to claim 1, characterized in that: Set the weight coefficient of the RCC feature parameters as α = {α1, α2,..., α s1}, and the weight coefficient of the MFCC feature parameters as β = {β1, β2,..., β s2}. By fusing the two, the weight matrix of the fused feature parameters η = {α1, α2,..., α s1 , β1, β2,..., β s2} can be obtained; The dimensionality reduction processing of the fused feature parameters adopts the following formula: Among them, σ between represents the mean variance between different characteristic parameters; σ within represents the mean variance between different samples of the same characteristic parameter; r f is the weight value of different characteristic parameters after dimensionality reduction; Where: the total number of samples is W, represents the mean value of the u-th dimensional component in the i-th class of samples, m u represents the mean value of the u-th dimensional component of all class samples, n i represents the number of samples in the i-th class of samples, represents the u-th dimensional feature parameter of the i-th class of samples.

8. The method for identifying transmission line defects based on voiceprint recognition according to claim 1, characterized in that: In step 3, the Bayesian classifier is trained using the transmission line sound signal and the dimensionality-reduced fused feature parameters. The posterior probability and prior probability of the Bayesian classifier are shown in the following formula: Where: P(A d |Y) is the posterior probability, P(A d ) is the probability of a certain defect occurring, P(x|A d ) is the prior probability, and P(x) is the probability of a certain defect feature occurring; The joint probability of the classifier is shown in the following formula: Where: P(A d |x1,x2,...,x L ) is the joint probability, P(x i |A d ) is the prior probability of the i-th classification, and L is the number of types of transmission line defects; The sound signal and the dimensionality-reduced fused feature parameters of the transmission line collected in real time pass through the trained Bayesian classifier to obtain the joint probability of various types of transmission line defects, and the discrimination and type of transmission line defects are obtained based on the joint probability.

9. A power transmission line defect recognition system based on voiceprint recognition for implementing the method according to any one of claims 1-8, comprising: A sound signal acquisition module, a sound signal processing module, and a background terminal; characterized in that: The sound signal acquisition module and the sound signal processing module are arranged on-site. The sound signal acquisition module is installed on the transmission tower, and the sound signals emitted during the operation of the transmission line are collected in real time through non-contact measurement, and the collected sound signal data is transmitted to the sound signal processing module; The sound signal processing module processes the received sound signal data, extracts the feature parameters, and transmits the extracted feature parameters to the background terminal; After receiving the sound signal feature parameters, the background terminal uses the Bayesian classifier to classify the data to obtain the result of the transmission line defect category.

10. An electronic device for the method for identifying transmission line defects based on voiceprint recognition, running the method for identifying transmission line defects based on voiceprint recognition according to any one of claims 1-8, characterized in that: It includes a processor, a communication interface, a memory, and a communication bus. Among them, the processor, the communication interface, and the memory complete mutual communication through the communication bus; The memory is used to store computer programs; The processor is used to execute the programs stored on the memory to implement the method for identifying transmission line defects based on voiceprint recognition.

11. A computer-readable storage medium, on which a computer program is stored, characterized in that: When the program is executed by a processor, it implements the method for identifying transmission line defects based on voiceprint recognition according to any one of claims 1-8.

Images