Insulator spontaneous explosion fault detection method and system based on sound data

Abstract

The invention relates to the technical field of power system monitoring and fault diagnosis, and discloses an insulator self-explosion fault detection method and system based on sound data, and the method comprises the steps: collecting the sound data of a transmission tower insulator during normal operation and self-explosion through a sound sensor; the data are sampled, signal processing is carried out by applying a sound preprocessing technology and a noise reduction algorithm, and the data are converted into a Mel spectrogram feature image through a Mel coefficient extraction process. And the BP neural network is used to identify and classify the generated Mel image, and real-time detection of the operation state of the insulator is realized, so that the self-explosion fault of the insulator is effectively identified. According to the method, the accuracy of the insulator self-explosion sound data and the rapidity of the BP neural network in the aspect of image recognition are integrated, and the detection effect on the insulator self-explosion fault is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring and fault diagnosis, and in particular to a method and system for detecting insulator self-explosion faults based on sound data. Background Art

[0002] In recent years, my country's power industry has shown a booming development trend. With the continuous growth of the national economy, the demand for electricity has steadily increased, prompting the power industry to usher in a comprehensive upgrade and expansion. The scale of the national power grid has further expanded. By 2024, the total length of 110 kV and above transmission lines has exceeded 1.5 million kilometers, and the coverage of the power grid has further expanded to remote areas. However, the rapid development of power construction has also brought a series of problems, especially in the operation of transmission lines. In particular, the external operating environment of the power grid has caused damage and shedding of key components of the transmission line, such as self-explosion of insulators, damage to insulators, and breakage of wires on the guy tower. These faults have seriously affected the normal operation of the transmission lines and caused huge losses to the social economy.

[0003] With the increase of transmission lines, the demand for transmission capacity and voltage level is also growing rapidly, so ensuring transmission safety is a key prerequisite for meeting the demand for power supply. However, insulators in transmission lines are easily damaged by various external factors, such as thunderstorms, strong winds, etc. These factors cause the aging of various devices on the line and may cause insulator failures, such as: Insulator contamination and pollution: Since the surface of the insulator is exposed to the air for a long time, it is easy to accumulate dust, salt, dirt, etc. These pollutants will reduce the surface resistance of the insulator, resulting in a decrease in insulation performance and increasing the risk of flashover and breakdown on the surface of the insulator; Insulator flashover and breakdown: In humid weather or extreme environmental conditions, pollutants on the surface of the insulator will promote the occurrence of flashover discharge and may even cause the breakdown of the insulator, thereby affecting the stable operation of the power grid; Insulator aging and deterioration: Insulator materials will age over time, causing their mechanical strength and insulation performance to decrease, thereby affecting their stability and reliability during power transmission; Insulator mechanical damage: For example, storms or other natural disasters may cause insulators to suffer mechanical vibration or direct damage, which may cause the insulator to rupture or break, requiring timely replacement or repair; Insulator electric shock: Natural electric shocks such as lightning may directly cause damage or breakdown of insulators, causing serious impacts on transmission lines and power facilities.

[0004] Therefore, as a vital electrical insulation and mechanical support device in the transmission line, it is very important to regularly inspect and maintain insulators. At present, insulator inspection is a vital part of power system maintenance. Its main purpose is to ensure that insulators can effectively prevent electrical equipment and circuits from being affected by moisture, pollution or other damaging factors during operation.

[0005] Traditional inspection methods require inspectors to carry handheld instruments and climb pole towers. Facing harsh geological environments, this not only affects work efficiency, but also poses a threat to the health of inspectors. In addition, manual inspections are labor-intensive, and the quality of inspections is easily affected by subjective factors. Especially in harsh environments, the scope of inspections is severely limited. In recent years, with the development of computer vision and image processing technology, visual inspection is moving towards intelligence and automation. High-resolution cameras combined with drones can automatically identify defects on the surface of insulators, such as using image segmentation and pattern recognition technology to detect cracks, dirt or foreign objects. Power grid companies have gradually introduced image detection to replace traditional manual inspection methods for power facility inspections. However, for small parts of transmission line insulators, it is difficult for drones to collect images and the recognition accuracy is not high. Therefore, the method of deploying sound sensors is used to accurately identify insulator self-explosion failures. Summary of the invention

[0006] In view of the above existing problems, the present invention is proposed.

[0007] Therefore, the present invention provides an insulator self-explosion fault detection method based on sound data, which can collect the sound signal of the insulator during operation in real time through the sound sensor, continuously monitor the state of the insulator, and detect abnormalities in time.

[0008] In order to solve the above technical problems, the present invention provides the following technical solutions: an insulator self-explosion fault detection method based on sound data, comprising: collecting sound signals of insulators during operation through sound sensors, preprocessing and endpoint detection of the collected insulator sound signals; converting them into Mel spectrum feature images through a Mel coefficient extraction process; and using a BP neural network to identify and classify the generated Mel images to detect the operating status of the insulators in real time.

[0009] As a preferred solution of the insulator self-explosion fault detection method based on sound data described in the present invention, wherein: the collecting of the sound signal of the insulator during operation includes collecting the normal operation sound of the insulator at the transmission line tower and the abnormal sound signal of the self-explosion respectively through the sound sensor, and the sampling frequency needs to be greater than or equal to twice the highest frequency of the insulator sound signal, so that the sampling value contains all the information of the original sound signal, and the sampled sound signal is restored to the original sound signal without distortion. The specific expression is as follows:

[0010] F s ≥2f

[0011] Among them, F s is the sampling frequency, and f is the sampling frequency of the insulator sound signal.

[0012] As a preferred solution of the insulator self-explosion fault detection method based on sound data described in the present invention, wherein: the preprocessing includes: after the sensor sound recognition device collects the sound data, the original sound signal contains environmental noise and silent segments, and the sound signal noise reduction algorithm and the sound signal endpoint detection algorithm are used to analyze and process the sound signal;

[0013] By decomposing the spectrum of the original sound signal and removing the part below the threshold according to the preset amplitude threshold, the influence of background noise is weakened or eliminated. The processed amplitude spectrum is combined with the original phase spectrum, and then inversely transformed to finally obtain a clear time domain signal;

[0014] The signal is divided into frames and then discrete Fourier transformed to obtain the amplitude:

[0015]

[0016]

[0017] Among them, X f (k) represents the amplitude of the i-th frame component, x i (n) indicates the i-th frame signal obtained by frame processing, n indicates an independent variable, j indicates an imaginary number, k indicates a signal, k=0, 1, ..., N-1, N indicates the length of each frame, XA i (k) represents x i (n) phase spectrum;

[0018] The environmental noise signal is framed, and the length of each frame is the same as the frame length N of the insulator sound signal. The discrete Fourier transform is also performed, and the result is Z f (k), assuming that the noise is divided into M frames, the average amplitude of the entire environmental noise is:

[0019]

[0020] Subtracting the amplitude of the noise from the amplitude of the insulator sound signal is the spectrum subtraction principle, and the amplitude spectrum XP f (k) is:

[0021]

[0022] The amplitude spectrum XP obtained by spectral subtraction f (k) and the phase spectrum XA of the original signal f(n) i (k), the inverse Fourier transform is the signal Xp after spectrum subtraction i (n).

[0023] As a preferred solution of the insulator self-explosion fault detection method based on sound data described in the present invention, wherein: the endpoint detection includes that when adjacent sampling points have different algebraic signs, they are called zero crossing points, the number of zero crossing points is called zero crossing rate, and the zero crossing rate feature extraction is to discriminate the signal amplitude in the time domain, and obtain an effective signal feature vector by detecting the change law of the signal zero crossing point;

[0024] The collected transmission tower insulator sound signal has the characteristics of short-term stability. The short-time average value zero-crossing rate is used to represent the sound characteristics. The calculation expression of the short-time average value zero-crossing rate is as follows:

[0025]

[0026] Among them, x n Indicates the sound signal of the insulator during operation; sgn[x n (n)] is a symbolic function, expressed as follows:

[0027]

[0028] Add the window function to the short-time average zero-crossing rate calculation, that is:

[0029]

[0030] The window function expression is as follows:

[0031]

[0032] Among them, N 1 Represents the frame length of the abnormal target sound signal.

[0033] As a preferred solution of the insulator self-explosion fault detection method based on sound data described in the present invention, the Mel coefficient extraction process includes using a spectrogram to capture the frequency and energy variation characteristics. The insulator sound signal is a non-stationary time-varying signal. After a short-term analysis and processing of the signal through a frame splitting and windowing operation, it is regarded as a stationary signal, and then the insulator audio containing rich acoustic features is extracted and utilized;

[0034] The spectrogram combines time domain information and frequency domain information and presents it in the form of image modeling. It includes the main acoustic characteristic parameters of frequency, amplitude, and energy. The Mel frequency cepstral coefficient feature collects high-frequency information using the Mel filter by simulating the human hearing mechanism. The audio signal with a frequency unit of Hz is converted to a signal with a Mel unit. The conversion formula is as follows:

[0035]

[0036] Where, f is the Hertz scale;

[0037] Use the Mel filter to filter each frame of signal, and use the output formula of the Mel filter bank:

[0038]

[0039] Where x(t) is the number after FFT transformation, m is the number of the Mel filter, and H is the Mel filter bank. m (k) is a series of triangular filters of equal height:

[0040]

[0041] Where, f(m) is the center frequency;

[0042] The insulator sound signal is preprocessed, including noise reduction, framing and windowing. After applying the window function to each frame, it is converted into a frequency domain signal through FFT and mapped to the Mel frequency scale using the Mel filter bank to generate the final Mel spectrogram.

[0043] As a preferred solution of the insulator self-explosion fault detection method based on sound data described in the present invention, wherein: the identification and classification of the generated Mel images using the BP neural network includes preparing Mel spectra of normal operation and self-explosion of insulators as training data sets, and the data covers various types of insulator operating conditions, including images of normal operation of insulators and images of self-explosion faults of insulators;

[0044] The color histogram is used to describe the color features of the image, the gray-level co-occurrence matrix is ​​used to extract the texture features of the image, and the Canny edge detection algorithm is used to calculate the rate of change of pixel values ​​in the image to detect boundary pixels.

[0045] As a preferred solution of the insulator self-explosion fault detection method based on sound data described in the present invention, wherein: the use of the BP neural network to identify and classify the generated Mel image also includes building a back propagation neural network model for image recognition, the input layer receives the feature vector as input, the hidden layer includes multiple neurons, and the output layer represents the classification result;

[0046] The prepared training data set is input into the neural network for training. The training process uses the back propagation algorithm to update the weights and biases in the network to minimize the error between the predicted results and the actual labels. The LM algorithm is used:

[0047] x s+1 =x s -(H+αI) -1 G

[0048] Among them, H is the matrix of multidimensional vectors, G is the first-order gradient of multidimensional vectors, xs is the function extreme value of the previous step, α is the step size, and I is the identity matrix;

[0049] Use the test data set to evaluate the performance of the trained neural network model. Input the test data into the network for prediction, compare it with the actual label, and calculate the accuracy and recall evaluation indicators to measure the performance of the model.

[0050] The audio data collected by the sound sensor is transformed by Mel coefficient extraction, and the Mel spectrum obtained after transformation is input into the trained neural network model for real-time target recognition. The prediction results are obtained by extracting features from the image and forward propagation of the neural network to realize the identification of insulator self-explosion faults.

[0051] Another object of the present invention is to provide an insulator self-explosion fault detection system based on sound data, which can effectively improve the automation and intelligent management level of the power system and is of great significance to ensuring the safe and stable operation of the power grid.

[0052] As a preferred solution of the insulator self-explosion fault detection system based on sound data described in the present invention, it includes: a sound collection module, a preprocessing module, a feature extraction module, a BP neural network training module, and a fault detection and identification module;

[0053] The sound collection module collects the sound signal of the insulator during operation through the sound sensor;

[0054] The preprocessing module performs noise reduction and endpoint detection processing on the collected original sound signal;

[0055] The feature extraction module extracts features from the preprocessed sound signal and converts it into a Mel-spectrogram feature image;

[0056] The BP neural network training module uses the BP neural network to train the extracted features to identify and classify the operating status of the insulator;

[0057] The fault detection and identification module extracts and transforms the real-time audio data collected by the sound sensor through Mel coefficients, and inputs the transformed Mel spectrogram into a trained neural network model to perform real-time target recognition.

[0058] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, steps of a method for detecting insulator self-explosion faults based on sound data are implemented.

[0059] A computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the computer program implements the steps of an insulator self-explosion fault detection method based on sound data.

[0060] Beneficial effects of the present invention: By combining the advantages of sound and image, this method can more accurately judge the insulator self-explosion fault, avoid the misjudgment and missed judgment existing in the traditional method, and improve the accuracy of detection. The transmission line insulator self-explosion fault detection method can quickly identify the insulator self-explosion fault and promptly issue an alarm to relevant staff, significantly improving the safety of transmission line operation. This method takes timely measures when a fault occurs, effectively ensuring the safety of power grid facilities and operators. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0062] Figure 1 A schematic flow chart of a method for detecting insulator self-explosion faults based on sound data provided by an embodiment of the present invention.

[0063] Figure 2 A flowchart of an image recognition process of an insulator self-explosion fault detection method based on sound data provided by an embodiment of the present invention.

[0064] Figure 3 A schematic diagram of working modules of an insulator self-explosion fault detection system based on sound data provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0065] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0066] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0067] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0068] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0069] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0070] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0071] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides an insulator self-explosion fault detection method based on sound data, comprising:

[0072] S1: The sound sensor is used to collect the sound signal of the insulator during operation, and the collected sound signal of the insulator is preprocessed and the endpoint detection is performed.

[0073] Furthermore, the collecting of the sound signal of the insulator during operation includes collecting the normal operation sound of the insulator at the transmission line tower and the abnormal sound signal of self-explosion respectively through the sound sensor, and the sampling frequency needs to be greater than or equal to twice the highest frequency of the insulator sound signal, so that the sampling value contains all the information of the original sound signal, and the sampled sound signal is restored to the original sound signal without distortion. The specific expression is as follows:

[0074] F s ≥2f

[0075] Among them, F s is the sampling frequency, and f is the sampling frequency of the insulator sound signal.

[0076] It should be noted that the pre-processing includes: after the sensor sound recognition device collects the sound data, the original sound signal contains environmental noise and silent segments, and the sound signal noise reduction algorithm and the sound signal endpoint detection algorithm are used to analyze and process the sound signal;

[0077] By decomposing the spectrum of the original sound signal and removing the part below the threshold according to the preset amplitude threshold, the influence of background noise is weakened or eliminated. The processed amplitude spectrum is combined with the original phase spectrum, and then inversely transformed to finally obtain a clear time domain signal;

[0078] The signal is divided into frames and then discrete Fourier transformed to obtain the amplitude:

[0079]

[0080]

[0081] Among them, X f (k) represents the amplitude of the i-th frame component, x i (n) indicates the i-th frame signal obtained by frame processing, n indicates an independent variable, j indicates an imaginary number, k indicates a signal, k=0, 1, ..., N-1, N indicates the length of each frame, XA i (k) represents x i (n) phase spectrum;

[0082] The environmental noise signal is framed, and the length of each frame is the same as the frame length N of the insulator sound signal. The discrete Fourier transform is also performed, and the result is Z i (k), assuming that the noise is divided into M frames, the average amplitude of the entire environmental noise is:

[0083]

[0084] Subtracting the amplitude of the noise from the amplitude of the insulator sound signal is the spectrum subtraction principle, and the amplitude spectrum XP f (k) is:

[0085]

[0086] The amplitude spectrum XP obtained by spectral subtraction i (k) and the phase spectrum XA of the original signal f(n) i (k), the inverse Fourier transform is the signal Xp after spectrum subtraction i (n).

[0087] It should also be noted that the endpoint detection includes that when adjacent sampling points have different algebraic signs, they are called zero crossing points, the number of zero crossing points is called the zero crossing rate, and the zero crossing rate feature extraction is to discriminate the signal amplitude in the time domain, and obtain an effective signal feature vector by detecting the change law of the signal zero crossing point;

[0088] The collected transmission tower insulator sound signal has the characteristics of short-term stability. The short-time average value zero-crossing rate is used to represent the sound characteristics. The calculation expression of the short-time average value zero-crossing rate is as follows:

[0089]

[0090] Among them, x n Indicates the sound signal of the insulator during operation; sgn[x n (n)] is a symbolic function, expressed as follows:

[0091]

[0092] Add the window function to the short-time average zero-crossing rate calculation, that is:

[0093]

[0094] The window function expression is as follows:

[0095]

[0096] Among them, N 1 Represents the frame length of the abnormal target sound signal.

[0097] S2: Converted into Mel-spectrogram feature image through Mel coefficient extraction process.

[0098] Furthermore, the Mel coefficient extraction process includes using a spectrogram to capture the changing characteristics of frequency and energy. The insulator sound signal is a non-stationary time-varying signal. After a short-term analysis and processing of the signal through a frame-by-frame windowing operation, it is regarded as a stationary signal, and then the insulator audio containing rich acoustic features is extracted and utilized;

[0099] The spectrogram combines time domain information and frequency domain information and presents it in the form of image modeling. It includes the main acoustic characteristic parameters of frequency, amplitude, and energy. The Mel frequency cepstral coefficient feature collects high-frequency information using the Mel filter by simulating the human hearing mechanism. The audio signal with a frequency unit of Hz is converted to a signal with a Mel unit. The conversion formula is as follows:

[0100]

[0101] Where, f is the Hertz scale;

[0102] Use the Mel filter to filter each frame of signal, and use the output formula of the Mel filter bank:

[0103]

[0104] Where x(t) is the number after FFT transformation, m is the number of the Mel filter, and H is the Mel filter bank. m (k) is a series of triangular filters of equal height:

[0105]

[0106] Where, f(m) is the center frequency;

[0107] The insulator sound signal is preprocessed, including noise reduction, framing and windowing. After applying the window function to each frame, it is converted into a frequency domain signal through FFT and mapped to the Mel frequency scale using the Mel filter bank to generate the final Mel spectrogram.

[0108] S3: Use BP neural network to identify and classify the generated Mel images and detect the operating status of insulators in real time.

[0109] Furthermore, the use of the BP neural network to identify and classify the generated Mel images includes preparing Mel spectra of normal operation and self-explosion of insulators as training data sets, and the data covers various types of insulator operating conditions, including images of normal operation of insulators and images of self-explosion faults of insulators;

[0110] The color histogram is used to describe the color features of the image, the gray-level co-occurrence matrix is ​​used to extract the texture features of the image, and the Canny edge detection algorithm is used to calculate the rate of change of pixel values ​​in the image to detect boundary pixels.

[0111] It should be noted that the use of the BP neural network to identify and classify the generated Mel images also includes constructing a back propagation neural network model for image recognition, wherein the input layer receives a feature vector as input, the hidden layer includes a plurality of neurons, and the output layer represents the classification result;

[0112] The prepared training data set is input into the neural network for training. The training process uses the back propagation algorithm to update the weights and biases in the network to minimize the error between the predicted results and the actual labels. The LM algorithm is used:

[0113] x s+1 =x s -(H+αI) -1 G

[0114] Among them, H is the matrix of multidimensional vectors, G is the first-order gradient of multidimensional vectors, x sis the function extreme value of the previous step, α is the step size, and I is the identity matrix;

[0115] Use the test data set to evaluate the performance of the trained neural network model. Input the test data into the network for prediction, compare it with the actual label, and calculate the accuracy and recall evaluation indicators to measure the performance of the model.

[0116] The audio data collected by the sound sensor is transformed by Mel coefficient extraction, and the Mel spectrum obtained after transformation is input into the trained neural network model for real-time target recognition. The prediction results are obtained by extracting features from the image and forward propagation of the neural network to realize the identification of insulator self-explosion faults.

[0117] Embodiment 2, the second embodiment of the present invention, is different from the previous embodiment in that:

[0118] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0119] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0120] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0121] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0122] Example 3, reference Figure 2 , which is an embodiment of the present invention, provides an insulator self-explosion fault detection system based on sound data, characterized by: including a sound collection module, a preprocessing module, a feature extraction module, a BP neural network training module, and a fault detection and identification module.

[0123] The sound collection module collects the sound signals of the insulator during operation through the sound sensor.

[0124] The preprocessing module performs noise reduction and endpoint detection on the collected original sound signal.

[0125] The feature extraction module extracts features from the preprocessed sound signal and converts it into a Mel-spectrogram feature image.

[0126] The BP neural network training module uses the BP neural network to train the extracted features to identify and classify the operating status of the insulator.

[0127] The fault detection and identification module extracts and transforms the real-time audio data collected by the sound sensor through the Mel coefficient, and inputs the transformed Mel spectrogram into the trained neural network model for real-time target recognition.

[0128] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for detecting insulator self-explosion faults based on sound data, characterized in that: include, The sound sensor is used to collect the sound signal of the insulator during operation, and the collected sound signal of the insulator is preprocessed and the endpoint detection is performed; Through the Mel coefficient extraction process, it is converted into a Mel spectrogram feature image; The BP neural network is used to identify and classify the generated Mel images to detect the operating status of the insulator in real time.

2. The method for detecting insulator self-explosion fault based on sound data according to claim 1, characterized in that: The collecting of the sound signal of the insulator during operation includes collecting the normal operation sound of the insulator at the transmission line tower and the abnormal sound signal of self-explosion respectively through the sound sensor, and the sampling frequency needs to be greater than or equal to twice the highest frequency of the insulator sound signal, so that the sampling value contains all the information of the original sound signal, and the sampled sound signal is restored to the original sound signal without distortion. The specific expression is as follows: F s ≥2f Among them, F s is the sampling frequency, and f is the sampling frequency of the insulator sound signal.

3. The insulator self-explosion fault detection method based on sound data according to claim 2, characterized in that: The preprocessing includes: after the sensor sound recognition device collects the sound data, the original sound signal contains environmental noise and silent segments, and the sound signal noise reduction algorithm and the sound signal endpoint detection algorithm are used to analyze and process the sound signal; By decomposing the spectrum of the original sound signal and removing the part below the threshold according to the preset amplitude threshold, the influence of background noise is weakened or eliminated. The processed amplitude spectrum is combined with the original phase spectrum, and then inversely transformed to finally obtain a clear time domain signal; The signal is divided into frames and then discrete Fourier transformed to obtain the amplitude: Among them, X i (k) represents the amplitude of the i-th frame component, x i (n) indicates the i-th frame signal obtained by frame processing, n indicates an independent variable, j indicates an imaginary number, k indicates a signal, k=0, 1, ..., N-1, N indicates the length of each frame, XA i (k) represents x i (n) phase spectrum; The environmental noise signal is framed, and the length of each frame is the same as the frame length N of the insulator sound signal. The discrete Fourier transform is also performed, and the result is Z i (k), assuming that the noise is divided into M frames, the average amplitude of the entire environmental noise is: Subtracting the amplitude of the noise from the amplitude of the insulator sound signal is the spectrum subtraction principle, and the amplitude spectrum XP i (k) is: The amplitude spectrum XP obtained by spectral subtraction i (k) and the phase spectrum XA of the original signal f(n) i (k), the inverse Fourier transform is the signal Xp after spectrum subtraction i (n).

4. The method for detecting insulator self-explosion fault based on sound data according to claim 3, characterized in that: The endpoint detection includes: when adjacent sampling points have different algebraic signs, they are called zero crossing points, the number of zero crossing points is called zero crossing rate, and zero crossing rate feature extraction is to discriminate the signal amplitude in the time domain, and obtain an effective signal feature vector by detecting the change law of the signal zero crossing point; The collected transmission tower insulator sound signal has the characteristics of short-term stability. The short-time average value zero-crossing rate is used to represent the sound characteristics. The calculation expression of the short-time average value zero-crossing rate is as follows: Among them, x n Indicates the sound signal of the insulator during operation; sgn[x n (n)] is a symbolic function, expressed as follows: Add the window function to the short-time average zero-crossing rate calculation, that is: The window function expression is as follows: Wherein, N1 represents the frame length of the abnormal target sound signal.

5. The method for detecting insulator self-explosion fault based on sound data according to claim 4, characterized in that: The Mel coefficient extraction process includes using a spectrogram to capture the changing characteristics of frequency and energy. The insulator sound signal is a non-stationary time-varying signal. After a short-term analysis and processing of the signal through a frame splitting and windowing operation, it is regarded as a stationary signal, and then the insulator audio containing rich acoustic features is extracted and utilized; The spectrogram combines time domain information and frequency domain information and presents it in the form of image modeling. It includes the main acoustic characteristic parameters of frequency, amplitude, and energy. The Mel frequency cepstral coefficient feature collects high-frequency information using the Mel filter by simulating the human hearing mechanism. The audio signal with a frequency unit of Hz is converted to a signal with a Mel unit. The conversion formula is as follows: Where, f is the Hertz scale; Use the Mel filter to filter each frame of signal, and use the output formula of the Mel filter bank: Where x(t) is the number after FFT transformation, m is the number of the Mel filter, and H is the Mel filter bank. m (k) is a series of triangular filters of equal height: Where, f(m) is the center frequency; The insulator sound signal is preprocessed, including noise reduction, framing and windowing. After applying the window function to each frame, it is converted into a frequency domain signal through FFT and mapped to the Mel frequency scale using the Mel filter bank to generate the final Mel spectrogram.

6. The method for detecting insulator self-explosion fault based on sound data according to claim 5, characterized in that: The method of using the BP neural network to identify and classify the generated Mel images includes preparing Mel spectra of normal operation and self-explosion of insulators as training data sets, wherein the data covers various types of insulator operating conditions, including images of normal operation of insulators and images of self-explosion faults of insulators; The color histogram is used to describe the color features of the image, the gray-level co-occurrence matrix is ​​used to extract the texture features of the image, and the Canny edge detection algorithm is used to calculate the rate of change of pixel values ​​in the image to detect boundary pixels.

7. The method for detecting insulator self-explosion fault based on sound data according to claim 6, characterized in that: The method of using the BP neural network to identify and classify the generated Mel images also includes constructing a back propagation neural network model for image recognition, wherein the input layer receives a feature vector as input, the hidden layer includes a plurality of neurons, and the output layer represents the classification result; The prepared training data set is input into the neural network for training. The training process uses the back propagation algorithm to update the weights and biases in the network to minimize the error between the predicted results and the actual labels. The LM algorithm is used: x s+1 =x s -(H+αI) -1 G Among them, H is the matrix of multidimensional vectors, G is the first-order gradient of multidimensional vectors, x s is the function extreme value of the previous step, α is the step size, and I is the identity matrix; Use the test data set to evaluate the performance of the trained neural network model. Input the test data into the network for prediction, compare it with the actual label, and calculate the accuracy and recall evaluation indicators to measure the performance of the model. The audio data collected by the sound sensor is transformed by Mel coefficient extraction, and the Mel spectrum obtained after transformation is input into the trained neural network model for real-time target recognition. The prediction results are obtained by extracting features from the image and forward propagation of the neural network to realize the identification of insulator self-explosion faults.

8. A system using the insulator self-explosion fault detection method based on sound data as claimed in any one of claims 1 to 7, characterized in that: It includes sound collection module, preprocessing module, feature extraction module, BP neural network training module, fault detection and identification module; The sound collection module collects the sound signal of the insulator during operation through the sound sensor; The preprocessing module performs noise reduction and endpoint detection processing on the collected original sound signal; The feature extraction module extracts features from the preprocessed sound signal and converts it into a Mel-spectrogram feature image; The BP neural network training module uses the BP neural network to train the extracted features to identify and classify the operating status of the insulator; The fault detection and identification module extracts and transforms the real-time audio data collected by the sound sensor through Mel coefficients, and inputs the transformed Mel spectrogram into a trained neural network model to perform real-time target recognition.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images