A partial discharge monitoring method based on voiceprint
Through sampling and processing of voiceprint signals, it is converted into digital signals to judge the local discharge of high-voltage equipment, which solves the problem of insufficient sensitivity and anti-interference of traditional methods, and achieves efficient local discharge detection.
Patent Information
- Application Number
- CN202110900753.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-08-06
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2041-08-06
AI Technical Summary
When monitoring local discharges of medium and high voltage equipment, traditional methods have insufficient sensitivity and anti-interference, making it difficult to effectively detect and judge the existence of local discharges.
Using a soundprint-based monitoring method, the soundprint signal is obtained for sampling, pre-emphasis, frame division, windowing, fast Fourier transform, noise reduction processing and basic spectrum subtraction, and converted to digital signals to determine whether there is local discharge in the device. .
It improves the sensitivity and anti-interference of monitoring, can accurately detect weak signals in complex environments, quantify the existence of local discharges, and has a small equipment and simple operation.
Smart Images

Figure CN113866570B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of electrical equipment, and in particular to a partial discharge monitoring method based on voiceprint. Background Art
[0002] At present, with the development of science and technology, there are more and more automated equipment. With less and less manual operation and management, the monitoring system of equipment is particularly important. For medium and high voltage equipment, partial discharge occurs from time to time. Partial discharge is mainly the discharge of internal insulation of transformers, mutual inductors and other high-voltage electrical equipment under the action of high voltage. This discharge only exists in the local position of the insulation and will not immediately form a breakdown or flashover of the entire insulation, so it is called partial discharge. The continuous development of partial discharge will gradually expand the deterioration damage of the insulation, eventually shortening the normal life of the insulation, reducing the short-term insulation strength, and even causing the entire insulation to break down. The traditional method is to use infrared detection or ultra-high frequency to monitor whether there is partial discharge, but these monitoring methods have certain shortcomings in terms of sensitivity and anti-interference. Summary of the invention
[0003] The present invention provides a partial discharge monitoring method based on voiceprint, which can effectively solve the above problems.
[0004] The present invention is achieved in that:
[0005] A partial discharge monitoring method based on voiceprint includes the following steps:
[0006] S1, obtain the voiceprint from the device under test, and intercept it in segments according to the set time length to obtain several segments of first voiceprint signals, and sample each first voiceprint signal to obtain multiple sampling signals; wherein the sampling frequency is fs, the amplitude of each sampling point is x(n), and n is the sampling point;
[0007] S2, performing pre-emphasis processing, frame processing, windowing processing and fast Fourier transform on each sampled signal to obtain a first spectrum;
[0008] S3, performing noise reduction processing and filtering processing on the first spectrum to obtain a second spectrum;
[0009] S4, respectively obtaining the energy P of each frame in the second spectrum j , and then calculate the average energy P of the j frame △ , where j is the number of sub-frames;
[0010] S5, judging whether there are L set periods within the set time length, and whether there are R pulses within each set period, if yes, judging that the device under test has partial discharge, otherwise judging that the device under test does not have partial discharge; wherein the pulse is to divide the energy P of the adjacent M framesj All satisfied: P j >β*P △ , where β is the energy coefficient.
[0011] As a further improvement, in step S2, according to the formula W(n)=(1-γ)-γcos(2πn / N), each frame signal after the frame processing is subjected to windowing processing, where N is the size of the window and γ is 0.46.
[0012] As a further improvement, in step S3, according to the formula: The first spectrum is subjected to denoising by basic spectral subtraction, wherein |Xj(k)|new represents the amplitude of each frame in the first spectrum after denoising, a is an over-subtraction factor, b is a gain compensation factor, a and b are both constants, 0≤k≤N-1, and D(k) represents the average energy value of the corresponding noise segment.
[0013] As a further improvement, in step S5, the value range of M is 3≤M≤15.
[0014] As a further improvement, in step S5, the value range of M is 5≤M≤10.
[0015] As a further improvement, in step S5, L≥15.
[0016] As a further improvement, in step S5, the value range of R is R≥2.
[0017] As a further improvement, in step S5, the value of R is 2.
[0018] As a further improvement, the set time length is 500ms.
[0019] As a further improvement, the set period is 20 ms.
[0020] The beneficial effects of the present invention are:
[0021] The present invention provides a partial discharge monitoring method based on voiceprint, which samples the acquired voiceprint, cuts it into segments of voiceprint signals for processing, and successively undergoes pre-emphasis, framing, windowing and other processing to first obtain a time domain signal that is better suitable for fast Fourier transform, wherein the pre-emphasis processing can compensate for the excessive attenuation of high-frequency components during the transmission process without affecting the signal-to-noise ratio; the framing processing is performed to make the voice signal meet the short-term stability and facilitate processing; and then the windowing processing is performed to make the framed signal more stable and more suitable for fast Fourier transform. After transforming to the frequency domain, the basic spectrum subtraction method is used to remove the noise component to obtain a clean frequency domain voiceprint signal. Then, by comparing the energy Pj of each frame with β*P△, the analog signal is converted into a digital signal 0 and 1. Then, a precisely set method is used to determine whether there is a partial discharge, that is, in each segment of the intercepted voiceprint signal, after the above processing, if the adjacent M frames are all set to 1, it is regarded as a pulse. If there are at least L set cycles in this segment of speech and at least R pulses in each set cycle, the device under test is considered to have a partial discharge. This monitoring method converts analog signals into digital signals, and the anti-interference performance is improved. By monitoring the number of pulses under certain conditions in each segment of the voiceprint signal, the monitoring of partial discharge is quantified, and the monitoring sensitivity is high. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0023] Figure 1 It is a flow chart of a partial discharge monitoring method based on voiceprint provided in an embodiment of the present invention.
[0024] Figure 2 It is a data processing flow chart before determining whether partial discharge exists provided by an embodiment of the present invention.
[0025] Figure 3 It is a conditional structure flow chart for judging partial discharge provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0026] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0027] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.
[0028] Reference Figure 1-3 As shown, a partial discharge monitoring method based on voiceprint includes the following steps S1-S5:
[0029] S1, obtain the sound from the device under test, and intercept it in segments according to the set time length to obtain several segments of the first voiceprint signal, and sample each first voiceprint signal to obtain multiple sampling signals; wherein, the sampling frequency is fs, the amplitude of each sampling point is x(n), and n is the sampling point. Specifically, the voiceprint containing the internal situation information of the device under test is obtained by the sensor arranged near the device under test, such as this, and the internal situation information of the device under test mainly refers to the information that the device under test contains partial discharge (PD) or does not contain partial discharge. It is segmented and intercepted according to the set time length. In this embodiment, the time for segmented interception can be set to 500ms, and the acquired sound is intercepted every 500ms, and each segment of the sound is processed in turn. Specifically, the use of sensors to obtain sound can be through long-distance probes or ultrasonic long-distance receivers, which can greatly increase the detection distance of the instrument and can detect distances of more than 30 meters.
[0030] S2, each sampled signal is subjected to pre-emphasis processing, frame processing, windowing processing and fast Fourier transform to obtain a first spectrum, wherein the spectrum of each frame is X j (k), k = 0, 1, 2, 3...N-1, where j represents the frame number and k represents the sequence number in each frame.
[0031] Specifically, each sampled signal is pre-emphasized according to the formula y(n)=x(n)-αx(n-1), where α is the pre-emphasis coefficient, y(n) is the amplitude of the nth sampling point after pre-emphasis; the value range of α is 0.9≤α≤1.0. In this embodiment, α is 0.97. Pre-emphasis mainly enhances the high-frequency component of the first voiceprint signal to compensate for the excessive attenuation of the high-frequency component during transmission, but does not affect the signal-to-noise ratio.
[0032] Then, each sampled signal after pre-emphasis is framed, which mainly refers to sampling each first voiceprint signal, pre-emphasis and then framing. The frame length of the frame is set to f1, the frame shift is set to f2, and j represents the number of frames, which is a positive integer. The frame length specifically refers to how many frames the data of the entire sampling point can be divided into, and the frame shift refers to the number of sampling points that cross between frames. In this embodiment, the frame length is 2ms and the frame shift is 1ms.
[0033] Then, the signal of each frame after framing is windowed using a Hamming window. This mainly refers to sampling, pre-emphasis, and framing each first voiceprint signal before windowing. Here, a Hamming window is used, according to the formula W(n) = (1-γ)-γcos(2πn / N), where N is the size of the Hamming window, and the N value is determined by the sampling frequency fs and the framing situation. The value of γ is 0.46, and W(n) represents the amplitude of the signal after windowing. The purpose of windowing the framed signal is to make the framed voiceprint signal more stable, so that the time domain signal can better meet the periodic requirements of the fast Fourier transform processing and reduce leakage.
[0034] Perform fast Fourier transform on each frame of voiceprint signal after windowing to obtain the first spectrum. The spectrum of each frame is X j (k), k = 0, 1, 2, 3 ... N-1, where j represents the frame number and k represents the sequence number in each frame. The time domain signal is transformed into a frequency domain signal.
[0035] S3, performing noise reduction and filtering on the first spectrum to obtain a second spectrum. Specifically, basic spectrum subtraction is used to remove noise, and the noise component is subtracted from the first spectrum. In step S3, according to the formula:
[0036] The first spectrum is subjected to basic spectrum subtraction denoising, wherein |X j(k)|new represents the amplitude of each frame in the first spectrum after noise reduction, a is the over-reduction factor, b is the gain compensation factor, a and b are constants, 0≤k≤N-1, D(k) represents the average energy value of the corresponding noise segment. The calculation method of D(k) is as follows: if the duration of the leading silent segment (noise segment) is IS and the corresponding number of frames is NIS, the average energy value of the noise segment can be calculated as
[0037]
[0038] After the first spectrum is subjected to noise reduction processing, relatively clean voiceprint information in the frequency domain is obtained. The first spectrum after noise reduction processing is filtered through a group of triangular filters designed according to the Mel scale to obtain a second spectrum.
[0039] S4, respectively obtaining the energy P of each frame in the second spectrum j , and then calculate the average energy P of the j frame △ , where j is the frame number, indicating the frame number.
[0040] Reference Figure 2 As shown, Figure 2 This is a data processing flow chart before determining whether there is a partial discharge in the above steps S1-S4.
[0041] S5, determining whether there are L set periods within the set time length, and whether there are R pulses within each set period, if yes, determining whether the device under test has partial discharge according to the first voiceprint signal, otherwise determining whether the device under test has no partial discharge according to the first voiceprint signal, wherein the pulse is the energy P of the adjacent M frames. j All satisfied: P j >β*P △ When P is set to 1 pulse, β is the energy coefficient. j >β*P △ When P j< β*P △ When , set to 0, thus completing the process of converting the analog signal with voiceprint information into a digital signal. The value range of M is 3≤M≤15, especially 5≤M≤10 is preferred. The value range of L≥15, especially L is preferably 15. The value range of R is R≥2, especially R=2 is preferred.
[0042] In this embodiment, refer to Figure 3As shown, the set duration is 500ms, and the corresponding set period is 20ms. For step S5, if the adjacent 5-10 frames are all set to 1, the corresponding adjacent 5-10 frames are set to 1 pulse. Under the 50HZ state, if there are two such pulses in one cycle, and there are 15 cycles in this 500ms voiceprint signal, then this 500ms voiceprint signal is regarded as a voiceprint signal with partial discharge.
[0043] The present invention samples the acquired voiceprint signal, cuts it into segments for processing, and then processes it through pre-emphasis, framing, windowing, etc., first obtaining a time domain signal that is better suitable for fast Fourier transform, wherein the pre-emphasis processing can compensate for the excessive attenuation of high-frequency components during transmission without affecting the signal-to-noise ratio; the framing processing is performed to make the voiceprint signal meet the short-term stability and facilitate processing; and then the windowing processing is performed to make the signal after framing more stable and more suitable for fast Fourier transform. After transforming to the frequency domain, the basic spectral subtraction method is used to remove the noise component to obtain a clean frequency domain voiceprint signal, and then the energy P of each frame is compared. j With β*P △ , converting analog signals into digital signals 0 and 1; then a precisely set method is used to determine whether there is partial discharge, that is: in each segment of the intercepted voiceprint signal, after the above processing, if the adjacent 5-10 frames are all set to 1, it is regarded as a pulse. If this segment of the voiceprint signal is calculated according to the frequency of 50Hz, there are at least 15 cycles and at least 2 pulses in each cycle, it is regarded as a partial discharge in the device under test. This monitoring method of converting analog signals into digital signals quantifies the monitoring of partial discharges. The monitoring sensitivity and anti-interference performance are relatively high. It can be used in complex environments and can detect weak signals. In addition, the partial discharge monitoring equipment based on the partial discharge monitoring method can be made into a smaller size, easy to carry, and simple to operate.
[0044] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A partial discharge monitoring method based on voiceprint, characterized in that: The following steps are involved: S1, obtain the voiceprint from the device under test, and intercept it in segments according to the set time length to obtain several segments of first voiceprint signals, and sample each first voiceprint signal to obtain multiple sampling signals; wherein the sampling frequency is fs, the amplitude of each sampling point is x(n), and n is the sampling point; S2, performing pre-emphasis processing, frame processing, windowing processing and fast Fourier transform on each sampled signal to obtain a first spectrum; S3, performing noise reduction processing and filtering processing on the first spectrum in sequence to obtain a second spectrum; wherein the first spectrum is subjected to noise reduction processing by basic spectrum subtraction; The first spectrum after the noise reduction process is filtered through a group of triangular filters designed according to the Mel scale to obtain a second spectrum; S4, respectively obtaining the energy P of each frame in the second spectrum j , and then calculate the average energy P of the j frame △ , where j is the number of sub-frames; S5, judging whether there are L set periods within the set time length, and whether there are R pulses within each set period, if yes, judging that the device under test has partial discharge, otherwise judging that the device under test does not have partial discharge; wherein the pulse is to divide the energy P of the adjacent M frames j All satisfied: P j >β*P △ It is set to 1 pulse, wherein β is the energy coefficient, the value range of M is 3≤M≤15, L≥15, the value range of R is R≥2, the set time length is 500ms, and the set period is 20ms.
2. According to claim 1, a partial discharge monitoring method based on voiceprint is characterized in that: In step S2, windowing is performed on each frame of the signal after the frame division process according to the formula W(n)=(1-γ)-γcos(2πn / N), where N is the size of the window and γ is 0.
46.
3. According to claim 2, a partial discharge monitoring method based on voiceprint is characterized in that: In step S3, according to the formula: The first spectrum is subjected to denoising by basic spectral subtraction, wherein |Xj(k)|new represents the amplitude of each frame in the first spectrum after denoising, a is an over-subtraction factor, b is a gain compensation factor, a and b are both constants, 0≤k≤N-1, and D(k) represents the average energy value of the corresponding noise segment.
4. According to claim 1, a partial discharge monitoring method based on voiceprint is characterized in that: In step S5, the value range of M is 5≤M≤10.
5. According to claim 1, a partial discharge monitoring method based on voiceprint is characterized in that: In step S5, the value of R is 2.
Citation Information
Patent Citations
Method for identifying local discharge signals of switchboard based on support vector machine model
CN102426835A
Partial discharge pulse detection system
CN104535902A
Supersonic wave based partial discharge detection method
CN108896878A
Acousto-optic combined detection method and system for partial discharging of a high-voltage cable
CN110275094A
Automatic monitoring method for analyzing abnormities of suckling piglets based on sound signals
CN112820275A
Cited By
A voiceprint monitoring method for substation main equipment based on speech recognition
CN115469170B
