Staff noise reduction method for partial discharge of ethylene propylene rubber cable

Abstract

The invention belongs to the technical field of discharge noise processing, and particularly relates to a staff noise reduction method for partial discharge of an ethylene propylene rubber cable. Comprising the following steps of: 1, processing a periodic narrow-band interference signal by adopting an energy ratio method, and removing the narrow-band interference in a partial discharge signal; 2, reprocessing the processed point signal by adopting a generalized S transformation mode; carrying out modular matrix algorithm transformation on the high-frequency partial discharge signal; 3, setting a basic value gamma by adopting a smaller threshold method, and obtaining the number of all narrowband interferences; 4, acquiring effective data of the center frequency; 5, wave crest data below the standard curve are set as interference frequency data, and the wave crest value is an interference wave crest; 6, extracting frequency peak values corresponding to all narrowband interference signals in the atlas according to the interference peak definition; and 7, setting a frequency spectral line corresponding to the peak value as kp. The method is more thorough in noise processing and has good application value.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of discharge noise processing, and particularly relates to a five-line spectrum noise reduction method for partial discharge of ethylene-propylene rubber cable. BACKGROUND

[0002] The ethylene-propylene rubber cable is used as the main pump power cable in some nuclear power plants due to excellent flexibility, waterproofness, heat resistance and anti-electromagnetic radiation characteristics. However, the cable will age after a long time of operation, thereby generating partial defects, and partial discharge will be generated in these tiny defects and continuously increase, which will damage the insulation and affect the safe and stable operation of the main pump over a long period of time. Therefore, it is necessary to monitor the partial discharge of the ethylene-propylene rubber cable.

[0003] When the cable partial discharge fault occurs, electricity, light, sound and heat will be generated, and the light, electricity and sound correspond to the electric pulse, light signal and ultrasonic signal. Therefore, the sensor is designed based on the signals during the detection of the cable, and the temperature sensor is combined for measurement, so as to obtain the cable partial discharge condition and analyze the cable insulation state. The commonly used partial discharge detection methods mainly include the pulse current method, ultrasonic detection method, high-frequency pulse current method, ultrahigh frequency detection method and temperature detection method, and the ultrasonic detection method is the commonly used sensor for the ethylene-propylene rubber cable detection.

[0004] Since the partial discharge signal of the ethylene-propylene rubber cable is very weak in the early stage of failure, the partial discharge detection sensor, especially the ultrasonic sensor, will receive complex interference signals under the operating condition during the partial discharge test of the ethylene-propylene rubber cable, and the weak partial discharge signal will be completely covered by the interference signal. How to reduce and eliminate the interference signal has become a research difficulty.

[0005] In view of the above situation, in order to solve the problem of collecting the partial discharge signal of the ethylene-propylene rubber main pump cable in the nuclear power plant, the environmental interference existing in the ethylene-propylene rubber cable detection process is deeply analyzed, and the interference is eliminated in combination with the analysis result, so that a new solution, that is, the five-line spectrum noise reduction method for the partial discharge of the ethylene-propylene rubber cable, is found. SUMMARY

[0006] The application aims to provide a five-line spectrum noise reduction method for the partial discharge of the ethylene-propylene rubber cable, which can more completely process the noise and has good application value.

[0007] To achieve the above object, the technical scheme adopted by the application is as follows:

[0008] A five-line spectrum noise reduction method for the partial discharge of the ethylene-propylene rubber cable, comprising the following steps:

[0009] First, the energy ratio method is used to process the periodic narrowband interference signal to remove the narrowband interference in the partial discharge signal.

[0010] Second, the processed point signal is reprocessed by using the generalized S transform method, and the high-frequency partial discharge signal is transformed by using the modulus matrix algorithm.

[0011] Third, a small threshold method is used to set a basic value γ to obtain the total number of narrowband interference.

[0012] Fourth, the center frequency effective data is obtained.

[0013] Fifth, the wave peak data below the standard curve is set as the interference frequency data, and the wave peak value is the interference wave peak.

[0014] Sixth, according to the interference wave peak definition, the frequency peak value corresponding to all narrowband interference signals in the atlas is extracted.

[0015] Seventh, the peak value corresponding to the frequency spectrum line is set as kp, the corresponding spectrum line amplitude is R(kp), and the amplitudes corresponding to the left and right sides of the unit function window spectrum line are intercepted: the values taken are R(kp+1) and R(kp-1).

[0016] Eighth, the complex data is converted, and the real number value of the complex number is obtained.

[0017] Ninth, according to the real number value, the correction coefficient is further calculated.

[0018] Tenth, the accurate interference frequency is obtained, and the accurate narrowband interference frequency fi' is derived through the correction coefficient.

[0019] Eleventh, the five-line spectrum method is used to derive the reconstructed curve, and since R(kp+1) and R(kp-1) are obtained, the parameters f i These parameters are combined with the five-line spectrum derivation point method to correct the submerged signal and derive the true approximate signal value.

[0020] First, the energy ratio algorithm is as follows:

[0021]

[0022] e i The energy ratio value of the i-th frequency window; ω i The midpoint frequency of the i-th frequency window; h is the frequency window width; P(ω) is the power spectrum energy; and ω is the midpoint frequency.

[0023] Second, the specific transformation algorithm is:

[0024]

[0025] where f is frequency; x(t) is energy function; ω(t-τ,f) is Gaussian window function; t, τ are time; signal curve of partial discharge signal containing noise is obtained by mode matrix algorithm, which contains frequency-amplitude square sum and mean curve.

[0026] The third step, the specific algorithm is based on:

[0027]

[0028] where T is the maximum point of the curve, σ is the standard frequency difference of noise; γ is a constant of 0.1, and N is the number of peaks of the frequency-amplitude square sum and mean curve.

[0029] The fourth step, the specific algorithm for obtaining the center frequency effective data is: the generalized S transform of the partial discharge signal is introduced again, the frequency-amplitude standard deviation curve diagram after the generalized S transform is obtained, and it is determined that the center frequency is calculated above 0.1 standard line, which is set as the center frequency effective data in the algorithm setting.

[0030] The fifth step, the peak data below the 0.1 standard curve is set as the interference frequency data.

[0031] The eighth step, the specific algorithm is:

[0032]

[0033] where M is the real number value of the complex number, real[] is the real number calculation formula of the complex number, and N is the number of peaks of the frequency-amplitude square sum and mean curve.

[0034] The ninth step, the algorithm for correcting coefficient is

[0035]

[0036] where δ is the correction value, N is the number of peaks of the frequency-amplitude square sum and mean curve, and M is the real number value of the complex number.

[0037] The tenth step,

[0038] (k, + 6) · fs / N

[0039] Where fi' is the i-th accurate frequency, δ is the correction value, N is the number of peaks of the frequency-amplitude square sum and mean curve, and fs is the sampling frequency.

[0040] The eleventh step, the specific five-line spectrum method is

[0041]

[0042] Wherein M' is a real number value of complex number, real【】 is a real number calculation formula of complex number, peak value corresponding frequency spectrum line is kp, N is the number of wave peak of frequency-amplitude square and mean value curve;Five corresponding values of R(kp), R(kp+1), R(kp-1), R(kp+0.5), R(kp-0.5) are replaced by the original values, and then the design of five-line spectrum signal recovery is formed;After transformation, the derived data is restored to the original curve, and replaces the frequency spectrum part of noise, so that the obtained signal is the signal of pulse interference.

[0043] The beneficial effects obtained by the present application are:

[0044] The present application is more thorough in noise processing, and if the project achieves the expected design, it will have good application value, especially for nuclear power plants using ethylene-propylene rubber in a large area, and the application value is more prominent. The specific innovation points are as follows: for the actual environment of ethylene-propylene rubber in nuclear power plants, the energy ratio method is used to process periodic narrow-band interference. For pulse interference, a double generalized S transform processing method is used, which can well identify all narrow-band noise. Another technical innovation for narrow-band noise is to transform the high-frequency partial discharge signal by matrix algorithm, and then process it by combining double generalized S transform. For the interference center frequency of ethylene-propylene rubber in nuclear power plants, the standard coefficient is set to 0.1, and the correction coefficient is derived. Through the correction coefficient, the accurate frequency fi of narrow-band interference is derived. The five-line spectrum method is designed to reconstruct the submerged curve and realize the recovery of the curve. The whole algorithm can effectively eliminate periodic narrow-band interference and white noise, and has strong interference processing capability in the nuclear power plant ethylene-propylene rubber main pump cable. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 It is a frequency-amplitude square and mean value curve graph of generalized S transform.

[0046] Figure 2 It is a frequency-amplitude standard deviation curve graph of re-generalized S transform.

[0047] Figure 3 It is a curve graph reconstructed by five-line spectrum method.

[0048] Figure 4 It is a signal graph of pulse interference elimination.

[0049] Figure 5 It is a noise elimination route graph of the present application. DETAILED DESCRIPTION

[0050] The present application will be described in detail below in combination with the drawings and specific embodiments.

[0051] A method for removing interference from a stave. Firstly, periodic narrowband interference is removed by an energy ratio method, then the amplitude and mean value graphs are obtained by a generalized S frequency transformation method, and then the standard ratio of the frequency amplitude is obtained. Then, the number of interferences is obtained by designing a threshold value, and then the frequency of the narrowband interference is evaluated in detail by designing a stave, and then the reconstruction is added to the original signal to achieve the effect of effective removal. Specifically, it is divided into five steps: first, the frequency peak value corresponding to the narrowband noise signal is obtained from the original signal; second, the M parameter is obtained; third, the correction value delta of M is obtained; fourth, the narrowband noise center frequency is obtained; fifth, the signal is analyzed by interpolation through the stave algorithm.

[0052] The source of the ethylene-propylene rubber main pump cable interference signal is mainly periodic narrowband interference, pulse interference and white noise. The periodic narrowband interference mainly refers to the narrowband interference emitted by wireless communication equipment, such as the radio frequency radiation interference generated when the communication equipment communicates wirelessly with the control center, and the electromagnetic interference emitted by the surrounding wireless communication towers and other communication equipment when working, broadcast signals, mobile phone signals, etc. Pulse signals can generally be divided into periodic pulse interference and random pulse interference. Random interference is generally caused by high-voltage circuit breaker switching, lightning weather interference. Periodic interference mainly refers to the periodic electromagnetic pulse radiation of electrical equipment. White noise interference is mainly caused by the thermal noise of electrical equipment. Thermal noise is generated by the random motion of micro-particles such as electrons, and the separation of space electrons existing in the air often generates thermal noise.

[0053] Under the influence of various interferences, general interference is not single interference, but often a combination of multiple interferences. Therefore, the noise processing method should be designed accordingly. The present application is designed according to the interference conditions in the nuclear power plant site.

[0054] There is periodic narrowband interference in the field, and such interference generally presents a sine or cosine waveform after signal processing. In actual PD signal detection, there is often a problem of narrowband interference and PD signal aliasing. The energy ratio method is generally used for interference signal denoising. The specific idea is as follows: the energy ratio method first needs to set a fixed frequency window width, and the frequency window is shifted point by point from the starting point of the signal. The midpoint of the frequency window is used as the boundary to divide the frequency band in the frequency window into two sections, and the power spectrum energy of the two sections is calculated and the ratio is compared. On the basis of the foregoing, the horizontal axis corresponding to the frequency is converted to the number of sampling points. The specific development algorithm of the energy ratio method is as follows:

[0055]

[0056] e i The energy ratio value corresponding to the i-th frequency window; ω i The midpoint frequency corresponding to the i-th frequency window; h is the frequency window width.

[0057] After the signal is transformed by the power spectrum algorithm, the narrowband interference appears as a pulse wave with a large energy value in the power spectrum, like a "peak". The partial discharge signal is relatively dispersed in the power spectrum, and has a small energy value, like a "hill". After the above algorithm formula transformation, the peak signal is removed, and the narrowband interference can be effectively suppressed.

[0058] The second step needs to process the pulse interference. Since the pulse interference is large in number and common, it is difficult to process in terms of noise. Here, the generalized S transform is used to reprocess the above processed electrical signal.

[0059] First, the high-frequency partial discharge signal is transformed by the modulus matrix algorithm. The specific transformation algorithm is

[0060]

[0061] f is the frequency; ω(t-π, f) is the Gaussian window function; t and τ are the time.

[0062] In this way, the generalized S transform frequency-amplitude square and mean value curve of the partial discharge signal containing noise is obtained. Then, a small threshold method is used, a basic value γ is set, γ is generally set to 0.1-0.5, and in combination with the ethylene-propylene rubber cable involved in the present patent, γ is taken as 0.1. The specific algorithm basis is:

[0063]

[0064] In this way, the total number of narrowband interferences can be obtained. There is a special case that the center frequency of the partial discharge signal will inevitably be calculated. In order to solve this problem, the generalized S transform of the partial discharge signal is introduced again to obtain the generalized S transform frequency-amplitude standard deviation curve. If it is above 0.1 standard line, it is identified as the center frequency calculated. This is not interference, and the center frequency effective data is set in the algorithm setting.

[0065] The wave peak data below the 0.1 standard curve is set as the interference wave peak.

[0066] The frequency peak value corresponding to all narrowband interference signals in the atlas is extracted.

[0067] The amplitudes corresponding to the left and right sides of the spectrum are intercepted. Here, the values are R(k p +1) and R(k p -1).

[0068] Through the algorithm, the complex data is converted, and the real number value of the complex number is obtained. The specific algorithm is:

[0069]

[0070] According to the above-mentioned real number value, the correction coefficient is further calculated. The algorithm of the correction coefficient is:

[0071]

[0072] Through the correction coefficient, the accurate frequency f of the narrowband interference is further derived i :

[0073] (k,+6)·fs / N

[0074] Since R(k p +1) and R(k p -1) are calibrated, f i These parameters are obtained, and the submerged signal is corrected by using the five-line spectrum derivation point method, and the true approximate signal value is derived. The specific five-line spectrum method is

[0075]

[0076] The curve reconstructed by the five-line spectrum method is shown in Figure 3 , and the horizontal line is the reconstructed spectrum line.

[0077] After transformation, the derived data is restored to the original curve to replace the noise spectrum part, so that the obtained signal is the signal with pulse interference removed.

[0078] A five-line spectrum noise reduction method for partial discharge of ethylene-propylene rubber cable, comprising the following steps:

[0079] First, the periodic narrowband interference signal is processed, and the energy ratio method is used to remove the narrowband interference in the partial discharge signal. The specific energy ratio algorithm is as follows:

[0080]

[0081] The energy ratio method is to compare the power spectrum energy difference of the same frequency window before and after the frequency band. e i The energy ratio value before and after the i-th frequency window; ω i The midpoint frequency in the i-th frequency window; P(ω) is the power spectrum energy; ω is the midpoint frequency; h is the frequency window width.

[0082] Second, the above-processed point signal is reprocessed by using the generalized S transform method.

[0083] The high-frequency partial discharge signal is transformed by using the modal matrix algorithm, and the specific transformation algorithm is:

[0084]

[0085] Where f is frequency; x(t) is energy function; ω(t-τ,f) is Gaussian window function; t, τ are time.

[0086] The signal curve of partial discharge signal containing noise is obtained by using the modal matrix algorithm, which contains the frequency-amplitude square and mean curve.

[0087] Thirdly, a small threshold method is used, a basic value γ is set, which is generally set to 0.1-0.5, and in combination with the ethylene-propylene rubber cable involved in the patent, γ is taken as 0.1, and the specific algorithm is based on:

[0088]

[0089] Where T is the maximum value point of the curve, σ is the standard frequency difference of noise; r is a constant 0.1, and N is the number of peaks of the frequency-amplitude square and mean curve.

[0090] In this way, the number of all narrow-band interferences can be obtained.

[0091] Fourthly, the center frequency effective data is obtained, specifically, the generalized S transform of the partial discharge signal is introduced again, the frequency-amplitude standard deviation curve after the generalized S transform is obtained, and the center frequency to be calculated is determined as being above the 0.1 standard line, which is not interference, and is set as the center frequency effective data in the algorithm setting.

[0092] Fifthly, the peak data below the 0.1 standard curve is set as the interference frequency data, and the peak value thereof is the interference peak.

[0093] Sixthly, according to the above definition of the interference peak, the frequency peak value corresponding to all narrow-band interference signals can be extracted in the atlas.

[0094] Seventhly, the peak value corresponding to the frequency spectrum is set as kp, the corresponding spectrum line amplitude is R(kp), and the amplitudes corresponding to the left and right sides of the unit function window spectrum line are intercepted, which are taken as R(kp+1) and R(kp-1).

[0095] Eighthly, through the algorithm, the complex data is converted, and the real number value of the complex number is obtained. The specific algorithm is:

[0096]

[0097] Where M is the real number value of the complex number, real【】 is the real number calculation formula of the complex number, and N is the number of peaks of the frequency-amplitude square and mean curve.

[0098] Ninthly, the correction coefficient is obtained, and the correction coefficient is further calculated according to the above real number value.

[0099] The algorithm of the correction coefficient is

[0100]

[0101] Wherein δ is the correction value, N is the number of peaks of the frequency-amplitude square and mean value curve, M is the real number value of the complex number.

[0102] The tenth step is to obtain the accurate frequency of interference, and then deduce the accurate frequency f of narrowband interference through the correction coefficient. i

[0103] (k, +6)fs / N

[0104] Wherein fi' is the accurate frequency (the i-th frequency), δ is the correction value, and N is the number of peaks of the frequency-amplitude square and mean value curve.

[0105] The tenth step is to deduce the reconstructed curve by using the five-line spectrum method. Since R(kp+1), R(kp-1) are calibrated, and fi' is obtained, combined with these parameters, the five-line spectrum is used to deduce the point method, correct the submerged signal, and deduce the true approximate signal value. The specific five-line spectrum method is

[0106]

[0107] Wherein M' is the real number value of the complex number, real【】 is the real number calculation formula of the complex number, the peak value corresponds to the frequency spectrum line kp, and N is the number of peaks of the frequency-amplitude square and mean value curve. Replace the original values with R(kp), R(kp+1), R(kp-1), R(kp+0.5), and R(kp-0.5) five corresponding values, and then form the design of five-line spectrum signal recovery.

[0108] The curve reconstructed by the five-line spectrum method is shown in the figure Figure 3 , and the horizontal line is the reconstructed spectrum line.

[0109] After transformation, the deduced data is restored to the original curve to replace the noise spectrum part, so that the obtained signal is the signal after removing the pulse interference.​

Claims

1. A five-line spectrum noise reduction method for partial discharge of EPDM cable, characterized by: The steps include: In the first step, the energy ratio method is used to process the signal of periodic narrowband interference and remove the narrowband interference in the partial discharge signal; In the second step, the processed point signal is reprocessed by using the generalized S-transform method; the high-frequency partial discharge signal is transformed by the module matrix algorithm; The third step is to use a smaller threshold method to set the basic value γ and obtain the total number of narrowband interferences; Step 4: Obtain valid center frequency data; Step 5: Set the peak data below the standard curve as interference frequency data, and its peak value as interference peak; Step 6: According to the definition of interference peak, extract the frequency peaks corresponding to all narrowband interference signals in the spectrum; Step 7: Set the frequency spectrum line corresponding to the peak to kp, the corresponding spectrum line amplitude to R(kp), and then intercept the amplitudes corresponding to the left and right sides of the unit window spectrum line: here the values ​​are R(kp+1) and R(kp-1); Step 8: Convert to complex data and then obtain the real value of the complex number; Step 9: further calculate the correction coefficient based on the real number value; The tenth step is to obtain the precise frequency of interference, and then derive the precise frequency of narrowband interference f through the correction coefficient. i '; In the eleventh step, the reconstruction curve is derived using the stave method. Since R(kp+1) and R(kp-1) are calibrated, f is obtained. i 'These parameters, combined with the five-line derivation point method, correct the submerged signal and deduce the true approximate signal value.

2. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 1 is characterized by: In the first step, the energy ratio algorithm is as follows: e i Corresponding to the ratio of the front and back energies in the i-th frequency window; ω i Corresponding to the midpoint frequency in the i-th frequency window; h is the frequency window width; P(ω) is the power spectrum energy; ω is the midpoint frequency.

3. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 2 is characterized by: In the second step, the specific transformation algorithm is: Where f is the frequency; x(t) is the energy function; ω(t-τ,f) is the Gaussian window function; t and τ are time. The signal curve of the noisy partial discharge signal is obtained through the modular matrix algorithm, which includes the frequency-amplitude square sum mean curve.

4. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 3 is characterized by: The third step, the specific algorithm is based on: Where T is the maximum point of the curve, σ is the standard frequency deviation of the noise; γ is a constant of 0.1, and N is the number of peaks in the frequency-amplitude square sum mean curve.

5. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 4 is characterized by: The fourth step is to obtain the valid data of the center frequency. Specifically, the generalized S transform of the partial discharge signal is introduced again to obtain the frequency-amplitude standard deviation curve after the generalized S transform. If it exceeds the standard line of 0.1, it is determined to be the calculated center frequency and set as the valid data of the center frequency in the algorithm settings.

6. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 5, characterized in that: Step 5: Set the peak data below the 0.1 standard curve as interference frequency data.

7. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 6, characterized in that: Step 8: The specific algorithm is: Where M is the real number value of the complex number, real【】 is the real number calculation formula of the complex number, and N is the number of peaks in the frequency-amplitude square sum mean curve.

8. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 7, characterized in that: In the ninth step, the algorithm for the correction coefficient is Where δ is the correction value, N is the number of peaks in the frequency-amplitude square sum mean curve, and M is a real number value that takes a complex value.

9. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 8, characterized in that: Step 10, f i ′=(k p +δ)·f s / N where f i ' is the i-th precise frequency, δ is the correction value, N is the number of peaks in the frequency-amplitude square sum mean curve; fs is the sampling frequency.

10. The five-line staff noise reduction method for partial discharge of EPDM cable according to claim 9, characterized in that: Step 11: The specific five-line notation method is Where M' is a real number value of a complex number, real【】 is a real number calculation formula for a complex number, the frequency spectrum line corresponding to the peak is kp, and N is the number of peaks in the frequency-amplitude square and mean curve; the five corresponding values ​​of R(kp), R(kp+1), R(kp-1), R(kp+0.5), and R(kp-0.5) are replaced with the original values, thereby forming a design for restoring the five-line spectrum signal; after the transformation, the derived data is restored to the original curve to replace the spectrum part of the noise, so that the signal obtained is the signal without pulse interference.

Images