A power transmission cable partial discharge detection method and system based on multi-physical field optimization

Abstract

The application relates to the technical field of power detection, and specifically discloses a power transmission cable partial discharge detection method and system based on multi-physical field optimization, which comprises the following steps: based on the multi-physical field prior parameter set of a power transmission cable, band-pass filtering an original mixed signal sequence to obtain a filtered signal sequence; performing discrete difference mapping on a discharge pulse characteristic waveform of the prior parameter set to obtain an adjacent point amplitude gradient and construct a fixed weight coefficient matrix; splitting the filtered signal sequence based on the waveform length to obtain even and odd subsequences; weighting and fusing the even subsequence according to the fixed weight coefficient matrix, combining the actual value of the odd subsequence to determine a predicted difference sequence; correcting the residual error of the even subsequence based on the predicted difference sequence to obtain a scale coefficient sequence; and finally performing attenuation compensation on the two types of sequences according to multi-band attenuation coefficients, and reconstructing a discharge pulse sequence after noise reduction through coefficient inversion; and the application can improve the detection accuracy of the power transmission cable partial discharge signal.

Description

Technical Field

[0001] This invention relates to the field of power detection technology, and in particular to a method and system for detecting partial discharge in power transmission cables based on multi-physics optimization. Background Technology

[0002] Currently, in the partial discharge detection of power transmission cables, basic filtering is generally performed directly on the acquired raw mixed signal. However, the processing does not take into account the multi-physics field characteristics of the cable, so clutter interference cannot be specifically eliminated. The environmental electromagnetic noise mixed in the raw mixed signal will continue to remain in the processed signal, and the details of the discharge pulse waveform corresponding to the partial discharge will be covered by noise. It is difficult to accurately capture the inherent characteristics of the pulse in the subsequent signal analysis stage.

[0003] Existing detection methods suffer from the following technical shortcomings: First, current filtering processes do not incorporate the multi-physics characteristics of the cable, making it impossible to determine the optimal filtering frequency band based on the inherent spectral characteristics of the standard discharge pulse, resulting in the inability to specifically eliminate clutter interference. Second, they lack processing logic that decomposes the signal based on the waveform regularity of the discharge pulse, failing to utilize amplitude gradients for weight allocation and adaptive residual correction. Third, the attenuation loss of multi-band signals generated during cable transmission cannot be quantitatively compensated. Existing methods typically use a single attenuation coefficient for overall amplification, failing to compensate for the differentiated attenuation of different frequency bands, leading to ineffective waveform distortion correction. Consequently, the reconstructed discharge pulse signal exhibits high distortion, directly reducing the reliability of partial discharge fault identification in transmission cables. Therefore, improving the detection accuracy of partial discharge signals in transmission cables has become an urgent problem to be solved. Summary of the Invention

[0004] In view of the above problems, the purpose of this invention is to provide a method and system for detecting partial discharge in power transmission cables based on multiphysics optimization, so as to solve the problem of detection accuracy of partial discharge signals in power transmission cables.

[0005] This invention provides a method for detecting partial discharge in power transmission cables based on multiphysics optimization, comprising: C1. A multi-physics prior parameter set based on the power transmission cable, wherein the multi-physics prior parameter set includes discharge pulse characteristic waveform and multi-band signal attenuation coefficient, frequency limiting parameter analysis is performed on the discharge pulse characteristic waveform to obtain the lower cutoff frequency and upper cutoff frequency of the discharge pulse characteristic waveform, and bandpass filtering is performed on the original mixed signal sequence of the power transmission cable according to the lower cutoff frequency and the upper cutoff frequency to obtain the filtered signal sequence of the power transmission cable; C2. Perform discrete difference mapping on the characteristic waveform of the discharge pulse to obtain the amplitude gradient of the neighboring points of the characteristic waveform of the discharge pulse, and construct the fixed weight coefficient matrix of the power transmission cable using the amplitude gradient of the neighboring points. C3. Using the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, determine the odd-even splitting start offset of the power transmission cable, and based on the odd-even splitting start offset, perform odd-even splitting on the filtered signal sequence to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence. C4. Based on the weight value of the current odd position in the fixed weight coefficient matrix, the even position subsequence is weighted and fused, and the predicted difference sequence of the power transmission cable is determined based on the fused current odd position standard value and the actual value of the odd position subsequence. C5. Based on the predicted difference sequence, perform residual correction on the even-numbered position subsequence to obtain the scale coefficient sequence of the power transmission cable; C6. Based on the multi-band signal attenuation coefficient, attenuation compensation is performed on the predicted difference sequence and the scale coefficient sequence, and coefficient inversion is performed on the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable.

[0006] Preferably, the multiphysics prior parameter set based on the power transmission cable includes a discharge pulse characteristic waveform and multi-band signal attenuation coefficients. Frequency limiting parameter analysis is performed on the discharge pulse characteristic waveform to obtain its lower and upper cutoff frequencies. Based on these lower and upper cutoff frequencies, bandpass filtering is applied to the original mixed signal sequence of the power transmission cable to obtain the filtered signal sequence. The process is as follows: The multi-physics prior parameter set and the original mixed signal sequence of the power transmission cable are obtained. The multi-physics prior parameter set includes the discharge pulse characteristic waveform and the multi-band signal attenuation coefficient. Frequency-limiting parameter analysis is performed on the multiphysics prior parameter set to obtain the lower cutoff frequency and upper cutoff frequency of the multiphysics prior parameter set; Based on the lower cutoff frequency and the upper cutoff frequency, the original mixed signal sequence is subjected to frequency window filtering to obtain the signal component of the power transmission cable, and the signal component is used as the filtered signal sequence of the power transmission cable.

[0007] Preferably, the process of performing discrete difference mapping on the discharge pulse characteristic waveform to obtain the neighboring point amplitude gradient of the discharge pulse characteristic waveform, and constructing the fixed weight coefficient matrix of the transmission cable using the neighboring point amplitude gradient, is as follows: Neighbor-point deviation quantization is performed on adjacent amplitude sampling points in the characteristic waveform of the discharge pulse to obtain the neighbor-point amplitude gradient sequence of the characteristic waveform of the discharge pulse. Each neighbor-point amplitude gradient in the neighbor-point amplitude gradient sequence is the amplitude difference between two adjacent amplitude sampling points. Based on the maximum absolute value of the neighboring point magnitude gradient sequence, the neighboring point magnitude gradients in the neighboring point magnitude gradient sequence are normalized to obtain the scale weight value sequence of the power transmission cable. Based on the sequential arrangement of the neighboring point amplitude gradients and the index positions of the odd-numbered sampling points in the characteristic waveform of the discharge pulse, the weight values ​​in the scale weight value sequence are allocated by odd-even projection to obtain the fixed weight coefficient matrix of the power transmission cable.

[0008] Preferably, the process of determining the odd-even splitting start offset of the transmission cable using the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, and then performing odd-even splitting on the filtered signal sequence based on the odd-even splitting start offset to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence, is as follows: The total number of amplitude sampling points in the characteristic waveform of the discharge pulse is taken as the waveform length of the characteristic waveform of the discharge pulse; Based on the original sampling time of the sampling points in the filtered signal sequence, the sampling points are time-series indexed to obtain the position index of the sampling points; Based on the waveform length, the parity split start offset of the filtered signal sequence is determined, wherein the parity split start offset is the integer part of the waveform length divided by two; Based on the position index and the odd-even splitting start offset, the sampling points are decoupled for odd and even positions to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence.

[0009] Preferably, the process of weighting and fusing the even-position subsequences based on the weight values ​​of the current odd-numbered positions in the fixed weight coefficient matrix, and determining the prediction difference sequence of the transmission cable based on the fused standard value of the current odd-numbered positions and the actual values ​​of the odd-numbered position subsequences, is as follows: In the fixed weight coefficient matrix, the weight value of the current odd position is determined based on the index number of the current odd position and the subsequence of the odd position; The position index of the current odd position is used to locate the neighbor index, thereby obtaining the left neighbor even position index and the right neighbor even position index of the current odd position; Based on the left neighbor even position index and the right neighbor even position index, the amplitude of the even position subsequence is sampled and extracted to obtain the left neighbor even position value and the right neighbor even position value of the even position subsequence; Based on the weight values, a coefficient weighting transformation is performed on the left neighbor even position value and the right neighbor even position value to obtain the left weighted value of the left neighbor even position value and the right weighted value of the right neighbor even position value. Then, the left weighted value and the right weighted value are linearly aggregated to obtain the current odd position standard value of the odd position subsequence. Based on the actual values ​​in the odd position subsequence that are the same as the current odd position index, error separation is performed on the current odd position standard value to obtain the prediction difference sequence of the power transmission cable.

[0010] Preferably, the formula for calculating the standard value of the current odd position is as follows: ; In the formula, The standard value for the current odd position. The index number is the position of the odd-numbered subsequence. The width of the even-numbered position neighborhood on one side of the even-numbered position subsequence. For the preset summation index variable, The first element in the fixed weight coefficient matrix Line number The element values ​​of the column, It is a natural constant.

[0011] Preferably, the process of performing residual correction on the even-numbered position subsequence based on the predicted difference sequence to obtain the scale coefficient sequence of the power transmission cable is as follows: The discharge pulse characteristic waveform is decomposed into time and frequency to obtain the energy distribution characteristics of the discharge pulse characteristic waveform, and the energy distribution characteristics are superimposed to obtain the total energy value of the discharge pulse characteristic waveform. Based on the length of the even-position subsequence, the total energy value is mean-mapped to obtain the residual correction coefficient corresponding to the even position in the even-position subsequence; Based on the predicted difference sequence, the residual correction coefficient is subjected to difference weighted modulation to obtain the correction increment of the current even position in the even position subsequence; Based on the current even-position value of the even-position subsequence, the correction increment is offset-corrected to obtain the scale coefficient value of the current even-position. Based on the index order in the even-numbered position subsequence, the scale coefficient values ​​are reconstructed to obtain the scale coefficient sequence of the power transmission cable.

[0012] Preferably, the process of attenuation compensation is performed on the predicted difference sequence and the scale coefficient sequence based on the multi-band signal attenuation coefficient, and coefficient inversion is performed on the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable, as follows: Based on the multi-band signal attenuation coefficient, attenuation compensation is performed on the prediction difference sequence and the scale coefficient sequence to obtain the compensated prediction difference sequence and the compensated scale coefficient sequence of the scale coefficient sequence. The energy distribution characteristics of the discharge pulse characteristic waveform are scaled to obtain the update coefficients of the discharge pulse characteristic waveform; The coefficient values ​​of the compensated prediction difference sequence are inversely coupled with the negative numbers of the updated coefficients to obtain the updated back-addition values ​​of the corresponding coefficient values ​​in the compensated prediction difference sequence. The updated back-addition values ​​are then superimposed on the corresponding positions of the compensated scale coefficient sequence to obtain the updated scale coefficient sequence of the compensated scale coefficient sequence. Based on the fixed weight coefficient matrix, subsequence inversion is performed on the updated scale coefficient sequence to obtain the original scale even-number position subsequence and the original scale odd-number position subsequence of the filtered signal sequence. Based on the odd-even order of the sampling point position index, the original scale even-numbered position subsequence and the original scale odd-numbered position subsequence are alternately merged to obtain the noise-reduced discharge pulse sequence of the power transmission cable.

[0013] Preferably, the process of attenuating the predicted difference sequence and the scaling factor sequence according to the multi-band signal attenuation coefficient to obtain the compensated predicted difference sequence and the compensated scaling factor sequence is as follows: The attenuation coefficient of the multi-band signal is interpreted by a multiplier parameter to obtain the attenuation multiplier value of the multi-band signal attenuation coefficient; Based on the temporal order of the predicted difference sequence, the predicted difference sequence is divided into sub-bands to obtain multiple sub-band coefficient segments of the predicted difference sequence. The coefficient values ​​in the plurality of sub-band coefficient segments and the attenuation ratio values ​​are adjusted by frequency band gain to obtain the compensated prediction difference sub-band segments of the prediction difference sequence; The compensated prediction difference sub-bands are fused to obtain the compensated prediction difference sequence of the prediction difference sequence. Based on the frequency sub-band division method in the compensated predicted difference sequence, the scaling coefficient sequence is divided into multiple sub-band coefficient segments of the scaling coefficient sequence; The coefficient values ​​in the sub-band coefficient segment are scaled and corrected by the attenuation ratio value to obtain the compensated scale coefficient sub-band segment of the scale coefficient sequence. The compensated scale coefficient sub-band segment is then reconstructed in the frequency domain to obtain the compensated scale coefficient sequence of the scale coefficient sequence.

[0014] The present invention also provides a power transmission cable partial discharge detection system based on multiphysics optimization, the system comprising: A bandpass filter module is used to perform frequency limiting parameter analysis on the discharge pulse characteristic waveform based on the multi-physics prior parameter set of the power transmission cable. The multi-physics prior parameter set includes discharge pulse characteristic waveform and multi-band signal attenuation coefficient. The lower cutoff frequency and upper cutoff frequency of the discharge pulse characteristic waveform are obtained by performing bandpass filtering on the original mixed signal sequence of the power transmission cable based on the lower cutoff frequency and the upper cutoff frequency to obtain the filtered signal sequence of the power transmission cable. The gradient weight module is used to perform discrete difference mapping on the discharge pulse characteristic waveform to obtain the neighboring point amplitude gradient of the discharge pulse characteristic waveform, and to construct the fixed weight coefficient matrix of the power transmission cable using the neighboring point amplitude gradient. The parity splitting module is used to determine the parity splitting start offset of the power transmission cable with the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, and to split the filtered signal sequence into even and odd positions based on the parity splitting start offset to obtain the even position subsequence and the odd position subsequence of the filtered signal sequence. The weighted prediction module is used to perform weighted fusion of the even-position subsequence based on the weight value of the current odd position in the fixed weight coefficient matrix, and to determine the prediction difference sequence of the power transmission cable based on the fused current odd position standard value and the actual value of the odd position subsequence. The residual correction module is used to perform residual correction on the even-numbered position subsequence based on the predicted difference sequence to obtain the scale coefficient sequence of the power transmission cable. The compensation inversion module is used to perform attenuation compensation on the predicted difference sequence and the scale coefficient sequence according to the multi-band signal attenuation coefficient, and to perform coefficient inversion on the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable.

[0015] As can be seen from the above technical solution, the present invention provides a method and system for detecting partial discharge of power transmission cables based on multi-physics field optimization. It relies on the prior parameters of multi-physics field to carry out hierarchical signal processing, which can accurately filter out interference noise in the original mixed signal, completely preserve the inherent waveform details of the discharge pulse, and refine the signal decomposition accuracy by customizing the weight matrix of waveform amplitude gradient and splitting the signal subsequence according to waveform length, so as to ensure that the split odd and even subsequences completely carry the feature information corresponding to the original signal.

[0016] By combining weighted value selection, residual correction and multi-band attenuation compensation, the amplitude loss caused by cable signal transmission is specifically compensated. The noise-reduced discharge pulse sequence obtained by coefficient inversion reconstruction can completely restore the original waveform characteristics of partial discharge. Based on this pulse sequence, the accuracy of cable partial discharge information acquisition can be steadily improved. Attached Figure Description

[0017] Other objects and results of the invention will become more apparent and readily understood by referring to the following description taken in conjunction with the accompanying drawings, and with a more complete understanding of the invention. In the drawings: Figure 1 This is a flowchart illustrating a method for detecting partial discharge in power transmission cables based on multiphysics optimization, according to an embodiment of the present invention. Figure 2 This is a functional block diagram of a power transmission cable partial discharge detection system based on multiphysics field optimization according to an embodiment of the present invention. Detailed Implementation

[0018] Existing partial discharge detection methods for power transmission cables do not incorporate prior information from multiple physics fields, making it difficult to specifically eliminate clutter interference and easily obscuring the details of the discharge pulse waveform. Furthermore, the lack of a signal decomposition and adaptive correction mechanism based on waveform characteristics makes it impossible to quantitatively compensate for the attenuation of multi-frequency band signals, resulting in high distortion of the reconstructed discharge pulse signal and insufficient accuracy in partial discharge detection.

[0019] To address the aforementioned problems, this invention provides a method and system for detecting partial discharge in power transmission cables based on multiphysics optimization. The specific embodiments of this invention will be described in detail below with reference to the accompanying drawings.

[0020] To illustrate the multiphysics-optimized method and system for detecting partial discharge in power transmission cables provided by this invention, Figure 1 An exemplary illustration is provided for a method for detecting partial discharge in power transmission cables based on multiphysics optimization according to an embodiment of the present invention; Figure 2 An exemplary illustration is provided for a power transmission cable partial discharge detection system based on multiphysics optimization according to an embodiment of the present invention.

[0021] The following description of exemplary embodiments is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques and equipment should be considered part of the specification.

[0022] Reference Figure 1 The diagram shown is a flowchart illustrating a multiphysics-optimized method for detecting partial discharge in power transmission cables according to an embodiment of the present invention. In this embodiment, the multiphysics-optimized method for detecting partial discharge in power transmission cables includes: C1. A multi-physics prior parameter set based on the power transmission cable, wherein the multi-physics prior parameter set includes discharge pulse characteristic waveform and multi-band signal attenuation coefficient, frequency limiting parameter analysis is performed on the discharge pulse characteristic waveform to obtain the lower cutoff frequency and upper cutoff frequency of the discharge pulse characteristic waveform, and bandpass filtering is performed on the original mixed signal sequence of the power transmission cable according to the lower cutoff frequency and the upper cutoff frequency to obtain the filtered signal sequence of the power transmission cable; In this embodiment of the invention, the multi-physics prior parameter set based on the power transmission cable includes a discharge pulse characteristic waveform and multi-band signal attenuation coefficients. Frequency limiting parameter analysis is performed on the discharge pulse characteristic waveform to obtain its lower and upper cutoff frequencies. Based on these lower and upper cutoff frequencies, bandpass filtering is applied to the original mixed signal sequence of the power transmission cable to obtain the filtered signal sequence. The process is as follows: The multi-physics prior parameter set and the original mixed signal sequence of the power transmission cable are obtained. The multi-physics prior parameter set includes the discharge pulse characteristic waveform and the multi-band signal attenuation coefficient. Frequency-limiting parameter analysis is performed on the multiphysics prior parameter set to obtain the lower cutoff frequency and upper cutoff frequency of the multiphysics prior parameter set; Based on the lower cutoff frequency and the upper cutoff frequency, the original mixed signal sequence is subjected to frequency window filtering to obtain the signal component of the power transmission cable, and the signal component is used as the filtered signal sequence of the power transmission cable.

[0023] The multiphysics prior parameter set was obtained by conducting standard partial discharge tests on a standard cable of the same type and specifications as the power transmission cable under test. The standard partial discharge test was conducted in an electromagnetic shielding room with a shielding effectiveness of not less than 90 dB, and a power frequency voltage consistent with the rated operating voltage of the power transmission cable under test was applied during the test.

[0024] The acquired multi-physics prior parameter set includes discharge pulse characteristic waveforms and multi-band signal attenuation coefficients.

[0025] The discharge pulse characteristic waveform is the complete time-domain waveform acquired when a standard cable generates a single standard partial discharge. The multi-band signal attenuation coefficient is the signal transmission attenuation value corresponding to each center frequency in the 10 kHz to 100 kHz frequency band for this type of standard cable.

[0026] The original mixed signal sequence was acquired by a high-frequency current sensor installed at the end of the power transmission cable to be tested.

[0027] The sampling rate of the high-frequency current sensor is set to 1 MHz, and the single acquisition time is set to 1 second.

[0028] The acquired raw mixed signal sequence contains partial discharge signals, power frequency interference signals, and environmental electromagnetic noise signals from the power transmission cable under test during operation. Discharge pulse characteristic waveforms are extracted from the multi-physics prior parameter set.

[0029] The discharge pulse characteristic waveform is transformed from the time domain to the frequency domain to obtain its frequency distribution range. The lower limit of the frequency distribution range is determined as the lower cutoff frequency of the multiphysics prior parameter set. The upper limit of the frequency distribution range is determined as the upper cutoff frequency of the multiphysics prior parameter set. A continuous frequency band is constructed using the upper and lower cutoff frequencies as boundaries. All signal components in the original mixed signal sequence that fall within this frequency band are retained. All signal components outside this frequency band are filtered out. The retained signal components are arranged according to the sampling time sequence of the original mixed signal sequence to obtain the signal component of the transmission cable. This signal component is used as the filtered signal sequence of the transmission cable.

[0030] The lower cutoff frequency and the upper cutoff frequency are derived from the discharge pulse characteristic waveform of a standard cable of the same type and specification as the power transmission cable under test, rather than being calculated in real time based on the spectrum of the currently acquired signal or fixedly set based on experience. This discharge pulse characteristic waveform reflects the inherent frequency distribution range of the standard partial discharge signal of this type of cable. Using this range as the frequency band boundary of the bandpass filter enables the filtering process to be precisely matched with the physical characteristics of the cable.

[0031] The beneficial effects include the ability to accurately define the filter frequency band that matches the partial discharge signal, specifically filtering out power frequency interference and environmental electromagnetic noise outside the frequency band of the original mixed signal sequence, fully preserving the effective components of the partial discharge signal, providing a pure basic signal for subsequent signal processing and discharge pulse reconstruction, and ensuring the accuracy and reliability of subsequent detection steps.

[0032] C2. Perform discrete difference mapping on the characteristic waveform of the discharge pulse to obtain the amplitude gradient of the neighboring points of the characteristic waveform of the discharge pulse, and construct the fixed weight coefficient matrix of the power transmission cable using the amplitude gradient of the neighboring points. In this embodiment of the invention, the process of performing discrete difference mapping on the characteristic waveform of the discharge pulse to obtain the amplitude gradient of the neighboring points of the characteristic waveform of the discharge pulse, and constructing the fixed weight coefficient matrix of the transmission cable using the amplitude gradient of the neighboring points, is as follows: Neighbor-point deviation quantization is performed on adjacent amplitude sampling points in the characteristic waveform of the discharge pulse to obtain the neighbor-point amplitude gradient sequence of the characteristic waveform of the discharge pulse. Each neighbor-point amplitude gradient in the neighbor-point amplitude gradient sequence is the amplitude difference between two adjacent amplitude sampling points. Based on the maximum absolute value of the neighboring point magnitude gradient sequence, the neighboring point magnitude gradients in the neighboring point magnitude gradient sequence are normalized to obtain the scale weight value sequence of the power transmission cable. Based on the sequential arrangement of the neighboring point amplitude gradients and the index positions of the odd-numbered sampling points in the characteristic waveform of the discharge pulse, the weight values ​​in the scale weight value sequence are allocated by odd-even projection to obtain the fixed weight coefficient matrix of the power transmission cable.

[0033] All amplitude sampling points of the discharge pulse characteristic waveform are sequentially numbered from 1 according to the acquisition time, with each number corresponding to a unique sampling point position index. Starting from the amplitude sampling point with position index 1, the amplitude difference between amplitude sampling points corresponding to every two consecutive position indices is calculated sequentially.

[0034] Each amplitude difference corresponds to a neighboring point amplitude gradient. All neighboring point amplitude gradients are arranged in the order of calculation to obtain the neighboring point amplitude gradient sequence of the discharge pulse characteristic waveform.

[0035] Iterate through all neighboring point magnitude gradients in the neighboring point magnitude gradient sequence, calculating the absolute value of each gradient. Compare the absolute values ​​of all neighboring point magnitude gradients and record the largest one as the maximum absolute value. Divide each neighboring point magnitude gradient in the sequence by this maximum absolute value to obtain the scale weight value corresponding to each gradient. Arrange all scale weight values ​​according to the order of their corresponding neighboring point magnitude gradients in the neighboring point magnitude gradient sequence to obtain the scale weight value sequence of the transmission cable.

[0036] Extract all sampling points with odd-numbered position indices from the characteristic waveform of the discharge pulse, and arrange them in ascending order of position index. Each odd-numbered sampling point corresponds to a row in a fixed weight coefficient matrix. For each odd-numbered sampling point, determine the amplitude gradient of its preceding and following neighboring points.

[0037] The amplitude gradient of the previous neighbor point is the amplitude gradient between the odd-numbered sampling point and the previous indexed sampling point, and the amplitude gradient of the next neighbor point is the amplitude gradient between the odd-numbered sampling point and the next indexed sampling point. If the odd-numbered sampling point is the first sampling point of the discharge pulse characteristic waveform, there is no amplitude gradient of the previous neighbor point, and the corresponding previous scale weight value is 0. If the odd-numbered sampling point is the last sampling point of the discharge pulse characteristic waveform, there is no amplitude gradient of the next neighbor point, and the corresponding next scale weight value is 0. The scale weight values ​​corresponding to the amplitude gradient of the previous neighbor point and the scale weight values ​​corresponding to the amplitude gradient of the next neighbor point are filled into the row corresponding to the odd-numbered sampling point in the fixed weight coefficient matrix, in the order of the previous one first and the next one last.

[0038] After all rows corresponding to the odd-numbered sampling points are filled, the resulting two-dimensional array is the fixed weight coefficient matrix of the power transmission cable.

[0039] The beneficial effect is that it can accurately extract the amplitude change characteristics of the characteristic waveform of the discharge pulse through discrete differential mapping, and obtain a uniform scale weight value after normalization. The fixed weight coefficient matrix constructed based on the sampling points at odd positions can accurately match the waveform change law of the discharge pulse, providing a standardized and targeted weight basis for the subsequent weighted fusion processing of signals, and ensuring the accuracy of subsequent signal prediction and correction steps.

[0040] C3. Using the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, determine the odd-even splitting start offset of the power transmission cable, and based on the odd-even splitting start offset, perform odd-even splitting on the filtered signal sequence to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence. In this embodiment of the invention, the process of determining the odd-even splitting start offset of the transmission cable using the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, and then performing odd-even splitting on the filtered signal sequence based on the odd-even splitting start offset to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence is as follows: The total number of amplitude sampling points in the characteristic waveform of the discharge pulse is taken as the waveform length of the characteristic waveform of the discharge pulse; Based on the original sampling time of the sampling points in the filtered signal sequence, the sampling points are time-series indexed to obtain the position index of the sampling points; Based on the waveform length, the parity split start offset of the filtered signal sequence is determined, wherein the parity split start offset is the integer part of the waveform length divided by two; Based on the position index and the odd-even splitting start offset, the sampling points are decoupled for odd and even positions to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence.

[0041] The number of amplitude sampling points in the characteristic waveform of the discharge pulse is counted, and this number is taken as the waveform length of the characteristic waveform of the discharge pulse. The sampling points of the filtered signal sequence are arranged in chronological order of the original acquisition time. Starting from the first sampling point, they are assigned consecutive integer numbers in chronological order. Each number corresponds to a position index of the sampling point, and the numbers start from 1 and increment. Each sampling point corresponds to a unique position index.

[0042] Divide the waveform length by 2 and take the integer part of the calculation result as the parity splitting start offset of the filtered signal sequence. The value of this offset is determined to ensure that the length of the split subsequence is consistent with the half-length of the discharge pulse characteristic waveform, matching the processing dimension of subsequent weighted fusion. Starting from the sampling point whose position index is equal to the parity splitting start offset plus 1, traverse all unskipped sampling points in the filtered signal sequence in sequence.

[0043] For each sample point traversed, calculate the difference between its position index and the starting offset of the parity split.

[0044] If the difference is odd, the amplitude of the sampled point is assigned to the odd-position subsequence. If the difference is even, the amplitude of the sampled point is assigned to the even-position subsequence. After all sampled points have been traversed, the amplitudes in the two subsequences are arranged in the order of the original sampling times to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence.

[0045] The initial offset for odd-even splitting is determined based on the waveform length of the discharge pulse characteristic waveform, rather than being arbitrarily set. This waveform length reflects the number of sampling points of the standard discharge pulse in the time domain. Using half of this length as the offset ensures that the dimensions of the even-position subsequences and odd-position subsequences after splitting precisely correspond to the row indices of the fixed weight coefficient matrix constructed in step C2 and the neighborhood range of the weighted fusion in step C4. This allows subsequent weighted fusion prediction and residual correction to perform accurate signal deviation correction based on the standard waveform.

[0046] The beneficial effects are that a unified splitting benchmark can be determined based on the inherent length of the discharge pulse characteristic waveform, ensuring that the structure of the odd and even subsequences is completely matched with the discharge pulse characteristic waveform, accurately splitting the odd and even position components of the signal and completely preserving the temporal correlation features of the signal, providing a dimensionally consistent and structurally corresponding processing basis for subsequent weighted fusion and residual correction, and improving the matching accuracy and execution efficiency of subsequent signal processing steps.

[0047] C4. Based on the weight value of the current odd position in the fixed weight coefficient matrix, the even position subsequence is weighted and fused, and the predicted difference sequence of the power transmission cable is determined based on the fused current odd position standard value and the actual value of the odd position subsequence. In this embodiment of the invention, the process of weighting and fusing the even-position subsequences according to the weight values ​​of the current odd-numbered positions in the fixed weight coefficient matrix, and determining the prediction difference sequence of the power transmission cable based on the fused standard value of the current odd-numbered positions and the actual values ​​of the odd-numbered position subsequences, is as follows: In the fixed weight coefficient matrix, the weight value of the current odd position is determined based on the index number of the current odd position and the subsequence of the odd position; The position index of the current odd position is used to locate the neighbor index, thereby obtaining the left neighbor even position index and the right neighbor even position index of the current odd position; Based on the left neighbor even position index and the right neighbor even position index, the amplitude of the even position subsequence is sampled and extracted to obtain the left neighbor even position value and the right neighbor even position value of the even position subsequence; Based on the weight values, a coefficient weighting transformation is performed on the left neighbor even position value and the right neighbor even position value to obtain the left weighted value of the left neighbor even position value and the right weighted value of the right neighbor even position value. Then, the left weighted value and the right weighted value are linearly aggregated to obtain the current odd position standard value of the odd position subsequence. Based on the actual values ​​in the odd position subsequence that are the same as the current odd position index, error separation is performed on the current odd position standard value to obtain the prediction difference sequence of the power transmission cable.

[0048] The formula for calculating the standard value of the current odd-numbered position is as follows: ; In the formula, The standard value for the current odd position. The index number is the position of the odd-numbered subsequence. The width of the even-numbered position neighborhood on one side of the even-numbered position subsequence. For the preset summation index variable, The first element in the fixed weight coefficient matrix Line number The element values ​​of the column, It is a natural constant.

[0049] Each element in the odd-position subsequence is assigned a consecutive integer index starting from 1. This index corresponds one-to-one with the row index of the fixed weight coefficient matrix, i.e., the row index of the odd-position subsequence. The element corresponds to the first element of the fixed weight coefficient matrix. OK.

[0050] It is the index number in the odd-position subsequence. The elements of the odd-position subsequence are numbered consecutively starting from 1 according to the original sampling time, and each number corresponds to a unique odd-position element. Each row of the fixed weight coefficient matrix contains two weight values, namely the left weight value and the right weight value. The left weight value is located in the first column of the row, and the right weight value is located in the second column of the row. It is the first in the fixed weight coefficient matrix Line number The element value of the column is obtained by normalizing the amplitude gradient of the neighboring points of the discharge pulse characteristic waveform and allocating it by parity projection, corresponding to the current odd position. The weight coefficients for even-numbered positions in the neighborhood. For the odd-numbered position subsequence, the index is... At the current odd position, extract the fixed weight coefficient matrix. The left and right weight values ​​of the row are used as the weight values ​​for the current odd-numbered positions.

[0051] It is the width of the even-numbered position neighborhood of the even-numbered position subsequence, and the value is 1. This value is based on the fact that the adjacent amplitude changes of the discharge pulse characteristic waveform are only affected by one even-numbered position sampling point on the left and right, which is consistent with the time-domain variation characteristics of the partial discharge pulse of the power transmission cable.

[0052] The original position index of the current odd position in the filtered signal sequence is equal to the odd-even split starting offset plus twice the current odd position index minus 1. The left neighboring even position index of the current odd position is the original position index of the current odd position minus 1, and the right neighboring even position index is the original position index of the current odd position plus 1. If the left neighboring even position index is less than the odd-even split starting offset plus 1, the left neighboring even position value is 0. If the right neighboring even position index is greater than the maximum position index of the filtered signal sequence, the right neighboring even position value is 0. The even position subsequence index corresponding to the left neighboring even position index is the integer result of the left neighboring even position index minus the odd-even split starting offset divided by 2. The even position subsequence index corresponding to the right neighboring even position index is the integer result of the right neighboring even position index minus the odd-even split starting offset divided by 2.

[0053] Extract the magnitude of the corresponding index position from the even-numbered position subsequence, and use it as the left neighbor even-numbered position value and the right neighbor even-numbered position value, respectively.

[0054] This is a pre-defined summation index variable used to iterate through all column elements of the current row in the fixed-weight coefficient matrix. The traversal ranges from 1 to twice the width of the single-sided neighborhood, covering all even-numbered neighboring positions to the left and right of the current odd-numbered position. The left weight is obtained by multiplying the left even-numbered position value by the left weight value. The right weight is obtained by multiplying the right even-numbered position value by the right weight value.

[0055] It is a natural constant with a value of 2.71828. This constant is a universal constant in the field of mathematics and is used to implement exponential weighting of neighborhood positions, so that neighborhood values ​​that are closer to the current odd position receive higher weight contributions.

[0056] Add the left weighted value to the right weighted value to get the weighted sum. Add the left weight value to the right weight value to get the weighted sum. If the weighted sum is 0, then the standard value for the current odd position is set to 0.

[0057] If the weight sum is not 0, the weighted sum is divided by the weight sum to obtain the current odd position standard value of the odd position subsequence.

[0058] It is the standard value of the current odd position, which is obtained by weighting the neighborhood values ​​of the even position subsequence with the corresponding weight values ​​of the fixed weight coefficient matrix. It is used to characterize the theoretical amplitude of the odd position signal derived from the law of discharge pulse waveform.

[0059] This formula is used to transform the discharge pulse waveform variation pattern contained in the fixed weight coefficient matrix into the theoretical calculated value of the odd position signal. By weighted summation and normalization of the even position values ​​in the neighborhood, the standard value of the current odd position that matches the characteristics of the standard discharge pulse is obtained, providing a benchmark value for the subsequent calculation of the prediction difference sequence.

[0060] Extracting the index of the odd-position subsequence The actual value is the amplitude of the sampling point corresponding to the original position index in the filtered signal sequence. Subtracting the standard value for the current odd position from this actual value yields the prediction difference for that odd position. Arranging all prediction differences sequentially according to the index order of the odd position subsequences results in the prediction difference sequence for the power transmission cable.

[0061] The beneficial effect is that it can accurately predict signals at odd positions by relying on the change law of discharge pulse waveform contained in the fixed weight coefficient matrix and combining the signal values ​​at even positions in the neighborhood. The quantified prediction difference is obtained by separating the error between the actual value and the predicted standard value, and the deviation characteristics between the signal and the standard discharge pulse are completely preserved. This provides a direct and accurate basis for the subsequent residual correction steps, ensuring the accuracy of subsequent signal denoising and reconstruction.

[0062] C5. Based on the predicted difference sequence, perform residual correction on the even-numbered position subsequence to obtain the scale coefficient sequence of the power transmission cable; In this embodiment of the invention, the process of performing residual correction on the even-numbered position subsequence based on the predicted difference sequence to obtain the scale coefficient sequence of the power transmission cable is as follows: The discharge pulse characteristic waveform is decomposed into time and frequency to obtain the energy distribution characteristics of the discharge pulse characteristic waveform, and the energy distribution characteristics are superimposed to obtain the total energy value of the discharge pulse characteristic waveform. Based on the length of the even-position subsequence, the total energy value is mean-mapped to obtain the residual correction coefficient corresponding to the even position in the even-position subsequence; Based on the predicted difference sequence, the residual correction coefficient is subjected to difference weighted modulation to obtain the correction increment of the current even position in the even position subsequence; Based on the current even-position value of the even-position subsequence, the correction increment is offset-corrected to obtain the scale coefficient value of the current even-position. Based on the index order in the even-numbered position subsequence, the scale coefficient values ​​are reconstructed to obtain the scale coefficient sequence of the power transmission cable.

[0063] The characteristic waveform of the discharge pulse is divided into consecutive time periods of equal length according to the time sequence, and the length of each time period is equal to the time interval of a single sampling point.

[0064] Calculate the square of the waveform amplitude for each time period. Arrange the square values ​​for all time periods in chronological order to obtain the energy distribution characteristics of the discharge pulse characteristic waveform. Add all the square values ​​in the energy distribution characteristics to obtain the total energy value of the discharge pulse characteristic waveform.

[0065] The total number of elements in the even-numbered position subsequence is counted, and the total energy value is divided by the total number of elements to obtain the residual correction coefficient corresponding to each even-numbered position in the even-numbered position subsequence. The predicted difference sequence has the same number of elements as the even-numbered position subsequence, and each even-numbered position corresponds to the predicted difference at the same index position in the predicted difference sequence.

[0066] Multiply the prediction difference corresponding to the current even-numbered position by the residual correction coefficient to obtain the correction increment for the current even-numbered position in the even-numbered position subsequence. Extract the amplitude of the current even-numbered position in the even-numbered position subsequence, and add this amplitude to the correction increment to obtain the scaling factor value for the current even-numbered position. Arrange the scaling factor values ​​corresponding to all positions in the even-numbered position subsequence according to their index order to obtain the scaling factor sequence of the power transmission cable.

[0067] The beneficial effects are that a unified residual correction benchmark can be determined based on the inherent energy distribution characteristics of the discharge pulse characteristic waveform, and the accurate deviation correction of the signal at each even position can be achieved by combining the predicted difference. The generated scale coefficient sequence completely preserves the temporal correlation and energy characteristics of the original signal, providing an accurate scale basis for subsequent attenuation compensation and discharge pulse reconstruction, and ensuring the fidelity and detection accuracy of the final reconstructed signal.

[0068] C6. Based on the multi-band signal attenuation coefficient, attenuation compensation is performed on the predicted difference sequence and the scale coefficient sequence, and coefficient inversion is performed on the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable.

[0069] In this embodiment of the invention, the process of attenuating the predicted difference sequence and the scaling coefficient sequence according to the multi-band signal attenuation coefficient, and then performing coefficient inversion on the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable is as follows: Based on the multi-band signal attenuation coefficient, attenuation compensation is performed on the prediction difference sequence and the scale coefficient sequence to obtain the compensated prediction difference sequence and the compensated scale coefficient sequence of the scale coefficient sequence. The energy distribution characteristics of the discharge pulse characteristic waveform are scaled to obtain the update coefficients of the discharge pulse characteristic waveform; The coefficient values ​​of the compensated prediction difference sequence are inversely coupled with the negative numbers of the updated coefficients to obtain the updated back-addition values ​​of the corresponding coefficient values ​​in the compensated prediction difference sequence. The updated back-addition values ​​are then superimposed on the corresponding positions of the compensated scale coefficient sequence to obtain the updated scale coefficient sequence of the compensated scale coefficient sequence. Based on the fixed weight coefficient matrix, subsequence inversion is performed on the updated scale coefficient sequence to obtain the original scale even-number position subsequence and the original scale odd-number position subsequence of the filtered signal sequence. Based on the odd-even order of the sampling point position index, the original scale even-numbered position subsequence and the original scale odd-numbered position subsequence are alternately merged to obtain the noise-reduced discharge pulse sequence of the power transmission cable.

[0070] The process of attenuating the predicted difference sequence and the scaling factor sequence according to the multi-band signal attenuation coefficient to obtain the compensated predicted difference sequence and the compensated scaling factor sequence is as follows: The attenuation coefficient of the multi-band signal is interpreted by a multiplier parameter to obtain the attenuation multiplier value of the multi-band signal attenuation coefficient; Based on the temporal order of the predicted difference sequence, the predicted difference sequence is divided into sub-bands to obtain multiple sub-band coefficient segments of the predicted difference sequence. The coefficient values ​​in the plurality of sub-band coefficient segments and the attenuation ratio values ​​are adjusted by frequency band gain to obtain the compensated prediction difference sub-band segments of the prediction difference sequence; The compensated prediction difference sub-bands are fused to obtain the compensated prediction difference sequence of the prediction difference sequence. Based on the frequency sub-band division method in the compensated predicted difference sequence, the scaling coefficient sequence is divided into multiple sub-band coefficient segments of the scaling coefficient sequence; The coefficient values ​​in the sub-band coefficient segment are scaled and corrected by the attenuation ratio value to obtain the compensated scale coefficient sub-band segment of the scale coefficient sequence. The compensated scale coefficient sub-band segment is then reconstructed in the frequency domain to obtain the compensated scale coefficient sequence of the scale coefficient sequence.

[0071] The multi-band signal attenuation coefficient is the signal transmission attenuation value corresponding to each center frequency in the 10 kHz to 100 kHz frequency band of the standard cable. The reciprocal of the attenuation value corresponding to each center frequency is taken to obtain the attenuation ratio value corresponding to that center frequency.

[0072] The frequency band from 10 kHz to 100 kHz is divided into 9 consecutive frequency sub-bands according to a frequency interval of 10 kHz. Each frequency sub-band corresponds to the coefficient segment of the same frequency range in the prediction difference sequence. The coefficients corresponding to each frequency sub-band are extracted in time domain order to obtain multiple sub-band coefficient segments of the prediction difference sequence.

[0073] Multiply all coefficient values ​​in each sub-band coefficient segment by the attenuation factor value of the corresponding frequency sub-band to obtain the compensated prediction difference sub-band segment of the prediction difference sequence.

[0074] Following the order of frequency from low to high, all compensated predicted difference sub-band segments are sequentially concatenated to obtain the compensated predicted difference sequence of the predicted difference sequence. Using the exact same frequency sub-band division method as the predicted difference sequence, the scaling coefficient sequence is divided into multiple sub-band coefficient segments, the same number as the predicted difference sequence.

[0075] Multiply all coefficient values ​​in each scale factor sub-band segment by the attenuation factor value of the corresponding frequency sub-band to obtain the compensated scale factor sub-band segment of the scale factor sequence.

[0076] Following the order of frequency from low to high, all compensated scaling factor sub-segments are sequentially spliced ​​together to obtain the compensated scaling factor sequence of the scaling factor sequence. The energy distribution characteristics of the discharge pulse characteristic waveform are the squared values ​​of the waveform amplitudes in each time interval arranged in chronological order. By dividing all values ​​in the energy distribution characteristics by the total energy value of the discharge pulse characteristic waveform, the update coefficients of the discharge pulse characteristic waveform are obtained.

[0077] Multiply the coefficient value at each position in the compensated predicted difference sequence by the negative of the corresponding position's update coefficient to obtain the update back-addition value at that position.

[0078] Each updated back-addition value is superimposed onto the coefficient value at the same index position in the compensated scaling coefficient sequence to obtain the updated scaling coefficient sequence of the compensated scaling coefficient sequence.

[0079] Based on the correspondence between the row indices of the fixed weight coefficient matrix and the odd-position subsequences, the updated scale coefficient sequence is split according to the parity of the indices. Elements with even indices form the original scale even-position subsequences, and elements with odd indices form the original scale odd-position subsequences.

[0080] Following the ascending order of the original sampling point location indices, elements with corresponding indices from the original scale odd-numbered position subsequence and the original scale even-numbered position subsequence are alternately arranged to obtain the noise-reduced discharge pulse sequence of the power transmission cable.

[0081] The beneficial effect is that the multi-band signal attenuation compensation is fundamentally different from the signal amplification or single-coefficient compensation in existing technologies. Existing technologies typically use a fixed attenuation coefficient to compensate for the entire signal, which cannot handle the physical fact that different frequency components attenuate to different degrees during cable transmission. By dividing the predicted difference sequence and scaling coefficient sequence into frequency sub-bands respectively, and using the pre-acquired attenuation ratio values ​​corresponding to each frequency band for compensation, the signal components of different frequency bands can achieve accurate amplitude recovery, thereby effectively eliminating waveform distortion caused by frequency band selective attenuation and improving the accuracy and reliability of partial discharge detection in power transmission cables.

[0082] As can be seen from the above embodiments, the multi-physics field-optimized partial discharge detection method for power transmission cables provided by the present invention achieves hierarchical signal processing by relying on multi-physics field prior parameters, accurately filters out environmental interference clutter and completely preserves the details of the discharge pulse waveform. Through the synergistic effect of waveform amplitude gradient customized weighting, odd and even signal splitting, residual adaptive correction and multi-band attenuation compensation, the original time domain and energy characteristics of the partial discharge signal are accurately restored, effectively improving the detection accuracy of partial discharge signals of power transmission cables, and providing a reliable basis for the early identification of partial discharge faults and safe operation monitoring of power transmission cables.

[0083] like Figure 2 The diagram shown is a functional block diagram of a power transmission cable partial discharge detection system 100 based on multiphysics optimization according to an embodiment of the present invention, including a bandpass filter module 101, a gradient weight module 102, an odd-even splitting module 103, a weighted prediction module 104, a residual correction module 105, and a compensation inversion module 106.

[0084] In this embodiment, the functions of each module are as follows: The bandpass filter module 101 is used to perform frequency limiting parameter analysis on the discharge pulse characteristic waveform based on the multi-physics prior parameter set of the power transmission cable. The multi-physics prior parameter set includes discharge pulse characteristic waveform and multi-band signal attenuation coefficient. The lower cutoff frequency and upper cutoff frequency of the discharge pulse characteristic waveform are obtained. Based on the lower cutoff frequency and the upper cutoff frequency, the original mixed signal sequence of the power transmission cable is bandpass filtered to obtain the filtered signal sequence of the power transmission cable. The gradient weight module 102 is used to perform discrete difference mapping on the discharge pulse characteristic waveform to obtain the neighboring point amplitude gradient of the discharge pulse characteristic waveform, and to construct the fixed weight coefficient matrix of the power transmission cable using the neighboring point amplitude gradient. The odd-even splitting module 103 is used to determine the odd-even splitting start offset of the power transmission cable with the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, and to perform odd-even splitting on the filtered signal sequence based on the odd-even splitting start offset to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence. The weighted prediction module 104 is used to perform weighted fusion of the even-position subsequence based on the weight value of the current odd position in the fixed weight coefficient matrix, and to determine the prediction difference sequence of the power transmission cable based on the fused current odd position standard value and the actual value of the odd position subsequence. The residual correction module 105 is used to perform residual correction on the even-numbered position subsequence based on the predicted difference sequence to obtain the scale coefficient sequence of the power transmission cable. The compensation inversion module 106 is used to perform attenuation compensation on the predicted difference sequence and the scale coefficient sequence according to the multi-band signal attenuation coefficient, and to perform coefficient inversion on the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable.

[0085] As can be seen from the above embodiments, the power transmission cable partial discharge detection system based on multi-physics field optimization provided by the present invention adopts a modular collaborative architecture. Each functional module independently executes the corresponding signal processing steps. The operating logic is clear and the execution efficiency is high. It can fully realize the automated detection of the entire process from the preprocessing of the original mixed signal to the reconstruction of the discharge pulse after noise reduction. It stably outputs accurate partial discharge detection results, providing reliable technical support for the condition monitoring and fault early warning of power transmission cables.

[0086] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0087] Finally, 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 it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for detecting partial discharge in power transmission cables based on multiphysics optimization, characterized in that, The method includes: C1. A multi-physics prior parameter set based on the power transmission cable, wherein the multi-physics prior parameter set includes discharge pulse characteristic waveform and multi-band signal attenuation coefficient, frequency limiting parameter analysis is performed on the discharge pulse characteristic waveform to obtain the lower cutoff frequency and upper cutoff frequency of the discharge pulse characteristic waveform, and bandpass filtering is performed on the original mixed signal sequence of the power transmission cable according to the lower cutoff frequency and the upper cutoff frequency to obtain the filtered signal sequence of the power transmission cable; C2. Perform discrete difference mapping on the characteristic waveform of the discharge pulse to obtain the amplitude gradient of the neighboring points of the characteristic waveform of the discharge pulse, and construct the fixed weight coefficient matrix of the power transmission cable using the amplitude gradient of the neighboring points. C3. Using the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, determine the odd-even splitting start offset of the power transmission cable, and based on the odd-even splitting start offset, perform odd-even splitting on the filtered signal sequence to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence. C4. Based on the weight value of the current odd position in the fixed weight coefficient matrix, the even position subsequence is weighted and fused, and the predicted difference sequence of the power transmission cable is determined based on the fused current odd position standard value and the actual value of the odd position subsequence. C5. Based on the predicted difference sequence, perform residual correction on the even-numbered position subsequence to obtain the scale coefficient sequence of the power transmission cable; C6. Based on the multi-band signal attenuation coefficient, attenuation compensation is performed on the predicted difference sequence and the scale coefficient sequence, and coefficient inversion is performed on the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable.

2. The method for detecting partial discharge in power transmission cables based on multiphysics optimization as described in claim 1, characterized in that, The multiphysics prior parameter set based on the power transmission cable includes discharge pulse characteristic waveforms and multi-band signal attenuation coefficients. Frequency limiting parameter analysis is performed on the discharge pulse characteristic waveform to obtain its lower and upper cutoff frequencies. Based on these frequencies, bandpass filtering is applied to the original mixed signal sequence of the power transmission cable to obtain the filtered signal sequence. The process is as follows: The multi-physics prior parameter set and the original mixed signal sequence of the power transmission cable are obtained. The multi-physics prior parameter set includes the discharge pulse characteristic waveform and the multi-band signal attenuation coefficient. Frequency-limiting parameter analysis is performed on the multiphysics prior parameter set to obtain the lower cutoff frequency and upper cutoff frequency of the multiphysics prior parameter set; Based on the lower cutoff frequency and the upper cutoff frequency, the original mixed signal sequence is subjected to frequency window filtering to obtain the signal component of the power transmission cable, and the signal component is used as the filtered signal sequence of the power transmission cable.

3. The method for detecting partial discharge in power transmission cables based on multiphysics optimization as described in claim 1, characterized in that, The process of performing discrete difference mapping on the characteristic waveform of the discharge pulse to obtain the amplitude gradient of the neighboring points of the characteristic waveform of the discharge pulse, and constructing the fixed weight coefficient matrix of the transmission cable using the amplitude gradient of the neighboring points, is as follows: Neighbor-point deviation quantization is performed on adjacent amplitude sampling points in the characteristic waveform of the discharge pulse to obtain the neighbor-point amplitude gradient sequence of the characteristic waveform of the discharge pulse. Each neighbor-point amplitude gradient in the neighbor-point amplitude gradient sequence is the amplitude difference between two adjacent amplitude sampling points. Based on the maximum absolute value of the neighboring point magnitude gradient sequence, the neighboring point magnitude gradients in the neighboring point magnitude gradient sequence are normalized to obtain the scale weight value sequence of the power transmission cable. Based on the sequential arrangement of the neighboring point amplitude gradients and the index positions of the odd-numbered sampling points in the characteristic waveform of the discharge pulse, the weight values ​​in the scale weight value sequence are allocated by odd-even projection to obtain the fixed weight coefficient matrix of the power transmission cable.

4. The method for detecting partial discharge in power transmission cables based on multiphysics optimization as described in claim 1, characterized in that, The process involves determining the odd-even splitting start offset of the power transmission cable using the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, and then performing odd-even splitting on the filtered signal sequence based on the odd-even splitting start offset to obtain even-position subsequences and odd-position subsequences of the filtered signal sequence. The process is as follows: The total number of amplitude sampling points in the characteristic waveform of the discharge pulse is taken as the waveform length of the characteristic waveform of the discharge pulse; Based on the original sampling time of the sampling points in the filtered signal sequence, the sampling points are time-series indexed to obtain the position index of the sampling points; Based on the waveform length, the parity split start offset of the filtered signal sequence is determined, wherein the parity split start offset is the integer part of the waveform length divided by two; Based on the position index and the odd-even splitting start offset, the sampling points are decoupled for odd and even positions to obtain the even-position subsequence and the odd-position subsequence of the filtered signal sequence.

5. The method for detecting partial discharge in power transmission cables based on multiphysics optimization as described in claim 1, characterized in that, The process of weighting and fusing the even-position subsequences based on the weight values ​​of the current odd-numbered positions in the fixed weight coefficient matrix, and determining the prediction difference sequence of the power transmission cable based on the fused standard value of the current odd-numbered positions and the actual values ​​of the odd-numbered position subsequences, is as follows: In the fixed weight coefficient matrix, the weight value of the current odd position is determined based on the index number of the current odd position and the subsequence of the odd position; The position index of the current odd position is used to locate the neighbor index, thereby obtaining the left neighbor even position index and the right neighbor even position index of the current odd position; Based on the left neighbor even position index and the right neighbor even position index, the amplitude of the even position subsequence is sampled and extracted to obtain the left neighbor even position value and the right neighbor even position value of the even position subsequence; Based on the weight values, a coefficient weighting transformation is performed on the left neighbor even position value and the right neighbor even position value to obtain the left weighted value of the left neighbor even position value and the right weighted value of the right neighbor even position value. Then, the left weighted value and the right weighted value are linearly aggregated to obtain the current odd position standard value of the odd position subsequence. Based on the actual values ​​in the odd position subsequence that are the same as the current odd position index, error separation is performed on the current odd position standard value to obtain the prediction difference sequence of the power transmission cable.

6. The method for detecting partial discharge in power transmission cables based on multiphysics optimization as described in claim 5, characterized in that, The formula for calculating the standard value of the current odd-numbered position is as follows: ； In the formula, The standard value for the current odd position. The index number is the position of the odd-numbered subsequence. The width of the even-numbered position neighborhood on one side of the even-numbered position subsequence. For the preset summation index variable, The first element in the fixed weight coefficient matrix Line 1 The element values ​​of the column, It is a natural constant.

7. The method for detecting partial discharge in power transmission cables based on multiphysics optimization as described in claim 1, characterized in that, The process of performing residual correction on the even-numbered position subsequence based on the predicted difference sequence to obtain the scale coefficient sequence of the power transmission cable is as follows: The discharge pulse characteristic waveform is decomposed into time and frequency to obtain the energy distribution characteristics of the discharge pulse characteristic waveform, and the energy distribution characteristics are superimposed to obtain the total energy value of the discharge pulse characteristic waveform. Based on the length of the even-position subsequence, the total energy value is mean-mapped to obtain the residual correction coefficient corresponding to the even position in the even-position subsequence; Based on the predicted difference sequence, the residual correction coefficient is subjected to difference weighted modulation to obtain the correction increment of the current even position in the even position subsequence; Based on the current even-position value of the even-position subsequence, the correction increment is offset-corrected to obtain the scale coefficient value of the current even-position. Based on the index order in the even-numbered position subsequence, the scale coefficient values ​​are reconstructed to obtain the scale coefficient sequence of the power transmission cable.

8. The method for detecting partial discharge in power transmission cables based on multiphysics optimization as described in claim 1, characterized in that, The process involves attenuation compensation of the predicted difference sequence and the scaling coefficient sequence based on the multi-band signal attenuation coefficient, followed by coefficient inversion of the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable. The procedure is as follows: Based on the multi-band signal attenuation coefficient, attenuation compensation is performed on the prediction difference sequence and the scale coefficient sequence to obtain the compensated prediction difference sequence and the compensated scale coefficient sequence of the scale coefficient sequence. The energy distribution characteristics of the discharge pulse characteristic waveform are scaled to obtain the update coefficients of the discharge pulse characteristic waveform; The coefficient values ​​of the compensated prediction difference sequence are inversely coupled with the negative numbers of the updated coefficients to obtain the updated back-addition values ​​of the corresponding coefficient values ​​in the compensated prediction difference sequence. The updated back-addition values ​​are then superimposed on the corresponding positions of the compensated scale coefficient sequence to obtain the updated scale coefficient sequence of the compensated scale coefficient sequence. Based on the fixed weight coefficient matrix, subsequence inversion is performed on the updated scale coefficient sequence to obtain the original scale even-number position subsequence and the original scale odd-number position subsequence of the filtered signal sequence. Based on the odd-even order of the sampling point position index, the original scale even-numbered position subsequence and the original scale odd-numbered position subsequence are alternately merged to obtain the noise-reduced discharge pulse sequence of the power transmission cable.

9. The method for detecting partial discharge in power transmission cables based on multiphysics optimization as described in claim 8, characterized in that, The process of attenuating the predicted difference sequence and the scaling factor sequence according to the multi-band signal attenuation coefficient to obtain the compensated predicted difference sequence and the compensated scaling factor sequence is as follows: The attenuation coefficient of the multi-band signal is interpreted by a multiplier parameter to obtain the attenuation multiplier value of the multi-band signal attenuation coefficient; Based on the temporal order of the predicted difference sequence, the predicted difference sequence is divided into sub-bands to obtain multiple sub-band coefficient segments of the predicted difference sequence. The coefficient values ​​in the plurality of sub-band coefficient segments and the attenuation ratio values ​​are adjusted by frequency band gain to obtain the compensated prediction difference sub-band segments of the prediction difference sequence; The compensated prediction difference sub-bands are fused to obtain the compensated prediction difference sequence of the prediction difference sequence. Based on the frequency sub-band division method in the compensated predicted difference sequence, the scaling coefficient sequence is divided into multiple sub-band coefficient segments of the scaling coefficient sequence; The coefficient values ​​in the sub-band coefficient segment are scaled and corrected by the attenuation ratio value to obtain the compensated scale coefficient sub-band segment of the scale coefficient sequence. The compensated scale coefficient sub-band segment is then reconstructed in the frequency domain to obtain the compensated scale coefficient sequence of the scale coefficient sequence.

10. A power transmission cable partial discharge detection system based on multiphysics optimization, characterized in that, The system is used to implement the multiphysics-optimized method for partial discharge detection of power transmission cables according to any one of claims 1-9, the system comprising: A bandpass filter module is used to perform frequency limiting parameter analysis on the discharge pulse characteristic waveform based on the multi-physics prior parameter set of the power transmission cable. The multi-physics prior parameter set includes discharge pulse characteristic waveform and multi-band signal attenuation coefficient. The lower cutoff frequency and upper cutoff frequency of the discharge pulse characteristic waveform are obtained by performing bandpass filtering on the original mixed signal sequence of the power transmission cable based on the lower cutoff frequency and the upper cutoff frequency to obtain the filtered signal sequence of the power transmission cable. The gradient weight module is used to perform discrete difference mapping on the discharge pulse characteristic waveform to obtain the neighboring point amplitude gradient of the discharge pulse characteristic waveform, and to construct the fixed weight coefficient matrix of the power transmission cable using the neighboring point amplitude gradient. The parity splitting module is used to determine the parity splitting start offset of the power transmission cable with the total number of amplitude sampling points in the characteristic waveform of the discharge pulse as the waveform length, and to split the filtered signal sequence into even and odd positions based on the parity splitting start offset to obtain the even position subsequence and the odd position subsequence of the filtered signal sequence. The weighted prediction module is used to perform weighted fusion of the even-position subsequence based on the weight value of the current odd position in the fixed weight coefficient matrix, and to determine the prediction difference sequence of the power transmission cable based on the fused current odd position standard value and the actual value of the odd position subsequence. The residual correction module is used to perform residual correction on the even-numbered position subsequence based on the predicted difference sequence to obtain the scale coefficient sequence of the power transmission cable. The compensation inversion module is used to perform attenuation compensation on the predicted difference sequence and the scale coefficient sequence according to the multi-band signal attenuation coefficient, and to perform coefficient inversion on the compensated coefficients to reconstruct the noise-reduced discharge pulse sequence of the power transmission cable.

Images