Cable magnetic field traveling wave head information extraction method, device, equipment and storage medium

Abstract

The present invention discloses a method, device, equipment, and storage medium for extracting traveling wave head information of a cable magnetic field, which is used to solve the technical problem of low accuracy of existing methods for extracting traveling wave head information of a cable magnetic field. The present invention comprises the following steps: obtaining an original cable magnetic field signal; performing wavelet multi-level decomposition on the original cable magnetic field signal to obtain wavelet domain signal components, wherein the wavelet domain signal components include a high-frequency detail signal and a low-frequency approximate signal; normalizing the high-frequency detail signal and the low-frequency approximate signal to obtain a multi-band normalized signal; calculating the time domain autocorrelation coefficient of the multi-band normalized signal to obtain a first layer of features; calculating the frequency domain autocorrelation coefficient of the wavelet domain signal components to obtain a second layer of features; fusing the first layer of features and the second layer of features to obtain a feature matrix; and extracting the traveling wave head information of the cable magnetic field according to the feature matrix.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a method, device, equipment and storage medium for extracting cable magnetic field traveling wave head information. Background Art

[0002] Power cables, as a key carrier of energy transmission in modern power systems, have a reliable operation that is directly related to the safety and stability of the power grid. However, cables operating in complex electromagnetic environments and subjected to multiple stresses over a long period of time inevitably face risks such as insulation aging, partial discharge, and even short-circuit failures. Failures can impact power quality at best, or even trigger widespread power outages, resulting in significant economic losses and social impact. Therefore, real-time monitoring of the operating status of power cables and early warning of faults are key technologies for ensuring the safe and stable operation of power systems. In recent years, the use of traveling wave signals generated by cable faults for fault location and diagnosis has garnered widespread attention. Cable fault traveling waves are transient electromagnetic waves that propagate through cable lines when a fault occurs. Their wavehead information contains rich fault characteristics. Accurately and rapidly extracting traveling wave headers is crucial for precise fault location, fault type identification, and cable insulation condition assessment. Currently, considerable progress has been made in the extraction of traveling wave headers. For example, methods based on time-domain threshold detection, wavelet transform, and morphological filtering have been applied to traveling wave header identification. These methods can, to a certain extent, achieve preliminary extraction of the wave head, laying the foundation for subsequent fault analysis. However, in actual application scenarios, the cable magnetic field signal is often interfered with by the complex electromagnetic environment. It is superimposed with multiple noise components such as background magnetic field fluctuations, environmental noise, equipment noise, and various transient interferences. This makes the traveling wave head characteristics submerged in the noise background, making it difficult to effectively identify and accurately extract them, making it difficult to meet the higher requirements of power systems for cable condition monitoring and fault diagnosis. Summary of the Invention

[0003] The present invention provides a method, device, equipment and storage medium for extracting cable magnetic field traveling wave head information, which are used to solve the technical problem of low accuracy of existing cable magnetic field traveling wave head information extraction methods.

[0004] The present invention provides a method for extracting cable magnetic field traveling wave head information, comprising:

[0005] Obtain the original signal of the cable magnetic field;

[0006] Performing wavelet multi-level decomposition on the cable magnetic field original signal to obtain wavelet domain signal components, wherein the wavelet domain signal components include high-frequency detail signals and low-frequency approximate signals;

[0007] Normalizing the high-frequency detail signal and the low-frequency approximation signal to obtain a multi-band normalized signal;

[0008] Calculating the time domain autocorrelation coefficient of the multi-band normalized signal to obtain the first layer of features;

[0009] Calculating the frequency domain autocorrelation coefficient of the wavelet domain signal component to obtain the second layer of features;

[0010] Fusing the first layer features and the second layer features to obtain a feature matrix;

[0011] The cable magnetic field traveling wave head information is extracted according to the characteristic matrix.

[0012] Optionally, the step of performing wavelet multi-level decomposition on the original cable magnetic field signal to obtain wavelet domain signal components includes:

[0013] The Daubechies wavelet odd function is used to perform wavelet multi-level decomposition on the original cable magnetic field signal to obtain the wavelet domain signal component; the formula is:

[0014]

[0015]

[0016] in, is the discrete signal obtained by sampling the original signal of the cable magnetic field, For the Level low-frequency approximation signal, For the High-frequency detail signal, is the total number of decomposition layers, is the signal length, b is the scaling factor, and c is the offset.

[0017] Optionally, the step of normalizing the high-frequency detail signal and the low-frequency approximation signal to obtain a multi-band normalized signal includes:

[0018] extracting a first-level high-frequency detail signal from the high-frequency detail signal;

[0019] Calculating the noise standard deviation based on the first-level high-frequency detail signal;

[0020] Calculating a noise threshold according to the noise standard deviation;

[0021] Obtain the absolute value of each level of high-frequency detail signal;

[0022] Calculate the denoised signal of each level of high-frequency detail signal according to the absolute value and the noise threshold;

[0023] Normalize the denoised signals corresponding to the high-frequency detail signals at each level and the low-frequency approximation signal to obtain a multi-band normalized signal.

[0024] Optionally, the multi-band normalized signal includes a multi-level frequency band signal; and the step of calculating the time domain autocorrelation coefficient of the multi-band normalized signal to obtain the first layer features includes:

[0025] Obtain the time lag value of the signal at each frequency band;

[0026] Calculating the time domain autocorrelation coefficients of the frequency band signals at each level according to the time lag value;

[0027] The time domain autocorrelation coefficients are arranged in the order of arrangement of the frequency band signals at each level to obtain the first layer features of the multi-band normalized signal.

[0028] Optionally, the step of calculating the frequency domain autocorrelation coefficient of the wavelet domain signal component to obtain the second layer features includes:

[0029] Obtaining frequency lag values ​​of each frequency band of the wavelet domain signal component;

[0030] Calculating the frequency domain autocorrelation coefficient of each frequency band according to the frequency lag value;

[0031] The frequency domain autocorrelation coefficients are arranged in the order of arrangement of the frequency band signals at each level to obtain the second layer features of the wavelet domain signal components.

[0032] Optionally, the step of fusing the first-layer features and the second-layer features to obtain a feature matrix includes:

[0033] Connect the time domain autocorrelation coefficients and frequency domain autocorrelation coefficients of the signals at each level of frequency band respectively to generate a fusion feature vector;

[0034] Connect the fused feature vectors of all frequency band signals to generate a feature matrix.

[0035] Optionally, the step of extracting cable magnetic field traveling wave header information according to the characteristic matrix includes:

[0036] Calculating the mean and standard deviation of the time domain autocorrelation coefficient;

[0037] Calculating a wave head detection threshold using the mean and standard deviation;

[0038] Traversing all the time domain autocorrelation coefficients and determining a set of candidate wave head positions in combination with the wave head detection threshold;

[0039] Calculating the time intervals between adjacent candidate wave head positions in the candidate wave head position set;

[0040] If the time interval is less than the preset minimum time interval, the candidate wave head positions with smaller amplitudes are eliminated from the adjacent candidate wave head positions to obtain a wave head position set;

[0041] Extracting amplitude information of each wave head position in the wave head position set to obtain an amplitude set;

[0042] Determine the maximum position of all the frequency domain autocorrelation coefficients as the main frequency component of the wave head;

[0043] The wave head position set, the amplitude set and the wave head main frequency component are output as cable magnetic field traveling wave head information.

[0044] The present invention also provides a device for extracting cable magnetic field traveling wave head information, comprising:

[0045] A cable magnetic field original signal acquisition module is used to acquire the cable magnetic field original signal;

[0046] A wavelet multi-level decomposition module is used to perform wavelet multi-level decomposition on the cable magnetic field original signal to obtain wavelet domain signal components, wherein the wavelet domain signal components include high-frequency detail signals and low-frequency approximate signals;

[0047] a normalization module, configured to normalize the high-frequency detail signal and the low-frequency approximation signal to obtain a multi-band normalized signal;

[0048] A first-layer feature calculation module, configured to calculate the time-domain autocorrelation coefficient of the multi-band normalized signal to obtain first-layer features;

[0049] A second-layer feature calculation module is used to calculate the frequency domain autocorrelation coefficient of the wavelet domain signal component to obtain the second-layer feature;

[0050] A fusion module, configured to fuse the first layer features and the second layer features to obtain a feature matrix;

[0051] The cable magnetic field traveling wave head information extraction module is used to extract the cable magnetic field traveling wave head information according to the characteristic matrix.

[0052] The present invention further provides an electronic device, comprising a processor and a memory:

[0053] The memory is used to store program code and transmit the program code to the processor;

[0054] The processor is configured to execute any one of the above methods for extracting cable magnetic field traveling wave head information according to instructions in the program code.

[0055] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute the cable magnetic field traveling wave head information extraction method as described in any one of the above items.

[0056] The above technical solution shows that the present invention has the following advantages: the present invention obtains the original cable magnetic field signal; performs wavelet multi-level decomposition on the original cable magnetic field signal to obtain wavelet domain signal components, which include high-frequency detail signals and low-frequency approximate signals; normalizes the high-frequency detail signals and low-frequency approximate signals to obtain a multi-band normalized signal; calculates the time domain autocorrelation coefficient of the multi-band normalized signal to obtain the first layer of features; calculates the frequency domain autocorrelation coefficient of the multi-band normalized signal to obtain the second layer of features; fuses the first layer features and the second layer features to obtain a feature matrix; and extracts the cable magnetic field traveling wave head information based on the feature matrix. This improves the accuracy of the cable magnetic field traveling wave head information extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0058] Figure 1 A flowchart of a method for extracting cable magnetic field traveling wave head information provided by an embodiment of the present invention;

[0059] Figure 2 A flow chart of a method for extracting cable magnetic field traveling wave head information provided by an embodiment of the present invention;

[0060] Figure 3 This is a structural block diagram of a cable magnetic field traveling wave head information extraction device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The embodiments of the present invention provide a method, device, equipment and storage medium for extracting cable magnetic field traveling wave head information, which are used to solve the technical problem of low accuracy of existing cable magnetic field traveling wave head information extraction methods.

[0062] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0063] See also Figure 1 , Figure 1A flowchart of the steps of a method for extracting cable magnetic field traveling wave head information provided by an embodiment of the present invention.

[0064] The present invention provides a method for extracting cable magnetic field traveling wave head information, which may specifically include the following steps:

[0065] Step 101, obtaining the original signal of the cable magnetic field;

[0066] In the embodiment of the present invention, the magnetic field signal around the cable can be collected by an electromagnetic sensor to obtain a continuous cable magnetic field original signal.

[0067] Step 102, performing wavelet multi-level decomposition on the original cable magnetic field signal to obtain wavelet domain signal components, where the wavelet domain signal components include high-frequency detail signals and low-frequency approximate signals;

[0068] After collecting the original cable magnetic field signal, the original cable magnetic field signal can be sampled to obtain a discrete original signal x[n], where n is the sampling point index. The calculation formula is as follows:

[0069]

[0070] in, is the discretized signal, n is the sampling point index, is the sampling period. Specifically, the original signal includes traveling wave signals, background magnetic field signals, environmental noise, equipment noise, transient interference signals, harmonic signals, high-frequency oscillation signals, signal changes caused by temperature or stress, and external electromagnetic coupling signals. It should be noted that the sensor's sampling frequency must satisfy the Nyquist sampling theorem to ensure undistorted signals.

[0071] Then, the discrete signal is decomposed by wavelet multi-level decomposition to obtain the wavelet domain signal components.

[0072] In one example, a multi-level wavelet decomposition is performed on a discrete original signal x[n], and a Daubechies wavelet basis function (such as db4 or db6) can be used. The multi-level wavelet decomposition formula is as follows:

[0073]

[0074]

[0075] in, is the discrete signal obtained by sampling the original signal of the cable magnetic field, For the Level low-frequency approximation signal, For the High-frequency detail signal, is the total number of decomposition layers, is the signal length, b is the scaling factor, and c is the offset.

[0076] In one example, the embodiment of the present invention preferably sets the decomposition level J=5, which is suitable for the spectrum characteristics of most cable magnetic field signals.

[0077] Specifically, the simulation experiment includes: simulating low-frequency background magnetic field changes; using high-frequency transient wave head characteristics as fault signals, such as short-circuit faults and high-impedance faults; adding Gaussian white noise to the noise signal, with a signal-to-noise ratio range of SNR=5-20dB.

[0078] Cable magnetic field signals collected from the power system cover a variety of operating conditions, such as normal operation, short circuit faults, and equipment startup;

[0079] The data sampling frequency is 1000 Hz, and wavelet decomposition is performed using decomposition levels J=4, J=5, and J=6 respectively. The Daubechies (db4) wavelet basis function is uniformly used. The Daubechies wavelet basis function (db4 or db6) is used for decomposition because it can effectively capture transient characteristics and suppress noise due to its compact support and high vanishing moment order.

[0080] The evaluation was conducted using wave head extraction accuracy, false alarm rate, signal-to-noise ratio improvement, and computational time;

[0081] The experimental steps include signal preprocessing, multi-band decomposition, extracting the characteristic matrix of each frequency band signal, wave head positioning and data recording;

[0082] The denoised signal was subjected to wavelet decomposition with J=4, J=5, and J=6 layers respectively. The experimental results are shown in Table 1-4 below:

[0083] Table 1 Wave head extraction accuracy table

[0084]

[0085] Table 2 False alarm rate table

[0086]

[0087] Table 3 Signal-to-noise ratio improvement table

[0088]

[0089] Table 4 Calculation schedule

[0090]

[0091] According to Table 1, J=5 has the highest average accuracy, which is 7.4% higher than J=4 and 3.3% higher than J=6. The accuracy of J=4 is lower, which may be due to insufficient frequency resolution, resulting in the failure to fully capture the high-frequency wave head features. The accuracy of J=6 is slightly lower than that of J=5, which may be because the excessive number of layers introduces redundant information, which in turn reduces the accuracy of feature extraction.

[0092] According to Table 2, J=5 has the lowest average false alarm rate, which is 4.3% lower than J=4 and 1.9% lower than J=6. The higher false alarm rate of J=4 may be due to insufficient noise suppression capability. The slightly higher false alarm rate of J=6 may be due to the excessive number of layers, which causes some low-frequency noise to be misidentified as wave heads.

[0093] According to Table 3, the average SNR improvement for J=5 is the largest, increasing by 2.5 dB compared to J=4 and by 1.2 dB compared to J=6. The SNR improvement for J=4 is lower, possibly because high-frequency noise is not completely removed. The SNR improvement for J=6 is slightly lower than that for J=5, possibly because some useful signals are over-smoothed due to the excessive number of layers.

[0094] According to Table 4, the average computation time for J=5 is moderate, increasing by 0.05 seconds compared to J=4 and decreasing by 0.08 seconds compared to J=6. J=4 has the shortest computation time, but its performance is poor. J=6 has the longest computation time, and its performance is not significantly better than that of J=5.

[0095] Experimental results show that J=5 shows excellent performance in terms of wave head extraction accuracy, false alarm rate, signal-to-noise ratio improvement and calculation time, so choosing a decomposition level of 5 is the most preferred choice of the present invention.

[0096] According to the sampling frequency And the decomposition layer number J, divide the frequency range of each frequency band, and divide it into multiple frequency band signals. The frequency range division formula is:

[0097]

[0098] in, For the The frequency range of the first frequency band, is the sampling frequency, is the number of the current decomposition level; for example, when the number of decomposition levels J=5, the corresponding frequency bands are divided as follows:

[0099] First level detail signal Represents high-frequency components, with a frequency range of ;

[0100] Second level detail signal Represents the mid-high frequency component, with a frequency range of ;

[0101] Third level detail signal Represents the intermediate frequency component, the frequency range is ;

[0102] Fourth level detail signal Represents the low and medium frequency components, with a frequency range of ;

[0103] Fifth-order approximation signal Represents the low-frequency component, with a frequency range of ;

[0104] Each frequency band signal corresponds to a different physical meaning. For example, the high-frequency component may contain wave head characteristics, while the low-frequency component may reflect the changes in the background magnetic field. It should be noted that in the multi-level wavelet decomposition process, since the signal after each layer of decomposition will be down-sampled, the length of each frequency band signal will gradually decrease. In order to ensure the consistency of the subsequent calculation of the time domain autocorrelation coefficient and the frequency domain autocorrelation coefficient, the present invention uses methods including but not limited to zero filling, symmetric extension, and no boundary processing to process the signal boundary to ensure that the length of all frequency band signals is consistent with the original signal length N.

[0105] Exemplarily, for the low-frequency approximate signal and high-frequency detail signal after decomposition at each layer, the signal length is restored to N by zero padding after downsampling; after the above processing, the length of all frequency band signals is maintained at N, thereby avoiding the difference in feature vector dimensions caused by inconsistent signal lengths.

[0106] Step 103: normalize the high-frequency detail signal and the low-frequency approximation signal to obtain a multi-band normalized signal;

[0107] After dividing the frequency band signals, the high-frequency detail signals and the low-frequency approximate signals can be normalized to obtain multi-band normalized signals.

[0108] In one example, step 103 may include the following sub-steps:

[0109] S31, extracting a first-level high-frequency detail signal from the high-frequency detail signal;

[0110] S32, calculating the noise standard deviation based on the first-level high-frequency detail signal;

[0111] S33, calculating a noise threshold according to the noise standard deviation;

[0112] S34, obtaining the absolute value of each level of high frequency detail signal;

[0113] S35, calculating a denoised signal of each level of high-frequency detail signal according to the absolute value and the noise threshold;

[0114] S36, normalizing the denoised signals corresponding to the high-frequency detail signals at each level to obtain a high-frequency normalized signal;

[0115] S37, normalizing the denoised signals and the low-frequency approximation signals corresponding to the high-frequency detail signals at each level to obtain a multi-band normalized signal.

[0116] In a specific implementation, after dividing the frequency band signal, a high-frequency detail signal is obtained, wherein the high-frequency detail signal includes a first-level high-frequency detail signal To the jth level high frequency detail signal , the maximum value of j is J-1, the first-level high-frequency detail signal Usually contains the most noise components, so by analyzing the first-level high-frequency detail signal Estimated noise standard deviation ; The noise standard deviation estimation formula is as follows:

[0117]

[0118] in, is the noise standard deviation, is the first-level high-frequency detail signal; it should be noted that 0.6745 here is a statistical constant used to convert the median to the standard deviation.

[0119] Noise Threshold The size of directly affects the denoising effect. If the threshold is too small, the noise may not be completely removed; if the threshold is too large, the useful features of the signal may be mistakenly deleted. In order to balance the two, a general empirical formula is used to calculate the threshold. , the calculation formula is as follows:

[0120]

[0121] in, is the noise threshold.

[0122] After determining the threshold After that, soft threshold processing is performed on each level of high-frequency detail signal. The core idea of ​​the soft threshold method is to shrink the signal amplitude while retaining the directionality of the signal. The soft threshold denoising formula is as follows:

[0123]

[0124] in, is the high-frequency detail signal after denoising, is the high-frequency detail signal, Represents a symbolic function.

[0125] After completing the soft threshold denoising, you can verify the denoising effect by observing the signal waveform to check whether the denoised signal retains the main features and whether the noise is significantly reduced;

[0126] Calculate the signal-to-noise ratio and evaluate the denoising effect by comparing the signal-to-noise ratio of the signal before and after denoising;

[0127] Spectrum analysis: Perform spectrum analysis on the signals before and after denoising to confirm whether high-frequency noise is effectively suppressed.

[0128] The low-frequency approximate signal does not need to be denoised. After the high-frequency detail signal is denoised, it is normalized with the denoised signal of the high-frequency detail signal to make its amplitude range uniform. The normalization formula is as follows:

[0129]

[0130] in, is the normalized signal, After denoising, Level band signal, is the signal mean, is the signal standard deviation.

[0131] Step 104: Calculate the time domain autocorrelation coefficient of the multi-band normalized signal to obtain the first layer of features;

[0132] After completing the normalization of the wavelet domain signal components, the time domain autocorrelation coefficient of the multi-band normalized signal can be calculated to obtain the first layer of features.

[0133] In one example, step 104 may include the following sub-steps:

[0134] S41, obtaining time lag values ​​of signals at each frequency band;

[0135] S42, calculating the time domain autocorrelation coefficients of the frequency band signals at each level according to the time lag value;

[0136] S43, arranging the time domain autocorrelation coefficients according to the arrangement order of the frequency band signals at each level to obtain the first layer features of the multi-band normalized signal.

[0137] In the embodiment of the present invention, the input signal is a multi-band normalized signal set ,in, Indicates the Normalized signal of the frequency band; according to the signal length , set the maximum time lag value :

[0138]

[0139] in, is the maximum time delay, the unit is sampling point; if the signal has obvious periodicity, the maximum time delay Cover at least one full cycle.

[0140] For each frequency band , calculate the time domain autocorrelation coefficient , the calculation formula is as follows:

[0141]

[0142] in, Indicates the The level band signal is at the hysteresis value The time domain autocorrelation coefficient under ; is the time lag value, and its value range is The numerator is the cross-covariance between the signal and its delayed version, and the denominator is used for normalization to ensure that the autocorrelation coefficient ranges between [1,1]. The time domain autocorrelation coefficient vector of all frequency bands is The signals of each frequency band are connected in order to form the first layer of features.

[0143] Step 105, calculating the frequency domain autocorrelation coefficient of the wavelet domain signal component to obtain the second layer features;

[0144] In a specific implementation, step 105 may include the following sub-steps:

[0145] S51, obtaining the frequency lag value of each frequency band of the wavelet domain signal component;

[0146] S52, calculating the frequency domain autocorrelation coefficient of each frequency band according to the frequency lag value;

[0147] S53, arranging the frequency domain autocorrelation coefficients according to the arrangement order of the frequency band signals at each level, and obtaining the second layer features of the wavelet domain signal components.

[0148] In the specific implementation, the input signal is the original frequency band signal set after wavelet decomposition ,in Indicates the High-frequency detail signal, Indicates the It should be noted that the original decomposed signal is used to preserve the original spectrum information of the signal, because denoising and normalization operations may change the spectrum structure of the signal.

[0149] Perform fast Fourier transform on each frequency band signal to obtain the frequency domain spectrum and extract the amplitude spectrum :

[0150]

[0151]

[0152] Among them, FFT is Fourier transform, For the The frequency domain spectrum of the high-frequency detail signal, For the The frequency domain spectrum of the low-frequency approximation signal, For the The amplitude spectrum of the high-frequency detail signal, For the The amplitude spectrum of the low-frequency approximation signal.

[0153] According to the spectrum length , set the maximum frequency hysteresis value , the calculation formula is as follows:

[0154]

[0155] in, Indicates the maximum frequency delay; if the spectrum has obvious periodicity, the maximum frequency delay Cover at least one full cycle.

[0156] For each frequency band , calculate the frequency domain autocorrelation coefficient , the calculation formula is as follows:

[0157]

[0158] in, Indicates the Frequency domain autocorrelation coefficient of the level frequency band signal at the lag value; is the frequency hysteresis value, and its value range is The numerator is the cross-covariance between the magnitude spectrum and its frequency-delayed version, and the denominator is used for normalization to ensure that the autocorrelation coefficient is in the range [1,1].

[0159] The frequency domain autocorrelation coefficient vector of all frequency bands The signals of each frequency band are connected in order to form the second layer of features.

[0160] Step 106: Fusing the first layer features and the second layer features to obtain a feature matrix;

[0161] In this embodiment of the present invention, step 106 may include the following sub-steps:

[0162] S61, respectively connecting the time domain autocorrelation coefficients and frequency domain autocorrelation coefficients of the frequency band signals at each level to generate a fusion feature vector;

[0163] S62, connecting the fusion feature vectors of all frequency band signals to generate a feature matrix.

[0164] In the specific implementation, for each frequency band , the time domain autocorrelation coefficient vector and frequency domain autocorrelation coefficient vector Connect to form a fusion feature vector :

[0165]

[0166] in, Indicates the The fusion feature vector of the frequency band contains the time domain and frequency domain autocorrelation information of the frequency band signal;

[0167] The fusion feature vector of all frequency bands Connect them in sequence to form a feature matrix. It should be noted that if multiple signal samples need to be processed, the feature vector of each sample is used as a row of the feature matrix to finally form the feature matrix.

[0168] Step 107: extracting cable magnetic field traveling wave header information according to the characteristic matrix.

[0169] In one example, step 107 may include the following sub-steps:

[0170] S71, calculate the mean and standard deviation of the time domain autocorrelation coefficient;

[0171] S72, calculate the wave head detection threshold using the mean and standard deviation;

[0172] S73, traversing all time domain autocorrelation coefficients and determining a set of candidate wave head positions in combination with a wave head detection threshold;

[0173] S74, calculating the time interval between adjacent candidate wave head positions in the candidate wave head position set;

[0174] S75, if the time interval is less than the preset minimum time interval, eliminating the candidate wave head positions with smaller amplitudes from the adjacent candidate wave head positions to obtain a wave head position set;

[0175] S76, extracting the amplitude information of each wave head position in the wave head position set to obtain an amplitude set;

[0176] S77, determining the maximum position of all frequency domain autocorrelation coefficients as the main frequency component of the wave head;

[0177] S78, outputting the wave head position set, amplitude set and main frequency components of the wave head as the cable magnetic field traveling wave head information.

[0178] In the specific implementation, the feature matrix contains the time domain and frequency domain autocorrelation coefficients, which can fully reflect the time series characteristics and frequency distribution characteristics of the signal.

[0179] Wave head detection includes:

[0180] According to the time domain autocorrelation coefficient in the feature matrix, the threshold of wave head detection is set :

[0181]

[0182] in, and represent the mean and standard deviation of the time domain autocorrelation coefficient respectively; It is an empirical parameter used to control the sensitivity of the threshold and is not specifically limited in the present invention. It should be noted that the selection of the threshold should be combined with the actual application scenario to avoid false positives or false negatives.

[0183] When choosing the empirical parameter k, it is usually necessary to define an objective function to evaluate the model performance. Common objective functions include: classification accuracy, mean square error, F1 score, AUC and other evaluation indicators. The goal is to find the optimal objective function. value;

[0184] In order to evaluate different The impact of the value on model performance is usually divided into a training set and a validation set. The specific method is as follows:

[0185] Empirical formulas are usually derived through analysis of experimental data. The following are the detailed steps of the derivation process:

[0186] Defining Candidates Value range, according to the problem scale and data characteristics, set the candidate A range of values. For example:

[0187]

[0188] in, is the number of training samples. Usually The maximum value does not exceed , to avoid overfitting or underfitting;

[0189] For each candidate Value, calculate the performance indicators of the model, such as classification accuracy; assuming there is Candidates value, we get a set of performance indicators:

[0190]

[0191] in, Indicates when performance indicators when .

[0192] By fitting the experimental data, we can obtain The relationship between the performance index and the performance index; Common fitting methods include: linear fitting and nonlinear fitting;

[0193] Linear fitting is used to assume and There is a linear relationship.

[0194] Nonlinear fitting is used to assume and A polynomial or other nonlinear relationship.

[0195] For example, suppose and It is a quadratic function relationship, and the following formula can be fitted:

[0196]

[0197] in, 、 、 is the fitting parameter, which is solved by the least square method.

[0198] According to the fitting formula, find the performance index Optimal Value. For example:

[0199]

[0200] in, is the optimal empirical parameter.

[0201] Iterate over the time domain autocorrelation coefficient vector of each signal sample , find all the values ​​exceeding the threshold Time point:

[0202]

[0203] in, Represents the set of candidate wave head positions.

[0204] For the candidate wave head position set Screen and remove redundant points or noise interference points:

[0205] Calculate the time interval between adjacent wave head positions. If the interval is less than the set minimum time interval , then the wave head position with larger amplitude is retained and the wave head position with smaller amplitude is eliminated, and finally the wave head position set is obtained. .

[0206] For each wave head position , extract the corresponding amplitude information:

[0207]

[0208] in, Represents a set of wave head amplitudes.

[0209] Using the frequency domain autocorrelation coefficient , analyze the main frequency components corresponding to the wave head:

[0210] Find the maximum position of the frequency domain autocorrelation coefficient , as the main frequency component of the wave head:

[0211]

[0212] It should be noted that the main frequency components reflect the spectrum characteristics of the wave head and can be used to distinguish different types of traveling wave signals.

[0213] According to the amplitude and frequency components of the wave head, the wave head is divided into the following categories:

[0214] High-frequency wave head is used to indicate that the frequency component is higher, corresponding to the fault traveling wave signal.

[0215] Low-frequency wave head is used to indicate low frequency components, corresponding to background magnetic field changes or slowly fluctuating signals.

[0216] Noise wave head is used to indicate that the amplitude is small and the frequency component is not obvious, which is caused by noise.

[0217] The wave head classification results are verified using known fault signal samples, and indicators such as classification accuracy and false alarm rate are calculated. It should be noted that the verification process can be completed using experimental data or simulation data to ensure the reliability and accuracy of the wave head extraction method.

[0218] Output wave head position set , amplitude set And the main frequency components It should be noted that this information can be directly used for subsequent fault diagnosis or signal analysis tasks.

[0219] The present invention obtains the original cable magnetic field signal; performs wavelet multi-level decomposition on the original cable magnetic field signal to obtain wavelet domain signal components, which include high-frequency detail signals and low-frequency approximate signals; normalizes the high-frequency detail signals and low-frequency approximate signals to obtain a multi-band normalized signal; calculates the time domain autocorrelation coefficient of the multi-band normalized signal to obtain the first layer of features; calculates the frequency domain autocorrelation coefficient of the multi-band normalized signal to obtain the second layer of features; fuses the first layer features and the second layer features to obtain a feature matrix; and extracts the cable magnetic field traveling wave head information based on the feature matrix. This improves the accuracy of the cable magnetic field traveling wave head information extraction.

[0220] See also Figure 2 , Figure 2 A flowchart of a method for extracting cable magnetic field traveling wave head information provided by an embodiment of the present invention.

[0221] First, the original cable magnetic field signal is acquired. Wavelet multi-level decomposition is then performed on the original signal, dividing it into frequency bands. The high-frequency signal within the frequency bands is then subjected to noise standard deviation estimation, noise threshold calculation, and soft threshold denoising to obtain denoised data. This data is then normalized with the low-frequency approximate signal to obtain a multi-band normalized signal. The time-domain autocorrelation coefficient of the multi-band normalized signal is then calculated to obtain the first-layer features. The frequency-domain autocorrelation coefficient of the wavelet frequency-domain signal components is then calculated to obtain the second-layer features. The first-layer and second-layer features are then fused to generate a feature matrix. Wavehead detection is then performed based on the feature matrix. A wavehead threshold is set to screen candidate wavehead locations. The candidate wavehead locations are then determined to meet the minimum time interval. If so, the larger wavehead location is retained. If not, redundant points or noise interference points are identified to extract the wavehead information. Finally, the wavehead location, amplitude, and frequency components are output.

[0222] See also Figure 3 , Figure 3 This is a structural block diagram of a cable magnetic field traveling wave head information extraction device provided by an embodiment of the present invention.

[0223] An embodiment of the present invention provides a device for extracting cable magnetic field traveling wave head information, comprising:

[0224] The cable magnetic field original signal acquisition module 301 is used to acquire the cable magnetic field original signal;

[0225] The wavelet multi-level decomposition module 302 is used to perform wavelet multi-level decomposition on the original cable magnetic field signal to obtain wavelet domain signal components, where the wavelet domain signal components include high-frequency detail signals and low-frequency approximate signals;

[0226] A normalization module 303 is used to normalize the high-frequency detail signal and the low-frequency approximation signal to obtain a multi-band normalized signal;

[0227] A first layer feature calculation module 304 is used to calculate the time domain autocorrelation coefficient of the multi-band normalized signal to obtain the first layer features;

[0228] The second layer feature calculation module 305 is used to calculate the frequency domain autocorrelation coefficient of the wavelet domain signal component to obtain the second layer feature;

[0229] A fusion module 306 is used to fuse the first layer features and the second layer features to obtain a feature matrix;

[0230] The cable magnetic field traveling wave header information extraction module 307 is used to extract the cable magnetic field traveling wave header information according to the characteristic matrix.

[0231] In this embodiment of the present invention, the wavelet multi-level decomposition module 302 includes:

[0232] The wavelet multi-level decomposition submodule is used to perform wavelet multi-level decomposition on the original cable magnetic field signal using Daubechies wavelet odd function to obtain the wavelet domain signal components; the formula is:

[0233]

[0234]

[0235] in, is the discrete signal obtained by sampling the original signal of the cable magnetic field, For the Level low-frequency approximation signal, For the High-frequency detail signal, is the total number of decomposition layers, is the signal length, b is the scaling factor, and c is the offset.

[0236] In this embodiment of the present invention, the normalization module 303 includes:

[0237] A first-level high-frequency detail signal extraction submodule, used to extract a first-level high-frequency detail signal from the high-frequency detail signal;

[0238] A noise standard deviation calculation submodule, used to calculate the noise standard deviation based on the first-level high-frequency detail signal;

[0239] A noise threshold calculation submodule is used to calculate the noise threshold according to the noise standard deviation;

[0240] Absolute value acquisition submodule, used to obtain the absolute value of each level of high-frequency detail signal;

[0241] A denoising signal calculation submodule is used to calculate the denoising signal of each level of high-frequency detail signal based on the absolute value and the noise threshold;

[0242] The normalization submodule is used to normalize the denoised signals and low-frequency approximation signals corresponding to the high-frequency detail signals at each level to obtain a multi-band normalized signal.

[0243] In this embodiment of the present invention, the first-layer feature calculation module 304 includes:

[0244] The time lag value acquisition submodule is used to obtain the time lag values ​​of the signals at each frequency band;

[0245] The time domain autocorrelation coefficient calculation submodule is used to calculate the time domain autocorrelation coefficients of the frequency band signals at each level according to the time lag value;

[0246] The first-layer feature calculation submodule is used to arrange the time domain autocorrelation coefficients according to the arrangement order of the frequency band signals at each level to obtain the first-layer features of the multi-band normalized signal.

[0247] In this embodiment of the present invention, the second-layer feature calculation module 305 includes:

[0248] The frequency lag value acquisition submodule is used to obtain the frequency lag value of each frequency band of the wavelet domain signal component;

[0249] The frequency domain autocorrelation coefficient calculation submodule is used to calculate the frequency domain autocorrelation coefficient of each frequency band according to the frequency lag value;

[0250] The second-layer feature calculation submodule is used to arrange the frequency domain autocorrelation coefficients according to the arrangement order of the frequency band signals at each level to obtain the second-layer features of the wavelet domain signal components.

[0251] In this embodiment of the present invention, the fusion module 306 includes:

[0252] A fusion feature vector generation submodule is used to connect the time domain autocorrelation coefficients and frequency domain autocorrelation coefficients of the frequency band signals at each level to generate a fusion feature vector;

[0253] The feature matrix generation submodule is used to connect the fusion feature vectors of all frequency band signals to generate a feature matrix.

[0254] In the embodiment of the present invention, the cable magnetic field traveling wave head information extraction module 307 includes:

[0255] The mean and standard deviation calculation submodule is used to calculate the mean and standard deviation of the time domain autocorrelation coefficient;

[0256] A wave head detection threshold calculation submodule is used to calculate the wave head detection threshold using the mean and standard deviation;

[0257] The candidate wave head position set determination submodule is used to traverse all time domain autocorrelation coefficients and determine the candidate wave head position set in combination with the wave head detection threshold;

[0258] The time interval calculation submodule is used to calculate the time interval between adjacent candidate wave head positions in the candidate wave head position set;

[0259] The wave head position set generation submodule is used to eliminate the candidate wave head positions with smaller amplitudes from the adjacent candidate wave head positions if the time interval is less than the preset minimum time interval, so as to obtain the wave head position set;

[0260] The amplitude set extraction submodule is used to extract the amplitude information of each wave head position in the wave head position set to obtain the amplitude set;

[0261] The wave head main frequency component determination submodule is used to determine the maximum position of all frequency domain autocorrelation coefficients as the wave head main frequency component;

[0262] The cable magnetic field traveling wave head information output submodule is used to output the wave head position set, amplitude set and wave head main frequency components as the cable magnetic field traveling wave head information.

[0263] An embodiment of the present invention further provides an electronic device, the device including a processor and a memory:

[0264] The memory is used to store program codes and transmit the program codes to the processor;

[0265] The processor is configured to execute the cable magnetic field traveling wave head information extraction method according to the instructions in the program code.

[0266] An embodiment of the present invention further provides a computer-readable storage medium, which is used to store program code, and the program code is used to execute the cable magnetic field traveling wave head information extraction method of the embodiment of the present invention.

[0267] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0268] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0269] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, apparatus, or computer program products. Thus, embodiments of the present invention may take the form of a fully hardware embodiment, a fully software embodiment, or an embodiment combining software and hardware. Furthermore, embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0270] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0271] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0272] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0273] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0274] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0275] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or terminal device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0276] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for extracting cable magnetic field traveling wave head information, characterized in that: include: Obtain the original signal of the cable magnetic field; Performing wavelet multi-level decomposition on the cable magnetic field original signal to obtain wavelet domain signal components, wherein the wavelet domain signal components include high-frequency detail signals and low-frequency approximate signals; Normalizing the high-frequency detail signal and the low-frequency approximation signal to obtain a multi-band normalized signal; Calculating the time domain autocorrelation coefficient of the multi-band normalized signal to obtain the first layer of features; Calculating the frequency domain autocorrelation coefficient of the wavelet domain signal component to obtain the second layer of features; Fusing the first layer features and the second layer features to obtain a feature matrix; Extracting cable magnetic field traveling wave head information according to the characteristic matrix; The multi-band normalized signal includes a multi-level frequency band signal; and the step of calculating the time domain autocorrelation coefficient of the multi-band normalized signal to obtain the first layer feature includes: Obtain the time lag value of the signal at each frequency band; Calculating the time domain autocorrelation coefficients of the frequency band signals at each level according to the time lag value; Arranging the time domain autocorrelation coefficients according to the arrangement order of the frequency band signals at each level to obtain the first layer features of the multi-band normalized signal; The step of calculating the frequency domain autocorrelation coefficient of the wavelet domain signal component to obtain the second layer features includes: Obtaining frequency lag values ​​of each frequency band of the wavelet domain signal component; Calculating the frequency domain autocorrelation coefficient of each frequency band according to the frequency lag value; The frequency domain autocorrelation coefficients are arranged in the order of arrangement of the frequency band signals at each level to obtain the second layer features of the wavelet domain signal components.

2. The method according to claim 1, characterized in that The step of performing wavelet multi-level decomposition on the original cable magnetic field signal to obtain wavelet domain signal components includes: The Daubechies wavelet odd function is used to perform wavelet multi-level decomposition on the original cable magnetic field signal to obtain the wavelet domain signal component; the formula is: ； ； in, is the discrete signal obtained by sampling the original signal of the cable magnetic field, For the Level low-frequency approximation signal, For the High-frequency detail signal, is the total number of decomposition layers, is the signal length, b is the scaling factor, and c is the offset.

3. The method according to claim 1, characterized in that The step of normalizing the high-frequency detail signal and the low-frequency approximation signal to obtain a multi-band normalized signal includes: extracting a first-level high-frequency detail signal from the high-frequency detail signal; Calculating the noise standard deviation based on the first-level high-frequency detail signal; Calculating a noise threshold according to the noise standard deviation; Obtain the absolute value of each level of high-frequency detail signal; Calculate the denoised signal of each level of high-frequency detail signal according to the absolute value and the noise threshold; Normalize the denoised signals corresponding to the high-frequency detail signals at each level and the low-frequency approximation signal to obtain a multi-band normalized signal.

4. The method according to claim 1, wherein The step of fusing the first layer features and the second layer features to obtain a feature matrix includes: Connect the time domain autocorrelation coefficients and frequency domain autocorrelation coefficients of the signals at each level of frequency band respectively to generate a fusion feature vector; Connect the fused feature vectors of all frequency band signals to generate a feature matrix.

5. The method according to claim 1, wherein The step of extracting cable magnetic field traveling wave head information according to the characteristic matrix includes: Calculating the mean and standard deviation of the time domain autocorrelation coefficient; Calculating a wave head detection threshold using the mean and standard deviation; Traversing all the time domain autocorrelation coefficients and determining a set of candidate wave head positions in combination with the wave head detection threshold; Calculating the time intervals between adjacent candidate wave head positions in the candidate wave head position set; If the time interval is less than the preset minimum time interval, the candidate wave head positions with smaller amplitudes are eliminated from the adjacent candidate wave head positions to obtain a wave head position set; Extracting amplitude information of each wave head position in the wave head position set to obtain an amplitude set; Determine the maximum position of all the frequency domain autocorrelation coefficients as the main frequency component of the wave head; The wave head position set, the amplitude set and the wave head main frequency component are output as cable magnetic field traveling wave head information.

6. A device for extracting cable magnetic field traveling wave head information, characterized in that: include: A cable magnetic field original signal acquisition module is used to acquire the cable magnetic field original signal; A wavelet multi-level decomposition module is used to perform wavelet multi-level decomposition on the cable magnetic field original signal to obtain wavelet domain signal components, wherein the wavelet domain signal components include high-frequency detail signals and low-frequency approximate signals; a normalization module, configured to normalize the high-frequency detail signal and the low-frequency approximation signal to obtain a multi-band normalized signal; A first-layer feature calculation module, configured to calculate the time-domain autocorrelation coefficient of the multi-band normalized signal to obtain first-layer features; A second-layer feature calculation module is used to calculate the frequency domain autocorrelation coefficient of the wavelet domain signal component to obtain the second-layer feature; A fusion module, configured to fuse the first layer features and the second layer features to obtain a feature matrix; A cable magnetic field traveling wave head information extraction module is used to extract the cable magnetic field traveling wave head information according to the characteristic matrix; Among them, the first-layer feature calculation module includes: The time lag value acquisition submodule is used to obtain the time lag values ​​of the signals at each frequency band; The time domain autocorrelation coefficient calculation submodule is used to calculate the time domain autocorrelation coefficients of the frequency band signals at each level according to the time lag value; The first-layer feature calculation submodule is used to arrange the time domain autocorrelation coefficients according to the arrangement order of the frequency band signals at each level to obtain the first-layer features of the multi-band normalized signal; The second-layer feature calculation module includes: The frequency lag value acquisition submodule is used to obtain the frequency lag value of each frequency band of the wavelet domain signal component; The frequency domain autocorrelation coefficient calculation submodule is used to calculate the frequency domain autocorrelation coefficient of each frequency band according to the frequency lag value; The second-layer feature calculation submodule is used to arrange the frequency domain autocorrelation coefficients according to the arrangement order of the frequency band signals at each level to obtain the second-layer features of the wavelet domain signal components.

7. An electronic device, characterized in that: The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the cable magnetic field traveling wave head information extraction method according to any one of claims 1 to 5 according to the instructions in the program code.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store program codes, and the program codes are used to execute the cable magnetic field traveling wave head information extraction method according to any one of claims 1 to 5.

Images