Power transmission line fault positioning system and method based on acoustic magnetic signal feature analysis

Abstract

The invention belongs to the technical field of electromagnetic fault positioning, and discloses a power transmission line fault positioning system and method based on acoustic magnetic signal feature analysis. The method comprises the following steps: performing linear correction processing on a magnetic field signal propagation speed of a magnetic field sensor and an ultrasonic signal propagation speed of an ultrasonic sensor through a dynamic time delay compensation mechanism, and synchronously acquiring a linearly corrected magnetic field signal and an ultrasonic signal after linear correction processing; performing strong electromagnetic interference suppression and transient pulse suppression on the linearly corrected magnetic field signal to obtain a first magnetic field signal, performing transient pulse suppression on the linearly corrected ultrasonic signal to obtain a first ultrasonic signal, and performing enhancement processing on the first ultrasonic signal and the first magnetic field signal to obtain a second magnetic field signal and a second ultrasonic signal; the method can effectively shorten the fault repair time, reduce the operation and maintenance cost, and guarantee the safe and stable operation of the power system.

Description

Technical Field

[0001] The present invention relates to the technical field of electromagnetic fault location, and more specifically, to a transmission line fault location system and method based on the analysis of acoustic magnetic signal characteristics. Background Art

[0002] When the traditional manual inspection method is applied to the fault location of transmission lines, the efficiency is low, it is difficult to quickly detect and locate the fault point, which seriously affects the rapid recovery of power grid faults.

[0003] Chinese Patent Application No. CN119044676A discloses a method for locating underground cable faults based on acoustic magnetic feature matching analysis and a ground penetrating sensing system. The steps include: Step 1: Generate and collect acoustic magnetic signals; Step 2: Signal preprocessing; Step 3: Calculate the distance to the fault point; Step 4: Repeat Steps 1 to 4 multiple times, and calculate the position coordinates of the fault point through the calculation results of multiple groups of distances to the fault point. This invention realizes the automatic calculation of the position of underground cable fault points through acoustic magnetic acquisition signals, reduces the labor cost of locating fault points, and greatly improves the efficiency and accuracy of cable fault location.

[0004] Although the above method can meet most scenarios, through the research and practical application of the above method and the existing technology, it is found that the above method and the existing technology have at least the following partial defects:

[0005] There are various electromagnetic devices and signals in underground cables, and their electromagnetic environment is more complex, making it difficult to extract effective fault signals from the complex electromagnetic background; there are also various pipelines and other cable facilities, which interfere with the magnetic field signal and sound signal collected by the sensing system, resulting in inaccurate fault location.

[0006] In view of this, the present invention proposes a transmission line fault location system and method based on the analysis of acoustic magnetic signal characteristics to solve the above problems. Summary of the Invention

[0007] In order to overcome the above defects of the prior art and to achieve the above object, the present invention provides the following technical solution: A transmission line fault location method based on the analysis of acoustic magnetic signal characteristics, including the following steps:

[0008] Perform linear correction processing on the propagation speed of the magnetic field signal of the magnetic field sensor and the propagation speed of the ultrasonic signal of the ultrasonic sensor through a dynamic time delay compensation mechanism, and synchronously collect the linearly corrected magnetic field signal and ultrasonic signal after linear correction processing;

[0009] Suppress strong electromagnetic interference and transient pulses from the linearly corrected magnetic field signal to obtain the first magnetic field signal, suppress transient pulses from the linearly corrected ultrasonic signal to obtain the first ultrasonic signal, and perform enhancement processing on the first ultrasonic signal and the first magnetic field signal to obtain the second magnetic field signal and the second ultrasonic signal;

[0010] Extract features from the second magnetic field signal and the second ultrasonic signal to obtain magnetic field features and ultrasonic features;

[0011] Use the ultrasonic features as the input of the fault identification model to obtain the mechanical vibration mode and the corresponding fault type;

[0012] Use the magnetic field features and ultrasonic features as the input of the dual-stream residual attention network to obtain the fused joint feature map, and fuse the joint feature maps of three adjacent monitoring terminals through the D-S evidence theory to calculate and obtain the three-dimensional coordinates of the fault point.

[0013] Further, the method for linearly correcting the propagation speed of the magnetic field signal of the magnetic field sensor and the propagation speed of the ultrasonic signal of the ultrasonic sensor includes:

[0014] Calculate the propagation speed of the magnetic field signal based on the relative magnetic permeability of the wire material and the relative node constant of the wire material;

[0015] Calculate the propagation speed of the ultrasonic signal based on the bulk modulus of the wire and the density of the wire material;

[0016] For different wire types, linearly correct the propagation speed of the magnetic field signal and the propagation speed of the ultrasonic signal in combination with the temperature correction coefficient and humidity correction coefficient of the wire.

[0017] Further, the training method of the dual-stream residual attention network includes:

[0018] Pre-collect R groups of training data. The training data includes acoustic-magnetic features and the corresponding joint feature maps. The acoustic-magnetic features include magnetic field features and ultrasonic features;

[0019] Taking the acousto-magnetic features as the input of the dual-stream residual attention network and the joint feature map as the output of the dual-stream residual attention network, with the goal of minimizing the error between the output joint feature map and the actual joint feature map, the network parameters of the dual-stream residual attention network are optimized by a nature-inspired optimization algorithm to obtain the network parameters corresponding to the minimum error between the joint feature map output by the dual-stream residual attention network and the actual joint feature map, and the dual-stream residual attention network constructed with the corresponding network parameters is used as the trained dual-stream residual attention network; the dual-stream residual attention network includes a magnetic field branch and an ultrasonic wave branch, the magnetic field branch includes an NG-layer ResNet and a channel attention module, the ultrasonic wave branch includes an NM-layer CNN+LSTM and a spatial attention module, and the outputs of the magnetic field branch and the ultrasonic wave branch are fused through a cross-attention mechanism to obtain the output joint feature map; the loss function of the dual-stream residual attention network includes a classification loss composed of cross entropy and a localization loss composed of mean square error;

[0020] The method for obtaining the three-dimensional coordinates of the fault point includes:

[0021] Based on the joint feature map, distance inversion is performed in combination with the fault location model to obtain the fault distance;

[0022] The fault distances calculated by three adjacent monitoring terminals are fused through the D-S evidence theory to calculate the three-dimensional coordinates of the fault point.

[0023] Further, the method for obtaining the first magnetic field signal includes:

[0024] Based on the transfer function of the adaptive notch filter, power frequency harmonic filtering is performed to obtain the filtered magnetic field signal;

[0025] Using the wavelet packet coefficients after threshold shrinkage processing to perform wavelet packet reconstruction on the filtered magnetic field signal to obtain the first magnetic field signal.

[0026] Further, the method for obtaining the wavelet packet coefficients after threshold shrinkage processing includes:

[0027] Select the wavelet basis function and the corresponding decomposition level;

[0028] According to the selected wavelet basis function and decomposition level, wavelet packet decomposition is performed on the filtered magnetic field signal to obtain wavelet packet coefficients;

[0029] With the goal of minimizing the weighted sum of the signal-to-noise ratio and the mean square error, a wavelet threshold is obtained by optimization based on a nature-inspired optimization algorithm;

[0030] The wavelet packet coefficients with absolute values less than the wavelet threshold are set to zero, and the wavelet packet coefficients with absolute values greater than or equal to the wavelet threshold are shrunk towards zero to obtain the wavelet packet coefficients after threshold shrinkage processing.

[0031] Further, the method for obtaining the second magnetic field signal and the second ultrasonic signal includes:

[0032] Dividing the collected acousto-magnetic signals into a training set, a validation set, and a test set according to a preset ratio; the acousto-magnetic signals include a first ultrasonic signal and a first magnetic field signal;

[0033] Selecting a U-Net structure as the generator, using a random noise vector as the input of the generator, and obtaining a generated acousto-magnetic signal having the same size and number of channels as the acousto-magnetic signal; selecting a convolutional neural network as the discriminator architecture, using the acousto-magnetic signal and the generated acousto-magnetic signal as the inputs of the discriminator, and obtaining the probability that the input data is the acousto-magnetic signal;

[0034] Randomly extracting M groups of acousto-magnetic signals from the training set, labeling them as 1, indicating real acousto-magnetic signals; the generator generates corresponding generated acousto-magnetic signals according to the input random noise vector, labeling them as 0, indicating generated acousto-magnetic signals; inputting the acousto-magnetic signals and the generated acousto-magnetic signals into the discriminator respectively, and the discriminator outputs a discrimination result; according to the output of the discriminator and the labels corresponding to the acousto-magnetic signals and the generated acousto-magnetic signals, aiming at minimizing the value of the loss function, optimizing the parameters of the generator and the discriminator through a nature-inspired optimization algorithm; wherein, the loss function is the sum of the generator loss and the discriminator loss, and the generator loss fuses the difference term between the generated acousto-magnetic signal and the real acousto-magnetic signal, the Maxwell equation constraint and the acoustic wave equation constraint; the discriminator loss is the cross-entropy loss; repeating the step of selecting acousto-magnetic signals for discrimination, and performing multiple rounds of iterative training until the generative adversarial network reaches a convergence state;

[0035] Inputting the acousto-magnetic signal into the generator of the generative adversarial network corresponding to the convergence state to generate an enhanced acousto-magnetic signal, and using the enhanced acousto-magnetic signal as the acousto-magnetic signal after enhancement processing, that is, the second magnetic field signal and the second ultrasonic signal.

[0036] Further, the method for obtaining the magnetic field characteristics includes:

[0037] Calculating the traveling wave mutation points of the second magnetic field signal as mutation point features;

[0038] Detecting the second magnetic field signal based on a magnetic field mutation point detection threshold; traversing the second magnetic field signal, when the difference between the magnetic field signals at adjacent time points exceeds the magnetic field mutation point detection threshold, extracting the corresponding magnetic field signal difference as the difference feature, and extracting the corresponding two adjacent time points for splicing as the time point feature;

[0039] Extracting the harmonic distortion rate based on the voltage amplitude of the second magnetic field signal as the distortion rate feature;

[0040] Extracting the spectral centroid based on the spectral value as the centroid feature;

[0041] Concatenate S mutation point features, S difference features, S time point features, S distortion rate features and S centroid features as the magnetic field features, where features less than S are padded with zeros.

[0042] Furthermore, the method for obtaining the ultrasonic features includes:

[0043] Step 1: Obtain the second ultrasonic signal g, decompose g into k modal components, and each modal component has a corresponding central frequency; construct a constrained variational model;

[0044] Step 2: Introduce Lagrange multipliers and quadratic penalty terms to transform the constrained variational problem into an unconstrained augmented Lagrangian function;

[0045] Step 3: Use the alternating direction method of multipliers to iteratively solve the augmented Lagrangian function;

[0046] Step 4: Update the modal functions;

[0047] Step 5: Update the central frequencies;

[0048] Step 6: Update the Lagrange multipliers;

[0049] Step 7: Repeat Step 4 - Step 6 until the change in each modal function is less than the convergence tolerance in two consecutive iterations, then stop the iteration;

[0050] Step 8: Based on Step 1 - Step 7, obtain K modal components, extract the time domain features and frequency domain features of each modal component. The time domain features include amplitude features and time features. The amplitude features are the mean amplitude, amplitude variance, maximum amplitude and minimum amplitude of the modal component, and the time features are the rise time, fall time and pulse width of the modal component; the frequency domain features include frequency features and energy features. The frequency features are the peak frequency, central frequency and bandwidth of the modal component, and the energy feature is the energy distribution of the modal component in different frequency bands; concatenate the time domain features and frequency domain features to obtain the ultrasonic features.

[0051] Furthermore, the method for obtaining the time features includes:

[0052] Preset an upper threshold and a lower threshold, traverse the corresponding modal component, obtain the first signal point greater than the lower threshold, continue to traverse the corresponding modal component from the first signal point, obtain the second signal point greater than the upper threshold, and calculate the product of the time difference between the second signal point and the first signal point and the sampling period as the rise time;

[0053] Traverse backward from the end position of the corresponding modal component to obtain the third signal point that is less than the upper threshold value. Continue to traverse the corresponding modal component from the third signal point to obtain the fourth signal point that is less than the lower threshold value. Calculate the product of the time difference between the fourth signal point and the third signal point and the sampling period as the falling time;

[0054] Obtain the maximum value and the minimum value of the modal component, calculate the intermediate value of the maximum value and the minimum value. Traverse the corresponding modal component to obtain the fifth signal point that is greater than the intermediate value. Continue to traverse the corresponding modal component from the fifth signal point to obtain the sixth signal point that is less than the upper threshold value. Calculate the product of the difference between the sixth signal point and the fifth signal point and the sampling period as the pulse width;

[0055] The methods for obtaining energy features include:

[0056] Calculate the sum of the energies corresponding to all frequency domain indices in the preset frequency band of the modal component, where the energy is the square of the amplitude.

[0057] A transmission line fault location system based on acousto-magnetic signal feature analysis, implementing the transmission line fault location method based on acousto-magnetic signal feature analysis, includes:

[0058] Correction acquisition module: Perform linear correction processing on the propagation speed of the magnetic field signal of the magnetic field sensor and the propagation speed of the ultrasonic signal of the ultrasonic sensor through a dynamic time delay compensation mechanism, and synchronously acquire the linearly corrected magnetic field signal and ultrasonic signal after the linear correction processing;

[0059] Signal processing module: Suppress strong electromagnetic interference and transient pulses for the linearly corrected magnetic field signal to obtain the first magnetic field signal, suppress transient pulses for the linearly corrected ultrasonic signal to obtain the first ultrasonic signal, and perform enhancement processing on the first ultrasonic signal and the first magnetic field signal to obtain the second magnetic field signal and the second ultrasonic signal;

[0060] Feature extraction module: Extract features from the second magnetic field signal and the second ultrasonic signal to obtain magnetic field features and ultrasonic features;

[0061] Fault identification module: Use the ultrasonic features as the input of the fault identification model to obtain the mechanical vibration mode and the corresponding fault type;

[0062] Fault location module: Use the magnetic field features and ultrasonic features as the input of the dual-stream residual attention network to obtain the fused joint feature map. Fuse the joint feature maps of three adjacent monitoring terminals through the D-S evidence theory, and calculate to obtain the three-dimensional coordinates of the fault point.

[0063] The technical effects and advantages of the transmission line fault location system and method based on acousto-magnetic signal feature analysis of the present invention:

[0064] The present invention combines physical drive and data drive technologies to achieve high-precision fault location. By utilizing the complementary characteristics of acoustic and magnetic signals, where the magnetic field signal reflects the mutation of fault current and the ultrasonic signal reflects the mechanical vibration characteristics, the effective magnetic field signal is extracted through strong electromagnetic interference suppression and transient pulse suppression. Additionally, the propagation speed difference is corrected through a dynamic time delay compensation mechanism to effectively eliminate environmental interference errors. At the same time, the loss function of the generator is constrained by introducing Maxwell's equations and acoustic wave equations to ensure that the synthesized signal conforms to physical laws and improve the reliability of feature extraction. Through energy-weighted fusion and an attention mechanism to dynamically allocate weights, a two-stream residual attention network is constructed. Combining dynamic calibration of the attenuation coefficient and the distance calculation formula significantly improves the location accuracy under complex working conditions. The present invention breaks through the limitations of low efficiency in traditional manual inspection and susceptibility to interference of a single signal, realizes rapid location and accurate identification of the fault point, provides key technical support for the intelligent operation and maintenance of the power grid, can effectively shorten the fault repair time, reduce the operation and maintenance cost, and ensure the safe and stable operation of the power system. Brief Description of the Drawings

[0065] Figure 1 It is a schematic flow chart of the transmission line fault location method based on acoustic and magnetic signal feature analysis of the present invention;

[0066] Figure 2 It is a schematic structural diagram of the two-stream residual attention network of the present invention;

[0067] Figure 3 It is a structural diagram of the transmission line fault location system based on acoustic and magnetic signal feature analysis of the present invention;

[0068] Figure 4 It is a schematic diagram of the interface of the correction acquisition module of the transmission line fault location system based on acoustic and magnetic signal feature analysis of the present invention;

[0069] Figure 5 It is a schematic diagram of the interface of the signal processing module of the transmission line fault location system based on acoustic and magnetic signal feature analysis of the present invention. Detailed Embodiments

[0070] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0071] Embodiment 1

[0072] Please refer to Figure 1As shown in the figure, the method for fault location of transmission lines based on acoustic and magnetic signal feature analysis in this embodiment includes the following steps:

[0073] Perform linear correction processing on the propagation speed of the magnetic field signal of the magnetic field sensor and the propagation speed of the ultrasonic signal of the ultrasonic sensor through a dynamic time delay compensation mechanism, and synchronously collect the linearly corrected magnetic field signal and ultrasonic signal after the linear correction processing;

[0074] The method for performing linear correction processing on the propagation speed of the magnetic field signal of the magnetic field sensor and the propagation speed of the ultrasonic signal of the ultrasonic sensor includes:

[0075] Calculate the propagation speed of the magnetic field signal based on the relative magnetic permeability of the wire material and the relative node constant of the wire material; such as the propagation speed of the magnetic field signal where c is the speed of light; μ r is the relative magnetic permeability of the wire material; ∈ r is the relative node constant of the wire material;

[0076] Calculate the propagation speed of the ultrasonic signal based on the bulk modulus of the wire and the density of the wire material; such as the propagation speed of the ultrasonic signal where VW is the bulk modulus of the wire; ρ is the density of the wire material;

[0077] For different wire types, perform linear correction on the propagation speed of the magnetic field signal and the ultrasonic signal in combination with the temperature correction coefficient and humidity correction coefficient of the wire: v adj = v base ×(1 + β×ΔWD + γ×ΔSD); where v adj is the standard propagation speed of the dynamically adjusted magnetic field signal or ultrasonic signal; β is the temperature correction coefficient; ΔWD is the temperature change value; γ is the humidity correction coefficient; ΔSD is the humidity change value; v base is the propagation speed of the magnetic field signal or ultrasonic signal.

[0078] The dynamic time delay compensation mechanism performs linear correction processing on the propagation speed of the magnetic field and ultrasonic signals, which can effectively compensate for the deviation caused by the influence of the signal propagation speed by the environment (such as temperature, humidity, etc.), ensure the time synchronization of the signals collected by different sensors, and avoid the error in the judgment of the fault time caused by the speed difference; and improve the accuracy of fault location. Based on the accurately corrected propagation speed, combined with the time difference of the acoustic and magnetic signals reaching different sensors, the position of the fault point can be calculated more accurately, reducing the positioning error range, and providing a reliable guarantee for the rapid repair and stable operation of the transmission line.

[0079] Magnetic field signals can sensitively capture the magnetic field changes generated by the sudden current mutation during a fault. Characteristics such as the mutation moment and the high-frequency harmonic distortion rate can be used to judge the fault type, providing key parameters for the positioning model, such as the attenuation coefficient, etc. Ultrasonic signals are closely related to the mechanical vibrations caused by faults. By analyzing characteristics such as the vibration energy integral and the main frequency offset, the fault location can be assisted in determination. Combining the two can utilize the differences in their propagation characteristics for spatio-temporal correction. Through methods such as energy weighted fusion in the later stage, the distance to the fault point can be comprehensively calculated, which can greatly improve the accuracy and reliability of fault location, providing strong support for the rapid repair and safe operation of transmission lines. Collecting the linearly corrected magnetic field signals and ultrasonic signals can effectively compensate for the time delay caused by the difference in propagation speed, making the two types of signals more accurately corresponding in time, so as to more accurately determine the fault occurrence moment; the corrected signals can reduce the distance calculation deviation caused by speed errors. Combining the respective advantages of acoustic and magnetic signals, based on accurate signal characteristics and propagation time differences, the fault point location can be calculated more precisely. In addition, these corrected signals can also enhance the recognition of fault characteristics, helping to extract effective fault information from complex environmental interferences, providing a more reliable basis for fault type discrimination, and improving the overall accuracy and reliability of transmission line fault location.

[0080] Perform strong electromagnetic interference suppression and transient pulse suppression on the linearly corrected magnetic field signal to obtain the first magnetic field signal, perform transient pulse suppression on the linearly corrected ultrasonic signal to obtain the first ultrasonic signal, and perform enhancement processing on the first ultrasonic signal and the first magnetic field signal to obtain the second magnetic field signal and the second ultrasonic signal;

[0081] The method for obtaining the first magnetic field signal includes:

[0082] Perform power frequency harmonic filtering processing based on the transfer function H(s) of the adaptive notch filter to obtain the filtered magnetic field signal; such as the transfer function where z is a complex variable; Y(z) is the output signal; X(z) is the input signal; μ is the filter parameter, which can be optimized through a nature-inspired optimization algorithm; ω0 is the notch frequency, ω0 = 2π × 50Hz;

[0083] In a discrete-time system, z -1 is the delay operator, z -1 Z(z) = x(n - 1), z -2 X(z) = x(n - 2); n is the sampling moment; x(n - 1) is the input signal corresponding to the sampling moment n - 1; x(n - 2) is the input signal corresponding to the sampling moment n - 2;

[0084] Obtain Y(z)(1 - 2μcos(ω0)z -1 + μ 2 z -2) = X(z)(1 - 2cos(ω0)z -1 + z -2 )); Converting the above equation to the time domain, the difference equation y(n) = 2μcos(ω0)y(n - 1) - μ 2 y(n - 2) + x(n) - 2cos(ω0)x(n - 1) + x(n - 2) is obtained; y(n) is the output signal corresponding to the sampling time n; y(n - 1) is the output signal corresponding to the sampling time n - 1, and x(n) is the input signal corresponding to the sampling time n;

[0085] Calculate the error e(n) = d(n) - y(n) between the output of the filter and the desired output through the least mean square algorithm; where d(n) is the desired output;

[0086] Update μ(n + 1) = μ(n) + α × e(n) × x(n) according to the LMS algorithm; where μ(n + 1) is the filter parameter corresponding to the sampling time n + 1; μ(n) is the filter parameter corresponding to the sampling time n; α is the step size factor used to control the speed of adaptive adjustment and can be obtained by optimizing through a nature-inspired optimization algorithm;

[0087] Take the obtained output signal y(n) as the magnetic field signal after filtering processing;

[0088] Use the wavelet packet coefficients after threshold shrinkage processing to perform wavelet packet reconstruction on the filtered magnetic field signal to obtain the first magnetic field signal.

[0089] Further suppress strong electromagnetic interference and transient pulses for the linearly corrected magnetic field signal, which can make the first magnetic field signal purer. Suppressing strong electromagnetic interference can avoid the interference of the surrounding complex electromagnetic environment on the magnetic field signal and ensure that the signal truly reflects the magnetic field characteristics generated by the transmission line fault, such as accurately capturing information such as the fault mutation moment and the high-frequency harmonic distortion rate. Transient pulse suppression can eliminate the short-term abnormal fluctuations in the signal and make the magnetic field signal characteristics more stable. The first magnetic field signal obtained in this way can provide more accurate parameters for fault location, improve the accuracy of fault type discrimination, and further improve the accuracy and reliability of the fault location model.

[0090] The method for obtaining the wavelet packet coefficients after threshold shrinkage processing includes:

[0091] Select the wavelet basis function and the corresponding decomposition level;

[0092] Perform wavelet packet decomposition on the filtered magnetic field signal according to the selected wavelet basis function and decomposition level to obtain wavelet packet coefficients;

[0093] Optimize to obtain the wavelet threshold based on a nature-inspired optimization algorithm with the goal of minimizing the weighted sum of the signal-to-noise ratio and the mean square error;

[0094] The wavelet packet coefficients with absolute values less than the wavelet threshold are set to zero, and the wavelet packet coefficients with absolute values greater than or equal to the wavelet threshold are shrunk towards zero to obtain the wavelet packet coefficients after threshold shrinkage processing.

[0095] The method for obtaining the first ultrasonic signal includes:

[0096] Performing wavelet packet reconstruction on the linearly corrected ultrasonic signal using the wavelet packet coefficients after threshold shrinkage processing to obtain the first ultrasonic signal. Transient pulse suppression is performed on the linearly corrected ultrasonic signal to remove abnormal pulses generated by external impacts or instantaneous changes inside the circuit. This enables the first ultrasonic signal to more accurately reflect the mechanical vibration conditions caused by transmission line faults, such as key features like vibration energy integration and main frequency offset more reliably. Based on the purer first ultrasonic signal, the relevant information of the acoustic wave energy peak can be determined more precisely. When combining with the magnetic field signal for fault location, better spatio-temporal correction can be achieved, making the energy weighted fusion result more reasonable, thereby improving the accuracy and stability of transmission line fault location.

[0097] The method for obtaining the second magnetic field signal and the second ultrasonic signal includes:

[0098] Step 1, data division: Divide the collected acoustic and magnetic signals into a training set, a validation set, and a test set according to a preset ratio; the acoustic and magnetic signals include the first ultrasonic signal and the first magnetic field signal;

[0099] Step 2, model construction: Select the U-Net structure as the generator, use a random noise vector as the input of the generator, and obtain the generated acoustic and magnetic signals with the same size and number of channels as the acoustic and magnetic signals; select a convolutional neural network as the discriminator architecture, use the acoustic and magnetic signals and the generated acoustic and magnetic signals as the inputs of the discriminator, and obtain the probability that the input data is the acoustic and magnetic signals;

[0100] Step 3, network training: Randomly extract M acoustic and magnetic signals from the training set, label them as 1, indicating real acoustic and magnetic signals; the generator generates corresponding generated acoustic and magnetic signals according to the input random noise vector, label them as 0, indicating generated acoustic and magnetic signals; input the acoustic and magnetic signals and the generated acoustic and magnetic signals into the discriminator respectively, and the discriminator outputs the discrimination result; according to the output of the discriminator and the labels corresponding to the acoustic and magnetic signals and the generated acoustic and magnetic signals, with the goal of minimizing the value of the loss function, optimize the parameters of the generator and the discriminator through the backpropagation algorithm; where the loss function is the sum of the generator loss and the discriminator loss, and the generator loss incorporates the constraints of Maxwell's equations and the acoustic wave equations; the discriminator loss is the cross-entropy loss; for example, the generator loss where, is the generated acoustic and magnetic signal; f real is the real acoustic and magnetic signal; ‖·‖2 is the L2 norm; ξ is the regularization parameter; is the curl of the magnetic field strength B in the first magnetic field signal; ∈ is the loss constant; is the partial derivative of the electric field strength E with respect to time t; is the Laplace operator of the sound pressure p of the first ultrasonic signal; is the second-order partial derivative of the sound pressure p with respect to time t; discriminator loss where is the mathematical expectation; D(·) is the discriminator function; is the weight parameter; is the spectral correlation loss; Repeat the above steps for multiple rounds of iterative training until the generative adversarial network reaches a convergence state.

[0101] Step 4, Acoustic-magnetic signal enhancement: Input the acoustic-magnetic signal into the generator of the generative adversarial network corresponding to the convergence state to generate an enhanced acoustic-magnetic signal, and use the enhanced acoustic-magnetic signal as the acoustic-magnetic signal after enhancement processing, that is, the second magnetic field signal and the second ultrasonic signal.

[0102] The second magnetic field signal and the second ultrasonic signal obtained by enhancing the first magnetic field signal and the first ultrasonic signal have improved signal amplitude and signal-to-noise ratio, making the fault characteristics more prominent. For example, features such as the fault mutation moment and high-frequency harmonic distortion rate in the second magnetic field signal are more obvious, which helps to more accurately identify the fault type; features such as the vibration energy integral and main frequency offset of the second ultrasonic signal are also easier to identify, and can more accurately determine the mechanical vibration situation caused by the fault. In addition, the enhanced signal can reduce misjudgment and missed judgment caused by weak signals, and provide more reliable basic data when calculating the fault point distance through multi-modal feature fusion, thus significantly improving the accuracy and reliability of transmission line fault location.

[0103] Extract magnetic field features and ultrasonic features from the second magnetic field signal and the second ultrasonic signal;

[0104] The methods for obtaining magnetic field features include:

[0105] Calculate the traveling wave mutation point of the second magnetic field signal as the mutation point feature, such as the traveling wave mutation point where A loop is the effective area of the magnetic field sensor; t is the time; B(t) is the magnetic field strength value at time t;

[0106] Based on the magnetic field mutation point detection threshold Detect the second magnetic field signal, where s is the magnetic field mutation point detection threshold coefficient; N is the number of samples of the magnetic field signal statistical features; B c is the magnetic field strength value of the c-th sample point; is the average magnetic field intensity of N sample points; traverse the second magnetic field signal, when the difference between the magnetic field signals at adjacent time points exceeds the magnetic field mutation point detection threshold, extract the corresponding magnetic field signal difference as the difference feature, and extract the corresponding two adjacent time points for splicing as the time point feature;

[0107] Extract the harmonic distortion rate based on the voltage amplitude of the second magnetic field signal as the distortion rate feature, such as the harmonic distortion rate where V h is the voltage amplitude of the h-th time; V1 is the voltage amplitude of the 1st time; N h is the highest harmonic order;

[0108] Extract the spectral centroid based on the spectral value as the centroid feature, such as the spectral centroid where f is the frequency; F is the maximum frequency; X(f) is the spectral value of the second magnetic field signal at frequency f;

[0109] Splice the S mutation point features, S difference features, S time point features, S distortion rate features and S centroid features as the magnetic field feature, where the features less than S are padded with zeros.

[0110] Magnetic field features such as traveling wave mutation points and high-frequency harmonic distortion rates can effectively reflect the electromagnetic changes during the fault moment, and can be used to accurately identify the fault type (such as lightning strike, broken strand, etc.) in the later stage, and then select appropriate attenuation coefficients and propagation speed correction rules for the positioning model, providing an important basis for calculating the distance of the fault point.

[0111] The methods for obtaining ultrasonic features include:

[0112] Obtain the second ultrasonic signal g, decompose g into k modal components, and each modal component has a corresponding central frequency ω k ; construct a constrained variational model; such as the constrained variational model The constraint condition is where u k is the k-th modal component; ω k is the central frequency of the k-th modal component u k ; K is the total number of modal components into which the second ultrasonic signal g is decomposed; k is the number of modal components into which the second ultrasonic signal g is decomposed; is the derivative operation with respect to time t; δ(t) is the Dirac function; j is the imaginary unit; * is the convolution operation; e is the mathematical constant; t is the time;

[0113] Introduce the Lagrange multiplier and the quadratic penalty term, and transform the constrained variational problem into an unconstrained augmented Lagrangian function; such as the augmented Lagrangian function λ is the Lagrange multiplier;

[0114] The augmented Lagrangian function is iteratively solved using the alternating direction method of multipliers;

[0115] Update the modal function: Fix ω k and λ, for each k, by solving Update u k ; is the k-th modal component after the h-th update; i is the number of modal components into which the ultrasonic signal g is decomposed;

[0116] Update the center frequency: Fix u k and λ, by solving Update ω k ; is the center frequency of the k-th modal component after the h-th update ;

[0117] The Lagrange multiplier is updated according to ; where λ h+1 is the Lagrange multiplier after the h-th update; λ h is the Lagrange multiplier before the h-th update; τ is the update step size;

[0118] Repeat the above update steps until, in two consecutive iterations, the changes in each modal function satisfy where tol is the convergence tolerance; is the k-th modal component before the h-th update;

[0119] Based on the above steps, K modal components u k are obtained, and the time-domain features and frequency-domain features of each modal component are extracted. The time-domain features include amplitude features and time features. The amplitude features are the mean amplitude, amplitude variance, maximum amplitude, and minimum amplitude of the modal component, and the time features are the rise time, fall time, and pulse width of the modal component; the frequency-domain features include frequency features and energy features. The frequency features are the peak frequency, center frequency, and bandwidth of the modal component, and the energy features are the energy distribution of the modal component in different frequency bands; the time-domain features and frequency-domain features are concatenated to obtain the ultrasonic features.

[0120] The ultrasonic features include the time-domain features and frequency-domain features of different modal components, etc. These features are closely related to the mechanical vibrations generated by faults and help to judge the fault location and nature from another dimension. By analyzing these features, the direction and energy distribution of the fault point can be determined. Combining with the magnetic field features, multi-modal information fusion can be achieved, the distance of the fault point can be calculated more accurately, the accuracy and reliability of fault location can be improved, and it is also helpful to evaluate the severity of the fault and provide a reference for subsequent maintenance decisions.

[0121] Methods for obtaining time characteristics include:

[0122] Preset an upper threshold and a lower threshold, traverse the corresponding modal component, obtain the first signal point greater than the lower threshold, continue to traverse the corresponding modal component from the first signal point, obtain the second signal point greater than the upper threshold, and calculate the product of the time difference between the second signal point and the first signal point and the sampling period as the rise time;

[0123] Traverse in reverse from the end position of the corresponding modal component, obtain the third signal point less than the upper threshold, continue to traverse the corresponding modal component from the third signal point, obtain the fourth signal point less than the lower threshold, and calculate the product of the time difference between the fourth signal point and the third signal point and the sampling period as the fall time;

[0124] Obtain the maximum and minimum values of the modal component, calculate the intermediate value of the maximum and minimum values, traverse the corresponding modal component, obtain the fifth signal point greater than the intermediate value, continue to traverse the corresponding modal component from the fifth signal point, obtain the sixth signal point less than the upper threshold, and calculate the product of the difference between the sixth signal point and the fifth signal point and the sampling period as the pulse width.

[0125] Methods for obtaining energy characteristics include:

[0126] Calculate the sum of the energies corresponding to all frequency domain indices in the preset frequency band of the modal component, where the energy is the square of the amplitude.

[0127] Take the ultrasonic characteristics as the input of the fault identification model to obtain the mechanical vibration mode and the corresponding fault type;

[0128] The training method of the fault identification model includes:

[0129] Pre-collect Q groups of type identification data, where the type identification data includes ultrasonic characteristics, mechanical vibration modes, and the corresponding fault types.

[0130] Take the ultrasonic characteristics as the input of the fault identification model, take the mechanical vibration mode and the corresponding fault type as the output of the fault identification model, and aim to minimize the error between the output mechanical vibration mode and the corresponding fault type and the actual mechanical vibration mode and the corresponding fault type. Optimize the network parameters of the fault identification model through a nature-inspired optimization algorithm to obtain the network parameters corresponding to the minimum error between the mechanical vibration mode and the corresponding fault type output by the fault identification model and the actual mechanical vibration mode and the corresponding fault type, and construct the fault identification model with the corresponding network parameters as the trained fault identification model.

[0131] Unique mechanical vibration patterns caused by different faults such as looseness and fracture can be accurately identified through ultrasonic features, the fault types can be determined, and the scope of location can be narrowed down. At the same time, the propagation characteristics of different vibration patterns can assist in judging the approximate location of the fault. Combining with other location methods can further accurately locate the fault, and it can also provide a basis for subsequent targeted maintenance, improving the maintenance efficiency and safety of transmission lines.

[0132] Taking the magnetic field features and ultrasonic features as the inputs of the dual-stream residual attention network, a fused joint feature map is obtained. Based on the joint feature maps of adjacent monitoring terminals, fusion is carried out through the D-S evidence theory, and the three-dimensional coordinates of the fault point are calculated; the joint feature maps include magnetic field features, acoustic wave features, and cross-modal correlation features; the magnetic field features include the traveling wave mutation moment, the high-frequency harmonic distortion rate, and the number of magnetic field polarity reversals; the acoustic wave features include the vibration energy integral, the main frequency offset, and the waveguide direction; the cross-modal correlation features include the timing alignment deviation between the magnetic field mutation and the acoustic wave energy peak and the weight distribution coefficient between modalities.

[0133] The training method of the dual-stream residual attention network includes:

[0134] Pre-collect R groups of training data. The training data includes acoustic and magnetic features and the corresponding joint feature maps. The acoustic and magnetic features include magnetic field features and ultrasonic features.

[0135] Taking the acoustic and magnetic features as the inputs of the dual-stream residual attention network and the joint feature maps as the outputs of the dual-stream residual attention network, with the goal of minimizing the error of the loss function, the network parameters of the dual-stream residual attention network are optimized through a nature-inspired optimization algorithm to obtain the network parameters corresponding to the minimum error between the joint feature map output by the dual-stream residual attention network and the actual joint feature map. The dual-stream residual attention network constructed with the corresponding network parameters is used as the trained dual-stream residual attention network; referring to Figure 2 , the dual-stream residual attention network includes a magnetic field branch and an ultrasonic branch. The magnetic field branch includes an NG layer (such as 5 layers) ResNet and a channel attention module. The ultrasonic branch includes an NM layer (such as 3 layers) CNN+LSTM and a spatial attention module. The outputs of the magnetic field branch and the ultrasonic branch are fused through a cross-attention mechanism to obtain the output joint feature map. The loss function of the dual-stream residual attention network includes a classification loss composed of cross-entropy and a localization loss composed of mean square error; such as the total loss Among them, is the classification loss; is the weight balance coefficient, which can be obtained by optimizing through a nature-inspired optimization algorithm; is the localization loss.

[0136] The dual-stream residual attention network can significantly improve the fault location accuracy by integrating magnetic field features and ultrasonic features and combining cross-modal correlation features (such as modal weight coefficients). The network uses the attention mechanism to dynamically allocate modal weights, capture the spatio-temporal correlation between acoustic and magnetic signals (such as the temporal alignment of magnetic field mutations and acoustic wave energy peaks), generate a joint feature map containing multi-modal complementary information, effectively suppress the interference of single-modal noise, enhance the recognition of fault features under complex working conditions, and provide a more comprehensive and reliable input basis for the subsequent location model.

[0137] The method for obtaining the three-dimensional coordinates of the fault point includes:

[0138] Based on the joint feature map and combined with the fault location model, perform distance inversion to obtain the fault distance; such as the fault distance Among them, ; is the attenuation coefficient, f THD is the total harmonic distortion rate; k material is the material correction factor; E source is the energy weighted fusion value, E mag is the magnetic field energy, E sonic is the acoustic wave energy, ζ is the weight distribution coefficient between modalities; E received is the received signal energy; v sonic is the acoustic wave signal propagation speed; Δt align is the timing deviation compensation value;

[0139] Fuse the fault distances calculated by three adjacent monitoring terminals through the D-S evidence theory to calculate the three-dimensional coordinates of the fault point; such as the three-dimensional coordinates of the fault point Among them, d r is the fault point distance calculated by the rth monitoring terminal; (x r , y r , z r ) is the three-dimensional coordinates of the rth monitoring terminal in space.

[0140] The D-S evidence theory constructs a multi-source evidence body by integrating the spatio-temporal information of multi-terminal acoustic and magnetic signals, reduces the single-point positioning error, uses the uncertainty reasoning ability of the evidence theory, synthesizes the confidence degrees of different terminals on the fault point position, and combines the signal propagation speed correction coefficient to accurately calculate the three-dimensional coordinates of the fault point; by eliminating the influence of environmental noise and signal attenuation through multi-terminal collaboration, it is especially suitable for fault location in long-distance transmission lines or complex terrains, and can significantly improve the reliability and robustness of the positioning results.

[0141] Embodiment 2

[0142] This embodiment provides an adaptive dynamic time-delay compensation mechanism, which predicts the temperature correction coefficient and humidity correction coefficient corresponding to the propagation speed of the magnetic field signal or the ultrasonic signal by introducing an LSTM network to predict the temperature and humidity.

[0143] The training method of the LSTM network includes:

[0144] Pre-collect L groups of type recognition data, where the type recognition data includes input training data and output training data; the input training data includes the first data group or the second data group, the first data group includes temperature, humidity and the propagation speed of the magnetic field signal, and the second data group includes temperature, humidity and the propagation speed of the ultrasonic signal; the output training data includes the corresponding temperature correction coefficient and humidity correction coefficient.

[0145] Take the input training data as the input of the LSTM network, take the output training data as the output of the LSTM network, aiming at minimizing the error between the output training data and the actual output training data, optimize the network parameters of the LSTM network through a natural inspiration optimization algorithm, obtain the network parameters corresponding to minimizing the error between the output training data output by the LSTM network and the actual output training data, and use the LSTM network constructed by the corresponding network parameters as the trained LSTM network.

[0146] Embodiment 3

[0147] Please refer to Figure 3 As shown, the transmission line fault location system based on the acoustic-magnetic signal feature analysis described in this embodiment includes:

[0148] Correction acquisition module: linearly correct the propagation speed of the magnetic field signal of the magnetic field sensor and the propagation speed of the ultrasonic signal of the ultrasonic sensor through the dynamic time-delay compensation mechanism, and synchronously acquire the linearly corrected magnetic field signal and ultrasonic signal after the linear correction process;

[0149] The steps of linearly correcting the propagation speed of the magnetic field signal of the magnetic field sensor and the propagation speed of the ultrasonic signal of the ultrasonic sensor are as follows:

[0150] Calculate the propagation speed of the magnetic field signal based on the relative magnetic permeability of the wire material and the relative node constant of the wire material;

[0151] Calculate the propagation speed of the ultrasonic signal based on the bulk modulus of the wire and the density of the wire material;

[0152] For different wire types, linearly correct the propagation speed of the magnetic field signal and the ultrasonic signal in combination with the temperature correction coefficient and humidity correction coefficient of the wire.

[0153] Since the dynamic time-delay compensation mechanism in Embodiment 1 is difficult to adapt to changes in environmental factors such as temperature and humidity in real time, the adaptive dynamic time-delay compensation mechanism of Embodiment 2 can be combined to optimize the temperature correction coefficient and humidity correction coefficient corresponding to the magnetic field signal propagation speed or ultrasonic signal propagation speed, so as to improve the accuracy of the correction results for the magnetic field signal propagation speed and ultrasonic signal propagation speed. Refer to Figure 4 Based on the LSTM network, the temperature correction coefficient and humidity correction coefficient corresponding to the magnetic field signal propagation speed and ultrasonic signal propagation speed are optimized respectively. The obtained temperature correction coefficient corresponding to the magnetic field signal propagation speed is 1.15 and the humidity correction coefficient is 0.88; the temperature correction coefficient corresponding to the ultrasonic signal propagation speed is 0.87 and the humidity correction coefficient is 1.05; the number of training times is 20,000 times, and the training accuracy reaches 98.5%, indicating that the model has been trained a lot and has high reliability.

[0154] Signal processing module: Perform strong electromagnetic interference suppression and transient pulse suppression on the linearly corrected magnetic field signal to obtain the first magnetic field signal, perform transient pulse suppression on the linearly corrected ultrasonic signal to obtain the first ultrasonic signal, and perform enhancement processing on the first ultrasonic signal and the first magnetic field signal to obtain the second magnetic field signal and the second ultrasonic signal;

[0155] The steps to obtain the first magnetic field signal are as follows:

[0156] Perform power frequency harmonic filtering processing based on the transfer function of the adaptive notch filter to obtain the filtered magnetic field signal;

[0157] Use the wavelet packet coefficients after threshold shrinkage processing to perform wavelet packet reconstruction on the filtered magnetic field signal to obtain the first magnetic field signal.

[0158] Among them, the steps to obtain the wavelet packet coefficients after threshold shrinkage processing are as follows:

[0159] Select the wavelet basis function and the corresponding decomposition level;

[0160] Perform wavelet packet decomposition on the filtered magnetic field signal according to the selected wavelet basis function and decomposition level to obtain wavelet packet coefficients;

[0161] With the goal of minimizing the weighted sum of the signal-to-noise ratio and the mean square error, optimize the wavelet threshold based on the natural inspiration optimization algorithm;

[0162] Set the wavelet packet coefficients with absolute values less than the wavelet threshold to zero, and shrink the wavelet packet coefficients with absolute values greater than or equal to the wavelet threshold towards zero to obtain the wavelet packet coefficients after threshold shrinkage processing. Refer to Figure 5, the initial threshold of the wavelet packet coefficients is set to 0.25, and the number of iterations is set to 100; the preset wavelet basis function is db4, the decomposition level is 4 layers, and the optimization state has converged. During the optimization process of the wavelet packet coefficients, the population size of the nature-inspired optimization algorithm is 50, the maximum number of generations is 200, the obtained optimized threshold is 0.487, the optimal fitness is 0.956, and the optimization converges to the 47th generation, with a convergence time of 8.5 seconds. Through the analysis of the wavelet packet coefficient distribution, a series of operations and results of threshold optimization are carried out. Real-time data monitoring ensures the dynamic tracking of the signal, and the settings of the optimization state and algorithm configuration parameters ensure the effectiveness and efficiency of the optimization process. The final obtained optimization results provide important parameter basis for subsequent signal processing and fault analysis.

[0163] The method for obtaining the first ultrasonic signal includes:

[0164] Performing wavelet packet reconstruction on the linearly corrected ultrasonic signal using the wavelet packet coefficients after threshold shrinkage processing to obtain the first ultrasonic signal.

[0165] The method for obtaining the second magnetic field signal and the second ultrasonic signal includes:

[0166] Dividing the collected acoustic-magnetic signals into a training set, a validation set, and a test set according to a preset ratio; the acoustic-magnetic signals include the first ultrasonic signal and the first magnetic field signal;

[0167] Selecting the U-Net structure as the generator, using the random noise vector as the input of the generator to obtain the generated acoustic-magnetic signal with the same size and number of channels as the acoustic-magnetic signal; selecting the convolutional neural network as the discriminator architecture, using the acoustic-magnetic signal and the generated acoustic-magnetic signal as the input of the discriminator to obtain the probability that the input data is the acoustic-magnetic signal;

[0168] Randomly extracting M groups of acoustic-magnetic signals from the training set, labeling them as 1, indicating real acoustic-magnetic signals; the generator generates corresponding generated acoustic-magnetic signals according to the input random noise vector, labeling them as 0, indicating generated acoustic-magnetic signals; inputting the acoustic-magnetic signals and the generated acoustic-magnetic signals into the discriminator respectively, and the discriminator outputs the discrimination result; according to the output of the discriminator and the labels corresponding to the acoustic-magnetic signals and the generated acoustic-magnetic signals, aiming at minimizing the value of the loss function, optimizing the parameters of the generator and the discriminator through the nature-inspired optimization algorithm; where the loss function is the sum of the generator loss and the discriminator loss, and the generator loss fuses the difference term between the generated acoustic-magnetic signal and the real acoustic-magnetic signal, the Maxwell equation constraint and the acoustic wave equation constraint; the discriminator loss is the cross-entropy loss; repeating the step of selecting acoustic-magnetic signals for discrimination, performing multiple rounds of iterative training until the generative adversarial network reaches the convergence state;

[0169] Input the acousto-magnetic signal into the generator of the generative adversarial network corresponding to the convergence state to generate an enhanced acousto-magnetic signal, and use the enhanced acousto-magnetic signal as the acousto-magnetic signal after enhancement processing, that is, the second magnetic field signal and the second ultrasonic signal.

[0170] Feature extraction module: Extract features from the second magnetic field signal and the second ultrasonic signal to obtain magnetic field features and ultrasonic features;

[0171] The method for obtaining magnetic field features includes:

[0172] Calculate the traveling wave mutation points of the second magnetic field signal as mutation point features;

[0173] Detect the second magnetic field signal based on the magnetic field mutation point detection threshold; traverse the second magnetic field signal. When the difference between the magnetic field signals at adjacent time points exceeds the magnetic field mutation point detection threshold, extract the corresponding magnetic field signal difference as the difference feature, and extract the corresponding two adjacent time points for splicing as the time point feature;

[0174] Extract the harmonic distortion rate based on the voltage amplitude of the second magnetic field signal as the distortion rate feature;

[0175] Extract the spectral centroid based on the spectral value as the centroid feature;

[0176] Splice the S mutation point features, S difference features, S time point features, S distortion rate features and S centroid features as the magnetic field features, where the features less than S are padded with zeros.

[0177] The method for obtaining ultrasonic features includes:

[0178] Step 1: Obtain the second ultrasonic signal g, decompose g into k modal components, and each modal component has a corresponding center frequency; construct a constrained variational model;

[0179] Step 2: Introduce Lagrange multipliers and quadratic penalty terms to transform the constrained variational problem into an unconstrained augmented Lagrangian function;

[0180] Step 3: Use the alternating direction multiplier method to iteratively solve the augmented Lagrangian function;

[0181] Step 4: Update the modal function;

[0182] Step 5: Update the center frequency;

[0183] Step 6: Update the Lagrange multiplier;

[0184] Step 7: Repeat Step 4 - Step 6 until the change of each modal function is less than the convergence tolerance in two consecutive iterations, and then stop the iteration;

[0185] Step 8: Based on Steps 1 - 7, obtain K modal components, and extract the time-domain features and frequency-domain features of each modal component. The time-domain features include amplitude features and time features. The amplitude features are the mean amplitude, amplitude variance, maximum amplitude, and minimum amplitude of the modal component, and the time features are the rise time, fall time, and pulse width of the modal component. The frequency-domain features include frequency features and energy features. The frequency features are the peak frequency, center frequency, and bandwidth of the modal component, and the energy features are the energy distribution of the modal component in different frequency bands. Concatenate the time-domain features and frequency-domain features to obtain the ultrasonic features.

[0186] The methods for obtaining the time features include:

[0187] Preset the upper threshold and lower threshold, traverse the corresponding modal component, obtain the first signal point greater than the lower threshold, continue to traverse the corresponding modal component from the first signal point, obtain the second signal point greater than the upper threshold, and calculate the product of the time difference between the second signal point and the first signal point and the sampling period as the rise time.

[0188] Traverse in reverse from the end position of the corresponding modal component, obtain the third signal point less than the upper threshold, continue to traverse the corresponding modal component from the third signal point, obtain the fourth signal point less than the lower threshold, and calculate the product of the time difference between the fourth signal point and the third signal point and the sampling period as the fall time.

[0189] Obtain the maximum and minimum values of the modal component, calculate the intermediate value of the maximum and minimum values, traverse the corresponding modal component, obtain the fifth signal point greater than the intermediate value, continue to traverse the corresponding modal component from the fifth signal point, obtain the sixth signal point less than the upper threshold, and calculate the product of the difference between the sixth signal point and the fifth signal point and the sampling period as the pulse width.

[0190] The methods for obtaining the energy features include:

[0191] Calculate the sum of the energies corresponding to all frequency-domain indices of the modal component in the preset frequency band, where the energy is the square of the amplitude.

[0192] Fault identification module: Use the ultrasonic features as the input of the fault identification model to obtain the mechanical vibration mode and the corresponding fault type.

[0193] The training method of the fault identification model includes:

[0194] Pre-collect Q groups of type identification data, where the type identification data includes ultrasonic features, mechanical vibration modes, and the corresponding fault types.

[0195] Taking the ultrasonic features as the input of the fault identification model, and the mechanical vibration mode and the corresponding fault type as the output of the fault identification model, aiming at minimizing the error between the output mechanical vibration mode and the corresponding fault type and the actual mechanical vibration mode and the corresponding fault type, optimizing the network parameters of the fault identification model through a nature-inspired optimization algorithm, obtaining the network parameters corresponding to the minimum error between the mechanical vibration mode and the corresponding fault type output by the fault identification model and the actual mechanical vibration mode and the corresponding fault type, and constructing the fault identification model with the corresponding network parameters as the trained fault identification model.

[0196] Fault location module: Taking the magnetic field features and ultrasonic features as the input of the dual-stream residual attention network, obtaining the fused joint feature map, and calculating the three-dimensional coordinates of the fault point by fusing the joint feature maps of three adjacent monitoring terminals through the D-S evidence theory.

[0197] The training method of the dual-stream residual attention network includes:

[0198] Pre-collecting R groups of training data, where the training data includes acoustic-magnetic features and the corresponding joint feature maps, and the acoustic-magnetic features include magnetic field features and ultrasonic features;

[0199] Taking the acoustic-magnetic features as the input of the dual-stream residual attention network, and the joint feature map as the output of the dual-stream residual attention network, aiming at minimizing the error between the output joint feature map and the actual joint feature map, optimizing the network parameters of the dual-stream residual attention network through a nature-inspired optimization algorithm, obtaining the network parameters corresponding to the minimum error between the joint feature map output by the dual-stream residual attention network and the actual joint feature map, and constructing the dual-stream residual attention network with the corresponding network parameters as the trained dual-stream residual attention network; the dual-stream residual attention network includes a magnetic field branch and an ultrasonic branch, the magnetic field branch includes an NG-layer ResNet and a channel attention module, the ultrasonic branch includes an NM-layer CNN+LSTM and a spatial attention module, and the outputs of the magnetic field branch and the ultrasonic branch are fused through a cross-attention mechanism to obtain the output joint feature map; the loss function of the dual-stream residual attention network includes a classification loss composed of cross-entropy and a localization loss composed of mean square error.

[0200] The method for obtaining the three-dimensional coordinates of the fault point includes:

[0201] Based on the joint feature map, performing distance inversion in combination with the fault location model to obtain the fault distance;

[0202] Fusing the fault distances calculated by three adjacent monitoring terminals through the D-S evidence theory to calculate the three-dimensional coordinates of the fault point.

[0203] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the said claims.

[0204] Finally: The above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall all be included within the protection scope of the present invention.

Claims

1. A power transmission line fault location method based on acoustic-magnetic signal feature analysis, characterized in that It includes the following steps: Through a dynamic time-delay compensation mechanism, linearly correct the propagation speeds of the magnetic field signals of the magnetic field sensor and the ultrasonic signals of the ultrasonic sensor, and synchronously collect the linearly corrected magnetic field signals and ultrasonic signals after the linear correction process; Suppress strong electromagnetic interference and transient pulses for the linearly corrected magnetic field signals to obtain the first magnetic field signals, suppress transient pulses for the linearly corrected ultrasonic signals to obtain the first ultrasonic signals, and enhance the first ultrasonic signals and the first magnetic field signals to obtain the second magnetic field signals and the second ultrasonic signals; Extract features from the second magnetic field signals and the second ultrasonic signals to obtain magnetic field features and ultrasonic features; Use the ultrasonic features as the input of the fault identification model to obtain the mechanical vibration mode and the corresponding fault type; Use the magnetic field features and ultrasonic features as the input of the dual-stream residual attention network to obtain a fused joint feature map, and fuse the joint feature maps of three adjacent monitoring terminals through the D-S evidence theory to calculate and obtain the three-dimensional coordinates of the fault point.

2. The power transmission line fault location method based on acoustic and magnetic signal feature analysis according to claim 1, characterized in that The method for linearly correcting the propagation speeds of the magnetic field signals of the magnetic field sensor and the ultrasonic signals of the ultrasonic sensor includes: Calculate the propagation speed of the magnetic field signal based on the relative magnetic permeability of the wire material and the relative node constant of the wire material; Calculate the propagation speed of the ultrasonic signal based on the bulk modulus of the wire and the wire material density; For different wire types, linearly correct the propagation speeds of the magnetic field signal and the ultrasonic signal in combination with the temperature correction coefficient and humidity correction coefficient of the wire.

3. The power transmission line fault location method based on acoustic and magnetic signal feature analysis according to claim 1, wherein The training method of the dual-stream residual attention network includes: Pre-collect R sets of training data. The training data includes acoustic-magnetic features and the corresponding joint feature maps. The acoustic-magnetic features include magnetic field features and ultrasonic features; Use the acoustic-magnetic features as the input of the dual-stream residual attention network and the joint feature map as the output of the dual-stream residual attention network. With the goal of minimizing the error between the output joint feature map and the actual joint feature map, optimize the network parameters of the dual-stream residual attention network through a nature-inspired optimization algorithm to obtain the network parameters corresponding to minimizing the error between the joint feature map output by the dual-stream residual attention network and the actual joint feature map, and use the dual-stream residual attention network constructed with the corresponding network parameters as the trained dual-stream residual attention network; The dual-stream residual attention network includes a magnetic field branch and an ultrasonic branch. The magnetic field branch includes an NG-layer ResNet and a channel attention module, and the ultrasonic branch includes an NM-layer CNN+LSTM and a spatial attention module. The outputs of the magnetic field branch and the ultrasonic branch are fused through a cross-attention mechanism to obtain the output joint feature map; The loss function of the dual-stream residual attention network includes a classification loss composed of cross-entropy and a localization loss composed of mean square error; The method for obtaining the three-dimensional coordinates of the fault point includes: Based on the joint feature map, perform distance inversion in combination with the fault location model to obtain the fault distance; Fuse the fault distances calculated by three adjacent monitoring terminals through the D-S evidence theory to calculate and obtain the three-dimensional coordinates of the fault point.

4. The transmission line fault location method based on acoustic-magnetic signal feature analysis according to claim 1, wherein The method for obtaining the first magnetic field signal includes: Performing power frequency harmonic filtering on the basis of the transfer function of an adaptive notch filter to obtain the filtered magnetic field signal; Performing wavelet packet reconstruction on the filtered magnetic field signal using the wavelet packet coefficients after threshold shrinkage processing to obtain the first magnetic field signal.

5. The method for fault location of a transmission line based on acoustic and magnetic signal feature analysis according to claim 3, wherein The method for obtaining the wavelet packet coefficients after threshold shrinkage processing includes: Selecting a wavelet basis function and the corresponding decomposition level; Performing wavelet packet decomposition on the filtered magnetic field signal according to the selected wavelet basis function and decomposition level to obtain wavelet packet coefficients; Taking the minimum of the weighted sum of the signal-to-noise ratio and the mean square error as the objective, and optimizing to obtain the wavelet threshold based on a nature-inspired optimization algorithm; Setting the wavelet packet coefficients with absolute values less than the wavelet threshold to zero, and shrinking the wavelet packet coefficients with absolute values greater than or equal to the wavelet threshold towards zero to obtain the wavelet packet coefficients after threshold shrinkage processing.

6. The power transmission line fault location method based on acoustic magnetic signal feature analysis according to claim 4, characterized in that The method for obtaining the second magnetic field signal and the second ultrasonic signal includes: Dividing the collected acousto-magnetic signals into a training set, a validation set, and a test set according to a preset ratio; the acousto-magnetic signals include the first ultrasonic signal and the first magnetic field signal; Selecting a U-Net structure as the generator, taking a random noise vector as the input of the generator to obtain a generated acousto-magnetic signal having the same size and number of channels as the acousto-magnetic signal; selecting a convolutional neural network as the discriminator architecture, taking the acousto-magnetic signal and the generated acousto-magnetic signal as the inputs of the discriminator to obtain the probability that the input data is the acousto-magnetic signal; Randomly extracting M groups of acousto-magnetic signals from the training set, labeling them as 1, indicating real acousto-magnetic signals; the generator generates corresponding generated acousto-magnetic signals according to the input random noise vector, labeling them as 0, indicating generated acousto-magnetic signals; inputting the acousto-magnetic signals and the generated acousto-magnetic signals into the discriminator respectively, and the discriminator outputs discriminant results; according to the output of the discriminator, and the labels corresponding to the acousto-magnetic signals and the generated acousto-magnetic signals, taking the minimum value of the loss function as the objective, optimizing the parameters of the generator and the discriminator through a nature-inspired optimization algorithm; wherein, the loss function is the sum of the generator loss and the discriminator loss, and the generator loss fuses the difference term between the generated acousto-magnetic signal and the real acousto-magnetic signal, the Maxwell equation constraint and the acoustic wave equation constraint; the discriminator loss is the cross-entropy loss; repeating the step of selecting acousto-magnetic signals for discrimination, performing multiple rounds of iterative training until the generative adversarial network reaches a convergence state; Inputting the acousto-magnetic signal into the generator of the generative adversarial network corresponding to the convergence state to generate an enhanced acousto-magnetic signal, and taking the enhanced acousto-magnetic signal as the enhanced processed acousto-magnetic signal, that is, the second magnetic field signal and the second ultrasonic signal.

7. The power transmission line fault location method based on acousto-magnetic signal feature analysis according to claim 1, wherein The method for obtaining the magnetic field characteristics includes: Calculating the traveling wave mutation points of the second magnetic field signal as mutation point characteristics; Detecting the second magnetic field signal based on a magnetic field mutation point detection threshold; traversing the second magnetic field signal, when the difference between the magnetic field signals at adjacent time points exceeds the magnetic field mutation point detection threshold, extracting the corresponding magnetic field signal difference as the difference characteristic, and extracting the corresponding two adjacent time points for splicing as the time point characteristic; Extracting the harmonic distortion rate as the distortion rate characteristic based on the voltage amplitude of the second magnetic field signal; Extracting the spectral centroid as the centroid characteristic based on the spectral value; Concatenate S mutation point features, S difference features, S time point features, S distortion rate features, and S centroid features as magnetic field features. Among them, features with less than S are padded with zeros.

8. The transmission line fault location method based on acoustic and magnetic signal feature analysis according to claim 1, wherein The methods for obtaining ultrasonic features include: Step 1: Obtain the second ultrasonic signal g, decompose g into k modal components, and each modal component has a corresponding central frequency; construct a constrained variational model; Step 2: Introduce Lagrange multipliers and quadratic penalty terms to transform the constrained variational problem into an unconstrained augmented Lagrangian function; Step 3: Use the alternating direction multiplier method to iteratively solve the augmented Lagrangian function; Step 4: Update the modal function; Step 5: Update the central frequency; Step 6: Update the Lagrange multiplier; Step 7: Repeat Step 4 - Step 6 until the change in each modal function is less than the convergence tolerance in two consecutive iterations, and then stop the iteration; Step 8: Based on Step 1 - Step 7, obtain K modal components, and extract the time-domain features and frequency-domain features of each modal component. The time-domain features include amplitude features and time features. The amplitude features are the mean amplitude, amplitude variance, maximum amplitude, and minimum amplitude of the modal component, and the time features are the rise time, fall time, and pulse width of the modal component; the frequency-domain features include frequency features and energy features. The frequency features are the peak frequency, central frequency, and bandwidth of the modal component, and the energy feature is the energy distribution of the modal component in different frequency bands; concatenate the time-domain features and frequency-domain features to obtain ultrasonic features.

9. The power transmission line fault location method based on acousto-magnetic signal feature analysis according to claim 8, wherein The methods for obtaining time features include: Preset an upper threshold and a lower threshold, traverse the corresponding modal component, obtain the first signal point greater than the lower threshold, continue to traverse the corresponding modal component from the first signal point, obtain the second signal point greater than the upper threshold, and calculate the product of the time difference between the second signal point and the first signal point and the sampling period as the rise time; Traverse in reverse from the end position of the corresponding modal component, obtain the third signal point less than the upper threshold, continue to traverse the corresponding modal component from the third signal point, obtain the fourth signal point less than the lower threshold, and calculate the product of the time difference between the fourth signal point and the third signal point and the sampling period as the fall time; Obtain the maximum and minimum values of the modal component, calculate the intermediate value of the maximum and minimum values, traverse the corresponding modal component, obtain the fifth signal point greater than the intermediate value, continue to traverse the corresponding modal component from the fifth signal point, obtain the sixth signal point less than the upper threshold, and calculate the product of the difference between the sixth signal point and the fifth signal point and the sampling period as the pulse width; The methods for obtaining energy features include: Calculate the sum of the energies corresponding to all frequency-domain indices of the modal component in the preset frequency band, where the energy is the square of the amplitude.

10. A transmission line fault location system based on acoustic-magnetic signal feature analysis, which implements the transmission line fault location method based on acoustic-magnetic signal feature analysis according to any one of claims 1-9, characterized in that, Include: Correction acquisition module: Perform linear correction processing on the propagation speed of the magnetic field signal of the magnetic field sensor and the propagation speed of the ultrasonic signal of the ultrasonic sensor through a dynamic time delay compensation mechanism, and synchronously acquire the linearly corrected magnetic field signal and ultrasonic signal after linear correction processing; Signal processing module: Suppress strong electromagnetic interference and transient pulses for the linearly corrected magnetic field signal to obtain the first magnetic field signal, suppress transient pulses for the linearly corrected ultrasonic signal to obtain the first ultrasonic signal, and perform enhancement processing on the first ultrasonic signal and the first magnetic field signal to obtain the second magnetic field signal and the second ultrasonic signal; Feature extraction module: Extract features from the second magnetic field signal and the second ultrasonic signal to obtain magnetic field features and ultrasonic features; Fault identification module: Use the ultrasonic features as the input of the fault identification model to obtain the mechanical vibration mode and the corresponding fault type; Fault location module: Use the magnetic field features and ultrasonic features as the input of the dual-stream residual attention network to obtain the fused joint feature map, fuse the joint feature maps of three adjacent monitoring terminals through the D-S evidence theory, and calculate to obtain the three-dimensional coordinates of the fault point.

Images