A Method and System for Partial Discharge Detection of Overhead Power Lines Based on Piezoelectric Array

Abstract

This invention relates to the field of power equipment condition monitoring technology, and discloses a method and system for detecting partial discharge (PD) in overhead power lines based on a piezoelectric array. The steps include: synchronously acquiring ultrasonic signals generated by partial discharge to obtain multi-channel time-domain signals; fusing the environmental sound velocity field calculated from environmental parameters with the measured sound velocity field at the measurement points obtained from array measurements to establish a dynamic sound velocity field model; performing time-frequency transformation on the time-domain signals and constructing a multi-constraint joint optimization objective function based on a PD waveform dictionary; combining the dynamic sound velocity field and the objective function, iteratively solving for the waveforms, locations, and attenuation parameters of multiple PD sources through blind source separation; finally, matching the dictionary to identify the discharge type, calculating the discharge intensity, and outputting a complete diagnostic result. This invention significantly improves the accuracy and reliability of PD detection in overhead power lines under complex environments by constructing an integrated technical system encompassing array signal acquisition, dynamic sound velocity field reconstruction, multi-constraint joint optimization, blind source separation, and full-dimensional diagnostic output.

Description

Technical Field

[0001] This invention relates to the field of power equipment condition monitoring technology, and more specifically, to a method and system for detecting partial discharge in overhead power lines based on piezoelectric arrays. Background Technology

[0002] As the core carrier of power grid transmission, the safe and stable operation of overhead power lines directly affects the overall reliability of the power system. Partial discharge is an important precursor to insulation degradation and fault initiation in line equipment. Timely and accurate detection of partial discharge signals and location of the discharge source are crucial for early fault investigation, prevention of line tripping, and even large-scale power outages. Piezoelectric sensor arrays, with their non-contact detection advantages, can effectively collect ultrasonic signals generated by partial discharge, becoming an important technical means for remote online monitoring of partial discharge in overhead power lines. However, existing partial discharge detection methods still have many technical shortcomings in practical engineering applications and cannot meet the detection needs of complex sites.

[0003] Overhead power lines are located in open outdoor environments with large diurnal temperature variations, significant differences in air pressure and relative humidity at different altitudes, and drastic seasonal climate changes. This causes significant spatiotemporal fluctuations in the speed of ultrasonic wave propagation due to environmental parameters. Existing detection methods generally use a fixed speed of sound for discharge source location calculations, which cannot adapt to dynamic changes in sound speed, resulting in large location errors and difficulty in accurately pinpointing the fault location. Furthermore, in actual transmission line operation, multiple discharge sources often operate simultaneously. Existing methods struggle to effectively separate aliased ultrasonic signals, only identifying the discharge source with the strongest signal, easily overlooking weaker discharge sources. Purely mathematical blind source separation techniques lack the constraints of the physical laws of ultrasonic wave propagation and prior knowledge of partial discharge waveforms, easily producing physically meaningless invalid solutions and resulting in poor reliability of separation results. Therefore, there is an urgent need to provide a piezoelectric array-based method and system for detecting partial discharge in overhead power lines to address these problems. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and system for partial discharge detection of overhead power lines based on piezoelectric arrays. It aims to solve the technical issues of inaccurate localization due to changes in ambient sound velocity and the difficulty in effectively separating multi-source aliasing signals in the partial discharge detection of overhead power lines. This invention significantly improves the accuracy and reliability of partial discharge detection of overhead power lines in complex environments by constructing an integrated technical system encompassing array signal acquisition, dynamic sound velocity field reconstruction, multi-constraint joint optimization, blind source separation solution, and full-dimensional diagnostic output.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows:

[0006] The method for detecting partial discharge in overhead power lines based on piezoelectric arrays includes the following steps:

[0007] S1. Arrange a piezoelectric sensor array in the area to be detected, and synchronously collect ultrasonic signals generated by partial discharge to obtain multi-channel time-domain signals;

[0008] S2. Obtain the environmental parameters of the detection area and calculate the ambient sound velocity field based on the environmental parameters; measure the ultrasonic wave propagation time between the piezoelectric sensor arrays and invert the sound velocity field at the measuring point; fuse the ambient sound velocity field and the sound velocity field at the measuring point to establish a dynamic sound velocity field model.

[0009] S3. Perform time-frequency transformation on the multi-channel time-domain signal to obtain the time-frequency domain signal; based on the partial discharge waveform dictionary, establish a joint optimization objective function that includes signal reconstruction error terms, spherical wave attenuation constraint terms, and waveform dictionary constraint terms;

[0010] S4. Based on the dynamic sound velocity field model and joint optimization objective function, blind source separation is performed on the time-frequency domain signal. The waveforms, spatial locations, and attenuation parameters of multiple local discharge sources are obtained through iterative optimization.

[0011] S5. Match the obtained partial discharge source waveforms with the partial discharge waveform dictionary to identify the discharge type; calculate the discharge intensity based on the waveform and attenuation parameters; and output the spatial location, discharge type, and discharge intensity of multiple partial discharge sources.

[0012] As a preferred embodiment of the present invention, the piezoelectric sensor array in S1 includes at least four piezoelectric sensors, the spacing between the sensors is set according to the monitoring range, and the frequency band covers the typical frequency band of partial discharge ultrasonic waves; the acquisition process adopts a unified clock to synchronously trigger multi-channel sampling, the sampling frequency meets the Nyquist sampling requirements, and the multi-channel time-domain signal obtained by synchronous sampling is organized into a matrix form as the input for subsequent processing.

[0013] As a preferred embodiment of the present invention, step S2 specifically includes:

[0014] S21. Obtain the temperature, air pressure, and relative humidity of the detection area, and calculate the ambient sound velocity field based on the physical relationship between environmental parameters and sound velocity.

[0015] S22. Select multiple pairs of sensors with known spacing in the piezoelectric sensor array, measure the arrival time difference of the ultrasonic signal, and obtain the sound velocity field at the measuring point based on the spacing and arrival time difference.

[0016] S23. The ambient sound velocity field and the sound velocity field at the measuring point are weighted and fused, with the weights allocated according to the spatial distance, to establish a dynamic sound velocity field model.

[0017] As a preferred embodiment of the present invention, the inversion process of the sound velocity field at the measuring point in S22 specifically includes:

[0018] Select at least three pairs of sensor pairs with known spacing in the piezoelectric sensor array, and ensure that there is a direct ultrasonic wave propagation path between the sensor pairs;

[0019] The arrival time difference of the ultrasonic signals received by each sensor is extracted, and the arrival time difference is determined by the peak value of the cross-correlation function of the two signals;

[0020] Divide the distance between each sensor pair by the corresponding time difference of arrival to obtain multiple sound speed calculation values;

[0021] The sound velocity field at the measurement point is obtained by performing least-squares fitting on multiple calculated sound velocity values.

[0022] As a preferred embodiment of the present invention, the partial discharge waveform dictionary library in S3 includes corona discharge waveform atoms, surface discharge waveform atoms, internal discharge waveform atoms, and suspended discharge waveform atoms; the time-domain expression of each waveform atom is composed of the product of the envelope function and the oscillation component, wherein the envelope function is selected as exponential decay type, Gaussian type, or double exponential type according to the discharge type, and the characteristic frequency of the oscillation component is set in the ultrasonic frequency band from 20kHz to 150kHz according to the discharge type.

[0023] As a preferred embodiment of the present invention, the joint optimization objective function in S3 integrates three constraints through weighting coefficients, namely the error between the array received signal and the model reconstructed signal, the deviation between the discharge power supply attenuation parameter and the spherical wave propagation theory, and the residual between the discharge power supply waveform and the sparse representation of the dictionary atoms.

[0024] As a preferred embodiment of the present invention, S4 employs an alternating iterative optimization method for blind source separation, specifically including the following steps:

[0025] S41. Set the number of partial discharge sources to be separated, and initialize the position, waveform and attenuation parameters of each partial discharge source;

[0026] S42. Fix the position and attenuation parameters of all partial discharge sources, and optimize and update the waveform of each partial discharge source according to the joint optimization objective function;

[0027] S43. Fix the waveforms of all partial discharge sources, and optimize and update the position and attenuation parameters of each partial discharge source according to the joint optimization objective function and dynamic sound velocity field model.

[0028] S44. Repeat steps S42 to S43 until the joint optimization objective function value converges, and output the final waveform, spatial location and attenuation parameters of each local amplifier power supply.

[0029] As a preferred embodiment of the present invention, in S5, the similarity between the partial discharge power source waveform obtained by normalizing the cross-correlation coefficient and various types of discharge atoms in the partial discharge waveform dictionary is calculated, and the category with the highest similarity is selected as the discharge type identification result; when the maximum similarity is less than a preset threshold of 0.5 to 0.7, it is determined to be an unidentified type.

[0030] As a preferred embodiment of the present invention, the calculation of the discharge intensity in S5 specifically includes the following steps:

[0031] S51. Extract the peak amplitude of the partial discharge power supply waveform;

[0032] S52. Based on the attenuation parameters and the propagation distance from the partial discharge source position to the reference position, attenuation compensation is performed on the peak amplitude to obtain the compensated amplitude.

[0033] S53. Calculate the apparent discharge quantity based on the compensated amplitude and compare it with the preset intensity level threshold to classify the discharge intensity level.

[0034] The present invention also provides a partial discharge detection system for overhead power lines based on a piezoelectric array, including an array signal acquisition module, a sound velocity field reconstruction module, an objective function construction module, a blind source separation and solution module, and a discharge diagnosis output module;

[0035] The array signal acquisition module is used to arrange a piezoelectric sensor array in the area to be detected, and synchronously acquire the ultrasonic signal generated by partial discharge to obtain multi-channel time domain signal.

[0036] The sound velocity field reconstruction module is used to acquire environmental parameters of the detection area, calculate the environmental sound velocity field based on the environmental parameters, measure the ultrasonic wave propagation time between the piezoelectric sensor arrays, invert the sound velocity field at the measuring point, and fuse the environmental sound velocity field and the sound velocity field at the measuring point to establish a dynamic sound velocity field model.

[0037] The objective function construction module is used to perform time-frequency transformation on multi-channel time-domain signals to obtain time-frequency domain signals; based on the partial discharge waveform dictionary library, a joint optimization objective function is established, which includes signal reconstruction error terms, spherical wave attenuation constraint terms, and waveform dictionary constraint terms;

[0038] The blind source separation solution module is used to perform blind source separation on time-frequency domain signals based on a dynamic sound velocity field model and a joint optimization objective function, and obtain the waveforms, spatial locations and attenuation parameters of multiple local discharge sources through iterative optimization solution;

[0039] The discharge diagnostic output module is used to match the separated partial discharge power source waveform with the partial discharge waveform dictionary to identify the discharge type; calculate the discharge intensity based on the waveform and attenuation parameters; and output the spatial location, discharge type, and discharge intensity of multiple partial discharge power sources.

[0040] The beneficial technical effects of this invention are:

[0041] This invention constructs a dynamic sound velocity field model by fusing the environmental sound velocity field calculated from environmental parameters with the sound velocity field at the measurement points obtained from the actual measurements of the sensor array. This model reflects the macroscopic sound velocity distribution in the detection area by incorporating temperature, air pressure, and relative humidity, while also correcting local sound velocity deviations using the measured signals from the sensor pairs. Furthermore, by allocating fusion weights according to spatial location, it achieves a smooth fusion of measured data and theoretical models. This model can adapt in real time to the spatiotemporal fluctuations in sound velocity caused by changes in environmental parameters, completely solving the positioning error problem caused by the fixed sound velocity assumption. It significantly improves the spatial positioning accuracy of power supplies under complex climatic conditions and can accurately pinpoint line fault points.

[0042] This invention constructs a joint optimization objective function that includes signal reconstruction error, spherical wave attenuation constraint, and waveform dictionary constraint. It organically integrates data fitting accuracy, the physical laws of ultrasonic wave propagation, and prior knowledge of partial discharge waveforms. It employs an alternating iterative optimization method for blind source separation. First, it fixes the position and attenuation parameters to optimize the waveform, then fixes the waveform to optimize the position and attenuation parameters, until the objective function converges. This ensures that the separation result conforms to physical laws and enhances the reliability of the solution by utilizing prior knowledge of the domain. It effectively achieves accurate separation of multi-source aliased partial discharge signals, can simultaneously identify multiple discharge sources, avoids the omission of weak discharge sources, and solves the problem of failure in multi-source scene detection in traditional methods.

[0043] This invention establishes a waveform dictionary library containing four typical discharge types: corona, surface, internal, and suspension. It achieves accurate identification of discharge types through normalized cross-correlation coefficients. Simultaneously, it combines the peak amplitude of the discharge source waveform, attenuation parameters, and propagation distance for attenuation compensation, calculates the apparent discharge quantity, and classifies intensity levels. It can simultaneously output the spatial location, discharge type, and discharge intensity of multiple discharge sources, realizing a comprehensive diagnosis of partial discharge faults, including location, classification, and quantification. Compared to existing single-location detection methods, this provides a comprehensive and direct decision-making basis for refined operation and maintenance and priority fault handling of overhead power lines. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating the present invention.

[0045] Figure 2 This is a schematic diagram of the detection system of the present invention.

[0046] Figure 3 This is a schematic diagram comparing the results of the present invention. Detailed Implementation

[0047] In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0048] Combination Figure 1-3 The present invention provides the following embodiments:

[0049] Example 1:

[0050] The method for detecting partial discharge in overhead power lines based on piezoelectric arrays includes the following steps:

[0051] S1. Arrange a piezoelectric sensor array in the area to be detected, and synchronously collect ultrasonic signals generated by partial discharge to obtain multi-channel time-domain signals;

[0052] S2. Obtain the environmental parameters of the detection area and calculate the ambient sound velocity field based on the environmental parameters; measure the ultrasonic wave propagation time between the piezoelectric sensor arrays and invert the sound velocity field at the measuring point; fuse the ambient sound velocity field and the sound velocity field at the measuring point to establish a dynamic sound velocity field model.

[0053] S3. Perform time-frequency transformation on the multi-channel time-domain signal to obtain the time-frequency domain signal; based on the partial discharge waveform dictionary, establish a joint optimization objective function that includes signal reconstruction error terms, spherical wave attenuation constraint terms, and waveform dictionary constraint terms;

[0054] S4. Based on the dynamic sound velocity field model and joint optimization objective function, blind source separation is performed on the time-frequency domain signal. The waveforms, spatial locations, and attenuation parameters of multiple local discharge sources are obtained through iterative optimization.

[0055] S5. Match the obtained partial discharge source waveforms with the partial discharge waveform dictionary to identify the discharge type; calculate the discharge intensity based on the waveform and attenuation parameters; and output the spatial location, discharge type, and discharge intensity of multiple partial discharge sources.

[0056] Furthermore, the piezoelectric sensor array in S1 contains at least four piezoelectric sensors, with the sensor spacing set according to the monitoring range, and the frequency band covering the typical frequency band of partial discharge ultrasound. The acquisition process uses a unified clock to synchronously trigger multi-channel sampling, and the sampling frequency meets the Nyquist sampling requirements. The multi-channel time-domain signals obtained by synchronous sampling are organized into a matrix form and used as input for subsequent processing.

[0057] Specifically, a piezoelectric sensor array is arranged in the area to be detected, the array containing... One piezoelectric sensor, among which The sensor spacing is determined based on the monitoring range, preferably between 0.5m and 5m. The sensor's center frequency is 40kHz, with a frequency band covering 20kHz to 150kHz, matching the partial discharge ultrasonic frequency band. The spatial coordinates of each sensor... Calibrate using measuring tools.

[0058] The acquisition system uses a unified clock for synchronous triggering to ensure time alignment of multi-channel signals. Sampling frequency. Set to 1MHz, which satisfies The sampling requirements are as follows. The analog-to-digital converter has a bit width of 12 to 16 bits.

[0059] Each sensor channel is sampled synchronously to obtain a time-domain signal sequence. ,in Channel number, For time sampling points, , Indicates the first The specific time value at each sampling moment. The multi-channel time-domain signal is organized as follows: 3D data matrix This serves as the input for subsequent processing modules.

[0060] Furthermore, the specific steps of S2 are as follows:

[0061] S21. Obtain the temperature, air pressure, and relative humidity of the detection area, and calculate the ambient sound velocity field based on the physical relationship between environmental parameters and sound velocity.

[0062] S22. Select multiple pairs of sensors with known spacing in the piezoelectric sensor array, measure the arrival time difference of the ultrasonic signal, and obtain the sound velocity field at the measuring point based on the spacing and arrival time difference.

[0063] S23. The ambient sound velocity field and the sound velocity field at the measuring point are weighted and fused, with the weights allocated according to the spatial distance, to establish a dynamic sound velocity field model.

[0064] Furthermore, the inversion process of the sound velocity field at the measuring point in S22 specifically includes:

[0065] Select at least three pairs of sensor pairs with known spacing in the piezoelectric sensor array, and ensure that there is a direct ultrasonic wave propagation path between the sensor pairs;

[0066] The arrival time difference of the ultrasonic signals received by each sensor is extracted, and the arrival time difference is determined by the peak value of the cross-correlation function of the two signals;

[0067] Divide the distance between each sensor pair by the corresponding time difference of arrival to obtain multiple sound speed calculation values;

[0068] The sound velocity field at the measurement point is obtained by performing least-squares fitting on multiple calculated sound velocity values.

[0069] Specifically, environmental parameter sensors are deployed in the detection area to acquire temperature data in real time. Atmospheric pressure and relative humidity The temperature sensor measures from -40℃ to 60℃ with an accuracy of ±0.5℃; the barometric pressure sensor measures from 80kPa to 110kPa with an accuracy of ±0.3kPa; and the humidity sensor measures from 0% to 100%RH with an accuracy of ±3%RH.

[0070] Calculate the ambient sound velocity field based on the physical relationship between environmental parameters and sound velocity. The relationship between sound speed and temperature, air pressure, and humidity is established using empirical formulas:

[0071] ;

[0072] in The unit is m / s. Temperature in Celsius This represents the relative humidity percentage. In the formula, 331.3 is the reference value for the speed of sound in dry air at 0℃, 0.606 is the temperature coefficient, and 0.0124 is the humidity coefficient.

[0073] This formula applies under standard atmospheric pressure; a correction term is introduced when the pressure deviates from the standard value. ,in The unit is kPa, and the coefficient 0.35 represents the change in sound speed caused by each 1 kPa deviation of air pressure from the standard value. The final ambient sound speed is... .

[0074] Ambient sound velocity field It reflects the macroscopic sound velocity distribution in the detection area and serves as the basic background value for the sound velocity field model.

[0075] Select from piezoelectric sensor array For sensor pairs with known spacing, the sound velocity at the measurement point is retrieved. The selection of the sensor pair satisfies the following conditions: there is a direct propagation path between the two sensors, and the spacing is... Precise measurements have been taken. Preferred. To improve the robustness of inversion for sensors.

[0076] Extract the first The time difference of arrival of the ultrasonic signals received by the sensor The time difference of arrival is estimated using the cross-correlation method: calculate the cross-correlation function of the two signals. ;

[0077] Among them, the first signal With the second signal The cross-correlation function is about the time delay variable. The function is called a function. Its value reflects the similarity between two signals at different time delays; the larger the value, the higher the matching degree between the two signals at that time delay. This is a time delay variable. This refers to the signal acquisition duration. For the first In the sensor, the ultrasonic time-domain signal received by the first sensor is time-varying. A continuously changing function. For the first In the sensor, the ultrasonic time-domain signal received by the second sensor is shifted forward on the time axis. The signal after a certain duration. The time delay corresponding to the peak value of the cross-correlation is... .

[0078] Calculate the speed of sound based on the relationship between spacing and time delay:

[0079] ;

[0080] in, For the first For the sensor, the precise and known distance between two sensors; For the first Estimated sound velocity between sensors.

[0081] right The calculated sound velocity values ​​are fitted using least squares to obtain the sound velocity field at the measurement point. :

[0082] ;

[0083] Solving for the given information That is, the arithmetic mean of the sound velocity estimates from multiple pairs of sensors.

[0084] in, The sound velocity is the fitting variable to be optimized. This represents the total number of sensor pairs involved in the calculation.

[0085] sound velocity field at measuring point It reflects the actual propagation speed of ultrasound within the detection area, including the effects of environmental non-uniformity and local disturbances.

[0086] Environmental sound velocity field With the sound velocity field at the measuring point Weighted fusion is performed to establish a dynamic sound velocity field model. The fusion weights are based on the spatial location and the distance from the center of the sensor array. The allocation uses a Gaussian decay function:

[0087] ;

[0088] in The attenuation scale parameter characterizes the range of influence of the sound velocity at the measuring point, and the preferred value is 0.3 to 0.5 times the maximum span of the sensor array.

[0089] any point in space The sound speed value is calculated through weighted fusion:

[0090] ;

[0091] in From this point to the array center The distance.

[0092] Dynamic sound field model Near the sensor array, the measured sound velocity is dominant, gradually transitioning to ambient sound velocity further away from the array, achieving a smooth fusion of measured data and theoretical models. When environmental parameters change, the ambient sound velocity... Real-time updates; when a new ultrasonic signal is received, the sound velocity at the measuring point... The dynamic sound velocity field model is then updated to adapt to the time-varying characteristics of the detection environment.

[0093] Furthermore, the partial discharge waveform dictionary library mentioned in S3 includes corona discharge waveform atoms, surface discharge waveform atoms, internal discharge waveform atoms, and suspended discharge waveform atoms; the time-domain expression of each waveform atom is composed of the product of the envelope function and the oscillation component, wherein the envelope function is selected as exponential decay type, Gaussian type, or double exponential type according to the discharge type, and the characteristic frequency of the oscillation component is set in the ultrasonic frequency band from 20kHz to 150kHz according to the discharge type.

[0094] The specific steps for performing time-frequency transformation on multi-channel time-domain signals to obtain time-frequency domain signals are as follows:

[0095] For multi-channel time-domain signal matrix Time-frequency transformation is performed to convert the signal from the time domain to the time-frequency domain to extract frequency features and time-varying characteristics. Short-Time Fourier Transform (STFT) is used for time-frequency analysis.

[0096] ;

[0097] in For the first The time-frequency domain signal of each channel, For window functions, For frequency, This is a dummy variable for integration time. A Hamming window is used, with a window length set to 512 to 1024 sampling points, corresponding to time lengths of 0.5 ms to 1 ms, achieving a balance between time and frequency resolution. The window overlap rate is set to 50% to ensure a smooth transition between the time and frequency domains.

[0098] After time-frequency transformation, the result is obtained Time-frequency domain signal of each channel This preserves the time evolution characteristics and spectral distribution information of the signal, providing a foundation for subsequent multi-source separation and waveform matching.

[0099] S3 also includes the following steps:

[0100] Establish a partial discharge waveform dictionary. It includes waveform atoms of four typical discharge types: corona discharge waveform atoms, surface discharge waveform atoms, internal discharge waveform atoms, and suspended discharge waveform atoms. The waveform atoms of each type of discharge were obtained through experimental measurement and theoretical modeling.

[0101] The time-domain expression of a waveform atom is given by the envelope function. With oscillating components The product is composed of:

[0102] ;

[0103] envelope function The amplitude attenuation characteristics of the discharge signal are characterized by selecting different forms according to the discharge type. Corona discharge and surface discharge employ an exponentially decaying envelope:

[0104] ;

[0105] in The initial amplitude, The attenuation coefficient is the corona discharge coefficient. Typical value to Surface discharge Typical value ; It is a time variable.

[0106] The internal discharge uses a Gaussian envelope:

[0107] ;

[0108] in At peak time, For time-stretching parameters, internal discharge Typical values ​​are 20 μs to 50 μs.

[0109] Suspension discharge employs a double exponential envelope:

[0110] ;

[0111] in and There are two attenuation coefficients. Floating discharge Typical value to Typical value to .

[0112] Oscillating components The frequency characteristics of the discharge signal are characterized by a sinusoidal oscillation.

[0113] ;

[0114] in For characteristic frequencies, This is the initial phase. The characteristic frequency is set within the ultrasonic frequency band according to the discharge type: corona discharge... Surface discharge at frequencies of 40 kHz to 60 kHz Internal discharge at 50kHz to 80kHz Suspension discharge at frequencies of 30kHz to 50kHz The frequency range is from 60kHz to 100kHz. The difference in frequency range stems from the different dominant frequencies of the sound waves generated by different discharge mechanisms.

[0115] Waveform dictionary library Composed of multiple waveform atoms The structure is formed by changing the envelope parameters and oscillation frequency to generate different atoms, thus creating an overcomplete dictionary. The number of atoms in the dictionary is preferably between 100 and 500.

[0116] Furthermore, the joint optimization objective function in S3 integrates three constraints through weighting coefficients: the error between the array received signal and the model reconstructed signal, the deviation between the discharge power supply attenuation parameter and the spherical wave propagation theory, and the residual between the discharge power supply waveform and the sparse representation of the dictionary atoms.

[0117] The joint optimization objective function includes:

[0118] The signal reconstruction error term is the error between the signal received by the piezoelectric sensor array and the reconstructed signal. The reconstructed signal is calculated based on the waveform, position, and dynamic sound velocity field model of the partial discharge source.

[0119] The spherical wave attenuation constraint term is the deviation between the attenuation parameter of the local discharge source and the theoretical value of spherical wave attenuation.

[0120] The waveform dictionary constraint term is the residual between the waveform of the partial discharge source and the representation of waveform atoms in the partial discharge waveform dictionary.

[0121] The joint optimization objective function integrates signal reconstruction error constraints, spherical wave attenuation constraints, and waveform dictionary constraints through a weighted summation method.

[0122] Specifically, a joint optimization objective function is established. The signal reconstruction error term, spherical wave attenuation constraint term, and waveform dictionary constraint term are integrated:

[0123] ;

[0124] in and These are weighting coefficients used to balance the importance of the three constraints.

[0125] Signal reconstruction error term Measuring the error between the signal received by the sensor array and the signal reconstructed by the model:

[0126] ;

[0127] in For the first The actual time-domain signal received by each sensor For the first The time-domain signal is reconstructed using a model of a single sensor. This represents the total number of channels in the piezoelectric sensor array, i.e., the total number of sensors.

[0128] The reconstructed signal is calculated based on assumed partial discharge source parameters: Assume... The first power supply, the second The location of the source is The waveform is Then the first The reconstructed signals from the sensors are:

[0129] ;

[0130] in The distance from the source to the sensor. Let be the three-dimensional spatial coordinates of the s-th local discharge source; Let be the three-dimensional spatial coordinates of the i-th piezoelectric sensor; To account for propagation delay, the integration path follows the direct propagation path. The sound speed value given by the dynamic sound speed field model. For the first Attenuation parameters of each source.

[0131] Spherical wave attenuation constraint The constrained attenuation parameters conform to the propagation laws of spherical waves:

[0132] ;

[0133] in Let be the theoretical attenuation coefficient of a spherical wave. For reference distance, the distance from the center of the sensor array to the source is typically taken. This constraint term affects the attenuation parameter. It is physically reasonable.

[0134] Waveform dictionary constraint terms The constrained source waveform can be sparsely represented by atoms in the dictionary:

[0135] ;

[0136] in For the first in the dictionary One atom, For the first A sparse representation coefficient vector of the power supply waveform. ; For the first The power supply waveform of the first discharge is related to the first The representation coefficients of each dictionary atom; The total number of atoms in the dictionary. This is the sparse regularization coefficient, with a value ranging from 0.01 to 0.5. Norm regularization term By penalizing the number of non-zero coefficients, we can improve the representation of the coefficient vector. The sparsity of the source waveform means that it is mainly represented by a linear combination of a few dictionary atoms, which reflects the typicality of the partial discharge waveform.

[0137] Weighting coefficient and Configure according to actual application requirements: The value ranges from 0.01 to 0.1. The value ranges from 0.1 to 1.0. When the speed of sound fluctuates significantly, it should be increased appropriately. To strengthen physical constraints; when the signal-to-noise ratio is low, it should be appropriately increased. This strengthens waveform matching constraints. In practical applications, weight values ​​can be optimized using cross-validation.

[0138] Joint optimization objective function Taking into account data fitting accuracy, physical propagation laws, and prior knowledge of discharge waveforms, a complete optimization criterion is provided for subsequent blind source separation.

[0139] Furthermore, in S4, an alternating iterative optimization method is used to solve for blind source separation, specifically including the following steps:

[0140] S41. Set the number of partial discharge sources to be separated, and initialize the position, waveform and attenuation parameters of each partial discharge source;

[0141] S42. Fix the position and attenuation parameters of all partial discharge sources, and optimize and update the waveform of each partial discharge source according to the joint optimization objective function;

[0142] S43. Fix the waveforms of all partial discharge sources, and optimize and update the position and attenuation parameters of each partial discharge source according to the joint optimization objective function and dynamic sound velocity field model.

[0143] S44. Repeat steps S42 to S43 until the joint optimization objective function value converges, and output the final waveform, spatial location and attenuation parameters of each local amplifier power supply.

[0144] Set the number of partial discharge sources to be separated Source number estimation methods include: after performing time-frequency transformation on the received signal, counting the number of independent peaks in the time-frequency energy distribution; or using information criteria (such as AIC, BIC) to evaluate the goodness of fit of the model under different source number assumptions, and selecting the source number that minimizes the criterion function. For typical overhead line scenarios, setting S=2 to 5 satisfies most multi-source discharge conditions.

[0145] Initialize the parameters of each power source, including its spatial location. Waveform and attenuation parameters superscript This represents the initial value. The initial position value is estimated using the energy centroid method, which calculates the energy of the signals received by each sensor. ,in For signal acquisition duration, For the first The partial discharge ultrasonic time-domain signal, synchronously acquired by each sensor, varies with time. A continuously changing function.

[0146] The sensor coordinates are weighted and averaged using energy as the weight to obtain the initial position estimate. When multiple sources need to be initialized, the time-frequency energy distribution is clustered, and each cluster center corresponds to the initial position of a source.

[0147] The initial waveform value is obtained from the waveform dictionary. Typical atoms were randomly selected from the sample. The initial value of the attenuation parameter was set according to the spherical wave attenuation theory. ,in This represents the distance from the initial position to the center of the sensor array.

[0148] The joint optimization objective function is solved using an alternating direction iterative optimization method. The optimization process divides the parameters into two groups, the first group being the source waveform. The second group is the source location. and attenuation parameters .

[0149] In the waveform optimization step, the positions of all sources are fixed. and attenuation parameters Optimize the waveforms of each source At this point, the joint optimization objective function This simplifies to a quadratic optimization problem involving the waveform. For the... Each source utilizes waveform dictionary constraints, where the waveform is represented as a linear combination of dictionary atoms. ,in For the first The representation coefficient vector of each source For the first The optimization objective is to find the sparse representation coefficients of the partial discharge source waveform for the k-th dictionary atom. :

[0150] ;

[0151] in, For the (n+1)th iteration, the... Optimization results of sparse representation coefficients for individual power supplies; superscript , The iteration number is used as an identifier, representing the iteration number. sequence The result of the next iteration; This is the traversal sequence number of the discharge source, used to traverse all discharge sources to be separated.

[0152] This optimization objective applies to all The received signals from each sensor are jointly fitted to ensure that the reconstructed signal matches the actual received signal on all channels. The sparse representation coefficients are solved using either the Orthogonal Matching Pursuit (OMP) algorithm or the Lasso algorithm.

[0153] In the position and attenuation parameter optimization step, the waveforms of all sources are fixed. Optimize location and attenuation parameters The optimization objective at this point is:

[0154] ;

[0155] in, For the (n+1)th iteration, the... The optimization results of the spatial location and attenuation parameters of the power supply. These are the weighting coefficients.

[0156] Optimization of position parameters involves nonlinear propagation delay. The calculation requires a dynamic sound velocity field model. Integral down the propagation path. Optimization is performed using gradient descent or a quasi-Newton method (such as L-BFGS). The gradient of the objective function with respect to location is calculated through numerical or automatic differentiation. Attenuation parameters... The optimization objective is to minimize Regarding the objective function... Taking the partial derivative and setting it to zero, and combining it with the spherical wave attenuation constraint, we obtain the analytical solution:

[0157] ;

[0158] This solution balances signal reconstruction accuracy with physical attenuation constraints. During position optimization, a dynamic sound velocity field model is used to calculate the propagation delay, which is discretized along the straight path from the source to the sensor. Each line segment has a length of [number]. ,in The value is preferably between 20 and 100 to balance calculation accuracy and efficiency. The sound speed is taken as the value at the midpoint, and the total time delay is... .

[0159] The iteration termination conditions include the convergence criterion and the maximum number of iterations. The convergence criterion is that the relative change in the objective function is less than a threshold. The expression is:

[0160] ;

[0161] , The total objective function values ​​for the nth and (n+1)th iterations; the convergence threshold. Set as to The maximum number of iterations is set to 50 to 200 to prevent infinite loops. Iteration stops when the convergence condition is met or the maximum number of iterations is reached.

[0162] After the iterative optimization converges, the final parameters of each local discharge power source are output, including the source waveform. Spatial location , , and attenuation parameters These parameters comprehensively describe the spatiotemporal characteristics and propagation attenuation characteristics of multiple discharge sources, providing a foundation for subsequent discharge type identification and intensity calculation.

[0163] Furthermore, in S5, the similarity between the partial discharge power source waveform obtained by normalizing the cross-correlation coefficient and various types of discharge atoms in the partial discharge waveform dictionary is calculated, and the category with the highest similarity is selected as the discharge type identification result; when the maximum similarity is less than the preset threshold of 0.5 to 0.7, it is determined to be an unidentified type.

[0164] The waveforms of each source obtained by blind source separation With partial discharge waveform dictionary The matching process identifies the discharge type by comparing the source waveform with various discharge atoms in the dictionary and selecting the category with the highest similarity as the identification result.

[0165] Similarity is measured using the normalized cross-correlation coefficient, for the th The source waveform is calculated, and its relationship with the first source waveform is determined. Typical atoms of discharge-like processes Similarity:

[0166] ;

[0167] in For signal acquisition duration, similarity The value ranges from 0 to 1, with a larger value indicating a better waveform match. Similarity is calculated for four types of discharges: corona discharge, surface discharge, internal discharge, and suspension discharge. The category corresponding to the highest similarity is then selected.

[0168] ;

[0169] Category 1 represents corona discharge, Category 2 represents surface discharge, Category 3 represents internal discharge, and Category 4 represents suspension discharge. The maximum similarity... If the value is less than the threshold of 0.6, it is considered an unidentified type and requires further manual analysis. The threshold of 0.6 is set based on the reliability requirements of waveform matching, and in practical applications, it can be adjusted within the range of 0.5 to 0.7 according to the required recognition accuracy.

[0170] Furthermore, the calculation of the discharge intensity in S5 specifically includes the following steps:

[0171] S51. Extract the peak amplitude of the partial discharge power supply waveform;

[0172] S52. Based on the attenuation parameters and the propagation distance from the partial discharge source position to the reference position, attenuation compensation is performed on the peak amplitude to obtain the compensated amplitude.

[0173] S53. Calculate the apparent discharge quantity based on the compensated amplitude and compare it with the preset intensity level threshold to classify the discharge intensity level.

[0174] The discharge intensity is calculated based on the source waveform and attenuation parameters, reflecting the severity of the discharge. The discharge intensity is characterized by the apparent discharge quantity, and the calculation process includes three steps: peak extraction, attenuation compensation, and intensity grading.

[0175] First, extract the source waveform. peak amplitude :

[0176] ;

[0177] The peak amplitude is the equivalent received amplitude of the discharge source at the sensor array, which already includes the attenuation effect during propagation.

[0178] To obtain the true amplitude at the discharge source, attenuation compensation is required. This is based on the attenuation parameters. and source location Propagation distance to the reference position Calculate the amplitude after compensation :

[0179] ;

[0180] The reference position is selected as the geometric center of the sensor array, and the reference distance is... The Euclidean distance from the source location to the array center:

[0181] ;

[0182] in( These are the geometric center coordinates of the sensor array (i.e., the array center position defined in the aforementioned sound velocity field model). The compensated amplitude. The influence of propagation distance differences is eliminated, reflecting the intrinsic strength of the discharge source.

[0183] Calculate the apparent discharge quantity based on the compensated amplitude. There is an empirical relationship between apparent discharge quantity and ultrasonic signal amplitude:

[0184] ;

[0185] in This is the proportionality coefficient, determined through a standard discharge power supply calibration experiment; a typical value is [value missing]. .index Reflecting nonlinear relationships, typical values ​​range from 1.2 to 1.5. For linear approximations, Take 1, at this time Apparent discharge quantity The unit is picocoulomb (pC).

[0186] The apparent discharge quantity is compared with a preset intensity level threshold to classify the discharge intensity level. The classification standard is set with reference to power industry standards, with slight discharge being the preferred option. Moderate discharge Severe discharge and dangerous discharge In practical applications, the threshold can be adjusted according to the voltage level and operating environment of the specific line.

[0187] Output complete diagnostic information for each partial discharge source, including spatial location, discharge type, and discharge intensity. For the first... A power supply outputs data organized into three-dimensional coordinates. , , Discharge type identification (Corona / Surface / Internal / Suspension) and Discharge Intensity And its severity level (mild / moderate / severe / dangerous).

[0188] The diagnostic results are presented in tabular or graphical form. Tables list the serial number, coordinates, type, and intensity of each source. Graphical representations mark the source locations in three-dimensional space, using different colors or symbols to distinguish discharge types, and using symbol size to indicate the discharge intensity level. The output provides a basis for operational and maintenance decisions; high-intensity discharge sources require priority handling.

[0189] Example 2:

[0190] The present invention also provides a partial discharge detection system for overhead power lines based on a piezoelectric array, including an array signal acquisition module, a sound velocity field reconstruction module, an objective function construction module, a blind source separation and solution module, and a discharge diagnosis output module;

[0191] The array signal acquisition module is used to arrange a piezoelectric sensor array in the area to be detected, and synchronously acquire the ultrasonic signal generated by partial discharge to obtain multi-channel time domain signal.

[0192] The sound velocity field reconstruction module is used to acquire environmental parameters of the detection area, calculate the environmental sound velocity field based on the environmental parameters, measure the ultrasonic wave propagation time between the piezoelectric sensor arrays, invert the sound velocity field at the measuring point, and fuse the environmental sound velocity field and the sound velocity field at the measuring point to establish a dynamic sound velocity field model.

[0193] The objective function construction module is used to perform time-frequency transformation on multi-channel time-domain signals to obtain time-frequency domain signals; based on the partial discharge waveform dictionary library, a joint optimization objective function is established, which includes signal reconstruction error terms, spherical wave attenuation constraint terms, and waveform dictionary constraint terms;

[0194] The blind source separation solution module is used to perform blind source separation on time-frequency domain signals based on a dynamic sound velocity field model and a joint optimization objective function, and obtain the waveforms, spatial locations and attenuation parameters of multiple local discharge sources through iterative optimization solution;

[0195] The discharge diagnostic output module is used to match the separated partial discharge power source waveform with the partial discharge waveform dictionary to identify the discharge type; calculate the discharge intensity based on the waveform and attenuation parameters; and output the spatial location, discharge type, and discharge intensity of multiple partial discharge power sources.

[0196] The following is an application example of the present invention:

[0197] A certain 220kV transmission line traverses mountainous and plain areas, with significant elevation variations and complex and variable climate conditions along the route. In the spring of 2025, the line repeatedly experienced abnormal discharge signals, leading maintenance personnel to suspect multiple insulator flashover hazards. However, traditional fixed sound velocity location methods were unable to accurately pinpoint the fault due to significant deviations caused by fluctuations in environmental sound velocity. Therefore, a method was adopted... Figure 2 The present invention, as shown in the flowchart, uses a piezoelectric array-based method for detecting partial discharge in overhead power lines.

[0198] A three-dimensional array containing eight piezoelectric sensors was deployed around the target tower, spanning approximately 4 meters and covering the insulator string area on both sides of the tower. The monitoring period was from afternoon to evening, during which the ambient temperature decreased from 28℃ to 18℃, the relative humidity increased from 45% to 75%, and the air pressure remained relatively stable at around 95 kPa. The system simultaneously acquired multi-channel ultrasonic signals at a sampling frequency of 1 MHz, with a single acquisition duration of 50 ms.

[0199] Temperature, air pressure, and humidity data are acquired in real time by environmental parameter sensors to calculate the ambient sound velocity field. Simultaneously, the sound velocity field at the measuring points is retrieved using the ultrasonic wave propagation time between the sensor arrays, and the two are fused to establish a dynamic sound velocity field model. During the detection process, the sound velocity gradually changes from 346 m / s to 342 m / s, fully demonstrating the impact of environmental changes on the sound velocity.

[0200] The acquired signals were subjected to short-time Fourier transform to obtain time-frequency domain signals. A joint optimization objective function was constructed and solved through alternating iterative optimization. After 68 iterations, the objective function converged, successfully separating three partial discharge sources. The discharge type was identified by matching the waveform with a dictionary, and the discharge intensity was calculated based on attenuation compensation.

[0201] To verify the effectiveness of the present invention, the detection results were compared with those of the traditional time delay difference positioning method (using a fixed sound velocity of 340 m / s), and the results are shown in Table 1.

[0202] Based on the test results, maintenance personnel addressed the issues at the B-phase crossarm joint and the A-phase insulator string, replacing aging joint hardware and cleaning contaminated insulators, thus eliminating the potential for discharge. Subsequent monitoring showed that the abnormal signal disappeared, verifying the accuracy of the test results.

[0203] This embodiment demonstrates that the present invention effectively compensates for sound speed fluctuations caused by environmental changes through a dynamic sound speed field model, significantly improving positioning accuracy compared to traditional fixed sound speed methods. Furthermore, the joint optimization framework successfully separates multiple discharge sources, overcoming the limitation of the traditional time delay difference method in multi-source scenarios where only the strongest signal can be identified, thus providing a reliable basis for on-site fault diagnosis.

[0204] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detection of partial discharges in overhead power lines based on a piezoelectric array, characterized in that, Includes the following steps: S1. Arrange a piezoelectric sensor array in the area to be detected, and synchronously collect ultrasonic signals generated by partial discharge to obtain multi-channel time-domain signals; S2. Obtain the environmental parameters of the detection area and calculate the ambient sound velocity field based on the environmental parameters; measure the ultrasonic wave propagation time between the piezoelectric sensor arrays and invert the sound velocity field at the measuring point; fuse the ambient sound velocity field and the sound velocity field at the measuring point to establish a dynamic sound velocity field model. S3. Perform time-frequency transformation on the multi-channel time-domain signal to obtain the time-frequency domain signal; based on the partial discharge waveform dictionary, establish a joint optimization objective function that includes signal reconstruction error terms, spherical wave attenuation constraint terms, and waveform dictionary constraint terms; S4. Based on the dynamic sound velocity field model and joint optimization objective function, blind source separation is performed on the time-frequency domain signal. The waveforms, spatial locations, and attenuation parameters of multiple local discharge sources are obtained through iterative optimization. S5. Match the obtained partial discharge source waveforms with the partial discharge waveform dictionary to identify the discharge type; calculate the discharge intensity based on the waveform and attenuation parameters; and output the spatial location, discharge type, and discharge intensity of multiple partial discharge sources.

2. The piezoelectric array based overhead power line arcing detection method of claim 1, wherein, The piezoelectric sensor array in S1 contains at least four piezoelectric sensors. The spacing between the sensors is set according to the monitoring range, and the frequency band covers the typical frequency band of partial discharge ultrasound. The acquisition process uses a unified clock to synchronously trigger multi-channel sampling. The sampling frequency meets the Nyquist sampling requirements. The multi-channel time-domain signals obtained by synchronous sampling are organized into a matrix form and used as input for subsequent processing.

3. The method for detecting partial discharge of overhead power lines based on piezoelectric arrays according to claim 1, characterized in that, The specific steps for S2 are as follows: S21. Obtain the temperature, air pressure, and relative humidity of the detection area, and calculate the ambient sound velocity field based on the physical relationship between environmental parameters and sound velocity. S22. Select multiple pairs of sensors with known spacing in the piezoelectric sensor array, measure the arrival time difference of the ultrasonic signal, and obtain the sound velocity field at the measuring point based on the spacing and arrival time difference. S23. The ambient sound velocity field and the sound velocity field at the measuring point are weighted and fused, with the weights allocated according to the spatial distance, to establish a dynamic sound velocity field model.

4. The method for detecting partial discharge of overhead power lines based on piezoelectric arrays according to claim 3, characterized in that, The inversion process of the sound velocity field at the measuring point in S22 specifically includes: Select at least three pairs of sensor pairs with known spacing in the piezoelectric sensor array, and ensure that there is a direct ultrasonic wave propagation path between the sensor pairs; The arrival time difference of the ultrasonic signals received by each sensor is extracted, and the arrival time difference is determined by the peak value of the cross-correlation function of the two signals; Divide the distance between each sensor pair by the corresponding time difference of arrival to obtain multiple sound speed calculation values; The sound velocity field at the measurement point is obtained by performing least-squares fitting on multiple calculated sound velocity values.

5. The method for detecting partial discharge of overhead power lines based on piezoelectric arrays according to claim 1, characterized in that, The partial discharge waveform dictionary library mentioned in S3 includes corona discharge waveform atoms, surface discharge waveform atoms, internal discharge waveform atoms, and suspended discharge waveform atoms; the time-domain expression of each waveform atom is composed of the product of the envelope function and the oscillation component, wherein the envelope function is selected as exponential decay type, Gaussian type, or double exponential type according to the discharge type, and the characteristic frequency of the oscillation component is set in the ultrasonic frequency band from 20kHz to 150kHz according to the discharge type.

6. The method for detecting partial discharge of overhead power lines based on piezoelectric arrays according to claim 1, characterized in that, The joint optimization objective function in S3 integrates three constraints through weighting coefficients: the error between the array received signal and the model reconstructed signal, the deviation between the discharge power supply attenuation parameter and the spherical wave propagation theory, and the residual between the discharge power supply waveform and the sparse representation of the dictionary atoms.

7. The method for detecting partial discharge of overhead power lines based on piezoelectric arrays according to claim 1, characterized in that, S4 employs an alternating iterative optimization method for blind source separation, specifically including the following steps: S41. Set the number of partial discharge sources to be separated, and initialize the position, waveform and attenuation parameters of each partial discharge source; S42. Fix the position and attenuation parameters of all partial discharge sources, and optimize and update the waveform of each partial discharge source according to the joint optimization objective function; S43. Fix the waveforms of all partial discharge sources, and optimize and update the position and attenuation parameters of each partial discharge source according to the joint optimization objective function and dynamic sound velocity field model. S44. Repeat steps S42 to S43 until the joint optimization objective function value converges, and output the final waveform, spatial location and attenuation parameters of each local amplifier power supply.

8. The method for detecting partial discharge of overhead power lines based on piezoelectric arrays according to claim 1, characterized in that, In S5, the similarity between the partial discharge source waveform obtained by normalizing the cross-correlation coefficient and various types of discharge atoms in the partial discharge waveform dictionary is calculated, and the category with the highest similarity is selected as the discharge type identification result; when the maximum similarity is less than the preset threshold of 0.5 to 0.7, it is determined to be an unidentified type.

9. The method for detecting partial discharge of overhead power lines based on piezoelectric arrays according to claim 1, characterized in that, The calculation of discharge intensity in S5 specifically includes the following steps: S51. Extract the peak amplitude of the partial discharge power supply waveform; S52. Based on the attenuation parameters and the propagation distance from the partial discharge source position to the reference position, attenuation compensation is performed on the peak amplitude to obtain the compensated amplitude. S53. Calculate the apparent discharge quantity based on the compensated amplitude and compare it with the preset intensity level threshold to classify the discharge intensity level.

10. A partial discharge detection system for overhead power lines based on piezoelectric arrays, characterized in that, It includes an array signal acquisition module, a sound velocity field reconstruction module, an objective function construction module, a blind source separation and solution module, and a discharge diagnosis output module; The array signal acquisition module is used to arrange a piezoelectric sensor array in the area to be detected, and synchronously acquire the ultrasonic signal generated by partial discharge to obtain multi-channel time domain signal. The sound velocity field reconstruction module is used to acquire environmental parameters of the detection area, calculate the environmental sound velocity field based on the environmental parameters, measure the ultrasonic wave propagation time between the piezoelectric sensor arrays, invert the sound velocity field at the measuring point, and fuse the environmental sound velocity field and the sound velocity field at the measuring point to establish a dynamic sound velocity field model. The objective function construction module is used to perform time-frequency transformation on multi-channel time-domain signals to obtain time-frequency domain signals; based on the partial discharge waveform dictionary library, a joint optimization objective function is established, which includes signal reconstruction error terms, spherical wave attenuation constraint terms, and waveform dictionary constraint terms; The blind source separation solution module is used to perform blind source separation on time-frequency domain signals based on a dynamic sound velocity field model and a joint optimization objective function, and obtain the waveforms, spatial locations and attenuation parameters of multiple local discharge sources through iterative optimization solution; The discharge diagnostic output module is used to match the separated partial discharge power source waveform with the partial discharge waveform dictionary to identify the discharge type; calculate the discharge intensity based on the waveform and attenuation parameters; and output the spatial location, discharge type, and discharge intensity of multiple partial discharge power sources.

Images