Transformer partial discharge ultrasonic three-dimensional positioning method, device and equipment
By arranging ultrasonic sensors on the outer wall of the transformer, filtering interference signals and processing signal envelopes, and combining this with propagation delay superposition, the problem of accurate three-dimensional localization of partial discharge inside the transformer was solved, achieving high-precision localization of partial discharge sources.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-04-03
AI Technical Summary
Existing technologies make it difficult to achieve precise three-dimensional localization of partial discharge inside transformers, especially when the core and windings are obstructed, leading to localization failure.
By arranging multiple ultrasonic sensors on the outer wall of the transformer to collect raw data, filtering interference signals, extracting the signal envelope and reversing the time axis, and generating an energy focusing signal based on the propagation delay superposition, the location of the partial discharge source is determined.
High-precision three-dimensional partial discharge location inside the transformer has been achieved, which can accurately locate the partial discharge source even when the iron core and windings are blocked, thus improving the accuracy and reliability of the location.
Smart Images
Figure CN121784468A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system technology, and in particular to a method, apparatus, and electronic device for ultrasonic three-dimensional localization of partial discharge in transformers. Background Technology
[0002] Power transformers are core equipment in the power grid, undertaking the crucial functions of voltage transformation and power transmission, and serving as the fundamental support for the power transmission and distribution system. Power transformers have complex internal structures, and their maintenance is large-scale and costly. A failure in a power transformer can cause incalculable economic losses and safety risks to the power grid system. Therefore, ensuring their operational reliability is of paramount importance to the overall safety of the power grid.
[0003] Partial discharge is a major early sign of transformer insulation degradation. It refers to a small, non-penetrating discharge phenomenon that occurs in a local area of the insulation structure of a power transformer because the electric field strength exceeds the breakdown field strength of the insulation medium. It has the characteristics of gradual and destructiveness. If partial discharge can be accurately located, transformer failures can be effectively prevented, operation and maintenance levels can be improved, and reliable power supply to the power grid can be guaranteed. Therefore, ultrasonic localization technology for partial discharge has become a core direction in the field of power equipment operation and maintenance.
[0004] The mainstream three-dimensional localization technology for partial discharge in transformers currently uses the Time Difference of Arrival (TDoA) method. The TDoA method calculates the coordinates of the partial discharge source by establishing a system of equations based on the time difference between the arrival times of ultrasonic signals at different sensors. However, this method relies on the accurate extraction of the arrival time of the direct wave. Since components inside the transformer, such as the core and windings, can easily block ultrasonic waves, some sensors may fail to receive the direct wave and only receive the reflected wave with a longer path. In such cases, the TDoA method cannot obtain an effective time difference, leading to localization failure.
[0005] Therefore, for power transformers, how to achieve precise location of partial discharge is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0006] The purpose of this invention is to provide at least one ultrasonic three-dimensional positioning method, device, and electronic device for transformer partial discharge, which can achieve high-precision three-dimensional positioning of transformer partial discharge.
[0007] To address the aforementioned technical problems, at least one embodiment of this application provides a method for ultrasonic three-dimensional localization of partial discharge in transformers, comprising: Acquire raw data, including partial discharge signals, collected by each ultrasonic sensor installed on the outer wall of the transformer; The original data is filtered for interference signals to obtain a preprocessed time-domain partial discharge signal array; Extract the signal envelope of the time-domain partial discharge signal array; Perform a time-axis reversal operation on the signal envelope to obtain the inverted envelope signal; The energy focusing signal is obtained by superimposing the inversion envelope signal on the propagation delay between each location and the sensor. The location corresponding to the signal with the highest energy in the energy-focused signal is determined as the location of the partial discharge source.
[0008] In one embodiment, extracting the signal envelope of the time-domain partial discharge signal array includes: By using a finite impulse response filter, an approximate simulation of a 90° phase shift is performed on the time-domain partial discharge signal array to generate an orthogonal accompanying signal array; An analytical signal is constructed based on the time-domain partial discharge signal array and the orthogonal adjoint signal array; The amplitude of the analyzed signal is extracted and used as the signal envelope.
[0009] In one embodiment, prior to the approximate simulation of the time-domain partial discharge signal array with a 90° phase shift using a finite impulse response filter, the method further includes: The fundamental coefficients of the finite impulse response filter are multiplied one by one with the corresponding indexed Hanning window function values to correct the fundamental coefficients; The Hanning window function value is determined based on w[n] = 0.5. 0.5×cos(2πn / (M 1) Determine the corresponding array index position; n is the array index of the coefficients of the finite impulse response filter, and M is the length of the finite impulse response filter.
[0010] In one embodiment, the step of performing an approximate simulation of the time-domain partial discharge signal array with a 90° phase shift using a finite impulse response filter includes: A linear convolution operation is performed on the time-domain partial discharge signal array and the coefficients of the finite impulse response filter to obtain the convolution result; The convolution result is truncated, and the middle part, which has the same length as the temporal partial discharge signal array, is retained as the orthogonal companion signal array.
[0011] In one embodiment, filtering the original data for interference signals includes: The original data is truncated according to the preset pulse observation window width to obtain a single-window partial discharge signal; In the single-window partial discharge signal, small-amplitude broad-spectrum clutter with amplitudes below the amplitude threshold is filtered out to obtain the first filtered signal; For the first filtered signal, interference signals whose frequencies are not within the effective frequency range are filtered out to obtain the second filtered signal; The second filtered signal is used as the preprocessed time-domain partial discharge signal array.
[0012] In one embodiment, superimposing the inverted envelope signal based on the propagation delay between each location and the sensor includes: Construct a numerical model of the propagation of partial discharge signals inside the transformer; A grid is established in the propagation numerical model; Calculate the signal propagation delay from each sensor to each grid point within the grid; The inversion envelope signals corresponding to each sensor are superimposed according to the propagation delay to obtain the focusing signal of each grid point, which is used as the energy focusing signal.
[0013] In one embodiment, calculating the signal propagation delay from each sensor to each grid point within the grid includes: Determine the internal medium of the transformer along the propagation path from each sensor to the corresponding grid point; Obtain the signal propagation speed corresponding to each of the aforementioned media; Divide the path into segments along the medium it passes through, calculate the ratio of the length of each segment to the corresponding medium propagation speed, and obtain the delay of each segment. The total signal propagation delay from the sensor to the corresponding grid point is obtained by summing the delays of each path segment.
[0014] In one embodiment, before determining the location corresponding to the signal with the highest energy in the energy-focused signal as the location of the partial discharge source, the method further includes: Determine the coordinates corresponding to the energy focusing signal and the sensor coordinates, and calculate the theoretical time difference between the arrival of the signal at any two sensors; Extract the actual time difference between the arrival of the signal at the two sensors from the time-domain partial discharge signal array; Determine whether the deviation between the theoretical time difference and the actual time difference exceeds a preset range; If the peak exceeds the limit, the corresponding energy focusing signal will be identified as a spurious peak and removed.
[0015] At least one embodiment of this application also provides an ultrasonic three-dimensional positioning device for partial discharge of transformers, comprising: The data acquisition unit is used to acquire raw data containing partial discharge signals collected by various ultrasonic sensors installed on the outer wall of the transformer. A signal filtering unit is used to filter interference signals from the original data to obtain a preprocessed time-domain partial discharge signal array. An envelope extraction unit is used to extract the signal envelope of the time-domain partial discharge signal array; An envelope inversion unit is used to perform a time-axis reversal operation on the signal envelope to obtain an inverted envelope signal. The delay superposition unit is used to superimpose the inversion envelope signal according to the propagation delay between each location and the sensor to obtain the energy focusing signal; The energy focusing unit is used to determine the location of the partial discharge source corresponding to the signal with the highest energy in the energy focusing signal.
[0016] At least one embodiment of this application also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the above-described ultrasonic three-dimensional localization method for partial discharge of transformers.
[0017] The ultrasonic three-dimensional localization method for partial discharge in transformers provided in this application collects raw data by arranging multiple ultrasonic sensors on the outer wall of the transformer. The raw data is first filtered to remove small-amplitude clutter and invalid frequency band interference, and then the signal envelope is extracted to filter out high-frequency alternating fluctuations. This dual processing preserves the core temporal characteristics of the partial discharge pulse while avoiding noise masking key information, solving the problem of positioning errors caused by noise when using the raw signal directly. Then, the signal envelope is time-reversed, and a signal backtracking path is constructed based on the reversibility of wave propagation. The energy-focusing signal is generated by superimposing the inverted envelope signal according to the propagation delay. This method is based on the reversibility of wave propagation and preserves the path timing and full path delay through the signal envelope. By calculating and matching the reflection path and superimposing to enhance energy focusing, the reflected wave is transformed into a positioning basis equivalent to the direct wave. Regardless of the path the ultrasonic wave takes to reach the sensor, as long as the signal can be captured and its propagation delay calculated, the location of the true partial discharge source with the highest energy can be found through inversion and superposition. This method does not rely on the direct wave. Even if the transformer core and windings block the ultrasonic wave, causing some sensors to only receive the reflected wave, the signal can still be synergistically enhanced at the location of the true partial discharge source through delay matching. The location corresponding to the signal with the highest energy in the focused signal is determined as the location of the partial discharge source. This determination result corresponds to the discrete spatial location in the propagation numerical model, which can be directly converted into specific spatial coordinates, making it easy for engineers to quickly locate the partial discharge source and formulate maintenance plans. Attached Figure Description
[0018] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.
[0019] Figure 1This is a flowchart of an ultrasonic three-dimensional localization method for partial discharge in a transformer, provided in one embodiment of this application; Figure 2 This is a schematic diagram of a positioning layout coordinate system provided in one embodiment of this application; Figure 3 This is a schematic diagram of a sensor installation provided in one embodiment of this application; Figure 4 This is a schematic diagram of a network structure provided in one embodiment of this application; Figure 5 This is a sensor receiving signal diagram provided in one embodiment of this application; Figure 6 This is a schematic diagram of an ultrasonic three-dimensional positioning device for partial discharge of a transformer provided in one embodiment of this application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the various embodiments of this application will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details have been provided in the various embodiments of this application to help readers better understand this application. However, the technical solutions claimed in this application can be implemented even without these technical details and various changes and modifications based on the following embodiments. The division of the various embodiments below is for the convenience of description and should not constitute any limitation on the specific implementation of this application. The various embodiments can be combined with and referenced by each other without contradiction.
[0021] This invention proposes a method for ultrasonic three-dimensional localization of partial discharge in transformers. The implementation details of the ultrasonic three-dimensional localization method for partial discharge in transformers in this embodiment are described below. The following implementation details are provided for ease of understanding and are not necessary for implementing this solution.
[0022] Example 1: The specific process of the ultrasonic three-dimensional localization method for partial discharge of transformers in this embodiment can be as follows: Figure 1 As shown, it includes: Step 101: Obtain the raw data containing partial discharge signals collected by each ultrasonic sensor installed on the outer wall of the transformer.
[0023] Multiple ultrasonic sensors are arranged at predetermined positions on the outer wall of the transformer. The sensors capture ultrasonic signals excited by partial discharge phenomena inside the transformer and propagating to the outer wall in real time. At the same time, initial data including the target partial discharge signal and on-site environmental noise (such as electromagnetic interference, equipment vibration noise, etc.) are collected as raw data.
[0024] The sensors are non-invasively deployed on the outer wall, eliminating the need to disassemble or modify the transformer and avoiding interference with normal equipment operation. Sensor placement can be referenced... Figure 2 The diagram showing the positioning layout coordinates and Figure 3 The sensor installation diagram shown uses the location of the outer wall with the largest amplitude of the obtained ultrasonic signal as the origin coordinate. 1-9 measuring points are symmetrically set on the plane of the outer wall where the origin is located (each measuring point is on the same plane). The horizontal distance between adjacent measuring points is set as a and the vertical distance is set as b (the specific values of a and b are determined through experimental summary). In actual testing, 3 or more measuring points need to be selected (any combination must include measuring point 1 corresponding to the origin) to meet the requirements of multi-source signal acquisition.
[0025] Before starting the raw data acquisition, you can first configure the core parameters of the sensor in the data acquisition software, including the sampling rate, sampling duration and signal range. At the same time, in an environment without partial discharge sources, the sensor collects environmental noise data. This noise data is used to distinguish between effective partial discharge signals and interference signals and to determine the signal filtering threshold. After the parameter configuration and noise acquisition are completed, start the data acquisition system and store the signals received by the sensor in real time as raw data.
[0026] Step 102: Filter the original data for interference signals to obtain a preprocessed time-domain partial discharge signal array.
[0027] Due to the complex operating environment, the raw data inevitably contains various interference signals unrelated to partial discharge (PD). If directly used for subsequent core operations such as signal envelope extraction and time inversion, the interference will mask the true characteristics of the PD signal (such as pulse amplitude and duration), leading to deviations or even failures in subsequent positioning calculations. In this step, the raw data collected by the ultrasonic sensor, containing the target PD signal and environmental noise, is filtered using specific methods to remove invalid interference components, ultimately obtaining a well-organized data array that retains only the time-domain characteristics of the PD signal, serving as the preprocessed time-domain PD signal array.
[0028] The specific filtering methods used can be set according to the type of interference in the actual application scenario, and this embodiment does not limit them.
[0029] Step 103: Extract the signal envelope of the time-domain partial discharge signal array.
[0030] The original time-domain partial discharge signal usually contains high-frequency alternating fluctuations caused by electromagnetic interference and sensor noise. These fluctuations can mask the time characteristics of the partial discharge pulse itself (such as the start time, peak time, and end time of the pulse). Subsequent time inversion positioning relies on accurate signal time information to achieve inversion superposition and focusing. If the original signal is used directly, it will lead to focusing deviation.
[0031] This step extracts the signal envelope from the time-domain partial discharge signal array, which reflects the change of partial discharge pulse amplitude over time. This envelope is a contour curve that characterizes the overall trend of amplitude variation. By extracting the signal envelope, noise-dominated high-frequency alternating components can be filtered out, retaining only the core time characteristics of the partial discharge pulse amplitude change, thus providing a clear signal reference for localization.
[0032] The specific signal envelope extraction method used is not limited in this embodiment. It can be implemented with reference to relevant technologies. For example, the peak detection method can be used, which involves setting a sliding window to traverse the time-domain partial discharge signal array, capturing the signal peak value in each window, and connecting the consecutive peak values in time order to form an envelope curve. Alternatively, the rectification low-pass filtering method can be used, which involves first rectifying the time-domain partial discharge signal array and then filtering out the residual high-frequency components through a low-pass filter to obtain a smooth envelope curve. This embodiment only introduces the above two methods as examples. Other implementation methods can be referred to the description in this embodiment and will not be elaborated here.
[0033] Step 104: Perform a time axis reversal operation on the signal envelope to obtain the inverted envelope signal.
[0034] The extracted partial discharge signal envelope (a contour curve characterizing the amplitude of the partial discharge pulse as a function of time) is subjected to reverse time-domain sequence adjustment, that is, the original time order of the signal envelope (e.g., from the initial time t0 to the end time t) is adjusted. N-1 The amplitude sequence [a0,a1,…,a] N-1 ]) Reverse arrangement from t N-1 The sequence to t0 [a N-1 The operation of generating an inverted envelope signal with reverse propagation time sequence characteristics [, ..., a1, a0] is to reverse the propagation direction of its time dimension while preserving the amplitude characteristics of the signal envelope.
[0035] The propagation of partial discharge signals from the internal partial discharge source of the transformer to the external sensor is a forward timing sequence. The key to time-reverse localization is to trace the signal back to the source point through reverse propagation. The reverse envelope signal can simulate the reverse propagation process of the partial discharge signal from the sensor to the inside of the equipment, providing the necessary timing basis for subsequent superposition based on propagation delay and energy focusing at the location of the partial discharge source.
[0036] Step 105: Based on the propagation delay between each location and the sensor, the inverted envelope signal is superimposed to obtain the energy focusing signal.
[0037] The internal space of a transformer is continuous and its structure is complex, making it impossible to directly calculate the propagation delay and signal superposition for all continuous spatial points. By discretizing the continuous space into a finite number of discrete grid points, i.e., each location, based on the constructed numerical model of partial discharge signal propagation inside the transformer, the signal propagation delay from each location to each ultrasonic sensor in the model is determined. Then, for each discrete location, the inversion envelope signal corresponding to each sensor is time-series calibrated according to its propagation delay with that location. Finally, the amplitudes of all calibrated inversion envelope signals are superimposed at that location to obtain the sum of signal energy values for each discrete location. These energy values together constitute an energy focusing signal that reflects the degree of signal convergence at each location, realizing the differentiated distribution of energy in spatial location.
[0038] When the partial discharge signal propagates from the real partial discharge source to each sensor, there is a fixed delay due to the different propagation paths. The inverted envelope signal is the reverse timing signal of the partial discharge signal. When the inverted envelope signals are superimposed at the real partial discharge source location according to the real delay, the amplitudes of all signals will mutually enhance each other due to the perfect timing match, forming an energy peak. At non-real partial discharge source locations, due to the delay mismatch, the superimposed signal amplitudes will cancel each other out or disperse, presenting only low energy values. Through this process, the energy concentration area can be locked from a large number of potential locations, providing a direct basis for judgment for partial discharge source localization.
[0039] like Figure 4 The diagram shows an energy slice. Finding 3-5 extreme points where the energy of the inverted signal converges is the preliminary location of the partial discharge source.
[0040] Step 106: Determine the location of the partial discharge source as the location of the signal with the highest energy in the energy focusing signal.
[0041] The real partial discharge source is the emission point of the initial ultrasonic signal. When it propagates to each sensor, it has a fixed path and delay. The inverted envelope signal is a simulation of the signal's reverse propagation. When superimposed at the real partial discharge source location according to the real propagation delay, the inverted signals of all sensors will be perfectly matched in timing and mutually reinforced in amplitude, forming the maximum energy value in the entire energy focusing signal. On the other hand, at non-real partial discharge source locations, because the propagation delay of the inverted signal does not match the original path, the superimposed energy will cancel each other out or disperse, presenting only a lower energy value. Therefore, the location corresponding to the maximum energy has a unique correspondence with the location of the real partial discharge source.
[0042] This step involves filtering out the set of energy data with the highest energy value from the energy focusing signal, and determining the potential partial discharge source location corresponding to the highest energy value as the actual partial discharge source location inside the transformer. By using the key indicator of energy magnitude, the actual location of the partial discharge source is locked from a large number of discrete candidate locations.
[0043] Based on the above introduction, the ultrasonic three-dimensional localization method for partial discharge in transformers provided in this embodiment acquires raw data by arranging multiple ultrasonic sensors on the outer wall of the transformer. The raw data is first filtered to remove small-amplitude clutter and invalid frequency band interference, and then the signal envelope is extracted to filter out high-frequency alternating fluctuations. This dual processing preserves the core temporal characteristics of the partial discharge pulse while avoiding noise masking key information, solving the problem of localization errors caused by noise when using the raw signal directly. Then, the signal envelope is time-axis reversed, and a signal backtracking path is constructed based on the reversibility of wave propagation. The energy-focusing signal is generated by superimposing the inverted envelope signal according to the propagation delay. This method, based on the reversibility of wave propagation, preserves the path timing and the entire path through the signal envelope. Delay calculation matches the reflection path, and superposition enhancement achieves energy focusing, transforming the reflected wave into a positioning basis equivalent to the direct wave. Regardless of the path the ultrasonic wave takes to reach the sensor, as long as the signal can be captured and its propagation delay calculated, the location of the true partial discharge source with the highest energy can be found through inversion and superposition. This method does not rely on the direct wave. Even if the transformer core and windings block the ultrasonic wave, causing some sensors to only receive reflected waves, the signal can still be synergistically enhanced at the location of the true partial discharge source through delay matching. The location corresponding to the signal with the highest energy in the focused signal is determined as the location of the partial discharge source. This determination result corresponds to the discrete spatial location in the propagation numerical model, which can be directly converted into specific spatial coordinates, making it easy for engineers to quickly locate the partial discharge source and formulate maintenance plans.
[0044] Example 2: In the above embodiments, extracting the signal envelope of the time-domain partial discharge (PD) signal array is a crucial step connecting signal purification and time inversion. Although large-amplitude clutter and invalid frequency band interference have been removed through interference signal filtering in the early stage, the preprocessed time-domain PD signal still contains high-frequency alternating fluctuations caused by electromagnetic coupling and inherent sensor noise. These fluctuations can mask the continuous time characteristics of the PD pulse. The above embodiments do not limit the specific method for extracting the signal envelope; traditional envelope extraction methods such as peak detection and rectification filtering can be used. However, these methods are prone to envelope discrepancies and discontinuities due to noise interference, or pulse timing distortion due to nonlinear phase distortion, making it difficult to meet the requirements of accurate timing reference for subsequent time axis inversion.
[0045] Therefore, this embodiment proposes an envelope extraction scheme. Step 103, extracting the signal envelope of the time-domain partial discharge signal array, can be performed according to the following steps: Step 31: Approximately simulate the time-domain partial discharge signal array with a 90° phase shift using a finite impulse response filter to generate an orthogonal accompanying signal array.
[0046] Using the preprocessed time-domain partial discharge signal array as the processing object, a finite impulse response (FIR) filter with linear phase characteristics is used to perform an approximately 90° phase shift processing on the signal. That is, the amplitude variation law, frequency components and time sequence length of the original signal are kept unchanged, and only the phase is shifted back by 90° as a whole. Finally, a new signal array orthogonal to the original time-domain partial discharge signal array and with matching characteristics is generated, namely the orthogonal companion signal array.
[0047] The frequency response of an FIR filter needs to lag all positive frequency components of the signal by 90° and lead all negative frequency components by 90°. This embodiment does not limit the specific parameter settings of the FIR filter. To deepen understanding, a filter length M and coefficients are introduced here. The calculation method.
[0048] First, the filter length M can be determined using empirical formulas, where M is an odd number (e.g., 32, 64, 127, etc.). One constraint formula is as follows:
[0049] in, Sampling frequency, The formula represents the highest frequency component in the input signal. This formula ensures that the frequency resolution of the filter is sufficient to match the frequency range of the input signal, avoiding the truncation or distortion of high-frequency signals due to insufficient length.
[0050] The core of a filter is its unit impulse response coefficient, which is obtained by applying a continuous kernel function. The filter coefficients are obtained by discretization and truncation. Let the array index of the filter coefficients be n, and the value range be n=0,1,…,M-1.
[0051] With the filter center as the symmetrical point, the corresponding relative center offset m = n - k0, where k0 is the index of the filter center. The filter coefficients are calculated using the relative center offset m as a variable, according to the following piecewise formula:
[0052] Step 32: Construct an analytical signal based on the time-domain partial discharge signal array and the orthogonal accompanying signal array.
[0053] The previously obtained time-domain partial discharge signal array is used as the real part, and the orthogonal accompanying signal array generated by achieving a 90° phase shift through an FIR filter is used as the imaginary part. The signal is analyzed according to the rules of complex signal composition to form a complex signal that has both real part amplitude characteristics and imaginary part phase characteristics. The amplitude change and phase information of the original time-domain signal are integrated into a unified signal carrier.
[0054] Step 33: Extract the amplitude of the analytical signal as the signal envelope.
[0055] The amplitude of the complex-form analytical signal is extracted and transformed into a real-value curve that intuitively reflects the amplitude profile of the partial discharge pulse.
[0056] Specifically, based on the rules of complex number operations, the modulus of the analytic signal constructed from the time-domain partial discharge signal (real part) and the orthogonal accompanying signal (imaginary part) can be calculated. The resulting single-value sequence that reflects the instantaneous amplitude of the analytic signal is the signal envelope that can characterize the overall trend of the amplitude change of the original time-domain partial discharge signal.
[0057] The signal envelope extraction method provided in this embodiment relies on the linear phase characteristics of the FIR filter and the inherent mathematical properties of the analytic signal. By achieving a 90° phase shift through the FIR filter, the timing information of the original time-domain signal can be completely preserved. The orthogonal accompanying signal and the original signal only have a phase difference and no timing misalignment. Through the magnitude calculation of the analytic signal, the high-frequency alternating components dominated by electromagnetic interference and sensor noise in the original signal are naturally canceled out. At the same time, the low-frequency amplitude change trend of the partial discharge pulse is completely preserved. The resulting envelope is smooth and clean and can restore the true amplitude profile of the pulse. It can also accurately and synchronously reflect the pulse time node of the original signal, providing a zero-deviation timing reference for subsequent time inversion positioning.
[0058] Step 31 can be further broken down into two steps: linear convolution and result truncation. First, linear convolution is performed on the time-domain partial discharge signal array and the corrected FIR filter coefficients to fully utilize the preservation effect of convolution operation on linear phase characteristics, ensuring that the signal only undergoes phase shift and no timing distortion during the operation. Then, the obtained convolution result is truncated in a targeted manner, retaining only the middle part with the same length as the original time-domain partial discharge signal array as the orthogonal companion signal array.
[0059] Linear convolution can fully transmit the phase modulation capability of the corrected filter, ensuring that the signal phase shift meets the 90° approximation requirement. Truncation of the middle portion can accurately eliminate distortion regions affected by boundary effects, while simultaneously ensuring that the lengths of the orthogonal adjoint signal array and the original time-domain signal array are completely equal, providing a foundation for constructing an analytical signal using both as real and imaginary parts. Of course, other methods can also be used, and this embodiment does not limit this approach.
[0060] Furthermore, while an ideal FIR filter can generate fundamental coefficients that theoretically meet the 90° phase shift requirement through specific algorithms, these coefficients are prone to severe transition band fluctuations and insufficient stopband attenuation in the frequency domain due to the Gibbs phenomenon. To effectively suppress noise and avoid amplifying noise interference during phase shifting, which could lead to deviations in phase shift accuracy, the fundamental coefficients of the filter can be optimized and corrected before using a finite impulse response (FIR) filter to approximate a 90° phase shift in the time-domain partial discharge signal array.
[0061] Therefore, this embodiment proposes to introduce a Hanning window function to correct the fundamental coefficients of the FIR filter. Specifically, this is achieved by multiplying the fundamental coefficients of the filter by the corresponding Hanning window function values one by one. The Hanning window function value is given by the formula w[n] = 0.5. 0.5×cos(2πn / (M 1)) The calculation yields n, where n represents the array index of the filter coefficients and M is the filter length. This function can provide a weight distribution that smoothly transitions from both ends to the middle of the coefficient array, avoiding frequency domain oscillations caused by abrupt changes in coefficients.
[0062] The smooth weighting characteristic of the Hanning window can significantly reduce frequency domain oscillations, making the filter's amplitude-frequency characteristics closer to the ideal state. It also significantly improves stopband attenuation, suppressing the amplitude of residual high-frequency noise to a lower level and reducing noise interference with the phase-shifted signal. Furthermore, this correction method only changes the coefficient amplitude without affecting the phase distribution, fully preserving the linear phase advantage of the FIR filter. This ensures that the corrected filter still achieves a 90° phase shift without time misalignment, guaranteeing that the orthogonal adjoint signal and the original time-domain signal only have a phase difference without timing misalignment. This correction step improves signal processing accuracy from the filter design stage, providing a high-quality filtering foundation for subsequent orthogonal adjoint signal generation, analytical signal construction, and envelope extraction.
[0063] It should be noted that the linear convolution process in step 31 and the Hanning window correction process can be used in combination, for example, the Hanning window can be used to balance the attenuation of the transition band and the stopband.
[0064]
[0065] Obtain the final filter coefficients .
[0066] Obtain the coefficient h after windowing win After [n], the discrete signal array is filtered by convolution operation, and the original signal array s[n]=[s0,s1,…,s N-1 The filtered orthogonal adjoint array q[n] is composed of s[n] and h win Linear convolution of [n]:
[0067] After convolution, the signal length becomes N+M 1. The middle N points need to be retained to finally obtain the orthogonal adjoint signal array q[n].
[0068] The original signal s(n) is combined with its orthogonal adjoint signal to obtain the analytic signal z[n].
[0069]
[0070] Where j is the imaginary unit, s[n] is the real part of the complex signal, and q[n] is the imaginary part.
[0071] The amplitude of the analytical signal z[n] is the envelope waveform of the partial discharge signal, and the envelope a[n] of the signal is:
[0072] Example 3: After acquiring the raw data containing partial discharge signals collected by the ultrasonic sensor, the raw data not only contains the target signal excited by the partial discharge source inside the transformer, but also contains a large number of interference components. These include small-amplitude, wide-spectrum clutter with random amplitudes and wide distribution generated by equipment vibration and electromagnetic coupling, as well as specific frequency interference from power grid harmonics and radiation from surrounding electronic equipment. These interference signals superimpose with the partial discharge signal in both the time and frequency domains. To avoid over-filtering and losing details of the partial discharge signal, and to thoroughly eliminate composite interference, preventing noise amplification or even masking of the core characteristics of the partial discharge pulse during subsequent signal envelope extraction, this embodiment proposes a phased and targeted interference filtering system. Specifically, step 102, which filters interference signals from the raw data, can be performed according to the following steps: Step 21: Extract the signal from the original data according to the preset pulse observation window width to obtain a single-window partial discharge signal.
[0073] The original data is truncated according to the preset pulse observation window width to obtain the single-window partial discharge signal. The window width is set based on the effective duration of the partial discharge pulse determined in the pre-experiment, which can ensure that the single partial discharge pulse is fully included and avoid multi-pulse aliasing or single-pulse truncation.
[0074] In this embodiment, the preset value of the pulse observation window width is not limited. It can be set according to different ultrasonic sensor types. For example, after selecting a suitable ultrasonic sensor, multiple partial discharge preliminary detections can be carried out in the experimental environment. Combined with the preset standard of subsequent signal filtering threshold, the actual effective duration t of a single effective partial discharge pulse signal (i.e., the complete duration from the pulse start time to the amplitude decay to the noise level) can be determined through data analysis. Based on this duration t, the pulse observation window width T needs to be reasonably correlated with t. The value range can usually be set to 1.2t~2t. This range can reserve a certain time redundancy to cope with the small fluctuations in the pulse duration and avoid the risk of signal aliasing caused by excessive window redundancy.
[0075] For example, in a preliminary experiment, analysis of the detection data showed that the duration t of a single effective partial discharge pulse signal was approximately 1 ms. (See details...) Figure 5The image shows a sensor receiving a signal. Based on this, the specific value range of the pulse observation window width T can be determined to be 1.2ms to 2ms. By extracting the original data based on this width, a single-window partial discharge signal containing a single effective partial discharge pulse, without aliasing or truncation, can be obtained.
[0076] Step 22: In the single-window partial discharge signal, filter out small-amplitude broad-spectrum clutter with amplitudes below the amplitude threshold to obtain the first filtered signal.
[0077] Even if a single-window partial discharge signal is captured through the pulse observation window, it is still mixed with a large number of small-amplitude spike clutter caused by electromagnetic radiation, sensor noise, and mechanical vibration of equipment. These clutter amplitudes are random but generally low. If they are not filtered, they are easily misjudged as partial discharge pulse feature points, interfering with the subsequent identification and extraction of the real partial discharge pulse, and thus affecting the overall positioning accuracy.
[0078] Therefore, this step first sets a signal filtering threshold, compares the amplitude of all signal points in the single-window partial discharge signal one by one, sets small amplitude noise with an amplitude less than the threshold to zero, and retains only signal components with an amplitude greater than the threshold as valid signals, and finally outputs the first filtered signal.
[0079] The first filtered signal is obtained by filtering out small-amplitude broad-spectrum clutter with amplitudes below a threshold from a single-window signal. The amplitude threshold can be calibrated using noise data collected in a partial discharge-free environment to accurately distinguish between small-amplitude interference and partial discharge signals. The specific value is not limited in this embodiment and can be based on targeted pre-experiments. For example, after selecting a suitable sensor and building a detection system, a pre-discharge experiment can be conducted on the transformer under test. By collecting the pre-experiment waveform and analyzing the amplitude characteristics of the actual partial discharge pulse, and combining this with the core principle in engineering practice of neither omitting weak partial discharge signals nor retaining invalid clutter, the threshold can be determined. Based on mature industry experience, the threshold is usually set to 10% of the peak amplitude of the actual partial discharge signal in the pre-experiment waveform (see [link to specific waveform and threshold annotation]). Figure 5 This value effectively filters out most small-amplitude broad-spectrum clutter while reserving sufficient recognition space for early weak partial discharge signals with lower amplitudes, preventing effective signals from being falsely filtered. Through this threshold setting and filtering operation, invalid clutter in a single-window signal can be quickly removed, allowing the first filtered signal to focus on signal components with true partial discharge characteristics, laying a high-purity signal foundation for the subsequent frequency filtering stage to accurately remove specific frequency interference.
[0080] It should be noted that this embodiment only uses the above amplitude threshold determination method as an example for introduction. The specific numerical setting is not limited in this embodiment. Other methods can refer to the introduction of this embodiment, and will not be repeated here.
[0081] Step 23: For the first filtered signal, filter out interference signals whose frequencies are not within the effective frequency range to obtain the second filtered signal.
[0082] Interference signals at the transformer site exhibit multi-frequency distribution characteristics: the first filtered signal not only contains the partial discharge target signal but also low-frequency interference from equipment mechanical vibration, as well as high-frequency noise from power grid harmonics and electronic equipment radiation. These interference components with frequencies different from the partial discharge signal will superimpose with the target signal in the time domain, easily leading to phase feature distortion during subsequent signal envelope extraction and affecting positioning accuracy. Therefore, after obtaining the first filtered signal by removing small-amplitude broad-spectrum clutter, precise stripping of frequency-dimensional interference is performed. Signals are filtered according to the effective frequency range, and interference frequencies outside the effective frequency range in the first filtered signal are filtered out to obtain the second filtered signal. This step can use a Butterworth filter to achieve frequency-dimensional interference filtering. This filter has the advantages of flat passband amplitude-frequency characteristics and uniform stopband attenuation, which can accurately block invalid frequency signals while minimizing amplitude distortion of the effective frequency band signal. Of course, other filters can also be used; this embodiment does not limit this.
[0083] The effective frequency range can be set in combination with the typical frequency characteristics of transformer partial discharge signals. It can usually be set to tens of kHz to hundreds of kHz, such as 40 kHz to 200 kHz. This frequency band is the typical effective frequency range of transformer oil-paper insulation system partial discharge signals. It covers the main frequency range of partial discharge signals of different defect types and avoids common low-frequency and high-frequency interference.
[0084] Step 24: Use the second filtered signal as the preprocessed time-domain partial discharge signal array.
[0085] This progressive filtering process first locks in the effective signal range, completely enclosing a single partial discharge pulse to avoid feature truncation and preventing aliasing caused by an excessively wide window. Then, amplitude threshold filtering provides precise removal of small-amplitude clutter, zeroing out broad-spectrum clutter with amplitudes below the threshold in the single-window signal. While eliminating random interference such as equipment vibration and sensor noise, it reserves identification space for early, weak partial discharge signals with lower amplitudes. Furthermore, it locks the effective frequency range into the typical dominant frequency range of the partial discharge signal, accurately filtering out low-frequency interference from mechanical vibration and high-frequency noise from power grid harmonics, while minimizing amplitude distortion in the effective frequency band. This method, by first locking in the effective signal range and then layering and removing different types of interference, achieves precise interference removal while preserving the integrity of the partial discharge signal to the maximum extent, avoiding the limitations of single filtering methods.
[0086] Example 4: Based on the above embodiments, after obtaining the inversion envelope signals corresponding to each sensor, the core task is to achieve energy focusing by superimposing the propagation delays. However, the iron core and windings inside the transformer are interspersed, and the propagation path of the partial discharge signal is complex. The propagation delay cannot be obtained through simple geometric calculation. In order to ensure the accurate extraction of delay data, this embodiment proposes a calculation method. Specifically, step 105 superimposing the inversion envelope signals based on the propagation delays from each location to the sensor can be performed according to the following steps: Step 51: Construct a numerical model of the propagation of partial discharge signal inside the transformer.
[0087] First, a numerical model of the propagation of partial discharge signal inside the transformer is constructed. Combining the geometric dimensions of the transformer and the acoustic parameters (sound velocity, attenuation coefficient) of the materials of each component (such as oil-paper insulation, iron core, and copper winding), the propagation law of the signal in the complex structure is restored by finite element method or ray tracing method, providing a physical basis for delay calculation.
[0088] Step 52: Create a mesh in the propagation numerical model.
[0089] A three-dimensional mesh is established in the model, and the continuous internal space of the transformer is discretized into a large number of regular mesh points. Each mesh point serves as the location of a potential partial discharge source, and the coordinate range covers the area where the sensor is located and the range of possible partial discharge sources. This allows the superposition operation to be focused from continuous space to discrete nodes, solving the problem that continuous space cannot be quantized for calculation.
[0090] Step 53: Calculate the signal propagation delay from each sensor to each grid point within the grid.
[0091] For each grid point, calculate its complete propagation delay to each sensor (including the total time for all possible paths) to ensure that the delay data is highly consistent with the actual signal propagation process.
[0092] Step 54: The inversion envelope signals corresponding to each sensor are superimposed according to the propagation delay to obtain the focusing signal of each grid point, which is used as the energy focusing signal.
[0093] Using grid points as units, the inversion envelope signals of each sensor are time-calibrated according to the propagation delay corresponding to that grid point and then superimposed to obtain the focused signal of each grid point. The focused signals of all grid points together constitute the energy focused signal.
[0094] The propagation numerical model under this method can accurately reproduce the influence of internal structure on signal propagation. Compared with the calculation of delay using empirical formulas, it can significantly improve the accuracy of delay calculation under reflection and diffraction paths, providing a reliable basis for superposition alignment. In addition, grid partitioning transforms the complex space into computable discrete nodes, which not only ensures the identification accuracy of partial discharge source location through reasonable grid density, but also controls the amount of computation. Then, the grid point-sensor performs one-to-one delay calculation to ensure that each inversion envelope signal can participate in superposition according to the actual delay of the corresponding path, avoiding energy focusing misalignment caused by delay deviation. Even if the signal propagates through multiple paths, it can form amplitude superposition enhancement at the grid point where the actual partial discharge source is located, greatly improving the identification of energy focusing signals.
[0095] However, the interior of the transformer is not a single homogeneous space. The path of the signal propagating from the grid point (potential partial discharge source) to the sensor will inevitably pass through various media with significantly different acoustic properties, such as transformer oil, iron core, windings, and transformer walls. The propagation speed of ultrasound varies greatly in different media. To further improve the accuracy of the delay calculation, this embodiment proposes a refined signal propagation delay calculation method. Specifically, step 53, calculating the signal propagation delay from each sensor to each grid point within the grid, can be performed according to the following steps: Step 531: Determine the internal medium of the transformer that each sensor passes through on its propagation path to the corresponding grid point.
[0096] First, the propagation path from the sensor to the target grid point is determined (by the numerical model using techniques such as ray tracing), and all media types and segment boundaries along the path are identified. The media may include, but are not limited to, transformer oil, iron core, windings, and transformer walls.
[0097] Step 532: Obtain the signal propagation speed corresponding to each medium.
[0098] Step 533: Divide the path into segments according to the medium it passes through, calculate the ratio of the length of each segment to the corresponding medium propagation speed, and obtain the delay of each segment.
[0099] For each grid point P(x,y,z), calculate the signal propagation delay from each sensor to that point. Taking the i-th sensor as an example, the coordinates S i (x i ,y i ,z i The straight-line distance from point P(x,y,z) to grid point P(x,y,z) is d. i .
[0100]
[0101] If the speed of signal propagation in a medium (such as transformer oil) is c, then the delay time is:
[0102] Step 534: Sum the delays of each path segment to obtain the total signal propagation delay from the sensor to the corresponding grid point.
[0103] For each grid point P(x,y,z), the inversion signals from all sensors are calculated and superimposed according to their respective delays to obtain the focused signal at that point. If P(x,y,z) is the actual location of the partial discharge source, the inversion signals from all sensors will be delayed by k. i Matching the original propagation path, the signal is superimposed at this point, forming the maximum amplitude; however, at non-partial discharge source locations, due to delay mismatch, the superimposed signal amplitude is smaller.
[0104] This calculation method clearly decomposes the medium types such as transformer oil, iron core, and windings along the propagation path, avoiding delay calculation deviations caused by ignoring differences in the acoustic characteristics of the medium. The segmented calculation mechanism ensures high accuracy of the delay data. After splitting the path according to the medium boundary, the delay of each segment is calculated one by one by the segment length / corresponding medium sound velocity. Each segment uses a precise propagation velocity matched to the medium. Compared with the coarse algorithm of straight distance × average sound velocity, this significantly reduces the accumulated error, making the total delay highly consistent with the actual signal propagation time. Of course, this embodiment is not limited to this, and other calculation methods can also be used. All of these can be referred to the description in this embodiment, which will not be elaborated here.
[0105] Example 5: Based on the above embodiments, although the reliability of energy focusing has been improved through multi-stage interference filtering and precise delay calculation, the propagation numerical model error caused by the complex internal structure of the transformer, as well as residual interference not completely eliminated in the previous signal processing, may still cause false high-energy signals, i.e., pseudo-peaks, to appear at some grid points of non-real partial discharge sources. The energy value of such pseudo-peaks may be close to or even exceed the energy value of the real partial discharge source, leading to misjudgment of the partial discharge source location. In response to this, this embodiment proposes to add a pseudo-peak removal operation before determining the location of the partial discharge source. Specifically, before determining the location corresponding to the signal with the highest energy in the energy focusing signal as the location of the partial discharge source in step 106, the following steps can be further performed: Step 107: Determine the coordinates of the energy focusing signal and the sensor coordinates, and calculate the theoretical time difference between the arrival of the signal at any two sensors.
[0106] Taking the grid points in the propagation numerical model as the core, the three-dimensional coordinates of each grid point (i.e., the candidate partial discharge source position corresponding to the energy focusing signal) are first determined. At the same time, the fixed coordinates of each ultrasonic sensor on the outer wall of the transformer are retrieved. Then, based on the previously calculated total signal propagation delay from the sensor to the grid point, any two sensors are selected as a group. The total propagation delay from the two sensors in the group to the target grid point is subtracted. The time difference obtained is the theoretical time difference of the signal from the grid point to the two sensors.
[0107] Step 108: Extract the actual time difference between the arrival of the signal at the two sensors from the time-domain partial discharge signal array.
[0108] Using a time-domain partial discharge signal array as the data source, for any two previously selected sensors, the actual time when the same partial discharge pulse signal arrives at the two sensors is captured by time-domain feature recognition methods such as peak detection and rising edge threshold triggering. For example, the moment when the signal amplitude first reaches the peak value or the moment when the amplitude rises to 70% of the peak value is used as the judgment standard. The actual time difference when the signal arrives at the two sensors is obtained by subtracting the two actual moments.
[0109] Step 109: Determine whether the deviation between the theoretical time difference and the actual time difference exceeds the preset range; if it does, determine the corresponding energy focusing signal as a spurious peak and remove it.
[0110] The deviation between the theoretical time difference and the actual time difference of the signal arriving at the two sensors is calculated. This deviation is then compared with a pre-set reasonable threshold range (determined based on the sensor detection accuracy and the allowable error of the propagation numerical model, usually in the microsecond range). If the calculated deviation exceeds this preset range, it indicates that the corresponding energy-focused signal does not meet the propagation law of the real partial discharge signal, and is judged as a false high-energy signal, i.e., a pseudo-peak. It is then removed from all energy-focused signals, and only candidate grid point signals with deviation values within the preset range are retained.
[0111] Specifically, let sensor S i S j The coordinates are (x i ,y i ,z i ), (x j ,y j ,z j The coordinates of the candidate peak point (the point corresponding to the energy focusing signal to be identified) are P(x,y,z): The theoretical time difference is calculated based on the direct route, and the formula is: ( From P to S i distance, Similarly); The actual time difference can be extracted from the sensor signal using the following formula: ( , For the signal to reach S i S j (The moment).
[0112] If P is the real source, but The deviation is minimal. If it is a spurious peak, the actual time difference is much greater than the theoretical value because the reflection path is longer. Therefore, the deviation between the two is significant.
[0113] Using an energy focusing algorithm, select 3-5 peak points with the highest energy ranking, denoted as P1, P2, ..., P3. k For each candidate peak, the theoretical time difference between any two sensors is calculated based on its coordinates and the sensor coordinates. This theoretical time difference is then compared with the extracted actual time difference. False peaks that do not meet the requirements are eliminated, and the point with the maximum energy among the remaining peak points is taken as the final partial discharge coordinate.
[0114] This method verifies the deviation between theoretical and actual time differences, and performs a secondary screening of candidate locations from the perspective of consistency with the physical laws of signal propagation. This ensures that the ultimately retained high-energy signals possess both energy advantages and conform to the propagation characteristics of real partial discharge signals, significantly reducing the risk of misjudgment in positioning. Furthermore, the selection logic for any two sensor groups supports multiple parallel verifications, and the reliability of false peak identification can be further improved by increasing the verification dimensions.
[0115] Example 6: This embodiment relates to an ultrasonic three-dimensional positioning device for partial discharge of a transformer. A schematic diagram of the device provided in this embodiment is shown below. Figure 6 As shown, it includes: a data acquisition unit 201, a signal filtering unit 202, an envelope extraction unit 203, an envelope inversion unit 204, a delay superposition unit 205, and an energy focusing unit 206.
[0116] The data acquisition unit 201 is used to acquire raw data containing partial discharge signals collected by each ultrasonic sensor installed on the outer wall of the transformer. The signal filtering unit 202 is used to filter interference signals from the original data to obtain a preprocessed time-domain partial discharge signal array. Envelope extraction unit 203 is used to extract the signal envelope of the time-domain partial discharge signal array; Envelope inversion unit 204 is used to perform time axis reversal operation on the signal envelope to obtain the inverted envelope signal; The delay superposition unit 205 is used to superimpose and invert the envelope signal based on the propagation delay between each location and the sensor to obtain the energy focusing signal; The energy focusing unit 206 is used to determine the location of the partial discharge source corresponding to the signal with the highest energy in the energy focusing signal.
[0117] It should be noted that the contents of the ultrasonic three-dimensional positioning device for partial discharge of transformers provided in this embodiment can be referred to in conjunction with the ultrasonic three-dimensional positioning method for partial discharge of transformers provided in the above embodiments, and the repeated parts will not be repeated in this embodiment.
[0118] In the ultrasonic three-dimensional positioning device for partial discharge of transformers provided in this embodiment, an inversion envelope signal that meets the positioning requirements is constructed by an envelope inversion unit. Combined with the accurate propagation delay calculated by the delay superposition unit based on the multi-medium propagation law, the inversion signal is aligned in time and superimposed with energy according to the real propagation characteristics. Finally, the energy focusing unit locks the position of maximum energy. The whole process strictly follows the physical laws of ultrasonic propagation, avoids positioning deviations caused by empirical estimation, and ensures the accuracy of partial discharge source location identification. At the same time, the non-invasive deployment adapts to field requirements. The sensor only needs to be set on the outer wall of the transformer to complete data acquisition. There is no need to disassemble or modify the equipment. This avoids the impact of invasive detection on the normal operation of the transformer and is applicable to transformers of different models and operating conditions, significantly improving the field application range and operability of the device.
[0119] Furthermore, it should be noted that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, to highlight the innovative aspects of this application, this embodiment does not introduce units that are not closely related to solving the technical problems proposed in this application; however, this does not mean that other units do not exist in this embodiment.
[0120] Example 7: Another embodiment of this application relates to an electronic device, such as... Figure 7 As shown, it includes: at least one processor 301; and a memory 302 communicatively connected to at least one processor 301; wherein the memory 302 stores instructions executable by at least one processor 301, which are executed by at least one processor 301 to enable at least one processor 301 to perform the steps of the ultrasonic three-dimensional positioning method for partial discharge of transformers in the above embodiments.
[0121] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.
[0122] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.
[0123] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing this application, and in practical applications, various changes can be made to them in form and detail without departing from the spirit and scope of this application.
Claims
1. A method for three-dimensional ultrasonic localization of partial discharge in a transformer, characterized in that, include: Acquire raw data, including partial discharge signals, collected by each ultrasonic sensor installed on the outer wall of the transformer; The original data is filtered for interference signals to obtain a preprocessed time-domain partial discharge signal array; Extract the signal envelope of the time-domain partial discharge signal array; Perform a time-axis reversal operation on the signal envelope to obtain the inverted envelope signal; The energy focusing signal is obtained by superimposing the inversion envelope signal on the propagation delay between each location and the sensor. The location corresponding to the signal with the highest energy in the energy-focused signal is determined as the location of the partial discharge source.
2. The ultrasonic three-dimensional localization method for partial discharge in transformers according to claim 1, characterized in that, Extracting the signal envelope of the time-domain partial discharge signal array includes: By using a finite impulse response filter, an approximate simulation of a 90° phase shift is performed on the time-domain partial discharge signal array to generate an orthogonal accompanying signal array; An analytical signal is constructed based on the time-domain partial discharge signal array and the orthogonal adjoint signal array; The amplitude of the analyzed signal is extracted and used as the signal envelope.
3. The ultrasonic three-dimensional localization method for partial discharge in transformers according to claim 2, characterized in that, Before performing an approximate simulation of the time-domain partial discharge signal array with a 90° phase shift using a finite impulse response filter, the method further includes: The fundamental coefficients of the finite impulse response filter are multiplied one by one with the corresponding indexed Hanning window function values to correct the fundamental coefficients; The Hanning window function value is determined based on w[n] = 0.
5. 0.5×cos(2πn / (M 1) Determine the corresponding array index position; n is the array index of the coefficients of the finite impulse response filter, and M is the length of the finite impulse response filter.
4. The ultrasonic three-dimensional localization method for partial discharge in transformers according to claim 2, characterized in that, The approximate simulation of the time-domain partial discharge signal array with a 90° phase shift using a finite impulse response filter includes: A linear convolution operation is performed on the time-domain partial discharge signal array and the coefficients of the finite impulse response filter to obtain the convolution result; The convolution result is truncated, and the middle part, which has the same length as the temporal partial discharge signal array, is retained as the orthogonal companion signal array.
5. The ultrasonic three-dimensional localization method for partial discharge in transformers according to claim 1, characterized in that, The original data is subjected to interference signal filtering, including: The original data is truncated according to the preset pulse observation window width to obtain a single-window partial discharge signal; In the single-window partial discharge signal, small-amplitude broad-spectrum clutter with amplitudes below the amplitude threshold is filtered out to obtain the first filtered signal; For the first filtered signal, interference signals whose frequencies are not within the effective frequency range are filtered out to obtain the second filtered signal; The second filtered signal is used as the preprocessed time-domain partial discharge signal array.
6. The ultrasonic three-dimensional localization method for partial discharge in transformers according to claim 1, characterized in that, The step of superimposing the inverted envelope signal based on the propagation delay between each location and the sensor includes: Construct a numerical model of the propagation of partial discharge signals inside the transformer; A grid is established in the propagation numerical model; Calculate the signal propagation delay from each sensor to each grid point within the grid; The inversion envelope signals corresponding to each sensor are superimposed according to the propagation delay to obtain the focusing signal of each grid point, which is used as the energy focusing signal.
7. The ultrasonic three-dimensional localization method for partial discharge in transformers according to claim 6, characterized in that, The calculation of the signal propagation delay from each sensor to each grid point within the grid includes: Determine the internal medium of the transformer along the propagation path from each sensor to the corresponding grid point; Obtain the signal propagation speed corresponding to each of the aforementioned media; Divide the path into segments along the medium it passes through, calculate the ratio of the length of each segment to the corresponding medium propagation speed, and obtain the delay of each segment. The total signal propagation delay from the sensor to the corresponding grid point is obtained by summing the delays of each path segment.
8. The ultrasonic three-dimensional localization method for partial discharge in transformers according to claim 1, characterized in that, Before determining the location of the partial discharge source corresponding to the signal with the highest energy in the energy-focused signal, the process also includes: Determine the coordinates corresponding to the energy focusing signal and the sensor coordinates, and calculate the theoretical time difference between the arrival of the signal at any two sensors; Extract the actual time difference between the arrival of the signal at the two sensors from the time-domain partial discharge signal array; Determine whether the deviation between the theoretical time difference and the actual time difference exceeds a preset range; If the peak exceeds the limit, the corresponding energy focusing signal will be identified as a spurious peak and removed.
9. A transformer partial discharge ultrasonic three-dimensional positioning device, characterized in that, include: The data acquisition unit is used to acquire raw data containing partial discharge signals collected by various ultrasonic sensors installed on the outer wall of the transformer. A signal filtering unit is used to filter interference signals from the original data to obtain a preprocessed time-domain partial discharge signal array. An envelope extraction unit is used to extract the signal envelope of the time-domain partial discharge signal array; An envelope inversion unit is used to perform a time-axis reversal operation on the signal envelope to obtain an inverted envelope signal. The delay superposition unit is used to superimpose the inversion envelope signal according to the propagation delay between each location and the sensor to obtain the energy focusing signal; The energy focusing unit is used to determine the location of the partial discharge source corresponding to the signal with the highest energy in the energy focusing signal.
10. An electronic device, characterized in that, include: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the ultrasonic three-dimensional localization method for partial discharge of transformers as described in any one of claims 1 to 8.