A transformer partial discharge acoustic-electric combined positioning method and system
By coupling high-frequency current and ultrasonic signals, a multimodal signal dataset was constructed and Fourier transform and particle swarm optimization were performed to solve the problem of low positioning accuracy of transformer partial discharge and achieve efficient and accurate positioning results.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHAANXI PUBLIC INTELLIGENT TECH CO LTD
- Filing Date
- 2026-05-11
- Publication Date
- 2026-06-09
AI Technical Summary
Existing transformer partial discharge location technology relies on a single signal, which is affected by electromagnetic interference and signal attenuation, resulting in low location accuracy, large errors in solving the location equations, and weak anti-interference ability, making it impossible to achieve high-precision monitoring.
By coupling the high-frequency current signal and ultrasonic signal of the transformer, a multimodal signal dataset is constructed. Using the leading edge of the initial pulse of the high-frequency current as the zero point of the time reference, a short-time Fourier transform is performed to construct the signal time spectrum matrix. Combining energy ratio analysis and particle swarm optimization, a hyperbolic positioning equation system is constructed to achieve precise positioning of the spatial coordinates of partial discharge.
It improves the accuracy of signal processing and the stability of positioning results, enabling rapid and accurate positioning of partial discharge in transformers and ensuring efficient monitoring of transformer safe operation.
Smart Images
Figure CN122171963A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of acoustic and electrical processing technology, and in particular to a method and system for combined acoustic and electrical localization of partial discharge in transformers. Background Technology
[0002] Traditional partial discharge location techniques for transformers often rely on a single electrical signal or a single ultrasonic signal for monitoring. Complex electromagnetic interference in the field and signal attenuation of the internal medium of the transformer can directly interfere with signal feature extraction, resulting in low accuracy in identifying the signal arrival time. The calculation of the acoustic-electric propagation time difference also has significant deviations, and the hyperbolic location model has large solution errors, making it impossible to accurately locate the partial discharge source.
[0003] Existing positioning methods do not couple and correlate multimodal signals or calibrate a unified time reference. The signal time-frequency analysis and energy discrimination process is redundant and inefficient. The solution of the positioning equations is easily trapped in local optima due to algorithm limitations. The positioning convergence speed is slow and the results are unstable. At the same time, the anti-interference ability is weak and noise signals cannot be effectively eliminated. The overall positioning efficiency and reliability are difficult to meet the high-precision monitoring requirements for the safe operation of transformers. Summary of the Invention
[0004] This invention provides a method and system for combined acoustic and electrical localization of partial discharge in transformers, in order to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides a method for combined acoustic and electrical localization of partial discharge in transformers, comprising: 1. In a partial discharge event of a transformer, the high-frequency current signal and ultrasonic signal of the transformer are coupled and labeled to obtain a multimode signal dataset of the transformer; 2. Mark the initial pulse leading edge of the high-frequency current signal as the time reference zero point; 3. Based on the time reference zero point, perform a short-time Fourier transform on the multimodal signal dataset to construct the signal time-frequency matrix of the transformer; 4. Based on the signal time-spectrum matrix, perform energy ratio analysis on the multimodal signal dataset to obtain the arrival times of the electrical signal and the acoustic signal of the transformer; 5. Based on the combined acoustic-electric time difference between the arrival times of the electrical signal and the acoustic signal, construct a hyperbolic positioning equation set for the ultrasonic signal; 6. Using the partial discharge spatial coordinates of the transformer as the initial value, perform particle swarm optimization on the hyperbolic positioning equations to obtain the target positioning result of the transformer.
[0006] In a preferred embodiment, during a partial discharge event in the transformer, the high-frequency current signal and ultrasonic signal of the transformer are coupled and labeled to obtain a multi-mode signal dataset of the transformer, including: Threshold monitoring of pulse current in the transformer is performed to determine partial discharge events in the transformer; In response to the triggering of the partial discharge event, the high-frequency current signal and ultrasonic signal of the transformer are collected within the same time window, respectively. The high-frequency current signal and the ultrasonic signal are timestamped to obtain the acquisition time tags of the high-frequency current signal and the ultrasonic signal; Based on the acquisition time tag, the high-frequency current signal and the ultrasonic signal corresponding to the same discharge event are associated and stored as a multimodal data pair; All the multimodal data pairs are aggregated in chronological order to form the multimodal signal dataset.
[0007] In a preferred embodiment, marking the initial pulse leading edge of the high-frequency current signal as a time reference zero point includes: Acquire a continuous waveform sequence of the high-frequency current signal in the time domain, and identify the starting point of the first rising edge in the continuous waveform sequence whose amplitude exceeds the range of background noise fluctuations; The time position of the rising edge start point is determined as the candidate leading edge moment; Starting from the candidate leading edge moment, extend a fixed short window backward to examine the continuous growth trend of the waveform amplitude within the window; If the continuous growth trend satisfies the monotonically increasing condition, then the candidate leading edge moment is confirmed as a valid initial pulse leading edge moment; The absolute time value corresponding to the effective initial pulse leading edge moment is set to zero, and the zero moment is used as the time reference zero point for all subsequent signal processing steps.
[0008] In a preferred embodiment, the step of performing a short-time Fourier transform on the multimodal signal dataset based on the time reference zero point to construct the signal time-spectrum matrix of the transformer includes: Starting from the zero point of the time reference, a continuous time series segment containing high-frequency current signals and ultrasonic signals is extracted from the multimodal signal dataset; Hamming window analysis was performed on the continuous time series segment to obtain the complex matrix of current and the complex matrix of sound wave for the continuous time series segment. The current complex matrix and the sound wave complex matrix are combined and sorted in chronological order to obtain the signal time spectrum matrix of the transformer.
[0009] In a preferred embodiment, performing Hamming window analysis on the continuous time series segment to obtain the complex matrix of the current and the complex matrix of the acoustic wave of the continuous time series segment includes: High-frequency current signal subsequence and ultrasonic signal subsequence are separated from the continuous time series segment; According to the preset window length, the high-frequency current signal subsequence and the ultrasonic signal subsequence are segmented and adjusted to obtain the boundary smooth current segmented sequence and the boundary smooth acoustic wave segmented sequence of the transformer. The adjacent boundary smooth current segment sequence and the boundary smooth acoustic wave segment sequence are weighted and superimposed to obtain the current segment sequence group and the acoustic wave segment sequence group of the continuous time series segment. Extract the real component sequence and the imaginary component sequence in the frequency domain from the current segmented sequence group and the sound wave segmented sequence group, respectively. The real component sequence and the imaginary component sequence are paired and combined, and the combined complex spectrum is arranged in chronological order to obtain the current complex matrix and the sound wave complex matrix of the continuous time sequence segment.
[0010] In a preferred embodiment, the step of performing energy ratio analysis on the multimodal signal dataset based on the signal time-spectrum matrix to obtain the arrival times of the electrical and acoustic signals of the transformer includes: The adaptive energy ratio threshold for each time window in the current complex matrix and the acoustic complex matrix is calculated to obtain the high-frequency current energy sequence and ultrasonic energy sequence of the partial discharge event. The formula for calculating the adaptive energy ratio threshold is as follows: ; In the formula, For the first The energy ratio of each time window to the threshold This represents the average background noise energy over the first 10 time windows. This is a preset proportional coefficient. The standard deviation of background noise energy for the first 10 time windows; The arrival time of the electrical signal is defined as the starting moment when the energy of the first three consecutive time windows in the high-frequency current energy sequence exceeds the corresponding adaptive energy ratio threshold. The arrival time of the acoustic signal is defined as the starting time when the energy of the first three consecutive time windows in the ultrasonic energy sequence exceeds the corresponding adaptive energy ratio threshold.
[0011] In a preferred embodiment, constructing the hyperbolic localization equations for the ultrasonic signal based on the combined acoustic-electric time difference between the arrival times of the electrical signal and the acoustic signal includes: The arrival time of the electrical signal is used as the reference time of the partial discharge event; The difference between the arrival time of the acoustic signal in the ultrasonic sensor channel and the reference time is evaluated to obtain the combined acoustic and electrical propagation time of the ultrasonic sensor channel. The three ultrasonic sensors with the shortest combined acoustic and electrical propagation time were selected and denoted as sensor A, sensor B, and sensor C, respectively. Based on the arrival time difference of the acoustic signals from sensor A and sensor B, and the arrival time difference of the acoustic signals from sensor A and sensor C, a hyperbolic positioning equation set for the ultrasonic signal is established.
[0012] In a preferred embodiment, the specific contents of the hyperbolic positioning equations are as follows: ; In the formula, The spatial coordinates of the local discharge source are to be determined. Let A be the spatial coordinates of sensor A. Let B be the spatial coordinates of sensor B. Let C be the spatial coordinates of sensor C. The speed at which ultrasound propagates in transformer oil. The time difference between the arrival times of the acoustic signals from sensor A and sensor B. Let be the time difference between the arrival times of the acoustic signals from sensor A and sensor C.
[0013] In a preferred embodiment, the step of using the partial discharge spatial coordinates of the transformer as initial values and performing particle swarm optimization on the hyperbolic positioning equations to obtain the target positioning result of the transformer includes: The partial discharge spatial coordinates of the transformer are obtained by solving the hyperbolic positioning equations using least squares. The partial discharge spatial coordinates are used as the current estimated position and substituted into the hyperbolic positioning equations. The deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference corresponding to the current estimated position is compared to obtain the degree of deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference. Based on the degree of deviation, the current estimated position is adjusted along multiple candidate directions in space to generate candidate updated positions for the transformer; The degree of deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference corresponding to the candidate update position is re-evaluated, and the candidate update position with the smallest new deviation degree is selected as the current estimated position for the next iteration. Repeat the adjustment and selection operations until the spatial distance between the current estimated positions obtained from two consecutive iterations is less than the preset convergence condition, and then determine the current estimated position obtained from the last iteration as the target positioning result of the transformer.
[0014] To address the above problems, the present invention also provides a transformer partial discharge acoustic-electric combined localization system, the system comprising: A multimodal signal coupling and marking module is used to couple and mark the high-frequency current signal and ultrasonic signal of the transformer in the partial discharge event of the transformer to obtain the multimodal signal dataset of the transformer. The time reference zero-point calibration module is used to mark the initial pulse leading edge moment of the high-frequency current signal as the time reference zero point; The time-spectrum matrix construction module is used to perform a short-time Fourier transform on the multimodal signal dataset based on the time reference zero point in order to construct the signal time-spectrum matrix of the transformer. The acoustic and electrical signal arrival time extraction module is used to perform energy ratio analysis on the multimodal signal dataset based on the signal time-spectrum matrix to obtain the arrival time of the electrical signal and the arrival time of the acoustic signal of the transformer. The hyperbolic positioning equations construction module is used to construct the hyperbolic positioning equations of the ultrasonic signal based on the combined acoustic-electric time difference between the arrival time of the electrical signal and the arrival time of the acoustic signal. The particle swarm iterative optimization localization module is used to perform particle swarm iterative optimization on the hyperbolic localization equation set with the partial discharge spatial coordinates of the transformer as the initial value, so as to obtain the target localization result of the transformer.
[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention constructs a standardized multimodal signal dataset by coupling and marking the high-frequency current signal and ultrasonic signal of partial discharge in a transformer. Using the leading edge of the initial pulse of the high-frequency current as a unified time reference zero point, and combining short-time Fourier transform to build a signal time spectrum matrix, the arrival times of electrical and acoustic signals are accurately extracted based on energy ratio analysis, effectively improving the accuracy and standardization of signal processing.
[0016] 2. This invention constructs a hyperbolic positioning equation system based on the combined acoustic and electrical time difference, and uses particle swarm optimization to solve the equation system, achieving rapid convergence and accurate determination of the spatial coordinates of partial discharge. This significantly improves the efficiency of acoustic and electrical joint positioning of transformer partial discharge, while ensuring the stability and reliability of the positioning results, providing efficient positioning technology support for transformer safe operation monitoring. Attached Figure Description
[0017] Figure 1This is a flowchart illustrating a method for combined acoustic and electrical localization of partial discharge in a transformer, provided in an embodiment of the present invention. Figure 2 A functional block diagram of a transformer partial discharge acoustic-electric combined positioning system provided in an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0019] This application provides a method for the combined acoustic and electrical localization of partial discharge in transformers. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for the combined acoustic and electrical localization of partial discharge in transformers can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.
[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a method for jointly locating partial discharge in a transformer using acoustic and electrical methods, according to an embodiment of the present invention. In this embodiment, the method includes: 1. In a partial discharge event of a transformer, the high-frequency current signal and ultrasonic signal of the transformer are coupled and labeled to obtain a multimode signal dataset of the transformer; In this embodiment of the invention, the step of coupling and marking the high-frequency current signal and ultrasonic signal of the transformer during a partial discharge event to obtain a multi-mode signal dataset of the transformer includes: Threshold monitoring of pulse current in the transformer is performed to determine partial discharge events in the transformer; In response to the triggering of the partial discharge event, the high-frequency current signal and ultrasonic signal of the transformer are collected within the same time window, respectively. The high-frequency current signal and the ultrasonic signal are timestamped to obtain the acquisition time tags of the high-frequency current signal and the ultrasonic signal; Based on the acquisition time tag, the high-frequency current signal and the ultrasonic signal corresponding to the same discharge event are associated and stored as a multimodal data pair; All the multimodal data pairs are aggregated in chronological order to form the multimodal signal dataset.
[0021] The pulse current signal at the transformer grounding wire location is continuously collected. The amplitude of the real-time pulse current signal is compared point by point with the preset partial discharge judgment threshold. When the amplitude of the pulse current signal reaches or exceeds the preset threshold, it is determined that the transformer is currently experiencing a partial discharge event. When the amplitude of the pulse current signal does not reach the preset threshold, it is determined that the transformer is currently not experiencing a partial discharge event.
[0022] When the partial discharge event of the transformer is determined, the synchronous acquisition program is started. According to the pre-set fixed time window range, the acquisition functions of the high-frequency pulse current sensor and the contact ultrasonic partial discharge sensor are simultaneously turned on. Within the fixed time window, the high-frequency current signal and ultrasonic signal of the transformer are completely acquired, ensuring that the acquisition time interval of the two signals is completely consistent.
[0023] During the acquisition of high-frequency current signals, a corresponding real-time acquisition time is matched for each set of high-frequency current signal data. This real-time acquisition time is then used as an identifier and attached to the corresponding high-frequency current signal data to form an acquisition time tag for the high-frequency current signal. Similarly, during the acquisition of ultrasonic signals, a corresponding real-time acquisition time is matched for each set of ultrasonic signal data. This real-time acquisition time is then used as an identifier and attached to the corresponding ultrasonic signal data to form an acquisition time tag for the ultrasonic signal.
[0024] The acquisition time tags of the high-frequency current signal and ultrasonic signal are checked one by one. High-frequency current signals and ultrasonic signals whose acquisition time tags fall within the same time window corresponding to a partial discharge event are selected, and these two sets of signals are bound and stored. The combined signals after binding constitute the multimodal data pair corresponding to the same discharge event. All generated multimodal data pairs are arranged sequentially according to the chronological order of the partial discharge events corresponding to each multimodal data pair. All sequentially arranged multimodal data pairs are then integrated and summarized. The overall dataset after integration and summarization is the transformer's multimodal signal dataset.
[0025] The beneficial effects are that pulse current threshold monitoring can accurately determine transformer partial discharge events. When a discharge event is triggered, high-frequency current signals and ultrasonic signals within the same time window are collected simultaneously. Acquisition time tags are added to the two types of signals respectively. Based on the tags, signals corresponding to the same discharge event are associated and stored as multimodal data pairs. Then, all data pairs are collected in chronological order to form a multimodal signal dataset, ensuring the synchronization and correlation of multimodal signals, and providing a standard and accurate data foundation for subsequent processing of transformer partial discharge.
[0026] 2. Mark the initial pulse leading edge of the high-frequency current signal as the time reference zero point; In this embodiment of the invention, marking the initial pulse leading edge of the high-frequency current signal as a time reference zero point includes: Acquire a continuous waveform sequence of the high-frequency current signal in the time domain, and identify the starting point of the first rising edge in the continuous waveform sequence whose amplitude exceeds the range of background noise fluctuations; The time position of the rising edge start point is determined as the candidate leading edge moment; Starting from the candidate leading edge moment, extend a fixed short window backward to examine the continuous growth trend of the waveform amplitude within the window; If the continuous growth trend satisfies the monotonically increasing condition, then the candidate leading edge moment is confirmed as a valid initial pulse leading edge moment; The absolute time value corresponding to the effective initial pulse leading edge moment is set to zero, and the zero moment is used as the time reference zero point for all subsequent signal processing steps.
[0027] Continuously collect and record all sampling data of high-frequency current signals in the time domain. Combine these sampling data arranged in chronological order to form a continuous and complete waveform sequence. Collect signal data in advance under stable operating conditions where the transformer does not experience partial discharge, thereby determining a fixed background noise fluctuation range. Compare the amplitude of each sampling point in the continuous waveform sequence with the background noise fluctuation range in chronological order to accurately find the starting point of the first signal rising edge whose amplitude exceeds the background noise fluctuation range.
[0028] Extract the specific time position corresponding to the starting point of the rising edge from the time axis of signal acquisition, and directly mark this time position as the candidate leading edge moment.
[0029] Starting from the time point corresponding to the candidate leading edge moment, a complete time interval is extended forward according to a pre-set fixed short-time window length. Waveform amplitude data corresponding to each sampling moment within this time interval is extracted point by point. The amplitudes of adjacent sampling points are compared sequentially according to time to fully verify the continuous increasing trend of waveform amplitude within the window. All adjacent amplitude data within the fixed short-time window are checked sequentially. When the amplitude data of each subsequent sampling point within the window is greater than the amplitude data of the previous sampling point, the continuous increasing trend of waveform amplitude is determined to meet the monotonically increasing condition. At this point, the candidate leading edge moment is officially confirmed as the valid initial pulse leading edge moment.
[0030] Extract the original absolute time value corresponding to the effective initial pulse leading edge moment in the signal acquisition system, reset the original absolute time value to zero, and use the reset zero moment as the time reference zero point used uniformly for all subsequent signal processing steps.
[0031] The beneficial effect is that by acquiring the continuous waveform sequence of high-frequency current signal and identifying the rising edge start point that exceeds the background noise fluctuation range, the candidate leading edge moment can be accurately determined. Then, by using a fixed short-time window to check the waveform amplitude growth trend, the effective initial pulse leading edge moment can be accurately determined. After assigning the value of this moment to zero, a unified time reference zero point is formed, providing a stable and accurate time reference standard for subsequent signal processing.
[0032] 3. Based on the time reference zero point, perform a short-time Fourier transform on the multimodal signal dataset to construct the signal time-frequency matrix of the transformer; In this embodiment of the invention, the step of performing a short-time Fourier transform on the multimodal signal dataset based on the time reference zero point to construct the signal time-spectrum matrix of the transformer includes: Starting from the zero point of the time reference, a continuous time series segment containing high-frequency current signals and ultrasonic signals is extracted from the multimodal signal dataset; Hamming window analysis was performed on the continuous time series segment to obtain the complex matrix of current and the complex matrix of sound wave for the continuous time series segment. The current complex matrix and the sound wave complex matrix are combined and sorted in chronological order to obtain the signal time spectrum matrix of the transformer.
[0033] Using a preset time reference zero point as the starting reference point for signal interception, this time reference zero point is a preset fixed time node in the process of acquiring transformer multi-mode signals. It can be selected as the normal operation start time of the transformer or the initial time when the signal monitoring system triggers the acquisition. Signal segments are continuously extracted from the multi-mode signal dataset. During the extraction process, it is ensured that the segment contains both high-frequency current signals and ultrasonic signals. The extracted signal segments are uninterrupted and without missing parts in the time dimension. After the interception is completed, a continuous time sequence segment containing high-frequency current signals and ultrasonic signals is obtained.
[0034] Hamming window analysis is performed on the extracted continuous time series segments. The specific operation of Hamming window analysis is to include the entire continuous time series segment within the coverage of the Hamming window, and to perform signal segmentation processing on the continuous time series segment through the Hamming window, dividing the continuous time series into several interconnected and non-overlapping fixed-length signal sub-segments. Each signal sub-segment completely retains the high-frequency current signal and ultrasonic signal information within the corresponding time period. Then, signal feature extraction is performed on each signal sub-segment. The amplitude and phase information of the high-frequency current signal within each sub-segment are extracted and integrated to form a current complex matrix. At the same time, the amplitude and phase information of the ultrasonic signal within each sub-segment are extracted and integrated to form a sound wave complex matrix. The current complex matrix and the sound wave complex matrix correspond one-to-one with the signal sub-segments of the continuous time series segment.
[0035] The time flow of the continuous time series segment is clearly defined, and this time flow is consistent with the time sequence during signal acquisition. That is, starting from the zero point of the time reference, the first signal sub-segment, the second signal sub-segment, and so on, corresponding to the last signal sub-segment of the continuous time series segment in sequence, are sequentially arranged. According to this time sequence, the complex current matrix and the complex acoustic wave matrix are combined and sorted. The elements in the complex current matrix corresponding to the same time node are arranged with the elements in the complex acoustic wave matrix. The element pairing and arrangement of all time nodes are completed in sequence. After all elements are arranged, a complete matrix is formed, which is the signal time spectrum matrix of the transformer.
[0036] Based on the differences in the physical properties of signal types within a continuous time series segment, all signal components with electrical characteristics are classified into one category to form an independent and complete high-frequency current signal subsequence, while all signal components with acoustic characteristics are classified into another category to form an independent and complete ultrasonic signal subsequence.
[0037] Based on the preset window length to avoid edge abrupt interference in the time-frequency analysis of transformer partial discharge signals, the high-frequency current signal subsequence is divided into multiple continuous segments according to a fixed number of data points. The signal values at both ends of each segment are adjusted in a gradual transition manner so that the signal values at the segment edges smoothly converge to the reference zero value, resulting in a boundary-smooth current segment sequence. The same segmentation rules and edge adjustment methods are used to process the ultrasonic signal subsequence so that the signal values at the edge of each segment smoothly transition to zero, resulting in a boundary-smooth acoustic wave segment sequence.
[0038] For each group of adjacent boundary smooth current segment sequences, a fixed ratio of signal amplitude superposition is used for fusion processing. During the superposition process, the signal change trend within each segment is preserved and no signal distortion is generated. After all adjacent segments are superimposed, a current segment sequence group is formed. Adjacent boundary smooth acoustic wave segment sequences are processed using the same fixed ratio of amplitude superposition, preserving the acoustic wave signal characteristics and completing the fusion of adjacent segments to form an acoustic wave segment sequence group.
[0039] Frequency domain feature decomposition is performed on each sequence in the current segmented sequence group. The pure real part signal component without the imaginary part of the phase is extracted from the decomposition result to form the real part sequence. Then, the pure imaginary part signal component without the real part of the amplitude is extracted to form the imaginary part sequence. The same decomposition and extraction method is used to process the acoustic wave segmented sequence group to obtain the corresponding acoustic wave real part sequence and acoustic wave imaginary part sequence.
[0040] The real component sequence and the imaginary component sequence of the current at the same frequency domain position are combined one-to-one to form a complex spectrum that can completely characterize the frequency domain features of the current signal. Then, all complex spectra are arranged in the order of time corresponding to each segment to obtain the complex matrix of the current for the continuous time series segment. The real component sequence and the imaginary component sequence of the sound wave at the same frequency domain position are combined to form a complex spectrum. After arranging them in the order of time, the complex matrix of the sound wave for the continuous time series segment is obtained.
[0041] The beneficial effects include the ability to accurately extract continuous time series segments containing high-frequency current signals and ultrasonic signals. Through Hamming window analysis and subsequent processing, the signal characteristics are fully preserved without distortion, and the complex current matrix, complex acoustic matrix, and various related sequences are accurately obtained. By reasonably determining the relevant parameters of the adaptive energy ratio threshold and calculating the threshold, the effective signals generated by background noise and partial discharge can be effectively distinguished. This provides reliable support for accurately determining the arrival time of electrical and acoustic signals, ensuring the accuracy and reproducibility of transformer partial discharge monitoring.
[0042] 4. Based on the signal time-spectrum matrix, perform energy ratio analysis on the multimodal signal dataset to obtain the arrival times of the electrical signal and the acoustic signal of the transformer; In this embodiment of the invention, the step of performing energy ratio analysis on the multimodal signal dataset based on the signal time-spectrum matrix to obtain the arrival times of the electrical signal and the acoustic signal of the transformer includes: The adaptive energy ratio threshold for each time window in the current complex matrix and the acoustic complex matrix is calculated to obtain the high-frequency current energy sequence and ultrasonic energy sequence of the partial discharge event. The formula for calculating the adaptive energy ratio threshold is as follows: ; In the formula, For the first The energy ratio of each time window to the threshold This represents the average background noise energy over the first 10 time windows. This is a preset proportional coefficient. The standard deviation of background noise energy for the first 10 time windows; The arrival time of the electrical signal is defined as the starting moment when the energy of the first three consecutive time windows in the high-frequency current energy sequence exceeds the corresponding adaptive energy ratio threshold. The arrival time of the acoustic signal is defined as the starting time when the energy of the first three consecutive time windows in the ultrasonic energy sequence exceeds the corresponding adaptive energy ratio threshold.
[0043] Energy data corresponding to each time window is extracted from the complex matrix of current and the complex matrix of sound wave. The energy data of the first 10 time windows are selected as background noise energy samples. The sum of these 10 background noise energy samples is calculated, and the summation result is divided by 10. The resulting value is the average background noise energy of the first 10 time windows. This average value directly reflects the average energy level of the background noise in the initial stage, providing a basic reference for subsequent threshold calculation.
[0044] The preset proportional coefficient is derived from historical operating data and long-term practical verification of transformer partial discharge monitoring. It is determined by combining the energy difference pattern between background noise and effective signal under different partial discharge scenarios in the past and through multiple comparative tests. Its function is to adjust the influence of the standard deviation of background noise energy on the threshold, so as to ensure that the threshold can accurately distinguish between background noise and effective signal generated by partial discharge.
[0045] The standard deviation of background noise energy for the first 10 time windows is calculated based on the background noise energy samples selected when calculating the mean background noise energy. First, the difference between each sample and the mean background noise energy is calculated. Then, each difference is squared. All squared differences are summed. The sum is divided by 10 to obtain the variance. Then, the square root of the variance is taken. The resulting value is the standard deviation of background noise energy, which reflects the dispersion of background noise energy in the first 10 time windows.
[0046] The core meaning of this formula is to calculate the first... The adaptive energy ratio threshold for each time window combines the average and dispersion of background noise energy from the first 10 time windows, and adjusts the threshold sensitivity using a preset proportional coefficient. This allows the threshold for each time window to adaptively match the current background noise level, avoiding misjudgments caused by fixed thresholds due to background noise fluctuations. This provides an accurate and reproducible criterion for determining whether the energy of each time window in the high-frequency current energy sequence and ultrasonic energy sequence exceeds the threshold, thereby helping to accurately extract the high-frequency current energy sequence and ultrasonic energy sequence of partial discharge events, and providing key evidence for determining the arrival time of electrical and acoustic signals.
[0047] Energy data corresponding to each time window in the complex current matrix is extracted. For each time window, the average level and dispersion of the background noise energy before that time window are considered, and then adjusted by a preset fixed ratio coefficient to obtain the adaptive energy ratio threshold for each time window. The energy data of each time window is compared with the corresponding threshold. Without discarding the energy data of any time window, all energy data are arranged in the order of the time windows to form the high-frequency current energy sequence of the partial discharge event. Using the same processing method, energy data of each time window in the complex acoustic matrix is extracted, and the adaptive energy ratio threshold for each time window is calculated. After comparing the energy data of each time window with the corresponding threshold, all energy data are arranged in the order of the time windows to obtain the ultrasonic energy sequence of the partial discharge event.
[0048] Starting from the first time window of the high-frequency current energy sequence, each of the three consecutive adjacent time windows is checked one by one. During the check, it is confirmed whether the energy of each time window exceeds the adaptive energy ratio threshold corresponding to that time window. Only when the energy of all three consecutive time windows reaches their respective adaptive energy ratio thresholds, and the group of three consecutive time windows is the first combination in the entire high-frequency current energy sequence to meet this condition, is the time corresponding to the earliest time window in the group of three time windows determined as the arrival time of the electrical signal.
[0049] Starting from the first time window of the ultrasonic energy sequence, three consecutive adjacent time windows are checked one by one. The check criteria are completely consistent with those of the high-frequency current energy sequence, that is, to confirm whether the energy of each time window exceeds the adaptive energy ratio threshold corresponding to that time window. Find the first combination in the entire ultrasonic energy sequence that satisfies the requirement that the energy of three consecutive time windows exceeds the corresponding adaptive energy ratio threshold, and determine the time corresponding to the first time window in the combination as the arrival time of the acoustic signal.
[0050] The beneficial effects are that it can combine the average level and dispersion of background noise energy, accurately calculate the adaptive energy ratio threshold of each time window by preset fixed ratio coefficient, without removing any energy data of time window and forming high-frequency current energy sequence and ultrasonic energy sequence in sequence, and adopt a unified screening standard of three consecutive adjacent time windows. The arrival time of electrical signal and acoustic signal can be determined by the combination that first meets the threshold condition, thereby improving the accuracy, reproducibility and anti-interference ability of signal arrival time determination.
[0051] 5. Based on the combined acoustic-electric time difference between the arrival times of the electrical signal and the acoustic signal, construct a hyperbolic positioning equation set for the ultrasonic signal; In this embodiment of the invention, the step of constructing a hyperbolic localization equation set for the ultrasonic signal based on the combined acoustic-electric time difference between the arrival times of the electrical signal and the acoustic signal includes: The arrival time of the electrical signal is used as the reference time of the partial discharge event; The difference between the arrival time of the acoustic signal in the ultrasonic sensor channel and the reference time is evaluated to obtain the combined acoustic and electrical propagation time of the ultrasonic sensor channel. The three ultrasonic sensors with the shortest combined acoustic and electrical propagation time were selected and denoted as sensor A, sensor B, and sensor C, respectively. Based on the arrival time difference of the acoustic signals from sensor A and sensor B, and the arrival time difference of the acoustic signals from sensor A and sensor C, a hyperbolic positioning equation set for the ultrasonic signal is established.
[0052] The specific contents of the hyperbolic positioning equations are as follows: ; In the formula, The spatial coordinates of the local discharge source are to be determined. Let A be the spatial coordinates of sensor A. Let B be the spatial coordinates of sensor B. Let C be the spatial coordinates of sensor C. The speed at which ultrasound propagates in transformer oil. The time difference between the arrival times of the acoustic signals from sensor A and sensor B. Let be the time difference between the arrival times of the acoustic signals from sensor A and sensor C.
[0053] The spatial coordinates of the partial discharge source are unknown parameters that need to be solved in the system of equations. These parameters are used to determine the specific spatial location where the partial discharge occurs inside the transformer and are the core solution object of the entire positioning calculation. The spatial coordinates of sensor A are obtained on-site by professional measuring tools with reference to the transformer's preset unified spatial reference plane when the sensor is actually installed in the transformer tank. After measurement, they are recorded as fixed three-dimensional position values.
[0054] The spatial coordinates of sensor B are measured precisely in the field after installation, using the same spatial reference system. The obtained values directly correspond to the actual installation position of sensor B inside the transformer and remain fixed. The spatial coordinates of sensor C are measured using the same method as the previous two sensors, relying on the unified spatial reference of the transformer to complete the field measurement. The obtained three-dimensional values accurately match the actual installation point of sensor C. The propagation speed of ultrasound in transformer oil is determined through multiple field measurements under standard experimental conditions. In transformer insulating oil at constant temperature and pressure, the time it takes for ultrasound to propagate a fixed distance is measured, and a stable value is taken as the standard propagation speed after multiple calculations.
[0055] The arrival time difference of the acoustic signals from sensor A and sensor B is calculated by subtracting the combined acoustic and electrical propagation times of the two sensors, directly reflecting the time interval between the two sensors receiving the ultrasonic signals. The arrival time difference of the acoustic signals from sensor A and sensor C is calculated using the same method, by subtracting the combined acoustic and electrical propagation times of the two sensors, and is used to reflect the time difference between the two sensors receiving the ultrasonic signals.
[0056] This set of equations combines the known sensor installation location, ultrasonic propagation speed, and measured acoustic signal arrival time difference to construct the spatial location constraint relationship of the partial discharge source. It transforms the acoustic signal propagation time difference into a spatial distance constraint condition. By solving this set of equations, the specific spatial coordinates of the partial discharge source can be obtained, enabling precise location of the partial discharge location inside the transformer and providing an accurate basis for transformer fault monitoring and location.
[0057] The arrival time of the electrical signal, which was determined in the previous stage, is selected as the reference time of the partial discharge event. This reference time is the time corresponding to the first time window in the high-frequency current energy sequence that the energy of three consecutive time windows exceeds the corresponding adaptive energy ratio threshold. It serves as the initial reference time for the detection of the electrical signal after the partial discharge event occurs, and is used to uniformly measure the propagation time of the acoustic signal in each ultrasonic sensor channel, providing a unified time reference standard for the subsequent calculation of the combined acoustic and electrical propagation time.
[0058] For each ultrasonic sensor channel, the arrival time of the corresponding acoustic signal is retrieved, and the time difference between the arrival time of the acoustic signal and the preset reference time is calculated. In the calculation process, the time value corresponding to the arrival time of the acoustic signal of the channel is directly subtracted from the time value corresponding to the reference time. The resulting time difference is the acoustic-electrical combined propagation time of the ultrasonic sensor channel. This time difference accurately reflects the actual time taken for the ultrasonic wave to propagate from the partial discharge power source to the sensor, and each ultrasonic sensor channel corresponds to a unique acoustic-electrical combined propagation time.
[0059] Collect the combined acoustic and electrical propagation times for all ultrasonic sensor channels, and compare all time values one by one. During the comparison, sort the time values in ascending order, and select the top three combined acoustic and electrical propagation times. The three corresponding ultrasonic sensors are labeled as sensor A, sensor B, and sensor C. The selection criterion is that the shorter the combined acoustic and electrical propagation time, the closer the sensor is to the local discharge source. Selecting the three closest sensors can improve the solution accuracy of the subsequent positioning equations and ensure the reliability of the positioning results.
[0060] The acoustic-electric propagation time corresponding to sensors A and B is retrieved, and the acoustic-electric propagation time of sensor A is subtracted from the acoustic-electric propagation time of sensor B to obtain the arrival time difference of the acoustic signals between sensors A and B. Then, the acoustic-electric propagation time corresponding to sensors A and C is retrieved, and the acoustic-electric propagation time of sensor A is subtracted from the acoustic-electric propagation time of sensor C to obtain the arrival time difference of the acoustic signals between sensors A and C. Based on these two time differences, and considering the propagation characteristics of ultrasound inside the transformer, the two time differences are respectively converted into distance differences between the sensors and the partial discharge source, thereby establishing a hyperbolic positioning equation set for the ultrasonic signal that can characterize the location of the partial discharge source.
[0061] The beneficial effects include the ability to clearly define the reference time of partial discharge events, providing a unified time reference for calculating the combined acoustic and electrical propagation time, accurately calculating the combined acoustic and electrical propagation time of each ultrasonic sensor channel, truly reflecting the actual time it takes for ultrasonic waves to propagate from the partial discharge source to the corresponding sensor, effectively improving the solution accuracy of subsequent positioning equations by selecting the sensor closest to the partial discharge source, and combining the time difference with the distance difference relationship to reliably establish a hyperbolic positioning equation set of ultrasonic signals characterizing the location of the partial discharge source, ensuring the accuracy and reliability of the partial discharge positioning results.
[0062] 6. Using the partial discharge spatial coordinates of the transformer as the initial value, perform particle swarm optimization on the hyperbolic positioning equations to obtain the target positioning result of the transformer.
[0063] In this embodiment of the invention, the step of using the partial discharge spatial coordinates of the transformer as initial values and performing particle swarm optimization on the hyperbolic positioning equations to obtain the target positioning result of the transformer includes: The partial discharge spatial coordinates of the transformer are obtained by solving the hyperbolic positioning equations using least squares. The partial discharge spatial coordinates are used as the current estimated position and substituted into the hyperbolic positioning equations. The deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference corresponding to the current estimated position is compared to obtain the degree of deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference. Based on the degree of deviation, the current estimated position is adjusted along multiple candidate directions in space to generate candidate updated positions for the transformer; The degree of deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference corresponding to the candidate update position is re-evaluated, and the candidate update position with the smallest new deviation degree is selected as the current estimated position for the next iteration. Repeat the adjustment and selection operations until the spatial distance between the current estimated positions obtained from two consecutive iterations is less than the preset convergence condition, and then determine the current estimated position obtained from the last iteration as the target positioning result of the transformer.
[0064] The hyperbolic positioning equations are solved using least squares. The process begins by calculating the difference between the theoretical and actual observed values for each equation. All differences are squared and summed. By progressively adjusting the partial discharge spatial coordinates, the sum of squares is minimized. The resulting coordinates are the obtained partial discharge spatial coordinates of the transformer, serving as the initial data for subsequent iterative optimization. These coordinates are then used directly as the current estimated position. Based on the positional relationships of the variables in the equations, the three spatial components of these coordinates are substituted into their corresponding positions in the hyperbolic positioning equations, ensuring accuracy. This combination of the estimated position and the equations provides the basis for subsequent deviation calculations.
[0065] First, based on the hyperbolic positioning equations after substituting the current estimated position, the theoretical acoustic-electric joint time difference corresponding to each sensor is calculated. This theoretical time difference is the ultrasonic propagation time derived from the current estimated position. Then, the actual acoustic-electric joint time difference obtained in the previous measurement is retrieved, and the theoretical time difference and the actual time difference of each group are compared one by one. The absolute value of the difference of each group is calculated, and then the overall deviation is obtained by integrating all the absolute values. This deviation directly reflects the deviation between the current estimated position and the actual local discharge power source position.
[0066] Based on the calculated deviation, and combined with the internal spatial structure of the transformer, six fixed candidate directions are preset, namely front and back, left and right, and up and down. Each direction corresponds to a fixed position adjustment step size, the size of which is determined according to the positioning accuracy requirements of the transformer. The three spatial components of the current estimated position are synchronously adjusted along each candidate direction. After adjustment, six different candidate update positions are generated, and each candidate update position corresponds to a unique spatial coordinate.
[0067] For each candidate update position, repeat the above operations of substituting into the system of equations, calculating the theoretical acoustic-electric joint time difference, and comparing it with the actual time difference. Re-evaluate the degree of deviation corresponding to each candidate update position, numerically compare the degree of deviation of all candidate update positions, select the candidate update position with the smallest degree of deviation, and determine the current estimated position to be used in the next iteration, so as to ensure that the iteration process moves in a direction that is closer to the actual position.
[0068] The above-described process of adjusting candidate update positions, re-evaluating the degree of deviation, and selecting the optimal candidate position is repeated. After each iteration, the spatial distance between the current estimated position and the current estimated position obtained in the previous iteration is calculated. The calculation method is the square root of the sum of the squares of the differences between the three spatial components of the two positions. The iteration continues until the spatial distance between the current estimated positions obtained in two consecutive iterations is less than the preset convergence condition. The convergence condition is set to one millimeter, which meets the accuracy requirements for transformer partial discharge positioning. At this time, the current estimated position obtained in the last iteration is determined as the target positioning result of the transformer. This result is the accurate location of the partial discharge power source inside the transformer.
[0069] The beneficial effect is that the spatial coordinates of the partial discharge of the transformer can be obtained by least squares solution, which serves as the initial basis for iterative optimization. After substituting these coordinates into the hyperbolic positioning equations, the deviation between the theoretical and actual acoustic-electric joint time difference can be accurately compared. Candidate updated positions are generated and the optimal position is selected by adjusting along the preset candidate direction. Through repeated iterations until the convergence condition is met, the accuracy and reliability of the transformer target positioning result are ensured, which meets the accuracy requirements of transformer partial discharge positioning. This provides support for the accurate positioning of partial discharge sources inside the transformer and helps in the effective monitoring and troubleshooting of transformer faults.
[0070] like Figure 2 The diagram shown is a functional block diagram of a transformer partial discharge acoustic-electric combined positioning system provided in an embodiment of the present invention.
[0071] The transformer partial discharge acoustic-electric joint localization system described in this invention can be installed in electronic devices. Depending on the functions implemented, the transformer partial discharge acoustic-electric joint localization system may include a multi-mode signal coupling marking module, a time reference zero-point calibration module, a time-spectrum matrix construction module, an acoustic-electric signal arrival time extraction module, a hyperbola localization equation system construction module, and a particle swarm iterative optimization localization module. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.
[0072] In this embodiment, the functions of each module / unit are as follows: The multimodal signal coupling tagging module is used to couple and tag the high-frequency current signal and ultrasonic signal of the transformer in the partial discharge event of the transformer to obtain the multimodal signal dataset of the transformer. The time reference zero-point calibration module is used to mark the initial pulse leading edge moment of the high-frequency current signal as the time reference zero point; The time-spectrum matrix construction module is used to perform a short-time Fourier transform on the multimodal signal dataset based on the time reference zero point in order to construct the signal time-spectrum matrix of the transformer. The acoustic and electrical signal arrival time extraction module is used to perform energy ratio analysis on the multimodal signal dataset based on the signal time-spectrum matrix to obtain the arrival time of the electrical signal and the arrival time of the acoustic signal of the transformer. The hyperbolic positioning equation construction module is used to construct the hyperbolic positioning equation system of the ultrasonic signal based on the combined acoustic-electric time difference between the arrival time of the electrical signal and the arrival time of the acoustic signal. The particle swarm iterative optimization positioning module is used to perform particle swarm iterative optimization on the hyperbolic positioning equation set with the partial discharge spatial coordinates of the transformer as the initial value, so as to obtain the target positioning result of the transformer.
[0073] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.
[0074] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.
[0075] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.
[0076] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0077] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.
[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims
1. A method for combined acoustic and electrical localization of partial discharge in a transformer, characterized in that, The method includes:
1. In a partial discharge event of a transformer, the high-frequency current signal and ultrasonic signal of the transformer are coupled and labeled to obtain a multimode signal dataset of the transformer; 2. Mark the initial pulse leading edge of the high-frequency current signal as the time reference zero point; 3. Based on the time reference zero point, perform a short-time Fourier transform on the multimodal signal dataset to construct the signal time-frequency matrix of the transformer; 4. Based on the signal time-spectrum matrix, perform energy ratio analysis on the multimodal signal dataset to obtain the arrival times of the electrical signal and the acoustic signal of the transformer; 5. Based on the combined acoustic-electric time difference between the arrival times of the electrical signal and the acoustic signal, construct a hyperbolic positioning equation set for the ultrasonic signal; 6. Using the partial discharge spatial coordinates of the transformer as the initial value, perform particle swarm optimization on the hyperbolic positioning equations to obtain the target positioning result of the transformer.
2. The method for combined acoustic and electrical localization of partial discharge in a transformer as described in claim 1, characterized in that, In the partial discharge event of the transformer, the high-frequency current signal and ultrasonic signal of the transformer are coupled and labeled to obtain a multi-mode signal dataset of the transformer, including: Threshold monitoring of pulse current in the transformer is performed to determine partial discharge events in the transformer; In response to the triggering of the partial discharge event, the high-frequency current signal and ultrasonic signal of the transformer are collected within the same time window, respectively. The high-frequency current signal and the ultrasonic signal are timestamped to obtain the acquisition time tags of the high-frequency current signal and the ultrasonic signal; Based on the acquisition time tag, the high-frequency current signal and the ultrasonic signal corresponding to the same discharge event are associated and stored as a multimodal data pair; All the multimodal data pairs are aggregated in chronological order to form the multimodal signal dataset.
3. The method for combined acoustic and electrical localization of partial discharge in a transformer as described in claim 1, characterized in that, Marking the initial pulse leading edge of the high-frequency current signal as the time reference zero point includes: Acquire a continuous waveform sequence of the high-frequency current signal in the time domain, and identify the starting point of the first rising edge in the continuous waveform sequence whose amplitude exceeds the range of background noise fluctuations; The time position of the rising edge start point is determined as the candidate leading edge moment; Starting from the candidate leading edge moment, extend a fixed short window backward to examine the continuous growth trend of the waveform amplitude within the window; If the continuous growth trend satisfies the monotonically increasing condition, then the candidate leading edge moment is confirmed as a valid initial pulse leading edge moment; The absolute time value corresponding to the effective initial pulse leading edge moment is set to zero, and the zero moment is used as the time reference zero point for all subsequent signal processing steps.
4. The method for combined acoustic and electrical localization of partial discharge in a transformer as described in claim 1, characterized in that, The step of performing a short-time Fourier transform on the multimodal signal dataset based on the time reference zero point to construct the signal time-frequency matrix of the transformer includes: Starting from the zero point of the time reference, a continuous time series segment containing high-frequency current signals and ultrasonic signals is extracted from the multimodal signal dataset; Hamming window analysis was performed on the continuous time series segment to obtain the complex matrix of current and the complex matrix of sound wave for the continuous time series segment. The current complex matrix and the sound wave complex matrix are combined and sorted in chronological order to obtain the signal time spectrum matrix of the transformer.
5. The method for combined acoustic and electrical localization of partial discharge in a transformer as described in claim 4, characterized in that, The step of performing Hamming window analysis on the continuous time series segment to obtain the complex matrix of current and the complex matrix of sound wave for the continuous time series segment includes: High-frequency current signal subsequence and ultrasonic signal subsequence are separated from the continuous time series segment; According to the preset window length, the high-frequency current signal subsequence and the ultrasonic signal subsequence are segmented and adjusted to obtain the boundary smooth current segmented sequence and the boundary smooth acoustic wave segmented sequence of the transformer. The adjacent boundary smooth current segment sequence and the boundary smooth acoustic wave segment sequence are weighted and superimposed to obtain the current segment sequence group and the acoustic wave segment sequence group of the continuous time series segment. Extract the real component sequence and the imaginary component sequence in the frequency domain from the current segmented sequence group and the sound wave segmented sequence group, respectively. The real component sequence and the imaginary component sequence are paired and combined, and the combined complex spectrum is arranged in chronological order to obtain the current complex matrix and the sound wave complex matrix of the continuous time sequence segment.
6. The method for combined acoustic and electrical localization of partial discharge in a transformer as described in claim 5, characterized in that, The step of performing energy ratio analysis on the multimodal signal dataset based on the signal time-spectrum matrix to obtain the arrival times of the electrical and acoustic signals of the transformer includes: The adaptive energy ratio threshold for each time window in the current complex matrix and the acoustic complex matrix is calculated to obtain the high-frequency current energy sequence and ultrasonic energy sequence of the partial discharge event. The formula for calculating the adaptive energy ratio threshold is as follows: ； In the formula, For the first The energy ratio of each time window to the threshold This represents the average background noise energy over the first 10 time windows. This is a preset proportional coefficient. The standard deviation of background noise energy for the first 10 time windows; The arrival time of the electrical signal is defined as the starting moment when the energy of the first three consecutive time windows in the high-frequency current energy sequence exceeds the corresponding adaptive energy ratio threshold. The arrival time of the acoustic signal is defined as the starting time when the energy of the first three consecutive time windows in the ultrasonic energy sequence exceeds the corresponding adaptive energy ratio threshold.
7. The method for combined acoustic and electrical localization of partial discharge in a transformer as described in claim 1, characterized in that, The process of constructing a hyperbolic localization equation set for the ultrasonic signal based on the combined acoustic-electric time difference between the arrival times of the electrical signal and the acoustic signal includes: The arrival time of the electrical signal is used as the reference time of the partial discharge event; The difference between the arrival time of the acoustic signal in the ultrasonic sensor channel and the reference time is evaluated to obtain the combined acoustic and electrical propagation time of the ultrasonic sensor channel. The three ultrasonic sensors with the shortest combined acoustic and electrical propagation time were selected and denoted as sensor A, sensor B, and sensor C, respectively. Based on the arrival time difference of the acoustic signals from sensor A and sensor B, and the arrival time difference of the acoustic signals from sensor A and sensor C, a hyperbolic positioning equation set for the ultrasonic signal is established.
8. The method for combined acoustic and electrical localization of partial discharge in a transformer as described in claim 7, characterized in that, The specific contents of the hyperbolic positioning equations are as follows: ； In the formula, The spatial coordinates of the local discharge source are to be determined. Let A be the spatial coordinates of sensor A. Let B be the spatial coordinates of sensor B. Let C be the spatial coordinates of sensor C. The speed at which ultrasound propagates in transformer oil. The time difference between the arrival times of the acoustic signals from sensor A and sensor B. Let be the time difference between the arrival times of the acoustic signals from sensor A and sensor C.
9. The method for combined acoustic and electrical localization of partial discharge in a transformer as described in claim 1, characterized in that, The step of using the partial discharge spatial coordinates of the transformer as initial values and performing particle swarm optimization on the hyperbolic positioning equations to obtain the target positioning result of the transformer includes: The partial discharge spatial coordinates of the transformer are obtained by solving the hyperbolic positioning equations using least squares. The partial discharge spatial coordinates are used as the current estimated position and substituted into the hyperbolic positioning equations. The deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference corresponding to the current estimated position is compared to obtain the degree of deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference. Based on the degree of deviation, the current estimated position is adjusted along multiple candidate directions in space to generate candidate updated positions for the transformer; The degree of deviation between the theoretical acoustic-electric joint time difference and the actual acoustic-electric joint time difference corresponding to the candidate update position is re-evaluated, and the candidate update position with the smallest new deviation degree is selected as the current estimated position for the next iteration. Repeat the adjustment and selection operations until the spatial distance between the current estimated positions obtained from two consecutive iterations is less than the preset convergence condition, and then determine the current estimated position obtained from the last iteration as the target positioning result of the transformer.
10. A transformer partial discharge acoustic-electric combined positioning system, characterized in that, The system for implementing the acoustic-electric combined localization method for partial discharge of a transformer as described in claim 1 includes: A multimodal signal coupling and marking module is used to couple and mark the high-frequency current signal and ultrasonic signal of the transformer in the partial discharge event of the transformer to obtain the multimodal signal dataset of the transformer. The time reference zero-point calibration module is used to mark the initial pulse leading edge moment of the high-frequency current signal as the time reference zero point; The time-spectrum matrix construction module is used to perform a short-time Fourier transform on the multimodal signal dataset based on the time reference zero point in order to construct the signal time-spectrum matrix of the transformer. The acoustic and electrical signal arrival time extraction module is used to perform energy ratio analysis on the multimodal signal dataset based on the signal time-spectrum matrix to obtain the arrival time of the electrical signal and the arrival time of the acoustic signal of the transformer. The hyperbolic positioning equations construction module is used to construct the hyperbolic positioning equations of the ultrasonic signal based on the combined acoustic-electric time difference between the arrival time of the electrical signal and the arrival time of the acoustic signal. The particle swarm iterative optimization localization module is used to perform particle swarm iterative optimization on the hyperbolic localization equation set with the partial discharge spatial coordinates of the transformer as the initial value, so as to obtain the target localization result of the transformer.