Transformer fault sound source positioning method and system based on orthogonal sharp wave beams

Abstract

The invention discloses a transformer fault sound source localization method and system based on orthogonal sharp beams, and belongs to the technical field of power equipment sound source localization and state detection.The method comprises the steps that N sound sensors are arranged in a three-dimensional space around a transformer in an array mode, a three-dimensional sound field measuring plane is formed, and sound pressure signals are collected and preprocessed; constructing a cross-spectrum matrix between the acoustic sensors based on the preprocessed acoustic pressure signals, and performing eigenvalue decomposition and principal component extraction on the cross-spectrum matrix; orthogonal wave beam output is calculated by combining the cross spectrum matrix after principal component extraction, a preliminary sound source image is generated, and grid points in the preliminary sound source image are screened by combining a preset sound source intensity threshold value; using a CLEAN-SC algorithm to carry out iteration sharpening on the screened sound source image, and when an iteration termination condition is satisfied, outputting a sound source localization image after iteration sharpening wave beam processing; and an area with the intensity higher than a set threshold value is identified in the sound source positioning map, the position of the fault source is analyzed, and a fault positioning report is generated.

Description

Technical Field

[0001] The present invention belongs to the technical field of sound source localization and condition detection of power equipment, and particularly relates to a transformer fault sound source localization method and system based on orthogonal beamforming. Background Art

[0002] As one of the core equipment in the power system, the operating state of a transformer is particularly crucial for the stability and reliability of the entire power grid. Therefore, the fault detection of transformers has become a research hotspot in this field. The transformer fault detection technology is constantly improving and developing. Traditional fault detection methods mainly include vibration signal detection, oil chromatography detection, spectral signal detection, gas component detection, and acoustic signal detection, etc. Vibration signal detection can detect the operating state of a transformer by collecting its vibration signals and can relatively accurately detect its operating defects. However, the disadvantage is that it requires contact detection. Acoustic signal detection can perform non-contact detection on the transformer through sound signals, with convenient operation and high detection efficiency. However, it cannot specifically locate the sound source position of the transformer.

[0003] Compared with other detection technologies, the characteristic of sound source localization technology is that it can not only analyze the operating state of a transformer but also intuitively reflect the position of the transformer sound source, while maintaining non-contact, anti-interference, etc. characteristics, which has obvious advantages for transformer condition detection. The sound source localization technology accurately locates and identifies the sound source by analyzing the frequency spectrum, waveform, and time-domain characteristics of the sound. It was initially widely used in fields such as automobiles and aviation, and also has good application prospects for the condition detection of power equipment.

[0004] When the existing sound source localization technology is used for localization, there are problems such as low resolution and long localization time. How to comprehensively improve the accuracy and calculation efficiency of fault sound source localization according to the characteristics of transformer fault sound sources still needs further research and optimization.

[0005] The prior art document 1 (CN114235366A) discloses an identification scheme based on the noise and vibration characteristics of a transformer in the field of transformer fault identification. By collecting the vibration noise sound pressure signals of the transformer, calculating the characteristic values, and using a BP neural network for fault diagnosis, it relies on the collection and analysis of vibration signals. In practical applications, there are problems such as being easily affected by external noise, the BP neural network relying on training data and having high requirements for computing resources, resulting in the influence on the identification accuracy, diagnosis reliability, and calculation efficiency.

[0006] The prior art document 2 ("Identification of Weak Sound Insulation Parts of Automobile Front Fender Based on CLEAN-SC Clear Beamforming" (Yang Yang, Chu Zhigang, Acoustics Technology, Vol. 34, No. 5, October 2015)) discloses a solution for identifying sound sources based on CLEAN-SC clear beamforming in the field of automotive acoustics. By using CLEAN-SC clear beams, this solution can improve resolution, attenuate sidelobes, more accurately identify single sound sources and incoherent sound sources, and experimentally determine the weak sound insulation parts and reasons of the automobile front fender. However, it has many problems. When identifying low-frequency noise, the resolution and accuracy are poor, making it difficult to meet the high-precision positioning requirements; the computational complexity is relatively high, forming a performance bottleneck in scenarios with high real-time requirements; in different frequency ranges and complex and dynamic sound field environments, the applicability is unstable, restricting its popularization and application. Summary of the Invention

[0007] To solve the technical problems existing in the prior art, the present invention provides a method and system for transformer fault sound source localization based on orthogonal clear beams.

[0008] The present invention adopts the following technical solutions.

[0009] In a first aspect of the present invention, there is provided a method for transformer fault sound source localization based on orthogonal clear beams, which is characterized by including the following steps: Taking the center of the transformer under test as the coordinate origin, arranging N acoustic sensors in an array in the three-dimensional space around the transformer to form a three-dimensional sound field measurement surface, collecting sound pressure signals and performing preprocessing; Based on the preprocessed sound pressure signals, construct a cross-spectral matrix between acoustic sensors, perform eigenvalue decomposition on the cross-spectral matrix, select the first K largest eigenvalues and their corresponding eigenvectors to reconstruct the reduced-dimensional cross-spectral matrix; calculate the orthogonal beam output in combination with the reduced-dimensional cross-spectral matrix to generate a preliminary sound source map, and screen the grid points in the preliminary sound source map in combination with a preset sound source intensity adaptive dynamic threshold; use the CLEAN-SC algorithm to perform iterative clarification processing on the screened sound source map, and when the iterative termination condition is met, output the sound source localization map after iterative clarification beam processing; Mark the area with intensity higher than the set threshold in the sound source localization map, analyze the position of the fault source in combination with the physical structure and operating state of the transformer under test, and generate a fault localization report.

[0010] Optionally, the preprocessing of the sound pressure signals includes: Frame and window the sound pressure signals, and convert each frame of time-domain signal into a frequency-domain signal through Fourier transform to obtain a frequency-domain signal vector , and perform filtering and noise reduction.

[0011] Optionally, the construction of the cross-spectral matrix includes: Based on the frequency-domain signal vector obtained after preprocessing Construct the cross-spectral matrix between acoustic sensors :

[0012] wherein, is the average number of frames; is the frequency-domain signal vector of the nth frame; is the conjugate transpose; Perform eigenvalue decomposition on the cross-spectral matrix to obtain a product form of three matrices, including an orthogonal matrix composed of eigenvectors, a diagonal matrix with diagonal elements being the eigenvalues arranged in descending order from largest to smallest, and the conjugate transpose of the orthogonal matrix; From the decomposed eigenvalues, select the top 20 eigenvalues with the largest values, extract their corresponding 20 eigenvectors, multiply each eigenvalue by the outer product matrix of its corresponding eigenvector, and sum the weighted values of all 20 outer product matrices to obtain the dimension-reduced cross-spectral matrix .

[0013] Optionally, the calculation of the orthogonal beam output based on the cross-spectral matrix after principal component extraction includes: Discretize the measurement surface into a grid surface containing M grid points, and delimit the sound source calculation plane based on the grid surface; Calculate the orthogonal beam output of each grid point in the sound source calculation plane to generate a preliminary sound source map; Screen the grid points in the preliminary sound source map according to a preset sound source intensity threshold.

[0014] Optionally, the discretizing the measurement surface into a grid surface containing M grid points and generating a preliminary sound source map includes: Discretize the measurement surface into a grid surface containing M grid points, the grid point set is , the sound source calculation plane contains M grid points, define a focusing surface for each grid point, and the focusing vector of the mth grid point is , ; The orthogonal beam output of the mth grid point is:

[0015] wherein, is the dimension-reduced cross-spectral matrix, is the conjugate transpose.

[0016] Optionally, the iterative clarification process of the screened sound source map using the CLEAN-SC algorithm includes: Set the filtered sound source map as the initial sound source map, and use the reduced-dimensional cross-spectral matrix as the initial value of the cross-spectral matrix; In the i-th iteration, search for the main lobe peak position in the current sound source map as the strongest sound source, and calculate the direction vector and the weighting vector based on this position; According to the contribution of the coherent source component of the current main lobe peak, remove this component from the cross-spectral matrix and update the cross-spectral matrix of the remaining sound source information; Based on the direction vector and the weighting vector, eliminate the contribution of the current main lobe peak from the sound source map to generate an updated sound source localization map; Store the beamforming data for beam sharpening determined by the gain factor and the phase angle parameter in each iteration; Terminate the iteration when the preset criterion is met, and output the sound source localization map after iterative beam sharpening processing.

[0017] Optionally, the main lobe peak search and direction / weighting vector calculation include: In the i-th iteration, in the sound source map search for the position of the main lobe peak :

[0018] Calculate the direction vector of :

[0019] where, is the distance between the position of the main lobe peak searched in the i i-th iteration and the n -th microphone; is the position of the n -th microphone; Calculate the weighting vector of : .

[0020] Optionally, the update of the cross-spectral matrix of the remaining sound source information includes: Assume that in the i-th iteration is the cross-spectral matrix caused by a single coherent source component :

[0021] where, is the beamforming output corresponding to the main lobe peak in the (i - 1)-th iteration; is the focusing vector corresponding to a single coherent source component in the i-th iteration; is Conjugate transpose;

[0022] According to , update the cross-spectrum matrix:

[0023] where is the cross-spectrum matrix at the i-th iteration, used to store the updated sound source information during the iteration process, is the weight coefficient for adjusting the attenuation ratio of the sound source components during the iteration process.

[0024] Optionally, the fault location analysis and report generation include: Extract continuous regions with intensity higher than the threshold from the clear sound source localization map based on the dynamic threshold, identify potential fault sources through image processing algorithms, align the sound source grid coordinate system with the three-dimensional physical structure model of the transformer, and establish the mapping relationship between the fault region and the actual components; Combine the sound source intensity distribution, direction vector, weighted vector, and the physical structure of the transformer to perform localization optimization and coherence analysis on the high-intensity region, and determine the specific location and priority of the fault source; Output a localization report including the coordinates, intensity, associated components, and fault type of the fault source, verify the localization result in combination with the operating state of the transformer, and deploy repair measures.

[0025] The second aspect of the present invention provides a transformer fault sound source localization system based on orthogonal beamforming. Based on the transformer fault sound source localization method described in the first aspect of the present invention, the system includes: A sensor array module for three-dimensional array arrangement of acoustic sensors and real-time acquisition of the sound pressure signal of the transformer; A signal preprocessing module for processing the original sound pressure signal, including frame division, windowing, Fourier transform, and filtering and noise reduction; A cross-spectrum matrix calculation module for constructing a cross-spectrum matrix and extracting principal component features; An orthogonal beamforming module for generating a preliminary sound source map and screening valid grid points; An iterative beamforming module for iteratively optimizing the sound source localization map through the OB-CLEAN-SC algorithm; A fault analysis module for associating the transformer structure to analyze the position and priority of the fault source; A report generation module for generating a fault source localization report.

[0026] Compared with the prior art, the beneficial effects of the present invention are reflected in: 1. Aiming at the problems of large computational complexity and poor real-time performance in the prior art, the present invention retains effective grid points through dynamic grid compression technology and extracts the main components of the cross-spectrum matrix, solves the problem of resource waste caused by traditional full-grid calculations, improves the computational efficiency, and enables the algorithm to run in real time in complex scenarios such as substations.

[0027] 2. Aiming at the problems of low sound source resolution and fuzzy positioning in the prior art, the present invention uses the orthogonal beamforming (OB) algorithm and iterative cleaning (CLEAN-SC) to solve the problems of strong sidelobe interference and low low-frequency resolution in traditional beamforming, and improves the spatial resolution of sound source localization.

[0028] 3. Aiming at the problems of noise sensitivity and high false / missing judgment rates in the prior art, the present invention uses non-contact acoustic detection and adaptive dynamic thresholds to solve the problems of vibration signals being vulnerable to electromagnetic interference and poor adaptability of fixed thresholds, improves the anti-interference ability in complex noise environments, and reduces the false detection rate.

[0029] 4. Aiming at the problems of lack of physical basis in the results and large deviation from the actual position of real components in the prior art, the present invention uses iterative updating of the cross-spectrum matrix based on the sound field propagation law and mapping alignment of the sound source map - transformer three-dimensional structure to solve the problem of mismatch between the positioning result and the actual component position caused by ignoring physical constraints in traditional methods, and improves the physical rationality of fault location and the value of maintenance guidance. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the sound source recognition space of the beamforming provided by the embodiment of the present invention; Figure 2 It is a flowchart of the OB-CLEAN-SC algorithm provided by the embodiment of the present invention; Figure 3 It is a flowchart of the method provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] Hereinafter, the preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings; it should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the protection scope of the present invention.

[0032] Figure 1 It is a schematic diagram of the sound source recognition space of the algorithm of the present invention, Figure 2 It is a flowchart of the OB-CLEAN-SC algorithm.

[0033] In Embodiment 1 of the present invention, a transformer fault location method based on an orthogonal cleaning beamforming algorithm is provided, as Figure 3 shown, and includes the following steps: Step 1: Taking the center of the transformer under test as the coordinate origin, arrange N acoustic sensors in an array in the three-dimensional space around the transformer to form a three-dimensional sound field measurement surface, collect sound pressure signals and perform preprocessing.

[0034] Preferably, in the said Step 1, the coordinate vector of the nth acoustic sensor is , and this coordinate vector is a three-dimensional coordinate; Preferably, the preprocessing of the sound pressure signal includes: Frame and window the sound pressure signal, convert each frame of time-domain signal into a frequency-domain signal through Fourier transform to obtain a frequency-domain signal vector , and perform filtering and noise reduction.

[0035] Specifically, in the localization of the transformer fault sound source, first, it is necessary to arrange N sensors in an array in the sound field to form a measurement surface for collecting sound pressure data near the transformer. The coordinates of each sensor are represented by a three-dimensional coordinate vector, and the coordinate origin is set at the center position of the transformer. The coordinate system adopts a three-dimensional Cartesian coordinate system, with the center of the transformer as the reference point, and the positions of each sensor of the transformer are defined as three-dimensional coordinates (x, y, z). The number N of these sensors is usually greater than the number of sound sources because in practical applications, the sensor array can cover multiple angles of the transformer to provide more comprehensive sound data collection. To ensure the accuracy of sound source localization, the measurement surface needs to be perpendicular to a certain axis direction of the plane where the sensor array is located, and this plane is defined as the sound source focusing grid point plane. The setting of the focusing grid point plane is determined according to the actual structure of the transformer and the possible positions of the fault sources. In the actual application of the transformer, the sensor array should be arranged in the core area of the transformer or other areas prone to failure, usually the control panel, terminal block or equipment joint around the transformer. These areas are prone to mechanical vibration or electrical faults and emit acoustic signals. The arrangement position of the sensor array should ensure that all possible sound source signals can be captured and the surrounding environmental noise can be suppressed. The setting of the sound source focusing grid point plane needs to be flexibly adjusted according to the specific transformer fault detection requirements and the possible positions of the sound sources to ensure that the detection accuracy of the fault sound source can be maximally improved.

[0036] Step 2: Based on the preprocessed sound pressure signals, construct a cross-spectral matrix between acoustic sensors, and perform eigenvalue decomposition and principal component extraction on the cross-spectral matrix.

[0037] Preferably, in the said Step 2, constructing the cross-spectral matrix includes: Based on the frequency-domain signal vector obtained after preprocessing, construct a cross-spectral matrix between acoustic sensors:

[0038] Among them, is the average number of frames, which is used to smooth random noise; is the frequency-domain signal vector of the nth frame; is the conjugate transpose of.

[0039] Preferably, in step 2, the eigenvalue decomposition and principal component extraction of the cross-spectrum matrix include: The diagonal elements of the cross-spectrum matrix are eigenvalues arranged in descending order. Select the top 20 largest eigenvalues to form the cross-correlation spectrum matrix .

[0040] Specifically, perform eigenvalue decomposition on the cross-spectrum matrix to obtain the product form of three matrices, including an orthogonal matrix composed of eigenvectors, a diagonal matrix with diagonal elements being eigenvalues arranged in descending order from large to small, and the conjugate transpose of the orthogonal matrix:

[0041] Among them, the diagonal elements of are eigenvalues arranged in descending order. From the decomposed eigenvalues, select the top 20 largest eigenvalues , and extract their corresponding 20 eigenvectors . Multiply the outer product matrix of each eigenvalue and its corresponding eigenvector, and sum the weights of all 20 outer product matrices to obtain the reduced-dimensional cross-spectrum matrix : .

[0042] Step 3: Calculate the orthogonal beam output in combination with the cross-spectrum matrix after principal component extraction, generate a preliminary sound source map, and screen the grid points in the preliminary sound source map in combination with a preset sound source intensity threshold.

[0043] Preferably, in step 3, calculating the orthogonal beam output based on the cross-spectrum matrix after principal component extraction includes: Step 3.1: Discretize the measurement surface into a grid surface containing M grid points, and delimit the sound source calculation plane based on the grid surface; Specifically, discretize the measurement surface into a grid surface containing M grid points. The grid point set is represented as , and the sound source calculation plane contains M grid points. Define a focusing surface for each grid point. The focusing vector of the mth grid point is , ; Step 3.2: Calculate the orthogonal beam output of each grid point in the sound source calculation plane to generate a preliminary sound source map; Specifically, the calculation of the orthogonal beam output of the mth grid point is:

[0044] Among them, is the cross-spectral matrix obtained by the principal component extraction in step 2, is the conjugate transpose of; Step 3.3, screening the grid points in the preliminary sound source map according to a preset sound source intensity threshold; Specifically, screening the grid points according to the preset sound source intensity threshold with the screening condition being:

[0045] retaining the grid points higher than the sound source intensity threshold as valid grid points; assigning the orthogonal beam output of the grid points that do not meet the screening condition to , ; It should be noted that the value range of the sound source intensity threshold is usually set between 10% and 30% of the lower limit of the signal dynamic range. The size of the sound source intensity threshold has an impact on OB-CLEAN-SC. When the sound source intensity threshold is larger, the main lobe width will become narrower, thereby improving the spatial resolution, but it may cause some low-energy valid signals to be ignored; when the sound source intensity threshold is smaller, the main lobe width becomes wider, which can capture more low-energy signals, but it will reduce the spatial resolution and may cause the sidelobe suppression effect to deteriorate; therefore, the setting of the sound source intensity threshold needs to balance between the main lobe width and the sidelobe suppression ability to ensure the accurate positioning of the fault sound source.

[0046] Further preferably, the sound source intensity threshold can be dynamically adjusted by an adaptive threshold algorithm based on the statistical characteristics of the sound source map. When screening the grid points in the preliminary sound source map, calculate the median and interquartile range of the intensity values of all grid points, and use the dynamic threshold formula , where is the median of the grid point intensity values, is the interquartile range of all grid point intensity values, and the coefficient can take 1.5 to cover 95% of the data in the normal distribution; represents the reference intensity of the background noise, reflects the noise fluctuation range. When the intensity of the fault signal is significantly higher than the background noise (IQR increases), automatically increases to focus on the valid signal.

[0047] Step 4: Use the CLEAN - SC algorithm to perform iterative clarification processing on the filtered sound source map. When the iterative termination condition is met, output the clear sound source localization map generated after iterative clarification beam processing.

[0048] Preferably, the iterative clarification processing of the filtered sound source map using the CLEAN - SC algorithm includes: (1) Initialize the sound source map Use the filtered sound source map as the initial sound source map , and the initial value of the cross - spectral matrix is .

[0049] (2) Main lobe peak search and direction / weighting vector calculation In the i - th iteration, search for the position of the main lobe peak in the sound source map : :

[0050] This position is the position of the strongest sound source in the current iteration; Calculate the direction vector of :

[0051] where is the position of the main lobe peak searched in the i - th iteration and the n - th microphone X n ; is the position of the n - th microphone; k is the wave number; Calculate the weighting vector of : .

[0052] (3) Cross - spectral matrix update Remove the contribution of the current main lobe peak from the cross - spectral matrix and update the remaining sound source information; The cross - spectral matrix for updating the remaining sound source information includes: The CLEAN - SC algorithm assumes that in the i - th iteration, when is the cross - spectral matrix caused by a single coherent source component :

[0053] where is the beamforming output corresponding to the main lobe peak in the (i - 1)-th iteration; is the focusing vector corresponding to a single coherent source component in the i-th iteration; is the conjugate transpose of;

[0054] According to , update the cross-spectral matrix:

[0055] where, is the cross-spectral matrix at the i-th iteration, used to store the updated sound source information during the iteration process, is the weight coefficient for adjusting the attenuation ratio of the sound source component during the iteration process.

[0056] (4) Update the sound source localization map according to the direction vector and the weighted vector Subtract the contribution of the peak source at during the (i - 1)-th iteration: During the (i - 1)-th iteration, although we have not calculated the result of the (i + 1)-th iteration, we can gradually subtract the contribution of the identified sound source through the design idea of the algorithm. Specifically, during each iteration process, the algorithm will calculate the sound source information identified in the current iteration (i.e., the i-th iteration), and update the sound source map according to this information. During the (i - 1)-th iteration, the contribution of the already identified sound source will be retained and gradually eliminated in the i-th iteration by optimizing the sound source map. For the part that has not entered the (i + 1)-th iteration, although it does not directly participate in the calculation, based on the previous sound source map (i.e., the result of the (i - 1)-th iteration), the approximate range of the current sound source position can be estimated, and by calculating the relative influence between the current sound source and the known sound sources, the contribution of the identified sound source can be subtracted. Reasonably speculate and gradually eliminate unnecessary contributions to ensure the stable convergence of the algorithm. Therefore, the not affected by this peak is:

[0057] where, is the weighted vector of the grid points meeting the screening conditions, then there is:

[0058]

[0059] In the formula, is the direction vector of the grid points meeting the screening conditions.

[0060] (5) Store the beam after clarity obtained in each iteration Store the beam after clarity obtained in each iteration :

[0061] In the formula, and are key parameters, where usually represents the gain factor, which is used to adjust the signal strength. Its value is dynamically adjusted according to the signal characteristics, noise environment, and algorithm objectives, and is usually set by minimizing the error or using an adaptive algorithm; while is the phase angle, representing the phase characteristics of the signal. Its value is based on the relative positions of the sensor arrays and the direction of signal arrival, and is usually determined by measuring the phase difference of the signals. And its value range is generally . and The specific values of

[0062] (6) Iteration termination condition According to the criterion , judge whether the algorithm meets the iteration termination condition, and the final output result is:

[0063] It should be noted that the OB-CLEAN-SC algorithm removes irrelevant noise in each round of calculation by iterating and optimizing the sound source map multiple times, and gradually improves the accuracy of sound source localization according to the characteristics and positions of known signal sources. In the finally output sound source localization map, the grid points near the sound source position will be displayed as high intensity values, indicating that there may be a fault source in these areas. In this way, based on the spatial distribution of the intensity values, the system can clearly display the fault source in the transformer and help the operation and maintenance personnel accurately locate the specific position of the fault source. In this way, OB-CLEAN-SC not only improves the accuracy of fault sound source localization, but also can effectively distinguish the main fault sound source from the secondary noise sources, avoiding misjudgment and missed judgment, and providing reliable technical support for subsequent maintenance work. Step 5, identify the areas with intensity higher than the set threshold in the clear sound source localization map generated in Step 4, analyze the position of the fault source in combination with the physical structure and operating status of the transformer under test, and generate a fault localization report.

[0064] Preferably, the said Step 5 includes: Step 5.1, identify the potential fault source positions by analyzing the high-intensity areas in the sound source map; The areas with higher intensity usually correspond to the fault sources inside or around the transformer; Exemplarily, set a dynamic threshold, usually taking 20% - 40% of the maximum intensity in the sound source map, which can be adjusted according to the transformer noise background; automatically identify the continuous regions in the sound source map with intensities greater than the threshold through image processing algorithms (such as region growing method, morphological filtering); align the grid coordinate system of the sound source map with the 3D CAD model or physical structure diagram of the transformer under test, establish the mapping relationship between the sound source grid points and the actual components, and determine the actual components corresponding to the high-intensity regions (such as windings, iron cores, bushings, etc.).

[0065] Step 5.2: Determine the specific location of the fault source according to the intensity distribution and direction vectors in the sound source map, combined with the physical structure of the transformer; for each grid point with a relatively high intensity, calculate its relevant direction vector and weighted vector to further improve the positioning accuracy; Exemplarily, for each high-intensity grid point, calculate its direction vector and weighted vector. Based on the calculation results, perform coherence analysis on adjacent high-intensity regions to determine whether the fault source is diffused or independent, then sort by intensity to clarify the processing priority and narrow down the range of the fault source.

[0066] Step 5.3: Generate a fault source location report, which includes the coordinates of the sound source location, the sound source intensity, the possible fault types and their locations, to help maintenance personnel quickly identify the fault source and perform repairs.

[0067] Specifically, judge the exact location of the fault source according to the high-intensity regions in the sound source map, combined with the operating status and structural characteristics of the transformer, and further confirm the fault source location; verify the positioning result to ensure that the identified fault source location is accurate. If necessary, deploy subsequent repair measures.

[0068] It should be noted that each grid point in the sound source localization map represents the intensity of a sound source. Areas with higher intensity usually indicate the location of the fault source. The areas with higher intensity can be identified by automatic recognition or by manual inspection by maintenance personnel of the sound source localization map. These areas usually correspond to the parts in the transformer where faults may exist. The exact positions of the high-intensity areas usually appear in the key parts of the transformer, such as the terminal blocks, the joint surfaces of the transformer, or other components that may fail. In the sound source localization map, areas with lower intensity may be irrelevant noise or background signals. Therefore, a suitable threshold γ needs to be set to filter out the signals with practical significance. By selecting a reasonable threshold, only those sound source intensities higher than the threshold will be retained for further analysis. This filtering step effectively avoids the influence of irrelevant noise signals on the localization accuracy and reduces the computational complexity at the same time. Through the filtered high-intensity areas, the next step is to conduct a refined analysis of the sound source distribution in these areas. With the specific intensity of each grid point in the sound source localization map, the maintenance personnel can locate the approximate position of the fault source. Further combining with the physical structure and operating status of the transformer, it can be confirmed whether the sound source comes from a potential fault point. During the localization process, it is possible to encounter interference from environmental noise or other irrelevant sources. By further clarifying the sound source map, redundant signals and noise will be suppressed, thereby further improving the reliability of the localization result. In each iteration process, the algorithm will optimize the identified sound source positions. Through repeated iterations, the accuracy of the sound source map is further improved. Each round of iteration will remove the contribution of the identified sound sources and adjust the localization of the remaining sound sources, ultimately making the position of the fault sound source more accurate. After multiple rounds of iteration, the final sound source map can clearly identify the position of the fault source inside or around the transformer. The high-intensity areas reflect the most likely fault sources, and these positions are the key areas for fault repair and maintenance. The final sound source map not only provides the accurate spatial position, but once the position of the fault source is determined, the system can generate a detailed fault localization report. The report will include the specific position of the fault source, the sound source intensity, the distinction from the background noise, and other possible fault-related information.

[0069] In Embodiment 2 of the present invention, a transformer fault sound source localization system based on orthogonal clarification beams is provided. Based on the method for localizing a transformer fault sound source using orthogonal clarification beams described in Embodiment 1 of the present invention, the system includes: A sensor array module for three-dimensionally arraying acoustic sensors and collecting the sound pressure signals of the transformer in real time; A signal preprocessing module for processing the original sound pressure signals, including frame segmentation, windowing, Fourier transform, and filtering for noise reduction; A cross-spectrum matrix calculation module for constructing a cross-spectrum matrix and extracting the principal component features; An orthogonal beamforming module for generating a preliminary sound source map and screening valid grid points; An iterative clarification module for iteratively optimizing the sound source localization map through the OB-CLEAN-SC algorithm; A fault analysis module for correlating the fault source location and priority through transformer structure analysis; A report generation module for generating a fault source localization report.

[0070] The present disclosure may be a system, a method, and / or a computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions thereon for causing a processor to implement various aspects of the present disclosure.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A transformer fault sound source localization method based on orthogonal beamforming, characterized in that It includes the following steps: Taking the center of the transformer under test as the coordinate origin, arranging N acoustic sensors in an array in the three-dimensional space around the transformer to form a three-dimensional sound field measurement surface, collecting sound pressure signals and performing preprocessing; Based on the preprocessed sound pressure signals, constructing a cross-spectral matrix between acoustic sensors, performing eigenvalue decomposition on the cross-spectral matrix, and selecting the first K largest eigenvalues and their corresponding eigenvectors to reconstruct the reduced-dimensional cross-spectral matrix; Combining the reduced-dimensional cross-spectral matrix to calculate the orthogonal beam output, generating a preliminary sound source map, and screening the grid points in the preliminary sound source map by combining a preset sound source intensity adaptive dynamic threshold; Using the CLEAN-SC algorithm to perform iterative clarification processing on the screened sound source map, and when the iterative termination condition is met, outputting the sound source localization map after iterative clarification beam processing; Identifying the areas with intensities higher than the set threshold in the sound source localization map, analyzing the position of the fault source in combination with the physical structure and operating state of the transformer under test, and generating a fault localization report.

2. A transformer fault sound source localization method based on orthogonal clarification beam according to claim 1, characterized in that: The collecting sound pressure signals and performing preprocessing includes: Frame and window the sound pressure signal, and convert each frame of time-domain signal into a frequency-domain signal through Fourier transform to obtain a frequency-domain signal vector , and perform filtering and noise reduction.

3. A transformer fault sound source localization method based on orthogonal clarification beam according to claim 2, characterized in that: The constructing a cross-spectral matrix between acoustic sensors based on the preprocessed sound pressure signals includes: Based on the frequency-domain signal vector obtained after preprocessing Construct the cross-spectrum matrix between acoustic sensors : Among them, is the average number of frames; is the frequency domain signal vector of the nth frame; is the conjugate transpose of; For the cross-spectrum matrix perform eigenvalue decomposition to obtain a product form of three matrices, including an orthogonal matrix composed of eigenvectors, a diagonal matrix with diagonal elements being eigenvalues arranged in descending order from largest to smallest, and the conjugate transpose of the orthogonal matrix; From the decomposed eigenvalues, select the top 20 eigenvalues with the largest numerical values, extract their corresponding 20 eigenvectors, multiply each eigenvalue by the outer product matrix of its corresponding eigenvector, and sum up the weighted values of all 20 outer product matrices to obtain the cross-spectrum matrix after dimensionality reduction .

4. A transformer fault sound source localization method based on orthogonal clarification beam according to claim 3, characterized in that: The combining the reduced-dimensional cross-spectral matrix to calculate the orthogonal beam output includes: Discretizing the measurement surface into a grid surface containing M grid points, and delimiting a sound source calculation plane based on the grid surface; Calculating the orthogonal beam output of each grid point in the sound source calculation plane to generate a preliminary sound source map; Screening the grid points in the preliminary sound source map according to a preset sound source intensity threshold.

5. A transformer fault sound source localization method based on orthogonal clarification beam according to claim 4, characterized in that: The discretizing the measurement surface into a grid surface containing M grid points and generating a preliminary sound source map includes: Discretize the measurement surface into a grid surface containing M grid points, and the grid point set is , the sound source calculation plane contains M grid points, define a focusing surface for each grid point, and the focusing vector of the m-th grid point is , ; The orthogonal beam output of the m-th grid point is: Among them, is the dimensionality-reduced cross-spectrum matrix, is the conjugate transpose of.

6. A transformer fault sound source localization method based on orthogonal clarification beam according to claim 5, characterized in that: The performing iterative clarification processing on the screened sound source map by using the CLEAN-SC algorithm includes: Setting the screened sound source map as the initial sound source map, and taking the reduced-dimensional cross-spectral matrix as the initial value of the cross-spectral matrix; In the i-th iteration, searching for the main lobe peak position in the current sound source map as the strongest sound source, and calculating the direction vector and the weighting vector based on this position; Removing this component from the cross-spectral matrix according to the contribution of the coherent source component of the current main lobe peak, and updating the cross-spectral matrix of the remaining sound source information; Based on the direction vector and the weighting vector, removing the contribution of the current main lobe peak from the sound source map to generate an updated sound source localization map; Storing the clarification beam data determined by the gain factor and the phase angle parameter in each iteration; Terminating the iteration when a preset criterion is met, and outputting the sound source localization map after iterative clarification beam processing.

7. A method for locating the fault sound source of a transformer based on an orthogonal beamforming method, characterized in that: The main lobe peak search and direction / weight vector calculation include: In the i-th iteration, search for the position of the main lobe peak in the sound source map and : Calculation of the direction vector : Among them, is the i peak position of the search main lobe in the th iteration, n and the distance between the th microphone; n is the position of the Calculation of the weighted vector : 。 8. A method for locating the fault sound source of a transformer based on an orthogonal beamforming method, characterized in that: The cross-spectral matrix for updating the remaining sound source information includes: Assume that in the $i$-th iteration is the cross-spectral matrix caused by a single coherent source component : Among them, is the beamforming output corresponding to the main lobe peak in the (i - 1)-th iteration; is the focusing vector corresponding to a single coherent source component in the i-th iteration; is the conjugate transpose of; According to , update the cross-spectrum matrix: Among them, is the cross-spectrum matrix at the i-th iteration, which is used to store the updated sound source information during the iteration process, is the weight coefficient for adjusting the attenuation ratio of the sound source component during the iteration process.

9. A method for locating the fault sound source of a transformer based on an orthogonal beamforming method, characterized in that: The fault location analysis and report generation include: Based on a dynamic threshold, extract continuous regions with intensities higher than the threshold from the clear sound source location map, identify potential fault sources through image processing algorithms, and align the sound source grid coordinate system with the three-dimensional physical structure model of the transformer to establish a mapping relationship between the fault region and the actual components; Combined with the sound source intensity distribution, direction vector, weight vector and the physical structure of the transformer, perform location optimization and coherence analysis on the regions with intensities higher than the threshold to determine the location and priority of the fault source; Output a location report containing the coordinates, intensity, associated components and fault type of the fault source, verify the location result in combination with the operating state of the transformer, and deploy repair measures.

10. A transformer fault sound source localization system based on orthogonal beamforming, based on the method for localizing a transformer fault sound source based on orthogonal beamforming according to any one of claims 1-9, characterized in that The system includes: A sensor array module for arranging acoustic sensors in a three-dimensional array and collecting the acoustic pressure signals of the transformer in real time; A signal preprocessing module for processing the original acoustic pressure signals, including frame division, windowing, Fourier transform and filtering for noise reduction; A cross-spectral matrix calculation module for constructing a cross-spectral matrix and extracting the principal component features; An orthogonal beamforming module for generating a preliminary sound source map and screening valid grid points; An iterative beamforming module for iteratively optimizing the sound source location map through the OB-CLEAN-SC algorithm; A fault analysis module for associating with the transformer structure to analyze the location and priority of the fault source; A report generation module for generating a fault source location report.

Images