Leakage detection and partial discharge digital imaging detection method and system based on acousto-optic fusion

Abstract

The invention relates to the technical field of nondestructive testing, in particular to a leak detection and partial discharge digital imaging detection method and system based on acousto-optic fusion, and the method comprises the following steps: based on channel microphone array sound wave data in a partial discharge signal suspicious region, extracting a sound wave abnormal section, positioning a sound source, matching image edge features, and synchronously marking; and analyzing the phase change of the multi-frequency signal to judge a sound source concentration area, tracking the moving trend of a disturbance point, and outputting an acousto-optic fusion positioning trend track. According to the method, the partial discharge feature recognition sensitivity is improved through high-frequency peak paragraph screening and dominant frequency recognition, the abnormal region positioning precision is enhanced in combination with image edge extraction and sound source space matching, and acousto-optic synchronous positioning and trend trajectory display are achieved through multi-frequency signal phase analysis and image frame disturbance tracking. Through fusion of frequency domain feature extraction, image recognition, dynamic comparison and other actions, the spatial precision of abnormal source recognition and the multi-source fusion analysis efficiency are improved, and the partial discharge traceability and dynamic monitoring capability are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of non-destructive testing, and in particular to a leak detection and partial discharge digital imaging detection method and system based on the fusion of sound and light. Background Art

[0002] The technical field of non-destructive testing includes technical means for detecting and evaluating internal or surface defects of a tested object by using physical methods without damaging the integrity of the material or structure. The core contents of this technical field include ultrasonic testing, ray testing, electromagnetic testing, acoustic emission testing, and infrared thermal imaging testing, etc. By obtaining and analyzing the changes of different physical quantities, the identification and analysis of material defects, fatigue, corrosion, and performance degradation are realized. The overall technical system covers multiple links such as signal acquisition, image recognition, waveform analysis, feature extraction, and visualization processing, and is widely used in equipment maintenance and safety assessment work in industries such as power, aviation, rail transit, and petrochemical.

[0003] Among them, the leak detection and partial discharge digital imaging detection method refers to a method for detecting and recording leakage and partial discharge phenomena inside the equipment or at the connection part through acoustic imaging means based on the signal characteristics in the audible and ultrasonic frequency bands. The technical matters targeted by this method cover the acquisition, filtering, and enhancement processing of multi-band acoustic signals, the graphical display of the sound source position by combining digital imaging technology, and the functions such as frequency band selection, real-time sound map synchronization display, image shooting, and video archiving through interface operations. The system cooperates with an acoustic acquisition device with an array sensing structure to form an image through the reconstruction of the spatial distribution of sound intensity, and realizes the detection and recording of leakage points and partial discharge sources.

[0004] In the prior art, acoustic signals are mostly collected and processed based on fixed frequency bands, lacking dynamic recognition of abnormal frequency bands, resulting in low detection sensitivity for non-typical signal characteristics. In terms of the combination of images and sound sources, conventional methods mostly display the superposition of acoustic intensities corresponding to a fixed imaging area, failing to achieve the accurate correspondence between the specific structure boundaries in the image and the spatial positions of the sound sources, and easily causing the problem of blurred image annotation. There is a lack of in-depth analysis of the phase and waveform delay of signals, making it difficult to accurately judge the correlation between multiple sound sources and the correspondence of the intensity center points, and easily causing misjudgment of the sound source distribution range. In terms of image change tracking, the existing means mainly rely on static image comparison, lacking the recognition and analysis of the perturbation trend of consecutive frames, and it is difficult to master the dynamic movement path of the discharge perturbation, affecting the accurate grasp of the evolution process of partial discharge behavior. The above deficiencies will cause an increase in the spatial error of the determination of the partial discharge position, a decrease in the timeliness of image feedback, and a weak abnormal source tracking ability in actual applications, thereby affecting the integrity of equipment status assessment and the reliability of safety judgment. Summary of the Invention

[0005] The object of the present invention is to solve the disadvantages existing in the prior art, and a leak detection and partial discharge digital imaging detection method and system based on the fusion of sound and light are proposed.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme: A leak detection and partial discharge digital imaging detection method based on the fusion of sound and light, comprising the following steps: S1: Based on the acoustic wave data of the channel microphone array within the suspicious area of the partial discharge signal, detect the time-domain waveform and frequency spectrum curve of each signal, screen the acoustic wave segments with high-frequency spikes, identify the corresponding main frequency range, and generate a list of acoustic wave marks for abnormal frequency bands; S2: Call the list of acoustic wave marks for abnormal frequency bands, extract the visible light image of the area where the sound source position is aligned, extract the edge features of the reflector in the image, compare the spatial coordinates with the sound wave signal positioning points, mark the position of the image area after matching, and obtain the image area with synchronized sound and light annotation; S3: Call the image area with synchronized sound and light annotation, perform waveform delay and phase shift analysis on the multi-band signals within the annotated area, screen the signal segments with consistent phase changes, judge the coincidence of the signal intensity peak and the center point of the image, judge whether the sound source distribution is concentrated, and obtain the position of the sound source concentration area; S4: Based on the position of the sound source concentration area, extract the backscattered wave image frames continuously captured within the area, compare the change range of the edge contours in the images, compare the length of the displacement boundary line between two frames, screen the frames with boundary disturbance mutations, and obtain the detection data of the disturbance points of the partial discharge image.

[0007] As a further scheme of the present invention, the list of acoustic wave marks for abnormal frequency bands includes the band number, main frequency range, and signal intensity. The image area with synchronized sound and light annotation includes the image positioning coordinates, edge feature labels, and annotated area numbers. The position of the sound source concentration area includes the concentration area coordinates, sound source density index, and signal consistency parameter. The detection data of the disturbance points of the partial discharge image includes the disturbance point number, boundary disturbance amplitude, and disturbance occurrence time.

[0008] As a further scheme of the present invention, the specific steps for obtaining the list of acoustic wave marks for abnormal frequency bands are as follows: S111: Based on the acoustic wave data of the channel microphone array within the suspicious area of the partial discharge signal, detect the waveform mutation position of each signal, extract the mutation time and paragraph position, and generate a mutation time interval paragraph; S112: Call the mutation time interval paragraph, identify the frequency peak rate and amplitude increase difference of the corresponding spectrum data, screen the qualified segments, and obtain the high-frequency mutation frequency band interval; S113: According to the high-frequency mutation frequency band interval, identify the main frequency of the band, jump amplitude, disturbance energy, and duration, and use the formula: ; Calculate the frequency band perturbation intensity value, determine whether it is an abnormal main frequency range, and generate a list of abnormal frequency band sound wave markers; Among them, represents the frequency band perturbation intensity value, is the total number of segments within the frequency band, is the main frequency of the th segment within the is the th segment within the jump amplitude of the is the th segment within the perturbation energy of the is the th segment within the duration of the

[0009] As a further solution of the present invention, the steps for obtaining the image area marked by sound and light synchronization are specifically as follows: S211: Call the list of abnormal frequency band sound wave markers, combine the sound pressure amplitude and the wave peak time series, locate the difference in the angle between the sound source direction and the image perspective, and obtain the boundary data set of the area corresponding to the sound source; S212: According to the boundary data set of the area corresponding to the sound source, extract the image blocks of the reflector in the image, screen the points within the specified edge intensity range, and obtain the reflector edge coordinate set; S213: Call the reflector edge coordinate set and the coordinate values of the sound wave positioning points, analyze the three-dimensional distance and density offset between the corresponding points, and use the formula: ; Calculate the difference degree of the spatial error matching value, combine the set of matching points with the difference degree lower than the sound and light synchronization threshold, establish a corresponding relationship, and obtain the image area marked by sound and light synchronization; Among them, is the coordinate value of the sound wave positioning point of the matching point pair , is the coordinate value of the reflector edge point of the matching point pair , is the density change value of the edge point area of the matching point pair , is the number of matching point pairs, is the difference degree of the spatial error matching value.

[0010] As a further solution of the present invention, the steps for obtaining the position of the sound source concentrated area are specifically as follows: S311: Call the sound-light synchronized labeled image area, analyze the multi-band signal sequences within the labeled area, identify the time delay and phase shift of each group of signals, and obtain the phase delay amount of the frequency band waveform. S312: According to the phase delay amount of the frequency band waveform, filter out the signal segments with consistent phase change trends, identify the average phase value of the waveform period, the peak amplitude, and the variance of the time point change for each signal segment, and extract the edge brightness change rate of the corresponding area in the image. Use the formula: ; Combine the set of signal segments smaller than the reference threshold to obtain the interval value of the phase-consistent signal segments. Among them, represents the average phase value of the th signal segment, is the peak amplitude of the th signal segment, is the variance of the time point change of the th signal segment, is the brightness change rate of the image area of the th signal segment, is the total number of signal segments, is the phase consistency matching degree value. S313: Call the interval value of the phase-consistent signal segments, judge the spatial coincidence between the position corresponding to the time point where the peak is located and the center point of the image, and combine the structural parameters of the joint area to judge the signal concentration area and obtain the position of the sound source concentration area.

[0011] As a further solution of the present invention, the steps for obtaining the local discharge image disturbance point position detection data are specifically as follows: S411: Based on the position of the sound source concentration area, extract consecutive backscattered wave image frames, uniformly perform inter-frame calibration, and identify the boundary differences of the image frames to obtain the edge pixel difference boundary value. S412: Call the edge pixel difference boundary value, and for the line segment of the boundary point distribution between image frame pairs, identify the ratio of the difference between the contour boundary line length and the starting frame boundary line length, and combine the edge point density and the boundary length extension rate of the image frame. Use the formula: ; Calculate the boundary disturbance growth trend value, filter out the image frame numbers with the disturbance growth trend value exceeding the threshold, and establish a disturbance mutation image frame sequence. Among them, represents the reference frame boundary line length, represents the current frame boundary line length, represents the number of edge points of the current frame, represents the number of edge points of the reference frame, Represents the variance of the edge point spacing in the current frame, is the growth trend value of boundary disturbance; S413: extracting pixel coordinates of edge disturbance intensity peaks in the image frames according to the disturbance mutation image frame sequence, and converting them into space coordinates according to physical dimensions and shooting angles to obtain partial discharge image disturbance point detection data.

[0012] As a further solution of the present invention, the method further comprises step S5: S5: calling the local discharge image disturbance point detection data, combining the time series characteristics of the local discharge digital imaging detection, tracking whether the disturbance point has a continuous movement trend in the continuous frames, comparing the distance offset value between the concentrated position of the sound source, judging whether the offset exceeds the preset area range, marking the movement trend trajectory, and outputting the acoustic-optical fusion positioning trend trajectory; The acoustic-optical fusion positioning trend trajectory includes a moving trajectory path, a trajectory offset value, and a trend change amplitude.

[0013] As a further solution of the present invention, the steps for obtaining the acoustic-optical fusion positioning trend trajectory are specifically as follows: S511: calling the disturbance point detection data of the partial discharge image, extracting the spatial coordinates and frame sequence numbers of the disturbance points in the continuous frames in combination with the time series features, tracking the coordinate change trend according to the numbers, eliminating the trajectory segments with jumps, retaining the disturbance point sequence with continuous changes, and marking it as an analysis target, and generating a disturbance trajectory change parameter set; S512: According to the coordinate values ​​in the disturbance trajectory change parameter set and the coordinates of the sound source center, the inter-frame distance offset of the disturbance point is identified, and compared with the spatial offset reference value, and the frame point sequence exceeding the reference range is extracted to obtain the offset abnormal distance sequence value; S513: Based on the frame sequence and the offset direction of the offset abnormal distance sequence value, the disturbance point path is connected in time order, and the acoustic and optical response coordinates are superimposed, the trajectory of the change of the acoustic and optical position relationship is marked, and the acoustic and optical fusion positioning trend trajectory is output.

[0014] The leak detection and partial discharge digital imaging detection system based on acoustic-optical fusion is used to perform the leak detection and partial discharge digital imaging detection method based on acoustic-optical fusion, and the system comprises: The acoustic wave recognition module extracts the high-frequency peak signal of the channel based on the acoustic wave data of the channel microphone array in the suspected area of ​​the partial discharge signal, and combines the sound source prone locations of the grounding knife switch, insulation gap, and connection contact of the high-voltage switch cabinet to screen the signal segments with concentrated overlaps in the main frequency interval to obtain the high-frequency directional interval of the sound source; The image annotation module extracts the visible light image within the corresponding area based on the high-frequency pointing range of the sound source, identifies the boundary of the reflector in the image, combines the spatial direction data of the sound source, compares the overlapping area between the position of the reflector boundary and the sound source direction, and obtains the sound source reflection positioning map block; The synchronous analysis module extracts the multi-band signals of the channels within the area based on the sound source reflection positioning map block, combines the structural layouts of the high-voltage cable terminal and the bus connection point, determines whether the peak signal intensity coincides with the center position of the map block, and obtains the position of the sound source concentration area; The disturbance detection module extracts the backscattered image frames taken continuously based on the position of the sound source concentration area, compares the change in the displacement length of the boundary line between the image frames, identifies the position of the image frame with prominent disturbance changes, and obtains the local discharge image disturbance point detection data; The trajectory extraction module extracts the horizontal and vertical coordinate sequences of the disturbance points in the continuous frames based on the local discharge image disturbance point detection data, determines whether there is a continuous offset direction, identifies the offset distance from the sound source structure concentration area, and outputs the acoustic-optic fusion positioning trend trajectory.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, by screening the high-frequency spike segments in the acoustic wave data and identifying the main frequency range, an acoustic wave marker list for abnormal frequency bands is established, effectively enhancing the recognition sensitivity of the local discharge signal characteristics. Extract the edge features of the reflector in the visible light image, and perform spatial matching with the acoustic wave positioning points to improve the accurate annotation ability of abnormal areas in the image. Combine the delay and phase shift analysis of multi-band signals to judge the coincidence between the peak signal intensity and the center point of the image, and accurately determine the position of the sound source with the assistance of structural features. Use the comparison of the change range of the edge contour in the image frame and the length of the displacement boundary line to identify the disturbance mutation points, increasing the accuracy and reliability of image disturbance recognition. Track the moving trend of the disturbance points, and determine whether it exceeds the preset range by comparing the distance offset value, realizing the synchronous positioning of acoustic-optic information and the visualization of the trend trajectory. The processing flow integrates technical actions such as acoustic wave frequency domain feature extraction, image edge recognition, signal phase analysis, and image frame change tracking, not only enhancing the spatial accuracy of abnormal source recognition, but also improving the effectiveness of multi-source data fusion analysis, strongly supporting the fast and accurate traceability and dynamic tracking ability of local discharge behavior. Description of the Drawings

[0016] Figure 1 It is a schematic diagram of the working process of the present invention; Figure 2 It is a flowchart for obtaining the acoustic wave marker list for abnormal frequency bands in the present invention; Figure 3 It is a flowchart for obtaining the acoustic-optic synchronous annotation image area in the present invention; Figure 4 It is the flowchart for obtaining the position of the sound source concentration area in the present invention; Figure 5 It is the flowchart for obtaining the detection data of the disturbance points of the partial discharge image in the present invention; Figure 6 It is the flowchart for obtaining the acoustic-optic fusion positioning trend trajectory in the present invention. Specific embodiments

[0017] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0018] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more, unless otherwise specifically defined.

[0019] Embodiment 1 Please refer to Figure 1 , the present invention provides a technical solution: a leakage detection and partial discharge digital imaging detection method based on acoustic-optic fusion, including the following steps: S1: Based on the acoustic wave data of the channel microphone array in the suspicious area of the partial discharge signal, detect the time-domain waveform and frequency spectrum curve of each signal, screen the acoustic wave segments with high-frequency spikes, identify the corresponding main frequency range, and generate a list of acoustic wave marks in the abnormal frequency band; S2: Call the list of acoustic wave marks in the abnormal frequency band, extract the visible light image of the area where the sound source position is aligned, extract the edge features of the reflector in the image, compare the spatial coordinates with the sound wave signal positioning points, mark the position of the image area after matching, and obtain the acoustic-optic synchronous annotation image area; S3: Call the acoustic-optic synchronous annotation image area, perform waveform delay and phase shift analysis on the multi-band signals in the annotated area, screen the signal segments with consistent phase changes, judge the coincidence of the signal intensity peak and the center point of the image, and combine the structural characteristics of the high-voltage switch cabinet joint area to judge whether the sound source distribution is concentrated, and obtain the position of the sound source concentration area; S4: Based on the position of the sound source concentration area, extract the backscattered wave image frames continuously captured within the area, compare the change range of the edge contours in the images, compare the lengths of the displacement boundary lines between two frames, screen for frames with sudden boundary perturbations, and obtain the detection data of the local discharge image perturbation points; S5: Call the detection data of the local discharge image perturbation points, combine with the time series characteristics of the partial discharge digital imaging detection, track whether there is a continuous movement trend of the perturbation points in consecutive frames, compare the distance offset value from the sound source concentration position, determine whether the offset exceeds the preset regional range, mark the movement trend trajectory, and output the acoustic-optic fusion positioning trend trajectory.

[0020] The abnormal frequency band acoustic wave marking list includes the frequency band number, main frequency range, and signal intensity. The acoustic-optic synchronous annotation image area includes the image positioning coordinates, edge feature labels, and annotation area numbers. The sound source concentration area position includes the concentration area coordinates, sound source density index, and signal consistency parameter. The detection data of the local discharge image perturbation points includes the perturbation point number, boundary perturbation amplitude, and perturbation occurrence time. The acoustic-optic fusion positioning trend trajectory includes the movement trajectory path, trajectory offset value, and trend change amplitude.

[0021] Please refer to Figure 2 , and the specific steps for obtaining the abnormal frequency band acoustic wave marking list are as follows: S111: Based on the acoustic wave data of the channel microphone array within the suspicious area of the partial discharge signal, detect the waveform mutation positions of each signal, extract the mutation time and paragraph position, and generate the mutation time interval paragraphs; The acoustic wave data collected by the microphone array reflects the spatio-temporal distribution characteristics of the acoustic waves in different channels. For example, in a typical partial discharge detection scenario of a power equipment, the staff can identify the operating state of the equipment through the acoustic wave characteristics of the equipment. Conduct time series analysis on the acoustic wave data to detect abnormal waveform changes in the data. For example, if the amplitude of the acoustic wave suddenly increases at a certain moment, it can be regarded as a preliminary sign of an abnormal signal. Mark the time point through software algorithms and record the acoustic wave data before and after the time point. This is the breakdown and refinement of the initial short sentence content to obtain the precise acoustic wave abnormal time period; for each marked time point, further analyze the surrounding data, identify the acoustic wave source and propagation path, and determine the approximate position of the sound source by comparing the similarity of the acoustic wave data in each channel. This process depends not only on the accuracy of the hardware but also on the accuracy of the acoustic wave data processing algorithm. The analysis helps to locate the specific section of the acoustic wave abnormality for subsequent detailed inspection and generate the mutation time interval paragraphs.

[0022] S112: Call the mutation time interval paragraphs, identify the frequency peak rate and amplitude increase difference of the corresponding spectrum data, screen for qualified segments, and obtain the high-frequency mutation frequency band interval; First, calculate the acoustic wave spectrum within the interval. Use the Fast Fourier Transform (FFT) to perform frequency-domain analysis on the acoustic wave signal to obtain a frequency distribution map. This is the breakdown and refinement of the first short sentence. In the monitoring of power equipment, spectrum analysis can reveal the abnormal operation of electrical equipment. The peaks in the spectrum diagram reflect the energy concentration at specific frequencies. The abnormal concentration of frequencies is related to equipment failures. Screen the key frequency bands according to the preset frequency peak rate and amplitude increase difference limit. For example, set the frequency peak to exceed 20% of the average value, and the amplitude increase to be significantly higher than the surrounding frequency bands. Through refined analysis, high-frequency mutation frequency bands significantly different from the normal operation state can be screened. The frequency band is an important indicator of potential equipment failures, and generate the high-frequency mutation frequency band interval.

[0023] S113: According to the high-frequency mutation frequency band interval, identify the main frequency, jump amplitude, disturbance energy, and duration of the frequency band, and use the formula: ; Calculate the disturbance intensity value of the frequency band, determine whether it is an abnormal main frequency interval, and generate a list of abnormal frequency band acoustic wave marks; Among them, represents the disturbance intensity value of the frequency band, is the total number of segments within the frequency band, is the main frequency of the th segment within the frequency band, is the jump amplitude of the th segment within the frequency band, is the disturbance energy of the th segment within the frequency band, Select multiple acoustic wave segments within the frequency band, extract the main frequency value, frequency jump amplitude, frequency disturbance energy, and disturbance duration of each segment. In the scenario of partial discharge monitoring of power equipment, for example, from a segment of acoustic wave data obtained from a microphone array around a certain transformer, 5 high-frequency mutation acoustic wave segments are identified, denoted as segment 1 to segment 5 respectively. Their main frequency values in kHz can be calculated by the Fast Fourier Transform to obtain the frequency at the maximum energy concentration in each segment. For example, the spectral main peak of segment 1 appears at 38 kHz, corresponding to , and the rest are , , , , and the frequency jump amplitude can be obtained by the difference between the main frequency and the main frequency of the previous frequency band. For example , , , , , the frequency perturbation energy is the integral value of the power spectral density within the frequency band, for example , , , , , the perturbation duration is the time during which the signal perturbation is maintained within the corresponding frequency band, for example , , , , ; Substitute the numerical values into the formula for calculation: ; ; ; ; The calculated value of the frequency band perturbation intensity , by comparing with the preset reference value on-site, if the reference empirical benchmark is 80 and since 98.231 is greater than this benchmark, then this frequency band is determined to be an abnormal frequency band. Subsequently, the position of this frequency band is bound to the time stamp and recorded as the abnormal frequency band sound wave mark. The dimension unification process in the formula is as follows: Both are in kHz, and their product is kHz², The unit is Pa². Since the sound pressure level is on a logarithmic scale and has an exponential response to frequency changes, and the frequency perturbation part has been normalized, the unit impact can be regarded as normalized. The final result is used to discriminate the abnormal trend intensity and generate a list of abnormal frequency band sound wave marks; By performing a weighted combination of the main frequency and the jump amplitude, the ability to express the jump concentration trend is enhanced. At the same time, introducing the ratio of the perturbation energy to the duration strengthens the discriminative sensitivity to the frequency perturbation intensity. This result indicates that there are high-amplitude jumps and high-energy perturbations in the sound wave within this frequency band, reflecting a relatively high degree of abnormality in the sound wave of this frequency band, which needs to be used as a key reference frequency band.

[0024] Please refer to Figure 3 , the specific steps for obtaining the image area with synchronized sound and light annotation are as follows: S211: Call the list of abnormal frequency band sound wave marks, combine the sound pressure amplitude and the wave crest time series, locate the difference in the angle between the sound source direction and the image perspective, and obtain the boundary data set of the area corresponding to the sound source; Call the peak time series and corresponding sound pressure amplitude in the acoustic signal, and combine the sound source position to align the visible light image of a specific area, so as to locate the angular difference between the sound source direction vector and the image viewing vector. For one embodiment, when detecting the structural integrity of a bridge, potential structural defect areas can be identified by detecting acoustic waves. The acoustic wave marker list can provide key time points and sound pressure data. The analysis of the angular difference between the sound source direction vector and the image viewing vector can help accurately align the sound source location with the image data. This alignment is achieved by calculating the angle difference between the two vectors and adjusting the angle of the camera. It is necessary to adjust the camera view according to the propagation direction of the acoustic wave and the acoustic wave reflection characteristics to ensure the alignment accuracy between the camera and the sound source, establish the range of the spatial coincidence area, and finally obtain the boundary data set of the area corresponding to the sound source. The data set will be used in subsequent image processing and analysis steps to identify and mark the areas in the image corresponding to the defects detected by the acoustic waves.

[0025] S212: According to the boundary data set of the area corresponding to the sound source, extract the image blocks of the reflector in the image, screen the points within the specified edge intensity range, and obtain the reflector edge coordinate set; Through image processing software, analyze the image blocks of the reflector formed by the reflection of the material edge in the image, calculate the gray gradient change value for the image blocks, and thus analyze the sharpness of the edges of the image blocks. For example, when detecting the surface defects of mechanical equipment, by extracting the reflection images of the equipment surface, calculate the gray gradient change value, the edge contour pixel density value, and the local image brightness dispersion value of each reflector. The weighted sum of the products of these three parameters is used as the judgment criterion to screen out the image blocks with sharpness values within the preset threshold range, and locate the edge point coordinate set. This process requires precise adjustment and optimization of image processing parameters to ensure the best visual results, improve the accuracy and efficiency of image analysis, and obtain the reflector edge coordinate set. The coordinate set provides key data for subsequent acoustic-optic synchronous positioning.

[0026] S213: Call the reflector edge coordinate set and the acoustic wave positioning point coordinate values, analyze the three-dimensional distance and density offset between the corresponding points, and use the formula: ; Calculate the difference degree of the spatial error matching value, combine the set of matching points with the difference degree lower than the acoustic-optic synchronous threshold, establish the corresponding relationship, and obtain the acoustic-optic synchronous labeled image area; Among them, is the acoustic wave positioning point coordinate value of the matching point pair , is the reflector edge point coordinate value of the matching point pair , is the edge point area density change value of the matching point pair , is the number of matching point pairs, is the difference degree of the spatial error matching value; When calling the edge coordinate set of the reflector and the three-dimensional coordinate values of the acoustic signal positioning points, it is necessary to normalize the format and unify the dimension of the coordinate data. First, convert the pixel coordinates of the reflector edge extracted from the image to physical three-dimensional coordinates. This conversion can be completed through the internal parameter matrix and viewing angle parameters of the camera. For example, the pixel coordinates of an edge point in a certain area are (520, 330), the image size is 1024×768, and the camera calibration parameters provide the focal length and imaging size, and the pixel value can be converted into physical coordinates in millimeters. Suppose the converted coordinates are (20.5, 13.2, 0), while the three-dimensional coordinate point data of the sound source collected in acoustic positioning is directly in millimeters. For example, a corresponding point is (19.6, 14.1, 0). At this time, the dimensions of the two are unified. Next, it is necessary to obtain the density change value of the edge point area , which is obtained by statistically analyzing the change amplitude of the pixel density in the neighborhood of the edge point. For example, taking the edge point as the center, count the number of edge pixels within a radius of 3 pixels, and then calculate the ratio with the total number of edge pixels in the surrounding area with a radius of 5 pixels. In the current example, the number of edge points in the neighborhood is 21, and the total area is 37, then the density change value is calculated as ; Substitute the parameters and calculate. Let the number of matching point pairs , and the three groups of points are respectively: Point pair 1: Acoustic point (19.6, 14.1), image point (20.5, 13.2), ; Point pair 2: Acoustic point (45.3, 28.4), image point (46.0, 27.5), ; Point pair 3: Acoustic point (78.9, 61.2), image point (79.4, 60.1), ; Then: ; ; ; ; ; ; The finally obtained difference degree of the spatial error matching value is 39.38. This value will be used to compare with the acousto-optic synchronization matching threshold. If the threshold is 45, then the current difference degree is lower than this value, and it is determined that the matching is effective. The matching points form the image area annotation position set, and the acousto-optic synchronization annotation image area is obtained; is the physical two-dimensional coordinate value of the acoustic positioning point, in millimeters, which is captured by the acoustic wave sensor and converted by the transmission system; is the two-dimensional coordinate value corresponding to the edge point of the image reflection sheet, which is obtained by converting the pixel points through the camera internal parameter matrix; is the density change value within the neighborhood of the image edge point, which is obtained by statistically analyzing the distribution difference of the pixel points after edge detection; is the number of valid matching point pairs, which is obtained by screening with the spatial distance less than a specific error range; is the acoustic-optic space matching error value, which represents the spatial synchronization degree between the reflection edge position and the sound source point, and is used to determine whether the matching is established; By combining the spatial offset between points and the intensity of local image structure change, the matching offset caused by image interference or local feature blur is avoided, making the mapping relationship of the acoustic-optic data more accurate. This result indicates that the spatial error between the reflection feature and the sound source positioning within the current image area is within the tolerance range, and the matching is valid, which can be used for image annotation and further regional analysis.

[0027] Please refer to Figure 4 , and the steps for obtaining the position of the sound source concentration area are specifically as follows: S311: Call the acoustic-optic synchronous annotation image area, analyze the multi-band signal sequence within the annotation area, identify the time delay and phase shift of each group of signals, and obtain the phase delay amount of the frequency band waveform; The processing of the multi-band signal sequence is particularly important in the diagnosis of power equipment, such as the insulation defect detection of transformers. By capturing the sounds and corresponding images emitted during the operation of power equipment, extracting the signal sequence from them, calculating the time delay and phase shift in each group of signals, this process requires normalizing the time series to eliminate the errors caused by different sampling frequencies, and then superimposing the time delay and phase shift through a specific algorithm. Considering the center frequency of the signal frequency band, adjust the size by multiplying the data by the inverse of the frequency. For example, in a detection, if the captured signal frequency is 500Hz, the calculated time delay value is 0.002 seconds, and the phase shift is 30 degrees, then the normalized delay amount can be represented as a specific value of the phase delay amount of the frequency band waveform. This value is used for subsequent signal quality evaluation and fault location identification to obtain the phase delay amount of the frequency band waveform, helping technicians determine whether the insulation materials inside the equipment are aging or damaged.

[0028] S312: According to the phase delay amount of the frequency band waveform, screen the signal segments with consistent phase change trends, identify the average phase value of the waveform cycle, the peak amplitude, and the variance of the time point change for each segment of the signal, and extract the edge brightness change rate of the corresponding area in the image. Use the formula: ; Combine the set of signal segments less than the reference threshold to obtain the phase-consistent signal segment interval value; Among them, represents the average phase value of the th signal segment, is the peak amplitude of the th signal segment, is the variance of the change in the time point of the th signal segment, is the change rate of the brightness of the image area of the th signal segment, is the total number of signal segments, is the phase consistency matching degree value; Screen the signal segments to identify the signal segments with consistent phase change trends. This process first divides each signal segment into cycles, extracts the average phase value, peak amplitude, and variance of the change in the time point within each cycle, and collects the change rate of the edge brightness in the corresponding area of the image. Set the number of signal segments to be screened as , and perform numerical calculations on three specific signal segments; The cycle average phase value of signal segment 1 , peak amplitude , variance of the change in the time point , change rate of the image edge brightness ; The corresponding values of signal segment 2 are , , , ; The corresponding values of signal segment 3 are , , , ; For unified units, all phase values are in degrees, amplitude values are in volts, variance values are in seconds squared, and brightness change rate values are in percentage of pixel gray value change, and all have been normalized; Calculate item by item as follows: Item 1: ; Item 2: ; Item 3: ; Sum result: ; Take the average value: ; Final consistency matching degree value: ; If the preset phase change reference threshold is 10,000, this value is less than the threshold, meeting the screening criteria, belonging to the consistency set, and obtaining the interval value of the phase-consistent signal segment; represents the average phase value within the signal cycle of the segment, in degrees, obtained by summing and averaging the phases corresponding to the sampling points of each signal cycle; is the variance of the position changes at each time point in this segment of the signal, reflecting the degree of periodic fluctuation, in square seconds, obtained by squaring the standard deviation; is the edge brightness change rate of the region corresponding to this segment of the signal in the image, in percentage gray scale change, calculated as the ratio of the mean change of the regional gray scale gradient to the gray scale change range of the entire image; is the number of signal segments to be screened, which is 3 here; is the phase consistency matching degree value, used to screen and determine whether the signal has homologous characteristics; By performing joint operations on the time-domain waveform features (average phase, amplitude), time fluctuation index (variance), and image features (brightness change rate), a multi-source data fusion metric is constructed on a unified scale, enabling phase screening to no longer be limited to a single signal channel or time-domain feature. This result indicates that the currently screened signal segments have high similarity under multiple metrics and can be grouped into the same phase feature set.

[0029] S313: Call the interval value of the phase-consistent signal segment, judge the spatial coincidence of the position corresponding to the time point of the wave peak with the center point of the image, and combine the joint area structure parameters to judge the signal concentration area to obtain the position of the sound source concentration area; Identifying the concentrated area of the sound source in electrical equipment is crucial for the maintenance of circuit breakers or switchgear in large power grids. By analyzing the time point positions of the maximum amplitudes of each signal and comparing the positions with the center point of the image through spatial projection, combined with the joint area structure characteristics of high-voltage equipment, such as the density of contacts, the proportion of the insulation wrapping area, and the number of components in the area, the concentrated area of the signal can be judged. For example, if in the joint area of a certain high-voltage switchgear, the spatial projection positions of multiple signal segments are highly coincident with the center point of the image, and the joint density in this area is high and the insulation material coverage is sufficient, this indicates that the detected discharge sound indeed comes from a specific position in this area, thus successfully obtaining the position of the sound source concentration area. The result provides accurate guidance for subsequent repair and maintenance work, ensuring the stable operation and safety of electricity.

[0030] Please refer to Figure 5 , and the steps for obtaining the detection data of the local discharge image disturbance point position are specifically as follows: S411: Based on the position of the sound source concentration area, extract consecutive backscattered wave image frames, uniformly perform inter-frame calibration, identify the boundary differences of the image frames, and obtain the edge pixel difference boundary value; In the fault diagnosis of high-voltage power equipment, especially in the detection of partial discharge, accurate positioning based on the position of the sound source concentration area is crucial. The consecutive backscattered wave image frames extracted from the sound source concentration area provide the original data for analysis. Through advanced image processing techniques, first perform spatial calibration of the image frames to ensure that all image frames are compared in the same spatial coordinate system. Use edge detection algorithms to extract the edge contours in each frame of the image. The contours are decisive for identifying the development of faults. For example, in an actual operation, when there are fine cracks in the equipment insulation layer, the cracks will be manifested as continuous changes in the edge contours in consecutive image frames. By comparing the edge contour differences between adjacent image frames, the expansion of the cracks can be accurately tracked, thus quickly and effectively locking the fault area and obtaining the edge pixel difference boundary value, which lays the foundation for the next detailed analysis.

[0031] S412: Call the edge pixel difference boundary value, and for the distribution line segment of the boundary points between pairs of image frames, identify the ratio of the difference between the length of the contour boundary line and the length of the starting frame boundary line, and combine the edge point density and boundary length extension rate of the image frame, and use the formula: ; Calculate the boundary disturbance growth trend value, screen the image frame numbers with the disturbance growth trend value exceeding the threshold, and establish a disturbance mutation image frame sequence; Among them, represents the reference frame boundary line length, represents the current frame boundary line length, represents the number of edge points in the current frame, represents the number of edge points in the reference frame, represents the variance of the edge point spacing in the current frame, is the boundary disturbance growth trend value; In the backscattered wave image frame sequence, extract any set of reference frames and current frames, and obtain the change of the boundary disturbance amount. The primary operation is to calculate the reference frame boundary line length and the current frame boundary line length , both of which are obtained by accumulating the lengths based on the Euclidean distance between adjacent pixel points in the edge pixel set. For example, if the edge line segment consists of 100 consecutive pixel points and the average distance between adjacent pixels is 1.2 pixel units, then the boundary line length is pixel units. Let the reference frame , the current frame grows due to edge perturbation to , secondly, it is necessary to obtain the number of edge points of the current frame and the reference frame. Let , , these two items are obtained by counting the number of points where the pixel gradient value is greater than the edge detection threshold. Then, calculate the variance of the edge point spacing of the current frame . This value is calculated based on the squared difference of the distances between adjacent points after sorting the edge points, and the mean value is taken as the variance. Assume (in units of pixel²). To eliminate the dimensional difference and scale influence, the normalized unit is uniformly adopted. All length and quantity parameters retain the pixel unit, and the variance remains in pixel square units without conversion; Substitute the numerical value to get:

[0032] Set the boundary perturbation threshold to 2.3, then the perturbation value of the current frame , which meets the mutation screening criteria. It is determined that this frame constitutes a perturbed mutation image frame, screened into the set, and a sequence of perturbed mutation image frames is established; By simultaneously introducing three participation factors: the relative change rate of the boundary length, the change rate of the edge pixel density, and the uniformity of the edge point distribution (variance), a composite perturbation value criterion under a unified dimension is constructed to avoid the risk of misjudgment caused by single-factor interference judgment based only on the boundary line difference or pixel quantity difference. The result shows that the current image frame fluctuates strongly under multi-dimensional perturbation terms and is determined to be a perturbed mutation frame, which can be used for subsequent extraction and positioning of image perturbation points.

[0033] S413: According to the sequence of perturbed mutation image frames, extract the pixel coordinates of the peak value of the edge perturbation intensity in the image frame, and convert them into spatial coordinates according to the physical size and shooting angle to obtain the detection data of the local discharge image perturbation point position; After identifying the sequence of perturbed mutation image frames, analyzing the specific perturbation point positions in the frames is the last step in diagnosing local discharge. The perturbation point positions correspond to the discharge points. By analyzing the positions of the perturbation points in the image in detail and combining the physical size and shooting angle information of the image, the image coordinates can be converted into the actual physical positions on the device, which not only provides the precise positions of the discharge points but also helps the maintenance team formulate more effective maintenance strategies and preventive measures. Thus, the detection data of the local discharge image perturbation point positions is finally obtained. The data provides decisive evidence for subsequent repairs and ensures the safe and stable operation of the power.

[0034] Please refer to Figure 6 , the specific steps for obtaining the trend trajectory of acoustic-optic fusion positioning are as follows: S511: Call the detection data of the disturbance point positions in the partial discharge images, combine with the time series features, extract the spatial coordinates of the disturbance points and the frame sequence numbers in consecutive frames, track the changing trend of the coordinates according to the numbers, eliminate the trajectory segments with jumps, retain the sequence of continuously changing disturbance points, mark them as the analysis targets, and generate a set of disturbance trajectory change parameters; Based on the recognition and tracking of the disturbance points in consecutive frames in image analysis, extract the spatial coordinates and time identifiers of the disturbance points in each frame. Each disturbance point coordinate includes the x and y position information. The data is analyzed and extracted from the original image sequence through image processing algorithms. For example, in the monitoring of power equipment, assume that during a continuous monitoring period, a partial discharge phenomenon occurs at the cable joint. This phenomenon is captured by a high-speed camera and recorded as a sequence of frames. The disturbance points in each frame are identified for their positions through specific image processing techniques, such as extracting the disturbance feature points through edge detection and region growing methods. Then, the movement trajectories of the disturbance points in consecutive frames are analyzed through a tracking algorithm. Assume that in the first frame, the disturbance point is located at (200, 300), and in the second frame, it moves to (205, 305). At this time, the continuity and directionality of its trajectory are confirmed through a comparison algorithm. For the trajectory segments that are not continuous or have significant jumps, they are eliminated. For example, if the point suddenly jumps to (500, 600) in the third frame, it is considered to be caused by noise or misidentification, and this data point should be deleted from the trajectory data. Retain the continuously changing data points for further analysis. The processed data point sequence can more truly reflect the behavioral characteristics of the disturbance source. The processed data set becomes a set of disturbance trajectory change parameters for further analysis. For example, in subsequent steps, the parameters will be used to calculate the relative movement between the disturbance source and the key components of the equipment, so as to evaluate the health status and failure risk of the equipment.

[0035] S512: According to the coordinate values and the sound source center coordinates in the set of disturbance trajectory change parameters, identify the inter-frame distance offset of the disturbance points, compare it with the spatial offset reference value, extract the frame point sequence beyond the reference range, and obtain the offset abnormal distance sequence value; Identify the inter-frame distance offset of the disturbance points. Set the sound source center as the set reference point. For example, in the fault diagnosis of a power transformer, the sound source center is a certain fault location of the transformer predetermined by the acoustic emission technology. The distance between each frame coordinate of the disturbance point and the sound source center is then compared with the preset spatial offset reference value. For example, if the set offset reference value is 10 pixels and the calculated distance is 15 pixels, it is considered that the offset of the disturbance point in this frame exceeds the normal range. By comparing the offset distances of all frames, screen out those frames that are continuously or repeatedly beyond the reference value. The frames represent abnormal states during the operation of the equipment. If frames beyond the reference are continuously monitored, it can be preliminarily judged as a fault point of the equipment. Mark the sequence of frames as abnormal, and further analyze its cause and fault type to obtain the offset abnormal distance sequence value.

[0036] S513: Based on the frame sequence and offset direction of the offset anomaly distance sequence values, connect the paths of the disturbance points in chronological order, superimpose the acousto-optic response coordinates, label the change trajectory of the acousto-optic position relationship, and output the acousto-optic fusion positioning trend trajectory; Connect the paths of the disturbance points in chronological order, and superimpose the acousto-optic response coordinates. First, arrange the disturbance points in all frames identified as abnormal in chronological order to construct a spatial path map of a time series. For example, when using fiber optic sensing technology to monitor the cable path of a power system, the optical images and acoustic signals of each frame can be captured synchronously. The position of the disturbance points can be analyzed by combining image processing technology and sound sensor data. The specific position of each disturbance point is compared and connected with the sound source response position. Through this method, the relative movement and change trend between the disturbance source and the sound source can be visualized, which is very useful for understanding the fault development process and early diagnosis. Label the change trajectory of the acousto-optic position relationship, providing an intuitive way to observe the device state and fault development.

[0037] The leak detection and partial discharge digital imaging detection system based on acousto-optic fusion is used to execute the above-mentioned leak detection and partial discharge digital imaging detection method based on acousto-optic fusion. The system includes: The acoustic wave identification module extracts the channel high-frequency spike signals based on the acoustic wave data of the channel microphone array within the suspicious area of the partial discharge signal, combines the sound source prone positions of the grounding knife switch, insulation gap, and connection contact of the high-voltage switchgear, filters the signal segments with concentrated and overlapping main frequency intervals, and obtains the high-frequency pointing interval of the sound source; The image annotation module extracts the visible light images within the corresponding area based on the high-frequency pointing interval of the sound source, identifies the boundary of the reflector in the image, combines the spatial direction data of the sound source, and compares the overlapping area between the boundary position of the reflector and the sound source direction to obtain the sound source reflection positioning tile; The synchronous analysis module extracts the multi-band signals of the channels within the area based on the sound source reflection positioning tile, combines the structural layout of the high-voltage cable terminal and the bus connection point, and determines whether the peak of the signal intensity coincides with the center position of the tile to obtain the position of the sound source concentration area; The disturbance detection module extracts the backscattered image frames taken continuously based on the position of the sound source concentration area, compares the change in the displacement length of the boundary line between the frames, identifies the position of the image frame with prominent disturbance changes, and obtains the local discharge image disturbance point position detection data; The trajectory extraction module extracts the horizontal and vertical coordinate sequences of the disturbance points in continuous frames based on the local discharge image disturbance point position detection data, determines whether there is a continuous offset direction, identifies the offset distance from the sound source structure concentration area, and outputs the acousto-optic fusion positioning trend trajectory.

[0038] The above are only the preferred embodiments of the present invention and do not limit the present invention in other forms. Any person skilled in the relevant art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as they do not depart from the technical solution content of the present invention, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An acoustic-optic fusion-based leak detection and partial discharge digital imaging detection method, characterized in that, Including the following steps: S1: Based on the acoustic wave data of the channel microphone array within the suspicious area of the partial discharge signal, detect the time-domain waveform and frequency spectrum curve of each signal, screen the acoustic wave segments with high-frequency spikes, identify the corresponding main frequency range, and generate a list of acoustic wave marks for abnormal frequency bands; S2: Call the list of acoustic wave marks for abnormal frequency bands, extract the visible light image of the area where the sound source position is aligned, extract the edge features of the reflector in the image, compare the spatial coordinates with the sound wave signal positioning point, mark the position of the image area after matching, and obtain the image area with synchronized acoustic and optical labeling; S3: Call the image area with synchronized acoustic and optical labeling, perform waveform delay and phase shift analysis on the multi-band signals within the labeled area, screen the signal segments with consistent phase changes, judge the coincidence of the signal intensity peak and the center point of the image, judge whether the sound source distribution is concentrated, and obtain the position of the sound source concentration area; S4: Based on the position of the sound source concentration area, extract the backscattered wave image frames continuously captured within the area, compare the change range of the edge contours in the images, compare the lengths of the displacement boundary lines between two frames, screen the frames with sudden boundary disturbances, and obtain the detection data of the disturbance points in the partial discharge image; 2. The leak detection and partial discharge digital imaging detection method based on the fusion of acousto-optic according to claim 1, wherein The list of acoustic wave marks for abnormal frequency bands includes the band number, main frequency range, and signal intensity. The image area with synchronized acoustic and optical labeling includes the image positioning coordinates, edge feature labels, and labeled area numbers. The position of the sound source concentration area includes the concentration area coordinates, sound source density index, and signal consistency parameter. The detection data of the disturbance points in the partial discharge image includes the disturbance point number, boundary disturbance amplitude, and disturbance occurrence time.

3. The leak detection and partial discharge digital imaging detection method based on the fusion of acousto-optic according to claim 1, wherein The specific steps for obtaining the list of acoustic wave marks for abnormal frequency bands are as follows: S111: Based on the acoustic wave data of the channel microphone array within the suspicious area of the partial discharge signal, detect the waveform mutation position of each signal, extract the mutation time and paragraph position, and generate a mutation time interval paragraph; S112: Call the mutation time interval paragraph, identify the frequency peak rate and amplitude increase difference of the corresponding frequency spectrum data, screen the qualified segments, and obtain the high-frequency mutation frequency band interval; S113: According to the high-frequency mutation frequency band interval, identify the main frequency of the band, jump amplitude, disturbance energy, and duration, and use the formula: ； Calculate the disturbance intensity value of the band, judge whether it is an abnormal main frequency interval, and generate a list of acoustic wave marks for abnormal frequency bands; Among them, represents the frequency band disturbance intensity value, is the total number of segments within the frequency band, is the main frequency of the th segment within the th frequency band, is the jump amplitude of the th segment within the th frequency band, is the disturbance energy of the th segment within the th frequency band, is the duration of the th segment.

4. The leak detection and partial discharge digital imaging detection method based on the fusion of acousto-optic according to claim 3, characterized in that, The specific steps for obtaining the image area with synchronized acoustic and optical labeling are as follows: S211: Call the list of acoustic wave marks for abnormal frequency bands, combine the sound pressure amplitude and the wave peak time series, locate the difference in the included angle between the sound source direction and the image perspective, and obtain the boundary data set of the area corresponding to the sound source; S212: According to the boundary data set of the area corresponding to the sound source, extract the reflector image blocks in the image, screen the points within the specified edge intensity range, and obtain the reflector edge coordinate set; S213: Call the reflector edge coordinate set and the sound wave positioning point coordinate values, analyze the three-dimensional distance and density offset between the corresponding points, and use the formula: ； Calculate the difference degree of the spatial error matching value, combine the set of matching points with a difference degree lower than the acoustic-optical synchronization threshold, establish a corresponding relationship, and obtain the image area with synchronized acoustic and optical labeling; Among them, is the coordinate value of the acoustic positioning point of the matching point pair , is the coordinate value of the edge point of the reflector of the matching point pair , is the density change value of the edge point area of the matching point pair , is the number of matching point pairs is the difference degree of the spatial error matching value 5. The leak detection and partial discharge digital imaging detection method based on the fusion of acoustic and optical signals according to claim 4, characterized in that, The specific steps for obtaining the position of the sound source concentration area are as follows: S311: Call the sound-light synchronized labeled image region, analyze the multi-band signal sequences within the labeled region, identify the time delay and phase shift of each group of signals, and obtain the phase delay amount of the frequency band waveform; S312: According to the phase delay amount of the frequency band waveform, screen the signal segments with consistent phase change trends, identify the average phase value of the waveform period, the peak amplitude, and the variance of the time point change for each signal segment, and extract the edge brightness change rate of the corresponding region in the image. Use the formula: ； Combine the set of signal segments smaller than the reference threshold to obtain the interval value of the phase-consistent signal segments; Among them, represents the average phase value of the th segment of the signal, is the peak amplitude of the th segment of the signal, is the variance of the change in the time points of the th segment of the signal, is the change rate of the brightness of the image area of the th segment of the signal, is the total number of signal segments, is the phase consistency matching degree value; S313: Call the interval value of the phase-consistent signal segments, judge the spatial coincidence between the position corresponding to the time point where the peak is located and the center point of the image, combine the structural parameters of the joint area to judge the signal concentration area, and obtain the position of the sound source concentration area.

6. The leak detection and partial discharge digital imaging detection method based on the fusion of acoustic and optical signals according to claim 5, wherein The steps for obtaining the detection data of the partial discharge image disturbance point positions are specifically as follows: S411: Based on the position of the sound source concentration area, extract consecutive backscattered wave image frames, uniformly perform inter-frame calibration, and identify the boundary differences of the image frames to obtain the edge pixel difference boundary value; S412: Call the edge pixel difference boundary value, for the line segment of the boundary point distribution between image frame pairs, identify the ratio of the difference between the contour boundary line length and the starting frame boundary line length, and combine the edge point density and the boundary length extension rate of the image frame. Use the formula: ； Calculate the boundary disturbance growth trend value, screen the image frame numbers with disturbance growth trend values exceeding the threshold, and establish a disturbance mutation image frame sequence; Among them, represents the length of the reference frame boundary line, represents the length of the current frame boundary line, represents the number of edge points of the current frame, represents the number of edge points of the reference frame, represents the variance of the edge point spacing of the current frame, is the boundary perturbation growth trend value; S413: According to the disturbance mutation image frame sequence, extract the pixel coordinates of the peak value of the edge disturbance intensity in the image frame, and convert them into spatial coordinates according to the physical size and shooting angle to obtain the detection data of the partial discharge image disturbance point positions.

7. The leak detection and partial discharge digital imaging detection method based on the fusion of acousto-optic according to claim 1, characterized in that, The method further includes step S5: S5: Call the detection data of the partial discharge image disturbance point positions, combine the time series characteristics of the partial discharge digital imaging detection, track whether there is a continuous movement trend of the disturbance points in consecutive frames, compare the distance offset value with the sound source concentration position, judge whether the offset exceeds the preset area range, mark the movement trend trajectory, and output the sound-light fusion positioning trend trajectory; The sound-light fusion positioning trend trajectory includes a movement trajectory path, a trajectory offset value, and a trend change amplitude.

8. The leak detection and partial discharge digital imaging detection method based on the fusion of acoustic and optical signals according to claim 7, characterized in that, The steps for obtaining the sound-light fusion positioning trend trajectory are specifically as follows: S511: Call the detection data of the partial discharge image disturbance point positions, combine the time series characteristics, extract the spatial coordinates and frame sequence numbers of the disturbance points in consecutive frames, track the coordinate change trend according to the numbers, eliminate the trajectory segments with jumps, retain the continuously changing disturbance point sequence, and mark it as the analysis target to generate a disturbance trajectory change parameter set; S512: According to the coordinate values and the sound source center coordinates in the disturbance trajectory change parameter set, identify the inter-frame distance offset of the disturbance points, compare it with the spatial offset reference value, extract the frame point sequence exceeding the reference range, and obtain the offset abnormal distance sequence value; S513: Connect the paths of the disturbance points in chronological order based on the frame sequence and the offset direction of the offset anomaly distance sequence values, superimpose the acousto-optic response coordinates, mark the change trajectory of the acousto-optic position relationship, and output the acousto-optic fusion positioning trend trajectory.

9. An acoustic-optic fusion-based leakage detection and partial discharge digital imaging detection system, characterized in that, The leak detection and partial discharge digital imaging detection method based on acousto-optic fusion according to any one of claims 1-8, the system comprising: Based on the acoustic wave data of the channel microphone array in the suspicious area of the partial discharge signal, the acoustic wave recognition module extracts the channel high-frequency spike signals, combines the source prone positions of the grounding knife switch, insulation gap, and connection contact of the high-voltage switch cabinet, and screens the signal segments with concentrated and overlapping main frequency intervals to obtain the high-frequency pointing interval of the sound source. Based on the high-frequency pointing interval of the sound source, the image annotation module extracts the visible light image in the corresponding area, identifies the boundary of the reflector in the image, combines the spatial direction data of the sound source, and compares the coincidence area between the boundary position of the reflector and the sound source direction to obtain the sound source reflection positioning tile. Based on the sound source reflection positioning tile, the synchronous analysis module extracts the multi-band signals of the channels in the area, combines the structural layout of the high-voltage cable terminal and the bus connection point, and determines whether the peak value of the signal intensity coincides with the center position of the tile to obtain the position of the sound source concentration area. Based on the position of the sound source concentration area, the disturbance detection module extracts the backscattered image frames taken continuously, compares the change in the displacement length of the boundary line between the frames, identifies the position of the image frame with prominent disturbance changes, and obtains the local discharge image disturbance point position detection data. Based on the local discharge image disturbance point position detection data, the trajectory extraction module extracts the horizontal and vertical coordinate sequences of the disturbance points in the continuous frames, determines whether there is a continuous offset direction, identifies the offset distance from the concentrated area of the sound source structure, and outputs the acousto-optic fusion positioning trend trajectory.

Images