A porcelain sleeve crack automatic detection method based on ultrasonic flaw detection

Abstract

This invention relates to the field of crack detection technology, specifically to an automatic crack detection method for ceramic bushings based on ultrasonic testing. The invention acquires images of the ceramic bushing under test on an automatic transmission device, determines the clamping position of the bushing, and extracts signal feature vectors containing time-domain, frequency-domain, and waveform morphology features from the ultrasonic echo signals at each detection position. These feature vectors are compared to determine the presence of cracks at each detection position, and the crack depth at each detection position where cracks are present is obtained. Combined with spatial correlation analysis, preliminary internal cracks are obtained, and their detection angle and detection surface position are determined. Simultaneously, a secondary detection is performed on the detection surface position to verify whether the spatial angle and length of the preliminary internal crack are acceptable. If the verification fails, it is repeated until it is successful, at which point a crack detection report is output. This effectively avoids misjudgment of crack parameters, improves crack detection accuracy, ensures the reliability of the detection report, and provides accurate data support for subsequent maintenance.

Description

Technical Field

[0001] This invention relates to the field of crack detection technology, and specifically to an automatic detection method for cracks in porcelain bushings based on ultrasonic testing. Background Technology

[0002] As a core insulation and support component in high-voltage power grids, the structural integrity of porcelain bushings directly determines the stability and safety of grid operation. If internal cracks are missed, they can easily lead to sudden penetrating breakdowns during operation, causing large-scale power outages. However, porcelain bushings often have complex umbrella-like shapes, requiring traditional inspection methods to balance full coverage with clamping and fixing compatibility. Furthermore, crack signals are susceptible to interference, making accurate crack identification and parameter measurement challenging. Therefore, an automated, high-precision crack detection method is urgently needed to solve these problems.

[0003] In the prior art, Chinese Patent Publication No. CN120721858A discloses a method for detecting internal cracks in ceramic sleeves based on acoustic emission detection. This method acquires initial acoustic emission signals, vibration signals, and electromagnetic interference signals, generates an interference feature vector, and performs adaptive interference cancellation. After extracting time-frequency features from the noise-reduced emission signal, dynamic feature analysis is performed to output crack presence indicators and type features. Finally, feature-weighted localization calculations are performed using spatial location parameters to generate the crack location coordinates. This method improves the detection reliability in complex environments through multi-source interference cancellation and dynamic feature analysis.

[0004] The existing technology has the following problems: 1. The existing technology does not have a dedicated clamping positioning and scanning path planning scheme for the special umbrella-shaped structure of the ceramic sleeve. It only achieves signal acquisition by arranging multiple sensors, which is prone to scanning blind spots or interference with detection at the clamping position. This results in incomplete detection coverage, causing some cracks to go undetected, creating safety hazards. At the same time, unreasonable clamping positions may cause damage to the ceramic sleeve or distortion of the detection signal.

[0005] 2. Existing technologies only rely on time and frequency features for crack identification and type determination, without integrating multi-dimensional features of time domain, frequency domain, and waveform morphology for accurate matching with standard feature libraries. This results in insufficient accuracy in crack identification, which can lead to false or false cracks. False cracks may cause unnecessary equipment replacement costs, while false cracks may allow cracked ceramic bushings to continue operating, creating safety risks.

[0006] 3. Existing technologies generate crack location coordinates by performing feature-weighted positioning calculations based on spatial position parameters. This can only achieve preliminary crack location and does not verify key parameters such as crack spatial direction and length in multiple dimensions. This may lead to misjudgment of crack parameters, resulting in a lack of accurate data support for subsequent maintenance decisions and affecting maintenance effectiveness. Summary of the Invention

[0007] The present invention aims to overcome the defects in the prior art and provide an automatic detection method for porcelain bushing cracks based on ultrasonic flaw detection, so as to achieve full coverage and high-precision detection of porcelain bushing cracks and accurately generate porcelain bushing crack detection reports.

[0008] The technical solution adopted by the present invention to solve its technical problem is: an automatic detection method for porcelain sleeve cracks based on ultrasonic flaw detection, including: S1, acquiring an image of the porcelain sleeve to be tested on an automatic transmission device, determining the clamping position of the porcelain sleeve to be tested, using an ultrasonic probe to perform a full-coverage scan of the porcelain sleeve to be tested, and acquiring ultrasonic echo signals at each detection position.

[0009] S2. Perform multi-dimensional feature extraction on the ultrasonic echo signals at each detection location to generate a signal feature vector containing time-domain features, frequency-domain features, and waveform morphology features. Then, retrieve the standard feature library of ceramic sleeve cracks for matching and comparison to determine whether there is a crack at each detection location.

[0010] S3. Screen the detection locations where cracks exist, obtain the crack depth location based on its ultrasonic echo signal, perform spatial correlation analysis on each crack depth location to obtain the preliminary internal crack, extract the spatial direction and length of the preliminary internal crack, and determine its detection angle and detection surface location.

[0011] S4. Control the clamping device to adjust the ceramic sleeve to the corresponding detection angle, detect the position of the detection surface, and verify whether the spatial angle and length of the preliminary internal crack are qualified.

[0012] S5. If the verification fails, the spatial correlation analysis and verification shall be performed again until the verification is successful and a crack detection report shall be output.

[0013] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention obtains the image of the porcelain sleeve to be tested on the automatic transmission device, determines the symmetrical clamping position of the porcelain sleeve to be tested corresponding to the non-umbrella skirt structure, and performs full coverage scanning in combination with the planned ultrasonic probe scanning path, effectively solving the problem of incomplete detection coverage, ensuring that all key areas of the porcelain sleeve can be detected, avoiding the cracks that are missed due to the scanning blind zone, and improving the comprehensiveness and reliability of the detection.

[0014] (2) The present invention performs multi-dimensional feature extraction on the ultrasonic echo signal at each detection location, generates a signal feature vector containing time domain features, frequency domain features and waveform morphology features, retrieves the standard feature library of ceramic sleeve cracks for matching and comparison, determines whether there is a crack at each detection location, achieves accurate crack identification through double similarity comparison, improves the accuracy of crack identification, effectively reduces the probability of crack misjudgment and missed judgment, and reduces equipment maintenance costs and safety risks.

[0015] (3) This invention obtains the crack depth location at each detection location where cracks exist, and obtains the preliminary internal crack by combining spatial correlation analysis. It extracts the spatial direction and length of the preliminary internal crack, determines its detection angle and detection surface location, provides accurate basis for subsequent secondary verification, solves the problem of ambiguous crack detection parameter characterization, improves the accuracy of crack detection, and provides reliable data support for maintenance decisions.

[0016] (4) This invention adjusts the ceramic sleeve to the detection angle and performs secondary detection on the detection surface position to verify whether the spatial angle and length of the initial internal crack are qualified. If the verification is not qualified, it is re-verified until the verification is qualified and then a crack detection report is output. This effectively avoids misjudgment of crack parameters, improves the accuracy of crack detection, ensures the reliability of the detection report, and provides accurate data support for subsequent maintenance. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0019] Figure 2 This is a schematic diagram of the steps for determining whether a crack exists at each detection location in this invention.

[0020] Figure 3 This is a schematic diagram illustrating the steps in this invention to verify whether the spatial angle and length of the initial internal crack are qualified. Detailed Implementation

[0021] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention. Furthermore, it should be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale.

[0022] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use. Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification.

[0023] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0024] Please see Figure 1 As shown, the present invention provides an automatic detection method for ceramic sleeve cracks based on ultrasonic flaw detection, including: S1, acquiring an image of the ceramic sleeve to be tested on an automatic transmission device, determining the clamping position of the ceramic sleeve to be tested, using an ultrasonic probe to perform a full-coverage scan of the ceramic sleeve to be tested, and acquiring ultrasonic echo signals at each detection position.

[0025] In one embodiment of the present invention, before detecting cracks in the ceramic sleeve, the problems of clamping interference and scanning blind spots must be solved. Therefore, the clamping position needs to be accurately located through image analysis first, and then a scanning path adapted to the umbrella skirt structure of the ceramic sleeve needs to be planned to ensure comprehensive detection coverage and the safety of the ceramic sleeve.

[0026] Considering that the ceramic sleeve adopts a complex umbrella-shaped structure, the umbrella-skirt area is uneven and is a critical insulation part. Direct clamping can easily cause damage to the surface of the ceramic sleeve, interfere with the transmission of ultrasonic signals, or cause scanning blind spots due to clamping obstruction. On the other hand, the non-umbrella-skirt structure area has a flat surface and strong mechanical stability, making it an ideal area for clamping. Therefore, it is necessary to prioritize the selection of safe clamping areas for non-umbrella-skirt structures.

[0027] Based on this, the determination of the clamping position of the ceramic sleeve to be tested is as follows: First, when the ceramic sleeve to be tested is conveyed into the image capturing area on the automatic transmission device, the appearance image of the ceramic sleeve to be tested is captured by the image acquisition device.

[0028] Then, edge contour extraction is performed on the acquired appearance image to identify the axial center line and outer surface contour of the ceramic sleeve to be tested.

[0029] Next, based on the outer surface contour of the ceramic sleeve to be tested, the clamping safety area corresponding to the non-umbrella structure of the ceramic sleeve to be tested is selected.

[0030] Finally, the geometric center point is obtained from the axial centerline of the porcelain bushing to be tested. Two clamping reference points symmetrical to the geometric center point are selected from the clamping safety area, and the spatial coordinates of the two clamping reference points are used as the clamping position of the porcelain bushing to be tested.

[0031] Preferably, in one embodiment of the present invention, the Canny edge detection algorithm is used to extract the edge contour of the acquired appearance image to obtain the outer surface contour of the ceramic sleeve to be tested, and the line connecting the centers of the two end faces of the ceramic sleeve is used as the axial center line. The Canny edge detection algorithm is an existing technology, and implementers can replace it with the Sobel algorithm, contour tracking algorithm, etc., according to actual needs, which will not be described in detail here.

[0032] After determining the clamping position of the ceramic sleeve to be tested, the ultrasonic probe scanning path needs to be planned to achieve comprehensive coverage of the outer surface of the sleeve without blind spots, while avoiding obstruction and interference from the clamping device. Considering that the ceramic sleeve has a cylindrical structure, a single axial or circumferential scan is prone to missed scans. Therefore, a combined axial and circumferential scanning mode is adopted. At the same time, when the scanning trajectory passes through the clamping position, the clamping device needs to be temporarily released to avoid obstructing the ultrasonic probe detection path.

[0033] Based on this, the ultrasonic probe is used to perform a full-coverage scan of the ceramic sleeve under test, as follows: First, based on the axial centerline and outer surface contour of the ceramic sleeve under test, the scanning path of the ultrasonic probe is planned. The scanning path includes an axial scanning trajectory along the axial centerline and a circumferential scanning trajectory around the axial centerline.

[0034] The second step is to control the ultrasound probe to perform coordinated scanning along the axial and circumferential scanning paths. When the scanning path reaches the first clamping position, the first clamping device is controlled to release and maintain a preset safe distance, ensuring that the ultrasound probe scans the area where the first clamping position is located without obstruction.

[0035] The third step is to control the first clamping device to reset and fix it after the ultrasonic probe has completed the first clamping position scan, and continue to complete the remaining scanning trajectory scan. When the remaining scanning trajectory reaches the second clamping position, the second clamping device is operated in the same way until the ceramic sleeve under test is fully covered by the scan.

[0036] This invention effectively avoids damage to the ceramic sleeve and signal interference during clamping by screening the clamping safe area corresponding to the non-umbrella structure of the ceramic sleeve to be tested, and then selecting the clamping position of the geometric center point on the symmetrical axial center line from the clamping safe area, thereby improving the clamping qualification rate. At the same time, by combining axial and circumferential coordinated scanning with dynamic avoidance of the clamping device, the entire surface of the ceramic sleeve is scanned, completely eliminating scanning blind spots and reducing the risk of missed crack detection.

[0037] This invention acquires images of the ceramic sleeve to be tested on an automatic transmission device, determines the symmetrical gripping position of the non-umbrella structure of the ceramic sleeve, and performs a full-coverage scan in combination with the planned ultrasonic probe scanning path. This effectively solves the problem of incomplete detection coverage, ensures that all key areas of the ceramic sleeve can be detected, avoids missed cracks caused by scanning blind spots, and improves the comprehensiveness and reliability of the detection.

[0038] S2. Perform multi-dimensional feature extraction on the ultrasonic echo signals at each detection location to generate a signal feature vector containing time-domain features, frequency-domain features, and waveform morphology features. Then, retrieve the standard feature library of ceramic sleeve cracks for matching and comparison to determine whether there is a crack at each detection location.

[0039] Considering that the original ultrasonic echo signal contains a lot of noise, it needs to be filtered first to retain the effective signal. The time domain features reflect the signal intensity and time distribution, the frequency domain features reflect the frequency composition of the signal, and the waveform morphology features can capture the sudden change pattern of the signal. The fusion of the three can comprehensively characterize the essential features of the crack signal, and combined with the standard feature library, a double similarity comparison is performed to effectively distinguish the crack signal from the normal signal.

[0040] Based on this, the signal feature vector generation method is as follows: S21, the ultrasonic echo signal at each detection position is filtered to obtain the filtered echo signal.

[0041] Preferably, in one embodiment of the present invention, a Gaussian filtering algorithm is used to filter the ultrasonic echo signal to remove high-frequency noise and electromagnetic interference signals, thereby obtaining the filtered echo signal.

[0042] S22. Extract time-domain features, frequency-domain features, and waveform morphology features from the filtered echo signal. The time-domain features include echo peak value, signal amplitude, rise time, and pulse width. The frequency-domain features include spectral peak value, main peak frequency, spectral bandwidth, and harmonic component ratio. The waveform morphology features include the number of inflection points and the number of slope abrupt changes.

[0043] It should be noted that the number of inflection points is calculated by taking the first derivative of the filtered echo signal to determine the slope of the signal. Points with a slope of 0 are identified as inflection points, and the number of inflection points is counted. The number of slope abrupt changes is the number of times the slope sign changes between adjacent intervals.

[0044] S23. Standardize the time-domain features, frequency-domain features, and waveform morphology features, and use all the processed features to form the signal feature vector for each detection position.

[0045] It should be noted that the Min-Max standardization method is used to standardize the time-domain features, frequency-domain features, and waveform morphology features, mapping the feature values ​​to the [0, 1] interval.

[0046] like Figure 2 As shown, the method for determining whether there is a crack at each detection location is as follows: First, collect porcelain sleeve samples with different crack types and different sizes from the historical porcelain sleeve detection database, extract the ultrasonic echo signals corresponding to the crack locations of all porcelain sleeve samples, construct the standard signal feature vectors of each crack type and different sizes, and form a porcelain sleeve crack standard feature library.

[0047] Secondly, the similarity between the signal feature vector at each detection location and the standard signal feature vectors of each crack type in the standard feature library of ceramic sleeve cracks of different sizes is calculated, and the signal feature vector at each detection location with the highest similarity to all standard signal feature vectors is selected.

[0048] Next, standard signal feature vectors of several normal porcelain sleeve samples are obtained from the historical porcelain sleeve detection database, the similarity between the standard signal feature vectors of each normal porcelain sleeve sample is calculated, and the similarity judgment threshold is determined based on the similarity.

[0049] Finally, the average similarity between the signal feature vector of each detection location and the standard signal feature vector of each normal porcelain sleeve sample is obtained. If the highest similarity corresponding to a certain detection location is greater than the similarity judgment threshold, and the average similarity corresponding to that detection location is less than the similarity judgment threshold, then it is determined that there is a crack at that detection location; otherwise, it is determined that there is no crack at that detection location.

[0050] Preferably, in one embodiment of the present invention, a cosine similarity algorithm is used to calculate the similarity between the signal feature vector of each detection position and the standard signal feature vector of each crack type at different sizes, wherein the cosine similarity value ranges from [0, 1], and the closer it is to 1, the higher the matching degree.

[0051] The similarity threshold is determined as follows: based on the similarity of the standard signal feature vectors among the normal porcelain sleeve samples, the average similarity of the standard signal feature vectors among the normal porcelain sleeve samples is obtained. and standard deviation ,Will As a similarity judgment threshold, it ensures that the similarity of normal samples is likely higher than the similarity judgment threshold, while the similarity of cracked samples is lower than the similarity judgment threshold.

[0052] In other embodiments of the present invention, the similarity judgment threshold can be dynamically adjusted according to the actual detection scenario. For example, in an environment with severe interference, the similarity judgment threshold can be appropriately reduced to improve detection sensitivity.

[0053] This invention extracts multi-dimensional features from the ultrasonic echo signals at each detection location, generating a signal feature vector that includes time-domain features, frequency-domain features, and waveform morphology features. It then retrieves a standard feature library of ceramic sleeve cracks for matching and comparison to determine whether a crack exists at each detection location. Through dual similarity comparison, it achieves accurate crack identification, improves the accuracy of crack identification, effectively reduces the probability of false and false crack detection, and lowers equipment maintenance costs and safety risks.

[0054] S3. Screen the detection locations where cracks exist, obtain the crack depth location based on its ultrasonic echo signal, perform spatial correlation analysis on each crack depth location to obtain the preliminary internal crack, extract the spatial direction and length of the preliminary internal crack, and determine its detection angle and detection surface location.

[0055] Considering that the propagation speed of ultrasound in ceramic sleeve material is constant, the crack depth can be calculated by the propagation time difference between the crack reflected wave and the bottom reflected wave; multiple depth locations of the crack have spatial correlation, and the crack outline can be delineated by cluster analysis; principal component analysis can effectively extract the spatial principal direction of the crack, providing a reference for the secondary detection angle.

[0056] Based on this, the method for obtaining the crack depth location is as follows: First, the ultrasonic echo signals corresponding to each detection location where a crack exists are screened, the crack reflection wave and the bottom reflection wave in the ultrasonic echo signal are identified, and the propagation time difference between the crack reflection wave and the bottom reflection wave is compared.

[0057] Then, based on the standard propagation speed and propagation time difference of ultrasound in the ceramic sleeve material, the crack depth corresponding to each detection location where a crack exists is calculated.

[0058] Finally, the location of the crack depth corresponding to each detection location where a crack exists is taken as the crack depth location.

[0059] It should be noted that in the ultrasonic echo signal, the crack reflection wave arrives earlier than the bottom reflection wave, and the bottom reflection wave is the latest strong reflection wave. Reflection wave identification is a prior art technique and will not be elaborated upon in this invention.

[0060] The preliminary method for obtaining internal cracks is as follows: coordinate calibration of each crack depth location in the same spatial coordinate system is performed to obtain the spatial coordinates of each crack depth location, and the spatial coordinates of each crack depth location are clustered into different cluster groups by neighbor clustering.

[0061] Based on the spatial coordinates of each crack depth location in each cluster group, a contour region containing all crack depth locations in each cluster group is delineated, and the contour region corresponding to each cluster group is taken as the initial internal crack.

[0062] Preferably, in one embodiment of the present invention, the K-means clustering algorithm is used to perform cluster analysis on the crack depth location coordinates, and the number of clusters K is set, for example, K=3. The implementer can also adjust it adaptively according to the number of cracks. The clustering distance threshold is the distance between adjacent detection locations. When the distance is less than or equal to the clustering distance threshold, they are classified into the same cluster group.

[0063] For each cluster, the convex hull algorithm is used to delineate the smallest convex polygon that contains the locations of all crack depths in the cluster, and this convex polygon is used as the initial internal crack.

[0064] It should be noted that the clustering algorithm can also be replaced by the DBSCAN algorithm, which is suitable for scenarios where the number of cracks is unknown; the convex hull algorithm is an existing technology and will not be discussed further.

[0065] After initially identifying the internal crack, this invention extracts its spatial direction and length, determines the detection angle and detection surface position for secondary detection, and the secondary detection must be carried out along the spatial direction of the crack to ensure accurate verification of crack parameters.

[0066] Based on this, the detection angle and the position of the detection surface are determined as follows:

[0067] S31. Based on the spatial coordinates of each initial internal crack corresponding to the crack depth position, obtain the centroid coordinates of each initial internal crack, and perform principal component analysis on them and the spatial coordinates of each crack depth position to obtain the spatial direction corresponding to each initial internal crack.

[0068] S32. Project the spatial coordinates of each initial internal crack corresponding to each crack depth position onto the spatial direction, obtain the maximum coordinate distance between each crack depth position, and take it as the length of each initial internal crack.

[0069] S33. Using the axial centerline of the porcelain sleeve to be tested as a reference baseline, obtain the angle between the spatial direction and the reference baseline, and use it as the detection angle. At the same time, obtain the position of the detection surface extending to the porcelain sleeve surface under this detection angle.

[0070] It should be noted that the spatial orientation of each preliminary internal crack is obtained as follows:

[0071] S311. Subtract the centroid coordinates of the corresponding initial internal crack from the spatial coordinates of the crack depth positions of each initial internal crack to obtain the decentralized data point set of each initial internal crack.

[0072] S312. Calculate the covariance matrix of each preliminary internal crack based on the decentralized data point set, and perform eigenvalue decomposition on the covariance matrix to obtain different eigenvalues ​​and corresponding unit eigenvectors.

[0073] S313. Compare different eigenvalues, select the unit eigenvector corresponding to the largest eigenvalue, and use it as the spatial direction.

[0074] This invention obtains the crack depth location at each detection position where a crack exists, combines spatial correlation analysis to obtain the preliminary internal crack, extracts the spatial direction and length of the preliminary internal crack, determines its detection angle and detection surface position, provides accurate basis for subsequent secondary verification, solves the problem of ambiguous crack detection parameter characterization, improves the accuracy of crack detection, and provides reliable data support for maintenance decisions.

[0075] S4. Control the clamping device to adjust the ceramic sleeve to the corresponding detection angle, detect the position of the detection surface, and verify whether the spatial angle and length of the preliminary internal crack are qualified.

[0076] In one embodiment of the present invention, the initially obtained crack parameters may be affected by factors such as detection angle and signal interference, and therefore errors are possible. Therefore, it is necessary to perform secondary detection verification by adjusting the detection angle to verify the accuracy of the crack space angle and length and avoid parameter misjudgment.

[0077] Considering that the initial detection is a conventional angle scan, it may not be able to fully capture the true extension direction and length of the crack. The secondary detection is carried out along the crack space direction, which can reduce the signal attenuation and distortion caused by the detection angle.

[0078] Based on this, such as Figure 3 As shown, the verification of whether the spatial angle and length of the preliminary internal crack are qualified includes: S41, based on the determined detection angle, combined with the current angle of the clamping device, calculating the target angle that the clamping device needs to adjust, and controlling the clamping device to adjust the ceramic sleeve to be tested to the target angle.

[0079] Preferably, in one embodiment of the invention, for example, the current angle of the gripping device. Based on the determined detection angle Calculate the target angle that the gripping device needs to adjust. If the detection angle is a spatial angle, it needs to be broken down into horizontal and vertical angle adjustments.

[0080] S42. The probe is used to perform a secondary scan of the surface position, and the ultrasonic echo signal of the secondary scan is collected. Based on the ultrasonic echo signal of the secondary scan, the spatial angle and length of the initial internal crack corresponding to the secondary scan are obtained in the same way.

[0081] S43. Compare the spatial angle and length of the preliminary internal crack with the spatial angle and length of the secondary scan. If the spatial angle is equal to the spatial angle of the secondary scan and the length is equal to the length of the secondary scan, then the verification of the spatial angle and length of the preliminary internal crack is qualified; otherwise, the verification of the spatial angle and length of the preliminary internal crack is unqualified.

[0082] S5. If the verification fails, the spatial correlation analysis and verification shall be performed again until the verification is successful and a crack detection report shall be output.

[0083] Considering that the reasons for unsuccessful verification may include clustering analysis errors, detection angle adjustment deviations, or signal interference, iterative verification requires repeating spatial correlation analysis and optimizing the detection angle until the verification is successful.

[0084] Preferably, in one embodiment of the present invention, the method of re-performing spatial correlation analysis and verification includes: if the secondary verification fails, adjusting the clustering parameters of the spatial correlation analysis, such as adjusting the clustering distance threshold or increasing or decreasing the number of clusters, re-performing the spatial correlation analysis of the crack depth location, obtaining the corrected preliminary internal crack and detection parameters, and performing iterative detection verification until the verification is qualified.

[0085] This invention performs secondary detection on the surface position by adjusting the ceramic sleeve to the detection angle, verifying whether the spatial angle and length of the initial internal crack are qualified. If the verification is unqualified, it is re-verified until it is qualified, and then a crack detection report is output. This effectively avoids misjudgment of crack parameters, improves the accuracy of crack detection, ensures the reliability of the detection report, and provides accurate data support for subsequent maintenance.

[0086] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0087] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0088] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0089] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0090] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An automatic detection method for ceramic sleeve cracks based on ultrasonic flaw detection, characterized in that, include: Images of the porcelain sleeve to be tested are acquired on the automatic transmission device to determine the clamping position of the porcelain sleeve to be tested. An ultrasonic probe is used to perform a full-coverage scan of the porcelain sleeve to be tested and to acquire ultrasonic echo signals at each detection position. Multidimensional feature extraction is performed on the ultrasonic echo signals at each detection location to generate signal feature vectors containing time-domain features, frequency-domain features, and waveform morphology features. The standard feature library of ceramic sleeve cracks is retrieved for matching and comparison to determine whether there are cracks at each detection location. Each detection location containing a crack is selected, and the crack depth is obtained based on its ultrasonic echo signal. Spatial correlation analysis is performed on each crack depth location to obtain preliminary internal cracks. The spatial direction and length of the preliminary internal cracks are extracted, and their detection angle and detection surface position are determined. The determination method is as follows: The coordinates of each crack depth location are calibrated in the same spatial coordinate system to obtain the spatial coordinates of each crack depth location. The spatial coordinates of each crack depth location are then clustered by proximity to form different cluster groups. Based on the spatial coordinates of each crack depth location in each cluster group, a contour region containing all crack depth locations in each cluster group is delineated, and the contour region corresponding to each cluster group is taken as the initial internal crack. Based on the spatial coordinates of each initial internal crack at its corresponding crack depth, the centroid coordinates of each initial internal crack are obtained. Principal component analysis is then performed on these coordinates and the spatial coordinates of each crack depth to obtain the spatial orientation of each initial internal crack. Project the spatial coordinates of each initial internal crack corresponding to each crack depth position onto the spatial direction, obtain the maximum coordinate distance between each crack depth position, and take it as the length of each initial internal crack. Using the axial centerline of the porcelain sleeve to be tested as a reference baseline, the angle between the spatial direction and the reference baseline is obtained, and this angle is used as the detection angle. At the same time, the position of the detection surface extending to the porcelain sleeve surface under this detection angle is obtained. The control clamping device is adjusted to the corresponding detection angle of the ceramic sleeve to be tested, and the position of the detection surface is detected to verify whether the spatial angle and length of the preliminary internal crack are qualified. If the verification fails, the spatial correlation analysis and verification will be performed again until the verification is successful and a crack detection report will be output.

2. The automatic detection method for ceramic sleeve cracks based on ultrasonic testing according to claim 1, characterized in that: The process of determining the clamping position of the ceramic sleeve to be tested is as follows: When the ceramic sleeve to be tested is conveyed into the image capturing area on the automatic transmission device, the appearance image of the ceramic sleeve to be tested is captured by the image acquisition device. Edge contours are extracted from the acquired appearance images to identify the axial centerline and outer surface contour of the ceramic sleeve to be tested. Based on the outer surface contour of the ceramic sleeve to be tested, select the clamping safety area corresponding to the non-umbrella structure of the ceramic sleeve to be tested. Obtain the geometric center point from the axial centerline of the porcelain bushing to be tested, and select two clamping reference points symmetrical to the geometric center point from the clamping safety area. Use the spatial coordinates of the two clamping reference points as the clamping position of the porcelain bushing to be tested.

3. The automatic detection method for ceramic sleeve cracks based on ultrasonic testing according to claim 2, characterized in that: The process of using an ultrasonic probe to perform a full-coverage scan of the ceramic sleeve under test is as follows: Based on the axial centerline and outer surface contour of the ceramic sleeve to be tested, the scanning path of the ultrasonic probe is planned. The scanning path includes an axial scanning trajectory along the axial centerline and a circumferential scanning trajectory around the axial centerline. The ultrasonic probe is controlled to perform coordinated scanning along the axial and circumferential scanning paths. When the scanning path reaches the first clamping position, the first clamping device is controlled to release and maintain a preset safe distance. After the ultrasonic probe completes the first clamping position scan, the first clamping device is reset and fixed, and the remaining scanning trajectory is continued. When the remaining scanning trajectory reaches the second clamping position, the second clamping device is operated in the same way until the ceramic sleeve under test is fully covered by the scan.

4. The automatic detection method for porcelain bushing cracks based on ultrasonic testing according to claim 1, characterized in that: The signal feature vector is generated in the following way: The ultrasonic echo signals at each detection location are filtered to obtain the filtered echo signals. Time-domain features, frequency-domain features, and waveform morphology features are extracted from the filtered echo signal. The time-domain features include echo peak value, signal amplitude, rise time, and pulse width. The frequency-domain features include spectral peak value, main peak frequency, spectral bandwidth, and harmonic component ratio. The waveform morphology features include the number of inflection points and the number of slope abrupt changes. The time-domain features, frequency-domain features, and waveform morphology features are standardized, and all the processed features are used to form the signal feature vectors for each detection location.

5. The automatic detection method for porcelain bushing cracks based on ultrasonic testing according to claim 1, characterized in that: The method for determining whether cracks exist at each detection location is as follows: We collected porcelain sleeve samples with different crack types and different sizes from the historical porcelain sleeve detection database, extracted the ultrasonic echo signals corresponding to the crack locations of all porcelain sleeve samples, and constructed standard signal feature vectors for each crack type and different sizes, thus forming a standard feature library of porcelain sleeve cracks. Calculate the similarity between the signal feature vector at each detection location and the standard signal feature vectors of each crack type in the standard feature library of ceramic sleeve cracks of different sizes, and select the signal feature vector at each detection location with the highest similarity to all standard signal feature vectors; Similarly, standard signal feature vectors of several normal porcelain sleeve samples are obtained from the historical porcelain sleeve detection database. The similarity between the standard signal feature vectors of each normal porcelain sleeve sample is calculated, and the similarity judgment threshold is determined based on the similarity. The average similarity between the signal feature vector of each detection location and the standard signal feature vector of each normal porcelain sleeve sample is obtained. If the highest similarity corresponding to a certain detection location is greater than the similarity judgment threshold and the average similarity corresponding to that detection location is less than the similarity judgment threshold, then it is determined that there is a crack at that detection location. Conversely, if no crack is found at the detection location, it is determined that there is no crack there.

6. The automatic detection method for porcelain bushing cracks based on ultrasonic testing according to claim 1, characterized in that: The method for obtaining the crack depth location is as follows: The ultrasonic echo signals corresponding to each detection location with cracks are screened, the crack reflection wave and the bottom reflection wave in the ultrasonic echo signal are identified, and the propagation time difference between the crack reflection wave and the bottom reflection wave is compared. Based on the standard propagation speed and propagation time difference of ultrasound in ceramic sleeve material, the crack depth corresponding to each detection location where cracks exist is calculated. The location of the crack depth corresponding to each detection location where a crack exists is taken as the crack depth location.

7. The automatic detection method for porcelain bushing cracks based on ultrasonic testing according to claim 1, characterized in that: The spatial orientation corresponding to each preliminary internal crack is obtained as follows: Subtract the centroid coordinates of the corresponding initial internal crack from the spatial coordinates of the crack depth of each initial internal crack to obtain the decentralized data point set of each initial internal crack. The covariance matrix of each preliminary internal crack is calculated based on the decentralized data point set. The covariance matrix is ​​then decomposed into eigenvalues ​​to obtain different eigenvalues ​​and corresponding unit eigenvectors. By comparing different eigenvalues, the unit eigenvector corresponding to the largest eigenvalue is selected and used as the spatial direction.

8. The automatic detection method for porcelain bushing cracks based on ultrasonic testing according to claim 1, characterized in that: The verification of whether the spatial angle and length of the preliminary internal crack are qualified specifically includes: Based on the determined detection angle, the target angle that the clamping device needs to be adjusted is calculated in combination with the current angle of the clamping device, and the clamping device is controlled to adjust the ceramic sleeve to be tested to the target angle. The surface position is scanned twice by an ultrasonic probe, and the ultrasonic echo signal of the second scan is collected. The spatial angle and length of the initial internal crack corresponding to the second scan are obtained based on the ultrasonic echo signal of the second scan. The spatial angle and length of the preliminary internal crack are compared with those of the secondary scan. If the spatial angle and length are equal, the verification of the spatial angle and length of the preliminary internal crack is qualified; otherwise, the verification of the spatial angle and length of the preliminary internal crack is unqualified.

Images