Automatic detection and quantitative evaluation method for damage of composite thin-wall shaft tube

Abstract

The invention discloses an automatic detection and quantitative evaluation method for damage of a composite thin-wall shaft tube. Comprising the following steps: designing a multi-axis water immersion type ultrasonic phased array automatic scanning device, and obtaining an ultrasonic A scanning original signal set of a complete detection area; background noise is suppressed by using a self-adaptive wavelet threshold de-noising method based on a progressive semi-soft threshold function, and effective signals are enhanced; gate amplitude data are extracted to reconstruct a damage grayscale image, and damage features are enhanced in combination with bilinear interpolation, morphological opening and closing reconstruction and a CLAHE algorithm; an improved Canny operator is adopted to extract a damage edge, breakpoints are repaired in combination with morphological processing, a damage area is filled with a flooding algorithm, and finally quantitative evaluation of the damage size is completed based on a pixel-physical size mapping relation. The invention provides a nondestructive detection scheme with high precision, high automation and strong adaptability, and can realize efficient, accurate and intelligent detection of composite material thin-wall shaft tube damage in the fields of aerospace and high-end manufacturing.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasonic phased array detection, and in particular to a method for automatic detection and quantitative evaluation of damage to a thin-walled shaft tube of a composite material. Background Art

[0002] Carbon fiber reinforced plastic (CFRP) composites (composites), due to their light weight, high strength, and excellent corrosion resistance, are becoming the material of choice for critical load-bearing components in the aerospace industry. They are particularly widely used in thin-walled axle tubes and other structures. These thin-walled axle tubes typically serve as aircraft drive shafts, propulsion system piping, and load-bearing support components. They significantly reduce structural weight while meeting the requirements of high-performance transmission and demanding service environments.

[0003] However, due to factors such as the diversity of material components and the complex geometric characteristics of components, they are prone to damage such as delamination, porosity, and debonding during manufacturing, assembly, and service. This significantly weakens the mechanical properties and structural integrity, endangering service safety. Therefore, throughout the entire life cycle of composite thin-walled shaft tubes, including design, processing, manufacturing, product acceptance, and post-maintenance, regular inspections using non-destructive testing technology are required to identify defects and damage levels, monitor the dynamic changes in damage, and provide information input for structural design optimization, repairability, and re-service safety assessment.

[0004] At present, conventional ultrasonic testing technology is mainly used for non-destructive testing of thin-walled composite shaft tubes. The damage is identified and reconstructed by decoupling the time-domain amplitude-phase characteristics of the pulse echo signal. However, this method has the following limitations when detecting complex curved surface structures: First, because thin-walled shaft tubes are large-curvature rotating body stacked structures, their geometric characteristics make it difficult for ultrasonic waves to be incident vertically on the detection interface. At the same time, the interface echo and the damage signal are prone to aliasing in the time-frequency domain, which seriously affects the detection signal-to-noise ratio and damage detection rate. Second, the quantitative assessment of damage relies on manual experience to set thresholds and feature interpretation, lacks the support of adaptive algorithms, and has low detection efficiency and poor result consistency, making it difficult to meet the needs of batch automated testing. Therefore, how to efficiently, reliably and flexibly realize the damage detection of thin-walled composite shaft tubes is an urgent problem to be solved. Summary of the Invention

[0005] To address the aforementioned challenges in the existing technology, the present invention proposes a method for automated damage detection and quantitative assessment of thin-walled composite material shaft tubes, aiming to improve detection efficiency, reliability, and intelligence. This method utilizes a multi-axis, water-immersed ultrasonic phased array automated scanning device to acquire the raw ultrasonic A-scan dataset. This method uses an adaptive wavelet threshold denoising algorithm to improve signal quality. Based on the two-dimensional damage grayscale image reconstructed from the denoised signal, damage image enhancement and edge segmentation techniques are used to extract and quantify damage features, enabling high-precision automated damage detection and quantitative assessment.

[0006] In order to achieve the above technical objectives, the present invention provides the following technical solutions:

[0007] A method for automatically detecting and quantitatively evaluating damage of a composite thin-walled shaft tube comprises the following steps:

[0008] S1. Design a multi-axis water immersion ultrasonic phased array automatic scanning device suitable for in-situ inspection of thin-walled composite material shaft tubes; the multi-axis water immersion ultrasonic phased array automatic scanning device includes: a mechanical structure module, a motion control module, and a signal acquisition module;

[0009] S2. Select optimized ultrasonic phased array detection process parameters and obtain the ultrasonic A-scan original signal of the complete detection area through the signal acquisition module in the multi-axis water-immersion ultrasonic phased array automatic scanning device;

[0010] S3, performing discrete wavelet transform on the obtained ultrasonic A-scan original signal, and using an adaptive wavelet threshold denoising method based on a progressive semi-soft threshold function to remove background noise in the signal;

[0011] S4, extracting gate amplitude data from the denoised ultrasonic A-scan signal, and obtaining the denoised reconstructed grayscale image I in a two-dimensional mapping manner. s , bilinear interpolation, morphological opening and closing reconstruction, and CLAHE algorithm are used to enhance the damage features, and the enhanced damage image I is obtained. ss ;

[0012] S5. Use the improved Canny operator to extract the damage edge, including multi-level threshold adaptive weighted median filtering, multi-directional gradient enhancement Sobel operator, non-maximum suppression, dual threshold detection and edge connection;

[0013] S6. Apply morphological processing to bridge edge breakpoints, then use the flooding algorithm to perform connected domain analysis on closed edges and fill the pixels in the damaged area; finally, complete the quantitative assessment of the damage size based on the pixel-physical size mapping relationship calibrated by the system.

[0014] Furthermore, in step S1, the mechanical structure module specifically includes:

[0015] A three-axis gantry linear guide, a shaft tube rotation tooling, a transducer clamping device and a platform bracket; the three-axis gantry linear guide is used to support, position and drive the linear array transducer, and is fixed to the platform bracket by bolts; the shaft tube rotation tooling is used to transmit torque and drive the composite material thin-walled shaft tube to rotate in situ; the transducer clamping device is mounted on the Z-axis guide slide to clamp the transducer and provide a stable acoustic coupling environment.

[0016] Furthermore, in step S1, the motion control module specifically includes:

[0017] Motion controller, servo motor, servo drive, grating ruler and proximity switch; the motion controller is used to coordinately control the scanning trajectory and scanning speed of each axis according to the detection requirements; the servo drive is used to convert the control signal generated by the controller into motor power; the servo motor is used as an actuator to drive the transducer to move according to the preset motion parameters; the grating ruler is used to feedback the actual motion position of the linear guide; the proximity switch is used for motion limit and origin position correction.

[0018] Furthermore, in step S1, the signal acquisition module specifically includes:

[0019] Ultrasonic phased array host and linear array transducer; the ultrasonic phased array host is used to control the emission and reception of ultrasonic waves; the linear array transducer is used to transmit and receive ultrasonic signals, and form a controllable sound beam through phased excitation of array units.

[0020] Furthermore, in step S3, the adaptive wavelet threshold denoising method based on the progressive semi-soft threshold function specifically includes:

[0021] S31. Select appropriate wavelet basis functions and decomposition scales, perform discrete wavelet transform on the original ultrasonic A-scan signal, decompose it into low-frequency approximate components and high-frequency detail components at different scales, and select detail coefficients W from them. j,k , the formula is:

[0022]

[0023] Where f(n) is the original signal of ultrasonic A-scan; j is the decomposition scale; k is the translation factor of the wavelet basis function; n is the discrete sampling point sequence; φ(.) is the wavelet basis function;

[0024] S32, using a progressive semi-soft threshold function to perform adaptive threshold processing on the high-frequency signal in the discrete wavelet coefficients; the formula of the progressive semi-soft threshold function is expressed as:

[0025]

[0026] in, is the estimated value of the original useful signal;j is the adaptive threshold for different decomposition layers, and its formula is expressed as:

[0027]

[0028] Among them, N j is the number of wavelet coefficients in the jth layer; σ j is the standard deviation of the noise signal in the jth layer, and its formula is expressed as:

[0029]

[0030] Among them, median(|W j,k |) is the wavelet coefficient W j,k median of absolute values;

[0031] S33, reconstruct the ultrasonic signal after wavelet denoising, As the detail coefficients after reconstruction, the approximation coefficients remain unchanged and the time domain waveform data is restored.

[0032] Furthermore, in step S4, the morphological opening and closing reconstruction specifically includes:

[0033] S41, denoising and reconstructing the grayscale image I by using a disk structure element SE with a radius of 2 pixels s Perform corrosion treatment to obtain the intermediate result I after corrosion e ;

[0034] S42, corroding image I e As the marker image, denoise and reconstruct the grayscale image I s As the mask image, the grayscale expansion of the marked image is iterated to obtain the reconstructed image I;

[0035] S43, using the same structural element SE to reconstruct the image I o Perform dilation processing to obtain the dilated image I d ;

[0036] S44, dilate the image I d As the marker image, the reconstructed image I o As the mask image, the grayscale erosion of the marked image is iterated repeatedly to finally obtain the closed reconstructed image I oc .

[0037] Furthermore, in step S5, the multi-level threshold adaptive weighted median filtering is specifically:

[0038] The enhanced damaged image I obtained in step S4 is ss According to the maximum filter window size n maxPerform boundary expansion and set the filtering window \(W\) with an initial size of \(n\times n\). n , and traverse all pixel points \(I\) ss (x, y) in sequence to obtain the maximum gray value \(I\) within the current filtering window range max , the minimum gray value \(I\) min , and the gray median value \(I\) mid ;

[0039] If \(I\) satisfies min \(<I\) mid \(<I\) max and \(I\) min \(<I\) ss (x, y) \(<I\) max , it is determined that there are no noise points in the current window, and the original gray value \(I\) of the current pixel is output ss (x, y); otherwise, increase the filtering window \(W\) with a single-step size of \(n\rightarrow n + 2\) n for hierarchical determination;

[0040] If the filtering window is further increased to \(n>n\) max , calculate the minimum gray difference \(I\) m , and the formula is expressed as:

[0041] \(I\) m \(=\min(|I\) ss (x, y)-I\) min |,|I\) ss (x, y)-I\) max |);

[0042] Define the noise judgment threshold \(T = k\cdot(I\) max -I\) min ). If \(I\) m \(<T\), retain the original gray value \(I\) ss (x, y), otherwise, it is judged as a noise point and calculated according to the multi-neighborhood gray value weighted fusion expression, and the calculation result is the filtered gray value;

[0043] Repeat the above steps until the filtering of all pixel points of \(I\) is completed ss ​​​​​​​​​​​​​​​is the corresponding weight.

[0047] Furthermore, in step S5, the multi-directional gradient enhancement Sobel operator is specifically:

[0048] The output image I of the multi-level threshold adaptive weighted median filter is sss Convolution operations are performed with 3×3 Sobel gradient templates in four directions of 0°, 45°, 90° and 135° respectively, and the convolution results are projected to the horizontal and vertical directions respectively, and the horizontal component G of the comprehensive gradient is accumulated. X (x,y) and the vertical component G Y (x,y), its expression is:

[0049]

[0050] Among them, g d (x, y) is the pixel gradient value obtained by convolution of the ultrasound image and the templates in four different directions;

[0051] Then calculate pixel I sss The gradient magnitude G(x,y) and gradient direction θ(x,y) of (x,y) are:

[0052]

[0053] Furthermore, in step S5, the dual threshold detection and edge connection are specifically as follows:

[0054] The maximum inter-class variance algorithm is used to adaptively obtain the low threshold T for edge point detection. L , high threshold T H It is set to 1.6 times of the former, and the pixels are divided into strong edge points according to the image gradient amplitude G(x,y), that is, G(x,y)≥T H , and weak edge points, namely T L ≤G(x,y)≤T H And the validity of edge points is determined through connected domain analysis to complete edge connection.

[0055] Based on the above technical solution, the present invention has at least the following beneficial effects:

[0056] The multi-axis water-immersion ultrasonic phased array automatic scanning device provided by the present invention adopts an axial I-shaped scanning method, which can flexibly adapt to the full coverage automatic detection of thin-walled shaft tubes of composite materials with different diameters, effectively improve the detection accessibility of the ultrasonic beam, and improve the vertical incidence quality of ultrasonic waves in complex curved surface structures, ensuring the integrity and accuracy of the detection data.

[0057] To address the problem of excessive speckle noise in ultrasonic testing signals of thin-walled composite tubes, this paper proposes an adaptive wavelet threshold denoising method based on a progressive semi-soft threshold function. This method maximizes the preservation of effective signal detail while adaptively suppressing background noise. Compared with traditional denoising methods, this method more effectively improves the signal-to-noise ratio and enhances the resolution of ultrasonic signals, providing high-quality input data for subsequent damage identification and quantitative assessment.

[0058] For the quantitative assessment of damage in thin-walled composite tubes, a damage edge segmentation and quantitative assessment method based on an improved Canny operator is proposed. A multi-level threshold adaptive weighted median filter is used to preprocess the damage image, balancing the preservation of damage edge detail and the suppression of high-frequency noise. A multi-directional gradient-enhanced Sobel operator is introduced to effectively enhance the convolutional template's ability to detect weak damage edges oriented in tilted, curved, or other non-orthogonal directions. The maximum inter-class variance algorithm is used to adaptively determine the threshold for edge point detection. This method dynamically adjusts the segmentation threshold based on the image's grayscale histogram, improving the robustness and adaptability of edge detection and avoiding false or missed detections due to changes in material properties or the detection environment.

[0059] The present invention overcomes the limitations of existing ultrasonic detection methods for thin-walled composite shaft tubes, such as low signal-to-noise ratio, large detection blind spots, and insufficient quantitative assessment capabilities, and proposes a high-precision, highly automated, and highly adaptable non-destructive testing solution that can achieve efficient, accurate, and intelligent detection of damage to thin-walled composite shaft tubes, providing reliable technical support for structural health monitoring of composite materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 Schematic diagram of the process of the method proposed in the present invention;

[0061] Figure 2 It is a multi-axis water immersion ultrasonic phased array automatic scanning device in the method proposed by the present invention;

[0062] Figure 3 Schematic diagram of an artificially damaged sample of a composite thin-walled shaft tube in the method proposed in the present invention;

[0063] Figure 4 Schematic diagram and actual process diagram of thin-walled shaft tube scanning in the method proposed by the present invention;

[0064] Figure 5 This is the C-scan imaging result of the composite material thin-walled shaft tube damage in the method proposed in the present invention;

[0065] Figure 6 This is the quantitative error curve of damage of composite thin-walled shaft tube in the method proposed in this invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of the present invention more clear, the following Figure 1-6 The present invention is further described in detail with reference to the accompanying drawings and Examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which this application belongs.

[0067] Although the steps in the present invention are arranged with numbers, they are not intended to limit the order of the steps. Unless the order of the steps is clearly stated or the execution of a step requires other steps as a basis, the relative order of the steps can be adjusted. It is understood that the term "and / or" used herein refers to and covers any and all possible combinations of one or more of the associated listed items.

[0068] The present invention proposes a method for automatic detection and quantitative evaluation of damage of composite thin-walled shaft tubes, using Figure 2 The multi-axis water-immersion ultrasonic phased array automatic scanning device shown can flexibly adapt to the full coverage automatic detection of thin-walled shaft tubes of composite materials with different diameters, and integrates adaptive wavelet threshold denoising and image processing technology to improve the signal-to-noise ratio and damage feature extraction capabilities, thereby achieving high-precision damage identification and quantitative assessment. While ensuring the reliability of detection, the present invention can significantly reduce the dependence on the experience of the detection personnel and improve the consistency and automation level of batch detection. The implementation process of the method proposed by the present invention is as follows Figure 1 As shown, specifically including:

[0069] S1. Design a multi-axis water immersion ultrasonic phased array automatic scanning device suitable for in-situ inspection of thin-walled composite material shaft tubes; the multi-axis water immersion ultrasonic phased array automatic scanning device includes: a mechanical structure module, a motion control module, and a signal acquisition module;

[0070] As a preferred embodiment, Figure 2 As shown, the mechanical structure module includes a three-axis gantry linear guide, a shaft tube rotation fixture, a transducer clamping device and a platform bracket; the three-axis gantry linear guide is used to support, position and drive the linear array transducer and is fixed to the platform bracket by bolts; the shaft tube rotation fixture is used to transmit torque and drive the composite thin-walled shaft tube to rotate in situ; the transducer clamping device is mounted on the Z-axis guide slide to clamp the transducer and provide a stable acoustic coupling environment;

[0071] The motion control module is used to control the predetermined scanning trajectory and coordinate the working rhythm, and includes: a motion controller, a servo motor, a servo driver, a grating ruler, and a proximity switch. The motion controller is used to coordinately control the scanning trajectory and scanning speed of each axis according to the detection requirements. The servo driver is used to convert the control signal generated by the controller into motor power. The servo motor is used as an actuator to drive the transducer to move according to the preset motion parameters. The grating ruler is used to feedback the actual motion position of the linear guide. The proximity switch is used for motion limit and origin position correction.

[0072] The signal acquisition module is used for sampling, storing and transmitting ultrasonic signals, and includes: an ultrasonic phased array host and a linear array transducer; the ultrasonic phased array host is used to control the transmission and reception of ultrasonic waves; the linear array transducer is used to transmit and receive ultrasonic signals, and form a controllable sound beam through phased excitation of the array unit.

[0073] S2. Select optimized ultrasonic phased array detection process parameters and obtain the ultrasonic A-scan original signal of the complete detection area through the signal acquisition module in the multi-axis water-immersion ultrasonic phased array automatic scanning device;

[0074] In this example, a double-layer polytetrafluoroethylene film with a single layer thickness of 0.08 mm was used as the layering medium. A standard specimen with artificial pre-embedded damage was prepared using a filament winding process. The axial length was 350 mm, the inner diameter was 100 mm, and the layer design was a symmetrical structure [902 / 0 / 902]. 2s , a total of 20 layers, of which the axial direction is 0°, the theoretical wall thickness is 3.0mm, and the actual thickness after autoclave curing is 2.8mm. The specific distribution of sample damage and the preparation process are as follows Figure 3 As shown in the figure, there are three pre-embedded depth positions, namely the shallow layer (between the 3rd and 4th layers, the corrected depth is 0.42mm), the middle layer (between the 8th and 9th layers, the corrected depth is 1.41mm) and the near-bottom layer (between the 17th and 18th layers, the corrected depth is 2.38mm). Each layer is set with four circular damages of different sizes, with diameters of 15mm, 12mm, 9mm and 6mm from large to small.

[0075] The optimized ultrasonic phased array linear array transducer parameters are: number of elements 64, center frequency 5MHz, element width 0.5mm, element gap 0.1mm, element length 10mm. During the detection process, an axial bow-shaped scanning method with alternating axial translation stepping of the transducer and in-situ rotation of the shaft tube is adopted, and the number of excitation elements of the line scan is set to 8, the element stepping is 1, and the single-point focusing depth is 2.8mm. The full coverage scan of the composite material thin-walled shaft tube is completed at a constant rotation speed of 3° / s, a translation stepping speed of 10mm / s and a sampling trigger interval of 0.1°. The ultrasonic A-scan original signal set of the complete detection area is obtained in pulse echo mode. The scanning method diagram and the actual scanning process are shown in the figure. Figure 4 shown.

[0076] S3, performing discrete wavelet transform on the obtained ultrasonic A-scan original signal, and using an adaptive wavelet threshold denoising method based on a progressive semi-soft threshold function to remove background noise in the signal;

[0077] As a preferred embodiment, step S3 specifically includes:

[0078] S31. Select appropriate wavelet basis functions and decomposition scales, perform discrete wavelet transform on the original ultrasonic A-scan signal, decompose it into low-frequency approximate components and high-frequency detail components at different scales, and select detail coefficients W from them. j,k , the formula is:

[0079]

[0080] Wherein, f(n) is the original signal of the ultrasonic A scan; j is the decomposition scale; k is the translation factor of the wavelet basis function; n is the discrete sampling point sequence; φ(.) is the wavelet basis function; in this embodiment, the db4 wavelet basis function and the decomposition scale of 4 are selected;

[0081] S32, using a progressive semi-soft threshold function to perform adaptive threshold processing on the high-frequency signal in the discrete wavelet coefficients; the formula of the progressive semi-soft threshold function is expressed as:

[0082]

[0083] in, is the estimated value of the original useful signal; j is the adaptive threshold for different decomposition layers, and its formula is expressed as:

[0084]

[0085] Among them, N j is the number of wavelet coefficients in the jth layer; σ j is the standard deviation of the noise signal in the jth layer, and its formula is expressed as:

[0086]

[0087] Among them, median(|W j,k |) is the wavelet coefficient W j,k median of absolute values;

[0088] S33, reconstruct the ultrasonic signal after wavelet denoising, As the reconstructed detail coefficients, the approximate coefficients remain unchanged, and the time domain waveform data is restored. Since the noise is mainly concentrated in the high-frequency part of the signal, in this embodiment, the threshold is calculated only based on the statistical characteristics of the detail coefficients, which can effectively remove the high-frequency noise while retaining the low-frequency part of the signal.

[0089] S4, extracting gate amplitude data from the denoised ultrasonic A-scan signal, and obtaining the denoised reconstructed grayscale image I in a two-dimensional mapping manner. s , bilinear interpolation, morphological opening and closing reconstruction, and CLAHE algorithm are used to enhance the damage features, and the enhanced damage image I is obtained. ss ;

[0090] In this embodiment, the gate width is set to 0.3 mm, and the center position is set to the damage pre-embedded depth of 0.42 mm, 1.41 mm, and 2.38 mm, respectively. After extracting the A-scan signal gate amplitude, the damage grayscale image is reconstructed in a two-dimensional mapping manner, and bilinear interpolation is performed on the damage image array direction (longitudinal direction) with a scaling factor of 3.0 to improve the uniformity of pixel distribution; isolated image noise and local artifacts are eliminated through morphological opening and closing reconstruction, and internal holes are filled; the sub-block size of the CLAHE algorithm is set to 4×4, and the cropping threshold is set to 1.2 to expand the dynamic range of the grayscale value and improve the contrast of the damaged area. Finally, the C-scan imaging and three-dimensional reconstruction results of the composite material thin-walled shaft tube damage mapped by Jet color ruler are obtained as shown in the figure. Figure 5 As shown;

[0091] As a preferred implementation, in this embodiment, the morphological opening and closing reconstruction specifically includes:

[0092] S41, denoising and reconstructing the grayscale image I by using a disk structure element SE with a radius of 2 pixels s Perform corrosion treatment to obtain the intermediate result I after corrosion e ;

[0093] S42, corroding image I e As the marker image, denoise and reconstruct the grayscale image I s As the mask image, the grayscale expansion of the marked image is iterated to obtain the reconstructed image I;

[0094] S43, using the same structural element SE to reconstruct the image Io Perform dilation processing to obtain the dilated image I d ;

[0095] S44, dilate the image I d As the marker image, the reconstructed image I o As the mask image, the grayscale erosion of the marked image is iterated repeatedly to finally obtain the closed reconstructed image I oc .

[0096] S5. Use the improved Canny operator to extract damage edges, including multi-level threshold adaptive weighted median filtering, multi-directional gradient enhancement Sobel operator, non-maximum suppression, dual threshold detection and edge connection;

[0097] As a preferred embodiment, in step S5, the multi-level threshold adaptive weighted median filtering is specifically:

[0098] The enhanced damaged image I obtained in step S4 is ss According to the maximum filter window size n max Expand the boundary and set the initial size of the filter window W to n×n n , for all pixels I ss (x, y) is traversed to obtain the maximum grayscale value I within the current filter window range max , minimum gray value I min And gray median I mid ;

[0099] If I min mid max And I min ss (x,y) max , then it is determined that there is no noise point in the current window, and the original gray value of the current pixel is output. ss (x,y); otherwise, increase the filter window W by a single step size n→n+2 n Conducting step-by-step judgment;

[0100] If the filtering window is further increased to n>n max , calculate the minimum grayscale difference I m , the formula is:

[0101] I m =min(|I ss (x,y)-I min |,|I ss (x,y)-I max |);

[0102] Define the noise judgment threshold T = k·(I​​​​max -I min ) In this embodiment, k = 0.6 is taken. If I m < T, retain the original gray value I ss (x, y), otherwise it is judged as a noise point and calculated according to the multi-neighborhood gray value weighted fusion expression. The calculation result is the filtered gray value;

[0103] Repeat the above steps until the filtering of all pixel points of I ss is completed, remove the extended boundary area, and obtain the filtered output image I sss ;

[0104] The formula expression of the multi-neighborhood gray value weighted fusion is:

[0105]

[0106] Among them, I i (i = 1, 2, 3, 4) are the gray averages of the upper, lower, left, and right four rectangular neighborhood windows centered on the current pixel point respectively; m0, m i are the corresponding weights;

[0107] The multi-directional gradient enhanced Sobel operator is specifically:

[0108] The output image I of the multi-level threshold adaptive weighted median filter sss is respectively convolved with the 3×3 Sobel gradient templates in the four directions of 0°, 45°, 90°, and 135°, and the convolution results are respectively projected onto the horizontal and vertical directions, and the horizontal component G X (x, y) and the vertical component G Y (x, y) of the comprehensive gradient are obtained by accumulation. The expression is:

[0109]

[0110] Among them, g d (x, y) is the pixel point gradient value obtained by convolving the ultrasonic image with the four different direction templates;

[0111] Then calculate the gradient amplitude G(x, y) and gradient direction θ(x, y) of the pixel point I sss (x, y) respectively as:

[0112]

[0113] It should be noted that in this embodiment, the calculation of the horizontal component, vertical component, and gradient magnitude using the multi-directional gradient enhancement Sobel operator is intended to enhance the detection of weak edges with multi-directional damage and further suppress interference from false edges and background noise. The gradient direction is calculated to determine the edge orientation for subsequent non-maximum suppression.

[0114] The dual threshold detection and edge connection are specifically as follows:

[0115] The maximum inter-class variance algorithm is used to adaptively obtain the low threshold T for edge point detection. L , high threshold T H It is set to 1.6 times of the former, and the pixels are divided into strong edge points according to the image gradient amplitude G(x,y), that is, G(x,y)≥T H , and weak edge points, namely T L ≤G(x,y)≤T H The validity of the edge points is determined by connected domain analysis to complete the edge connection; at this time, the damaged edge of the composite thin-walled shaft tube is obtained.

[0116] S6. Apply morphological processing to bridge edge breakpoints, then use the flooding algorithm to perform connected domain analysis on closed edges and fill the pixels in the damaged area; finally, complete the quantitative assessment of the damage size based on the pixel-physical size mapping relationship calibrated by the system.

[0117] In the embodiment, the damage edge segmentation and quantitative evaluation method based on the improved Canny operator is used to quantitatively evaluate the delamination damage of the composite thin-walled shaft tube with different damage characteristics (damage depth, damage size), and the quantitative error of each delamination damage relative to the embedded size standard value is obtained as follows: Figure 6 As shown. Due to the influence of the array geometry and the acoustic beam focusing mechanism, the focal area of the linear array transducer has an asymmetric elliptical distribution. The acoustic energy in the scanning direction is dispersed, while the coherent superposition effect improves the focusing in the element stepping direction, resulting in detection results closer to the true value. Therefore, in practical applications, the quantitative results in the element stepping direction should be used as the standard. Evaluation results show that the quantitative error of the present invention in the dimensional measurement of delamination damage in thin-walled composite shaft tubes is less than 1.4 mm, demonstrating high detection accuracy and reliability.

[0118] In summary, the method proposed in the present invention can realize efficient, high-precision and automated damage detection of thin-walled composite shaft tubes, reduce dependence on the experience of inspectors, reduce the impact of human factors on the test results, and improve the consistency and reliability of batch testing. It is suitable for composite material structural health monitoring in the aerospace and high-end manufacturing fields.

[0119] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

[0120] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A method for automatic detection and quantitative assessment of damage in composite thin-walled shaft tubes, characterized in that: The specific steps include: S1. Design a multi-axis water immersion ultrasonic phased array automatic scanning device suitable for in-situ inspection of thin-walled composite material shaft tubes; the multi-axis water immersion ultrasonic phased array automatic scanning device includes: a mechanical structure module, a motion control module, and a signal acquisition module; S2. Select optimized ultrasonic phased array detection process parameters and obtain the ultrasonic A-scan original signal of the complete detection area through the signal acquisition module in the multi-axis water-immersion ultrasonic phased array automatic scanning device; S3, performing discrete wavelet transform on the obtained ultrasonic A-scan original signal, and using an adaptive wavelet threshold denoising method based on a progressive semi-soft threshold function to remove background noise in the signal; S4, extracting gate amplitude data from the denoised ultrasonic A-scan signal, and obtaining the denoised reconstructed grayscale image I in a two-dimensional mapping manner. s Then, bilinear interpolation, morphological opening and closing reconstruction, and CLAHE algorithm are used to enhance the damage features to obtain the enhanced damage image I ss ; S5. Use the improved Canny operator to extract the damage edge. The Canny operator includes a multi-level threshold adaptive weighted median filter, a multi-directional gradient enhancement Sobel operator, non-maximum suppression, dual threshold detection, and edge connection connected in sequence. S6. Apply morphological processing to bridge edge breakpoints, then use the flooding algorithm to perform connected domain analysis on closed edges and fill the pixels in the damaged area; finally, complete the quantitative assessment of the damage size based on the pixel-physical size mapping relationship calibrated by the system.

2. The method for automatic detection and quantitative assessment of damage of a composite thin-walled shaft tube according to claim 1, characterized in that: In step S1, the mechanical structure module specifically includes: A three-axis gantry linear guide, a shaft tube rotation tooling, a transducer clamping device and a platform bracket; the three-axis gantry linear guide is used to support, position and drive the linear array transducer, and is fixed to the platform bracket by bolts; the shaft tube rotation tooling is used to transmit torque and drive the composite material thin-walled shaft tube to rotate in situ; the transducer clamping device is mounted on the Z-axis guide slide to clamp the transducer and provide a stable acoustic coupling environment.

3. The method for automatic detection and quantitative assessment of damage of a composite thin-walled shaft tube according to claim 1, characterized in that: In step S1, the motion control module specifically includes: Motion controller, servo motor, servo drive, grating ruler and proximity switch; the motion controller is used to coordinately control the scanning trajectory and scanning speed of each axis according to the detection requirements; the servo drive is used to convert the control signal generated by the controller into motor power; the servo motor is used as an actuator to drive the transducer to move according to the preset motion parameters; the grating ruler is used to feedback the actual motion position of the linear guide; the proximity switch is used for motion limit and origin position correction.

4. The method for automatic detection and quantitative assessment of damage of a composite thin-walled shaft tube according to claim 1, characterized in that: In step S1, the signal acquisition module specifically includes: Ultrasonic phased array host and linear array transducer; the ultrasonic phased array host is used to control the emission and reception of ultrasonic waves; the linear array transducer is used to transmit and receive ultrasonic signals, and form a controllable sound beam through phased excitation of array units.

5. The method for automatic detection and quantitative evaluation of damage of a composite thin-walled shaft tube according to claim 1, characterized in that: In step S3, the adaptive wavelet threshold denoising method based on the progressive semi-soft threshold function specifically includes: S31, select appropriate wavelet basis function and decomposition scale, perform discrete wavelet transform on the original ultrasonic A-scan signal, decompose it into low-frequency approximate components and high-frequency detail components at different scales; select detail coefficient W from them j,k , the formula is: Where f(n) is the original signal of ultrasonic A-scan; j is the decomposition scale; k is the translation factor of the wavelet basis function; n is the discrete sampling point sequence; φ(.) is the wavelet basis function; S32, using a progressive semi-soft threshold function to perform adaptive threshold processing on the high-frequency signal in the discrete wavelet coefficients; the formula of the progressive semi-soft threshold function is expressed as: in, is the estimated value of the detail coefficient; j is the adaptive threshold for different decomposition layers, and its formula is expressed as: Among them, N j is the number of wavelet coefficients in the jth layer; σ j is the standard deviation of the noise signal in the jth layer, and its formula is expressed as: Among them, median(|W j,k |) is the wavelet coefficient W j,k median of absolute values; S33, reconstruct the ultrasonic signal after wavelet denoising, As the detail coefficients after reconstruction, the approximation coefficients remain unchanged and the time domain waveform data is restored.

6. The method for automatic detection and quantitative evaluation of damage of a composite thin-walled shaft tube according to claim 1, characterized in that: In step S4, the morphological opening and closing reconstruction specifically includes: S41, denoising and reconstructing the grayscale image I by using a disk structure element SE with a radius of 2 pixels s Perform corrosion treatment to obtain the intermediate result I after corrosion e ; S42, corroding image I e As the marker image, denoise and reconstruct the grayscale image I s As the mask image, the grayscale expansion of the marked image is iterated to obtain the reconstructed image I; S43, using the same structural element SE to reconstruct the image I o Perform dilation processing to obtain the dilated image I d ; S44, dilate the image I d As the marker image, the reconstructed image I o As the mask image, the grayscale erosion of the marked image is iterated repeatedly to finally obtain the closed reconstructed image I oc .

7. The method for automatic detection and quantitative evaluation of damage of a composite thin-walled shaft tube according to claim 1, characterized in that: In step S5, the multi-level threshold adaptive weighted median filtering is specifically as follows: The enhanced damaged image I obtained in step S4 is ss According to the maximum filter window size n max Expand the boundary and set the initial size of the filter window W to n×n n , for all pixels I ss (x, y) is traversed to obtain the maximum grayscale value I within the current filter window range max , minimum gray value I min And the gray median I mid ; If I min mid max And I min ss (x,y) max , then it is determined that there is no noise point in the current window, and the original gray value of the current pixel is output. ss (x,y); otherwise, increase the filter window W by a single step size n→n+2 n Conducting step-by-step judgment;​​​​ If the filtering window is further increased to n>n max , calculate the minimum grayscale difference I m , the formula is: I m =min(|I ss (x,y)-I min |,|I ss (x,y)-I max |); Define the noise judgment threshold \(T = k\cdot(I max -I min )\). If \(I m < T\), retain the original gray value \(I ss (x, y)\). Otherwise, judge it as a noise point and calculate according to the multi-neighborhood gray value weighted fusion expression. The calculation result is the gray value after filtering; Repeat the above steps until I am done ss Filter all pixels and remove the extended boundary area to obtain the filtered output image I sss ; The formula for weighted fusion of multi-neighborhood grayscale values is: Among them, I i (i=1,2,3,4) are the grayscale averages of the four rectangular neighborhood windows above, below, left and right, centered on the current pixel; m0, m i is the corresponding weight.

8. The method for automatic detection and quantitative evaluation of damage of a composite thin-walled shaft tube according to claim 1, characterized in that: In step S5, the multi-directional gradient enhancement Sobel operator is specifically: The output image I of the multi-level threshold adaptive weighted median filter is sss Convolution operations are performed with 3×3 Sobel gradient templates in four directions of 0°, 45°, 90° and 135° respectively, and the convolution results are projected to the horizontal and vertical directions respectively, and the horizontal component G of the comprehensive gradient is accumulated. X (x,y) and the vertical component G Y (x,y), its expression is: Among them, g d (x, y) is the pixel gradient value obtained by convolution of the ultrasound image and the templates in four different directions; Then calculate pixel I sss The gradient magnitude G(x,y) and gradient direction θ(x,y) of (x,y) are:

9. The method for automatic detection and quantitative evaluation of damage of a composite thin-walled shaft tube according to claim 1, characterized in that: In step S5, the dual threshold detection and edge connection are specifically as follows: The maximum inter-class variance algorithm is used to adaptively obtain the low threshold T for edge point detection. L , high threshold T H It is set to 1.6 times of the former, and the pixels are divided into strong edge points according to the image gradient amplitude G(x,y), that is, G(x,y)≥T H , and weak edge points, namely T L ≤G(x,y)≤T H And the validity of edge points is determined through connected domain analysis to complete edge connection.

Images