Non-reference damage detection method for metal thin-wall structure based on synchronous guided wave excitation

Abstract

The invention relates to the technical field of nondestructive testing, and discloses a metal thin-wall structure non-reference damage detection method based on synchronous guided wave excitation, which comprises the following steps: S1, building a metal thin-wall structure damage detection platform to obtain a time domain receiving signal; s2, carrying out sparse sampling on the time domain receiving signal; s3, constructing a dictionary matrix by using the frequency dispersion curve; s4, constructing a compressed sensing equation; s5, obtaining a reconstruction signal, and establishing a matching image; and S6, obtaining a damage point according to the grid value. Excitation signals are sent out in a synchronous excitation mode, only one measuring point is used for receiving the signals, damage positioning is carried out according to the compressed sensing principle, the limitation that a baseline signal is used as a contrast is avoided, and the problem that the baseline signal is difficult to obtain in an actual application scene is solved.

Description

Technical Field

[0001] The invention relates to the technical field of nondestructive testing, and in particular to a reference-free damage detection method for a metal thin-wall structure based on synchronous guided wave excitation. Background Art

[0002] With the rapid development of industries such as aerospace, automobile, and energy, practical engineering applications have continuously put forward new requirements for the stiffness and strength of materials. However, due to the complex process of thin-wall manufacturing and the harsh service environment, thin-walled composite structures are prone to damage such as cracks, delamination, and debonding under impact and alternating loads, which will significantly reduce the service life of the equipment. Therefore, in order to ensure the safe operation of equipment and prevent serious accidents, it is very important to locate and evaluate the damage of thin-walled structures.

[0003] Nondestructive testing technology is an effective means to ensure the safe operation of equipment. Technological progress has spawned new detection schemes, and multi-sensor array imaging technology has gradually attracted the attention of scholars at home and abroad. However, there are many disadvantages and limitations in using multi-sensor arrays to locate damage. For example, the dense deployment of multiple sensors may cause superposition or interference between signals, reducing signal quality; multiple pairs of sensor arrays generate a large amount of data, and the signal processing process is complicated; it is difficult to ensure the delay synchronization of the signals received by each sensor in the time domain, which is not convenient for signal fusion and processing, etc. These limitations affect the accuracy of damage location. In order to solve the above-mentioned limiting factors, the present invention proposes a method for establishing sparse reconstruction of guided wave fields based on compressed sensing, which can achieve the goal of accurately reconstructing the complete wave field with a small number of measurement point signals. The principle of compressed sensing is to use the sparsity of the signal to reconstruct the complete signal with a small amount of randomly sampled data, accurately and effectively reconstruct the sparse signal from highly undersampled measurements, greatly reduce the number of sampling points, and achieve efficient data measurement. In this context, the present invention proposes a benchmark-free damage detection method for thin-walled structures based on synchronous guided wave excitation. The damage is located by sparsely reconstructing the guided wave signal under synchronous excitation. The data acquisition process only requires one signal excitation and a single-point receiving signal, and there is no need to measure the baseline signal. The compressed sensing principle is used to reconstruct the guided wave signal to achieve damage location. Summary of the invention

[0004] The purpose of the present invention is to realize the rapid detection of thin-walled structure damage by using only the guided wave signal of a single measuring point and the compressed sensing algorithm. The present invention provides a metal thin-walled structure non-reference damage detection method based on synchronous guided wave excitation, which can accurately locate the position of the damage in the metal thin-walled structure and greatly improve the efficiency of the signal measurement process. Specifically, the method comprises the following steps:

[0005] S1, build a metal thin-wall structure damage detection platform to obtain the time domain receiving signal x(t);

[0006] Paste L + 1 PZTs on the thin - walled metal structure, where one PZT is the measurement point and the other L PZTs are excitation points, and L is a positive integer greater than 1; the signal generator generates the original damage - detection signal, which is amplified and input to the excitation points as the excitation signal to generate guided waves in the thin - walled structure, and the oscilloscope collects the signal through the measurement point to obtain the time - domain received signal x(t);

[0007] S2. Perform sparse sampling on the time - domain received signal x(t);

[0008] Perform Fourier transform on the time - domain received signal x(t) to solve the corresponding frequency - domain received signal X(ω), and perform random sampling to obtain the sparse measurement signal Y(ω);

[0009] S3. Construct a dictionary matrix using the dispersion curve;

[0010] Construct a dictionary matrix using the dispersion curve. The dictionary matrix A is:

[0011]

[0012] where M and N represent the number of rows and columns respectively, and M < N. The element in the m - th row and n - th column of the dictionary matrix A is a mn , a mn = e jr(n)k(m) , j is the imaginary unit, r(n) represents the n - th element in the guided - wave propagation distance matrix, and k(m) represents the m - th wave number of the A0 mode of the guided - wave signal;

[0013] S4. Construct a compressive - sensing equation;

[0014] S5. Obtain the reconstructed signal Establish a matching image;

[0015] According to the compressive - sensing equation, use the convex - optimization linear - programming problem to obtain the reconstructed signal

[0016] Match the elements at the same positions in the reconstructed signal and the guided - wave propagation distance matrix r. Take r as the abscissa, as the ordinate to obtain the matching image of the guided - wave propagation distance and the reconstructed signal;

[0017] S6. Obtain the damage point according to the grid value;

[0018] In the XY coordinate system established on the plane where the thin - walled metal structure is located, after meshing the thin - walled metal structure, calculate the sum of the distances S i (x,y) from the center point of each grid to each excitation point and the measurement point. In the matching image, search for the corresponding i (x,y) as the abscissa The vertical coordinate value is the S of each grid center point. i (x, y) to get the grid value P(x, y) of each grid:

[0019] According to the grid value, two grids with the largest grid value are obtained, one of which is the measuring point and the other is the damage point. Since the position of the measuring point is known, the position of the damage point is obtained.

[0020] Preferably, the metal thin-walled structure in S1 is an aluminum plate.

[0021] Preferably, the number of excitation points L in S1 is 5.

[0022] Preferably, in S2, sparse sampling is performed on the time domain received signal x(t); specifically:

[0023] (1) Perform Fourier transform on the time domain received signal x(t) to obtain the corresponding frequency domain received signal X(ω):

[0024]

[0025] Where ω is the frequency, t is the time, and i is the imaginary unit;

[0026] (2) Randomly sample the frequency domain received signal X(ω) to obtain the sparse measurement signal Y(ω):

[0027] Y(ω)={Y(1), Y(2), …, Y(k), …, Y(M)} (2)

[0028] Y(k)=X(ω k ) (3)

[0029] Where Y(k) is the frequency point ω of the received signal X(ω) in the frequency domain k The signal obtained by random sampling at k is the kth sampling frequency point, k = 1, 2, …, M; M is the total number of samples.

[0030] Preferably, in S3, the waveguide propagation distance matrix r is a 1*N matrix, the element values ​​in the waveguide propagation distance matrix are evenly distributed from small to large, the interval distance of each element value is 1 / N, the value starts from non-zero, and the maximum value must exceed the maximum waveguide distance from the excitation point to the damage point and then to the measurement point.

[0031] Preferably, the compressed sensing equation in S4 is:

[0032]

[0033] Where Y(ω) is the sparse measurement signal obtained by randomly sampling the frequency domain received signal X(ω), A is the dictionary matrix, To reconstruct the signal, is the transpose of the reconstructed signal.

[0034] Preferably, in S5, a convex optimization linear programming problem is used to obtain the reconstructed signal, specifically:

[0035] The linear optimization problem is described as:

[0036]

[0037] Among them, ||·||1 represents the l1 norm, ||·||2 represents the l2 norm, λ1 and λ2 are the first weight parameter and the second weight parameter for regulating the weights of the l1 norm and the l2 norm, respectively, and the reconstructed signal It is a one-dimensional matrix of 1*N.

[0038] Preferably, the S6 further includes: each grid is displayed using a different color according to the grid value, and the positions of the measuring points and the damage points are intuitively displayed according to the displayed grid image.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. The present invention adopts a synchronous excitation method to collect signals, and the signal collection process is efficient and more in line with practical engineering applications.

[0041] 2. The present invention realizes signal reconstruction based on the principle of compressed sensing, and only uses one measuring point to receive the signal for damage location. The operation efficiency is high, and the damage of thin-walled structures can be quickly located.

[0042] 3. The present invention avoids the limitation of using baseline signals as controls and gets rid of the difficulty of obtaining baseline signals in actual application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of the benchmark-free damage detection method for metal thin-walled structures based on synchronous guided wave excitation;

[0044] Figure 2 Schematic diagram of the metal thin-wall structure damage detection platform built;

[0045] Figure 3 A schematic diagram of the layout of the excitation points and the measuring points in the embodiment;

[0046] Figure 4 It is the time domain signal waveform of the signal received at the measuring point;

[0047] Figure 5 It is a schematic diagram of the guided wave propagation model;

[0048] Figure 6 A schematic diagram of a reconstructed signal in a matching image in an embodiment;

[0049] Figure 7 This is a damage location effect diagram obtained by multiplication in the embodiment;

[0050] Figure 8 This is a damage location effect diagram obtained by using the addition method in the embodiment. DETAILED DESCRIPTION

[0051] In order to better understand the technical solution of the present invention, the specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings need not be drawn to scale unless otherwise specified.

[0052] The present invention is a non-reference damage detection method for metal thin-walled structures based on synchronous guided wave excitation, such as Figure 1 As shown, the specific steps are as follows:

[0053] S1, build a thin-walled structure damage detection platform and obtain the time domain received signal x(t).

[0054] like Figure 2 As shown, a metal thin-walled structure damage detection platform is built, which is specifically implemented as follows: L+1 lead zirconate titanate piezoelectric ceramic sheets (PZT) are pasted on the metal thin-walled structure, one PZT is a measuring point, and the other L PZTs are excitation points, where L is a positive integer greater than 1; a signal generator generates an original damage detection signal, which is then amplified by a power amplifier, and the amplified signal is input to the excitation point as an excitation signal to generate a guided wave in the thin-walled structure, and an oscilloscope collects signals through the measuring points to obtain a time domain received signal x(t).

[0055] In this embodiment, the metal thin-walled structure uses an aluminum plate with a thickness of 1 mm and a size of 200 mm*200 mm as an example. Five excitation points are set on the surface of the aluminum plate to be measured. These five excitation points input excitation signals at the same time. The specific layout of the measurement points and signal excitation points is as follows: Figure 3 As shown, the XY coordinate system is established on the plane where the aluminum plate is located, and the origin is usually located at the lower left corner of the aluminum plate. The lead zirconate titanate piezoelectric ceramic piece at the measuring point of the aluminum plate is connected to an oscilloscope, and the guided wave generated by the excitation signal is collected by the oscilloscope to obtain the time domain receiving signal x(t). The waveform of the time domain receiving signal in this embodiment is as follows Figure 4 As shown in Figure 2, some of the guided waves generated by the excitation signal in the metal thin-walled structure pass through the damage point, while others do not. Figure 5As shown, the time-domain received signal collected by the measurement point is the superposition of all guided waves that have not passed through the damaged point and the guided waves that have passed through the damaged point.

[0056] S2, perform sparse sampling on the time-domain received signal x(t).

[0057] One of the prerequisites for using the principle of compressive sensing is that the original signal must be sparse, that is, the signal has fewer non-zero elements. However, in actual situations, the time-domain received signal x(t) is not sparse. Therefore, sparse sampling is first performed on it to meet the prerequisite conditions of compressive sensing. The specific operation is as follows:

[0058] (1) Perform Fourier transform on the time-domain received signal x(t) to solve the corresponding frequency-domain received signal X(ω):

[0059]

[0060] where ω is the frequency, t is the time, and i is the imaginary unit.

[0061] (2) Perform random sampling on the frequency-domain received signal X(ω) to obtain the sparse measurement signal Y(ω):

[0062] Y(ω) = {Y(1), Y(2), …, Y(k), …, Y(M)} (2)

[0063] Y(k) = X(ω k ) (3)

[0064] where Y(k) is the signal obtained by randomly sampling at the frequency point ω k of the frequency-domain received signal X(ω), ω k is the k-th sampling frequency point, k = 1, 2, …, M; M is the total number of samplings.

[0065] S3, construct a dictionary matrix using the dispersion curve.

[0066] Construct a dictionary matrix using the dispersion curve. The dictionary matrix A is:

[0067]

[0068] where M and N represent the number of rows and columns respectively, and M < N. The element in the m-th row and n-th column of the dictionary matrix A is a mn , a mn = e jr(n)k(m) , j is the imaginary unit, r(n) represents the n-th element in the guided wave propagation distance matrix, and k(m) represents the m-th wave number of the A0 mode of the guided wave signal.

[0069] In this embodiment, M = 400 and N = 1000 are set. Since the size of the aluminum plate is 200mm * 200mm, it can be known that the maximum guided wave distance from the excitation point through the damage point to the measurement point will not exceed 600mm. Therefore, the guided wave distance value is set between 0 and 600mm. In the guided wave propagation distance matrix r which is a 1*N matrix, the interval distance of each element value is 1 / N. Therefore, in this embodiment, the guided wave propagation distance matrix r is a one-dimensional matrix of 1*1000, and the interval distance of each element value is set to 600 / 1000. Therefore, r = [0.6, 1.2, …, 600]. According to the above description, the elements in the guided wave propagation distance matrix r are evenly distributed from small to large, starting from a non-zero value, and the maximum value must exceed the maximum guided wave distance from the excitation point through the damage point to the measurement point. The guided wave signal A0 mode is a one-dimensional sequence obtained according to the material and thickness of the metal thin-walled structure. When M = 400, the first 400 values are taken to form a one-dimensional matrix of 1*400.

[0070] S4. Construct a compressive sensing equation.

[0071] The compressive sensing equation is:

[0072]

[0073] where Y(ω) is the sparse measurement signal obtained by randomly sampling the frequency-domain received signal X(ω), A is the dictionary matrix, is the reconstructed signal, is the transpose of the reconstructed signal.

[0074] S5. Obtain the reconstructed signal

[0075] Since M < N, the above equation is an underdetermined equation and cannot be directly solved. Therefore, the problem is usually transformed into a convex optimization linear programming problem to obtain the reconstructed signal. This linear optimization problem is described as:

[0076]

[0077] where ||·||1 represents the l1 norm, ||·||2 represents the l2 norm, λ1 and λ2 are the first weight parameter and the second weight parameter that respectively regulate the weights of the l1 norm and the l2 norm. The two weight parameters will affect the computational complexity and the sparsity of the result, and can be adjusted according to the actual signal. In the signal processing process of this article, λ1 = 100 and λ2 = 1 are taken for parameter adjustment.

[0078] Solve formula (5) to obtain the reconstructed signal which is a one-dimensional matrix of 1*N.

[0079] Substitute Match the element at the same position in r, using r as the horizontal coordinate. As the ordinate, the matching image of the waveguide propagation distance and the reconstructed signal is plotted. In this embodiment, the matching image of the waveguide propagation distance and the reconstructed signal is shown in FIG. Figure 6 As shown, the blue curve is the reconstructed signal Reconstructing the signal The value of the element in is called the sparsity value. In the matching image, the horizontal axis with a higher value represents the distance between the excitation point and the measured point obtained by the solution (l0), or the distance between the excitation point and the measured point after passing through the damage point (l1+l2).

[0080] To prove the correctness of the above theory, the aluminum plate is used as the test material in this embodiment. Figure 4 As shown in the figure, we can know not only the distance between the excitation point and the measuring point (l0), but also the distance between the excitation point and the measuring point after passing the damage point (l1+l2). Figure 6 The horizontal coordinates in the matching image are marked with purple circles, and it can be found that they are indeed Figure 6 The horizontal coordinates of the higher vertical coordinates in the matching image coincide with each other ( Figure 6 There are only 9 purple circles in the figure because two distances overlap), which shows that the theory used in this method is correct.

[0081] S6, get the damage point according to the grid value.

[0082] In the XY coordinate system established on the plane where the metal thin-walled structure is located, after the metal thin-walled structure is meshed, the sum of the distances S from each mesh center point to each excitation point and measurement point is calculated. i (x,y), where S i (x, y) represents the sum of the distances from the grid center point (x, y) to the i-th stimulus point and the measured point. i (x,y) is the horizontal coordinate to search for the corresponding The vertical coordinate value. Set the S of each grid center point i (x, y) to get the grid value P(x, y) of each grid:

[0083]

[0084] Where L is the number of excitation points, S i (x,y) represents the sum of the distances from the grid center point (x,y) to the i-th excitation point and the measurement point.

[0085] According to the grid value, the two grids with the largest grid value are directly obtained, one of which is the measuring point and the other is the damage point. Since the position of the measuring point is known, the position of the damage point is obtained. In order to display the position of the damage point more intuitively, different colors are usually used according to the grid value to intuitively display the measuring point and the damage point.

[0086] In this embodiment, the size of the aluminum plate is 200mm*200mm. To ensure that the aluminum plates are all within the grid, a grid of 250mm*250mm is established in the XY coordinate system. The size of each grid is 1mm*1mm, so a 250*250 grid is established. The size of each grid is determined according to the actual situation and determines the subsequent accuracy. The smaller the grid, the greater the accuracy, but the amount of calculation also increases accordingly. Calculate the sum of the distances from the center point of each grid to each excitation point and measurement point. In this embodiment, there are 5 excitation points and 1 measurement point, so each grid can get 5 values. Use these 5 values ​​as the horizontal coordinates to search for the corresponding The vertical coordinate value is obtained according to formula (6) to obtain the grid value P(x, y) of each grid. The grid is displayed in different colors according to the grid value, and the following is obtained: Figure 7 The final damage location shown in the figure shows that according to the XY coordinates, the coordinates of the highlighted area are (50mm, 50mm) and (150mm, 150mm). Since the coordinates of the measuring point are (50mm, 50mm), the coordinates of the damage point are (150mm, 150mm). Figure 2 The coordinates of the measuring points and the damage correspond to each other, which proves the effectiveness of this method.

[0087] It should be pointed out here that theoretically, if the S of each grid is used i The addition of (x, y) can also obtain the measurement points and damage points. The resulting grid diagram will be displayed as multiple ellipses centered on the measurement points and excitation points, such as Figure 8 As shown in the figure, the coordinates of the superposition of multiple ellipses are the damage locations, but according to Figure 8 It can be seen that the measuring points and the damaged points are not obvious when the addition method is adopted, so the present invention finally adopts the multiplication method to amplify the difference between the damaged position and the grid values ​​of other positions.

[0088] The embodiments described above are only descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for non-reference damage detection of metal thin-walled structures based on synchronous guided wave excitation, characterized in that: It includes the following steps: S1, build a metal thin-wall structure damage detection platform to obtain the time domain receiving signal x(t); L+1 PZTs are pasted on a metal thin-walled structure, one of which is a measuring point, and the other L PZTs are excitation points, where L is a positive integer greater than 1; a signal generator generates an original damage detection signal, which is amplified and input to the excitation point as an excitation signal to generate a guided wave in the thin-walled structure, and an oscilloscope collects signals from the measuring point to obtain a time domain receiving signal x(t); S2, sparse sampling of the time domain received signal x(t); Perform Fourier transform on the time domain received signal x(t) to obtain the corresponding frequency domain received signal X(ω), and perform random sampling to obtain the sparse measurement signal Y(ω); S3, constructing a dictionary matrix using dispersion curves; The dictionary matrix is ​​constructed using the dispersion curve. The dictionary matrix A is: where M and N represent the number of rows and columns respectively, and M < N. The element in the m-th row and n-th column of the dictionary matrix A is a mn , a mn = e jr (n)k(m) , j is the imaginary unit, r(n) represents the n-th element in the guided wave propagation distance matrix, and k(m) represents the m-th wave number of the A0 mode of the guided wave signal; S4, construct the compressed sensing equation; S5, get the reconstructed signal Create matching images; According to the compressed sensing equation, a convex optimization linear programming problem is used to obtain the reconstructed signal. The reconstructed signal Match the elements at the same position in the waveguide propagation distance matrix r, and use r as the horizontal coordinate. As the ordinate, the matching image of the waveguide propagation distance and the reconstructed signal is obtained; S6, get the damage point according to the grid value; In the XY coordinate system established on the plane where the metal thin-walled structure is located, after the metal thin-walled structure is meshed, the sum of the distances S from each mesh center point to each excitation point and measurement point is calculated. i (x,y), in the matching image S i (x,y) is the horizontal coordinate to search for the corresponding The vertical coordinate value is the S of each grid center point. i (x, y) to get the grid value P(x, y) of each grid: According to the grid value, two grids with the largest grid value are obtained, one of which is the measuring point and the other is the damage point. Since the position of the measuring point is known, the position of the damage point is obtained.

2. The method for non-reference damage detection of metal thin-walled structures based on synchronous guided wave excitation according to claim 1 is characterized in that: The metal thin-walled structure in S1 is an aluminum plate.

3. The method for non-reference damage detection of metal thin-walled structures based on synchronous guided wave excitation according to claim 1 is characterized in that: The number of excitation points in S1 is L=5.

4. The method for non-reference damage detection of metal thin-walled structures based on synchronous guided wave excitation according to claim 1 is characterized in that: In S2, sparse sampling is performed on the time domain received signal x(t); Specifically: (1) Perform Fourier transform on the time domain received signal x(t) to obtain the corresponding frequency domain received signal X(ω): Where ω is the frequency, t is the time, and i is the imaginary unit; (2) Randomly sample the frequency domain received signal X(ω) to obtain the sparse measurement signal Y(ω): Y(ω)={Y(1),Y(2),…,Y(k),…,Y(M)} (2) Y(k)=X(ω k ) (3) Where Y(k) is the frequency point ω of the received signal X(ω) in the frequency domain k The signal obtained by random sampling at k is the kth sampling frequency point, k = 1, 2,…, M; M is the total number of samples.

5. The method for non-reference damage detection of metal thin-walled structures based on synchronous guided wave excitation according to claim 1 is characterized in that: In S3, the waveguide propagation distance matrix r is a 1*N matrix, the element values ​​in the waveguide propagation distance matrix are uniformly distributed from small to large, the interval distance of each element value is 1 / N, the value starts from non-zero, and the maximum value must exceed the maximum waveguide distance from the excitation point to the damage point and then to the measurement point.

6. The method for non-reference damage detection of metal thin-walled structures based on synchronous guided wave excitation according to claim 1 is characterized in that: The compressed sensing equation in S4 is: Where Y(ω) is the sparse measurement signal obtained by randomly sampling the frequency domain received signal X(ω), A is the dictionary matrix, To reconstruct the signal, is the transpose of the reconstructed signal.

7. The method for non-reference damage detection of metal thin-walled structures based on synchronous guided wave excitation according to claim 1 is characterized in that: The reconstructed signal is obtained by using a convex optimization linear programming problem in S5, specifically: The linear optimization problem is described as: Among them, ||·||1 represents the l1 norm, ||·||2 represents the l2 norm, λ1 and λ2 are the first weight parameter and the second weight parameter for regulating the weights of the l1 norm and the l2 norm, respectively, and the reconstructed signal It is a one-dimensional matrix of 1*N.

8. The method for non-reference damage detection of metal thin-walled structures based on synchronous guided wave excitation according to claim 1 is characterized in that: The S6 also includes: Each grid is displayed in different colors according to the grid value, and the locations of the measuring points and damage points are intuitively displayed based on the displayed grid image.

Images