Mechanical structure damage positioning method based on near-field frequency space sparse decomposition

Abstract

The invention discloses a mechanical structure damage positioning method based on near-field frequency space sparse decomposition, and the method comprises the following steps: S1, carrying out the frame-by-frame processing and fast Fourier transform of an original acoustic emission signal, obtaining all frequency estimations of the original acoustic emission signal, and planning a sub-band according to the frequency estimations; s2, constructing an objective function of a near-field frequency space sparse decomposition algorithm based on a particle swarm optimization algorithm; s3, L1 norm-singular value decomposition is carried out on the multiple snapshot acoustic emission sub-bands, and background noise and the number of redundant snapshots are reduced; s4, constructing an array manifold matrix for each sub-band, and establishing an integral function of the over-complete dictionary array; and S5, summing the spatial spectrums of all the signal sub-bands to obtain the spatial spectrum of the acoustic emission signal so as to realize acoustic emission source positioning. According to the invention, detail features of the acoustic emission signal can be comprehensively extracted, the positioning efficiency and accuracy are significantly improved, and detailed information in the signal can be effectively extracted.

Description

Technical Field

[0001] The present invention relates to the technical field of structural damage detection, and in particular to a mechanical structure damage location method based on near-field frequency space sparse decomposition. Background Art

[0002] Mechanical structures are usually prone to damage or fracture due to factors such as cyclic loading, stress corrosion, and welding processes. More specifically, local damage to mechanical structures is accelerated by stress concentration effects, resulting in shutdown of the entire mechanical system and even casualties in actual engineering applications. In order to reduce maintenance costs and extend the remaining service life, the proposal of advanced and effective non-destructive testing (NDT) technology or structural health monitoring (SHM) methods has become a research focus for scholars and maintenance personnel. In particular, acoustic emission (AE), as the most commonly used structural health monitoring method, has been widely used to locate and identify defects (such as fatigue cracks and pitting) in simple mechanical structures.

[0003] At present, many studies on the localization and identification of damage acoustic emission (AE) sources in simple mechanical structures have been successful, mainly including methods such as fast Fourier transform (FFT), empirical mode decomposition (EMD), singular value decomposition (SVD) and empirical wavelet transform (EWT). However, these methods have some obvious shortcomings when analyzing wide-band AE signals generated by complex mechanical structures. The research on methods for locating damage AE sources in complex mechanical structures currently mainly includes methods such as deep learning, Bayesian methods, Delta T mapping and sparse decomposition (SD). In addition, sparse decomposition is a nonlinear and non-stationary analysis method that approximates the original signal by constructing a sparse representation to extract detailed information. However, although these methods can roughly extract AE features and locate the damaged AE sources in complex mechanical structures, they also have some obvious disadvantages: the existing sparse decomposition (SD) method cannot effectively extract sufficient feature information when processing the original AE signal; the algorithm runs slowly due to the need for a large number of training data sets and the complexity of the algorithm architecture, which reduces the efficiency of locating the damaged AE source; the search area is large, resulting in reduced source location accuracy; the AE signals in complex mechanical structures often have nonlinear and non-stationary characteristics, which are difficult to be effectively processed by traditional linear methods. Specifically, the traditional SD method cannot effectively extract AE features directly from the original AE signal. Summary of the invention

[0004] In order to solve the problems of insufficient acoustic emission extraction, low efficiency, insufficient accuracy and difficulty in processing nonlinear and non-stationary acoustic emission signals in the existing complex mechanical structure damage localization, the present invention proposes a mechanical structure damage localization method based on near-field frequency space sparse decomposition. By introducing the near-field frequency space sparse decomposition (NFSSD) method, the feature extraction capability, algorithm efficiency and positioning accuracy of acoustic emission source localization in complex mechanical structures are systematically improved to solve the above problems.

[0005] The present application discloses a mechanical structure damage location method based on near-field frequency space sparse decomposition, comprising the following steps:

[0006] S1, processing the original acoustic emission signal frame by frame and performing fast Fourier transform, obtaining all frequency estimates of the original acoustic emission signal, and dividing the sub-bands according to the frequency estimates;

[0007] S2, construct the objective function of the near-field frequency space sparse decomposition algorithm based on the particle swarm optimization algorithm;

[0008] S3, performing L1 norm-singular value decomposition on the acoustic emission sub-bands of multiple snapshots to reduce the number of background noise and redundant snapshots;

[0009] S4, constructing an array popularity matrix for each subband and establishing a product function of the overcomplete dictionary array;

[0010] S5. Sum the spatial spectra of all signal subbands to obtain the spatial spectrum of the acoustic emission signal, thereby realizing the location of the acoustic emission source.

[0011] Preferably, the S1 comprises the following steps:

[0012] According to the non-stationarity of the acoustic emission signal, the original acoustic emission signal is processed frame by frame to obtain a multi-quick signal, and an L-frame data set X with a length of T and a frame shift of T / 2 is obtained. l (t)(l=1,2,…,L), where each frame can be regarded as a short-term stationary signal. Then the data set X is obtained by fast Fourier transform. l (t) and divide the sub-bands according to these frequency estimates, where each frequency point corresponds to the center frequency of the sub-band. The frame of the time domain signal obtained by the array acoustic emission sensor is converted into a frequency domain model, and the expression is as follows:

[0013] X l (f) = A Nl (x,y,f)S Nl (f)+N l (f); (1)

[0014] in,

[0015]

[0016] X l (f) and S Nl (f) is the array received data set after fast Fourier transform of each frame, N l (f) is the noise vector, A Nl (x, y, f) is the direction matrix of each discrete point on the N grid of each frame data set, x is X l (f) is the row index of the matrix, y is the number of sampling frequency points, that is, the number of discrete points of the signal spectrum in the frequency domain, f is the acoustic emission signal with noise, and m is X l (f) The number of rows in the matrix, N0 is the number of subbands.

[0017] Preferably, S2 comprises the following steps:

[0018] Based on the ability of particle swarm optimization algorithm to realize global search in continuous space, the objective function of near-field frequency space sparse decomposition (NFSSD) algorithm is constructed by using particle swarm optimization algorithm, thereby improving the 1-matching performance of the best matching element. Specifically, particle swarm optimization algorithm regards individuals in the population as particles searching in multidimensional space and characterizes these particles by position, velocity and fitness function.

[0019] Let f = f s +f n , where f s is the original acoustic emission waveform, f n is the original acoustic emission noise, so the objective function can be defined as:

[0020]

[0021] Among them, g i Iterates over the elements in the dictionary D.

[0022] After the objective function is initialized, the position information and velocity of the particle will be continuously updated in the iteration process according to the fitness function. Specifically, when the number of iterations increases, the dimension of the optimal overcomplete dictionary gradually increases, which significantly increases the complexity of the algorithm. Therefore, the Hermitian inversion method is used to reconstruct the signal, and the expression is as follows:

[0023]

[0024] Among them, H k is a Hermitian matrix expressed in partitioned form, 0 k is the zero vector, ρ k is a scalar regularization term, and η kis the key component of the recursive update matrix, d yk is the error vector, the superscript T indicates transpose;

[0025] When an inverse matrix is ​​defined as When , formula (5) can be rewritten into a new form:

[0026]

[0027] in, represents the inverse matrix at k-1 steps, h k is the gradient vector, b k is the intermediate vector of the update direction.

[0028] It can be clearly seen from equation (5) that when the signal is reconstructed, the current matrix inversion can be expressed as a linear combination of the previous inversion results, thereby improving the efficiency of matrix inversion. Let the overcomplete dictionary be converted into the currently updated matrix D Δk =[D Δk- 1g yk ], where g yk is the best matching element after the current iteration update, so the matrix D Δk The update relationship is Θ Δk =[Θ Δk-1 d yk ], where Θ Δk =φψ Δk , d yk =φg yk , φ is the observation matrix, ψ Δk is the transformation matrix.

[0029] Therefore, let Constructed in a cascade manner, we finally get the following Hermitian matrix:

[0030]

[0031] According to formula (5), we get Thus, by k Substitute into equation (6) to calculate the inverse recursive formula, and then obtain the matrix used for signal reconstruction by the near-field frequency-space sparse decomposition algorithm.

[0032] Preferably, S3 comprises the following steps:

[0033] Assume that there are P given acoustic emission sources located at coordinates (x p ,y p ),p=1,2,…,P, the original acoustic emission signal is decomposed at point N0 by fast Fourier transform to obtain N l (f), where X l(f) is equivalent to the frequency snapshot after decomposition of each sub-band, represents the center frequency of the N0th sub-band;

[0034] Therefore, the P acoustic emission source direction matrices with dimensions of m×p are expressed as follows:

[0035]

[0036] Among them, A p (x,y,f j ) is a column vector of m×1 guiding quantity. By sampling the original acoustic emission signal in the frequency domain, the corresponding central frequencies f j (j=1,2,…,N0) multiple narrowband data sets;

[0037] Introducing narrowband characteristic acoustic emission signal X(f j ) to determine the location of the acoustic emission source. Based on the sparsity of the acoustic emission source, N grids are sampled in the entire near field area, and an over-complete sparse array popular basic matrix A is constructed for the sampling points in the x-direction and y-direction. N (x,y,f j ), the expression is as follows:

[0038]

[0039] Among them, α m (x N ,y N ,f j ) represents the amplitude function of the signal, x N Indicates the horizontal coordinate of the Nth spatial position, y N Represents the vertical coordinate of the Nth spatial position.

[0040] Preferably, S4 comprises the following steps:

[0041] Solve the spatial spectrum of each subband signal in the entire spatial domain and construct a convex optimization model to process A N (x,y,f j ) to achieve the joint estimation of different frequency domain snapshots at a single frequency point, the expression is as follows:

[0042]

[0043] in,‖·‖ F represents the Frobenius norm, X sv (f j ) represents the center frequency f j The spatial domain signal observation matrix, S sv (f j ) represents the center frequency f jThe sparse signal estimation coefficient matrix under , λ is the regularization parameter, Indicates the center frequency f j The regularized norm solution of .

[0044] Preferably, S5 comprises the following steps:

[0045] By combining non-sparse 2-norm with frequency domain multi-snap sampling and using sparse 1-norm for spatial sampling, the spatial spectrum of each subband signal in the entire spatial region is calculated by the following formula:

[0046] P(x,y,f j )=|S N (x,y,f j )| 2 ; (11)

[0047] All spatial spectra calculated by equation (11) are summed up and the spectrum peak is found to realize the location of the acoustic emission source, that is:

[0048]

[0049] Among them, S N (x,y,f j ) represents the position (x, y) and the center frequency f j The signal spectrum components below.

[0050] Beneficial effects of the present invention:

[0051] (1) The present invention can deeply fuse frequency information and sub-band information at multiple scales, comprehensively extract detailed features of acoustic emission signals, and solve the problem of insufficient feature extraction in traditional methods.

[0052] (2) The present invention optimizes the algorithm structure by integrating frequency information and sub-band decomposition in parallel, significantly improves the operating efficiency of the positioning algorithm, and overcomes the low efficiency defect of the existing method.

[0053] (3) The present invention can accurately construct the array manifold matrix and effectively solve the spatial spectrum, significantly reducing the search range, improving the accuracy of positioning, and solving the problem of insufficient positioning accuracy in traditional methods.

[0054] (4) The NFSSD method proposed in the invention is a nonlinear and non-stationary analysis method that can effectively extract detailed information from the signal by constructing a sparse representation, thus overcoming the limitations of traditional methods in processing nonlinear and non-stationary signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic flow chart of a mechanical structure damage location method based on near-field frequency space sparse decomposition according to an embodiment of the present invention;

[0056] Figure 2 It is the overall framework of the test platform of the embodiment of the present invention;

[0057] Figure 3 Schematic diagram of the time domain waveform and Fourier spectrum of the acoustic emission (AE) signal from sensor 1 to sensor 6 according to an embodiment of the present invention;

[0058] Figure 4 This is a schematic diagram of the decomposition result of the acoustic emission (AE) signal from sensor 1 to sensor 6 according to an embodiment of the present invention;

[0059] Figure 5 It is a schematic diagram of the acoustic emission (AE) counting process of the first laser cladding layer from sensor 1 to sensor 6 according to an embodiment of the present invention;

[0060] Figure 6 Schematic diagram of acoustic emission (AE) reconstructed time domain waveform and Fourier spectrum from sensor 1 to sensor 6 according to an embodiment of the present invention;

[0061] Figure 7 Schematic diagram of coordinate-based positioning results of the laser cladding acoustic emission (AE) source under different process parameters of an embodiment of the present invention: (a) overview, (b) first laser cladding layer, (c) second laser cladding layer, (d) third laser cladding layer, (e) comparison results;

[0062] Figure 8 Schematic diagram of the comparison results between the embodiments of the present invention and the traditional positioning method: (a) PAC AE acquisition system, (b) triangulation positioning method, (c) deep learning, (d) Bayesian method, (e) SD, (f) SD-SVD;

[0063] Fig. 9 Schematic diagram of root mean square (RMS) positioning error of different methods according to an embodiment of the present invention. DETAILED DESCRIPTION

[0064] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and examples.

[0065] Sparse decomposition (SD) is essentially equivalent to an overcomplete dictionary (OCD) with narrowband and broadband characteristics. The basic principle of SD is considered to be a solution to the signal representation problem. The representation method of the SD architecture mainly includes two stages: orthogonal matching pursuit (OMP) and basis pursuit (BP). Specifically, compared with BP, OMP is an effective weak signal feature extraction method, especially suitable for signals with strong background noise. Therefore, the embodiments of the present application mainly focus on the sparse decomposition of acoustic emission signals based on OMP.

[0066] The present application embodiment discloses a mechanical structure damage location method based on near-field frequency space sparse decomposition, the process is as follows: Figure 1 As shown, the following steps are included:

[0067] S1. Process the original acoustic emission signal frame by frame and perform fast Fourier transform (FTT) to obtain all frequency estimates of the original acoustic emission signal, and divide the sub-bands according to the frequency estimates.

[0068] According to the non-stationarity of the acoustic emission signal, the original acoustic emission signal is processed frame by frame to obtain a multi-quick signal, and an L-frame data set X with a length of T and a frame shift of T / 2 is obtained. l (t)(l=1,2,…,L), where each frame can be regarded as a short-term stationary signal. Then the data set X is obtained by fast Fourier transform. l (t) and divide the sub-bands according to these frequency estimates, where each frequency point corresponds to the center frequency of the sub-band. The frame of the time domain signal obtained by the array acoustic emission sensor is converted into a frequency domain model, and the expression is as follows:

[0069] X l (f) = A Nl (x,y,f)S Nl (f)+N l (f); (1)

[0070] in,

[0071]

[0072] X l (f) and S Nl (f) is the array received data set after fast Fourier transform of each frame, N l (f) is the noise vector, A nl (x, y, f) is the direction matrix of each discrete point on the N grid of each frame data set, x is X l (f) is the row index of the matrix, y is the number of sampling frequency points, that is, the number of discrete points of the signal spectrum in the frequency domain, f is the acoustic emission signal with noise, and m is X l (f) The number of rows in the matrix, N0 is the number of subbands.

[0073] S2. Construct the objective function of the near-field frequency space sparse decomposition algorithm based on the particle swarm optimization algorithm (PSO).

[0074] Since each orthogonal matching pursuit (OMP) iteration in traditional sparse decomposition (SD) needs to traverse the dictionary D = {g i,i=1,2,…,Q} to find the best matching element of the remaining signal, and the Gabor dictionary obtained by parameter discretization is usually selected as an over-complete dictionary (OCD). Therefore, the computational overhead is significantly increased, thereby reducing the redundancy of the Gabor dictionary. In order to solve this problem, the embodiment of the present application is based on the ability of the particle swarm optimization algorithm to realize global search in continuous space, and uses the particle swarm optimization algorithm to construct the objective function of the near-field frequency space sparse decomposition (NFSSD) algorithm, thereby improving the matching performance of the best matching element. Specifically, the particle swarm optimization algorithm regards individuals in the population as particles searching in a multidimensional space, and characterizes these particles by position, velocity, and fitness functions. The objective function construction process of the near-field frequency space sparse decomposition algorithm is as follows:

[0075] Let f = f s +f n , where f s is the original acoustic emission waveform, f n is the original acoustic emission noise, so the objective function can be defined as:

[0076]

[0077] Among them, g i Iterates over the elements in the dictionary D.

[0078] After the objective function is initialized, the position information and velocity of the particle will be continuously updated in the iteration process according to the fitness function. Specifically, when the number of iterations increases, the dimension of the optimal overcomplete dictionary gradually increases, which significantly increases the complexity of the algorithm. Therefore, the Hermitian inversion method is used to reconstruct the signal, and the expression is as follows:

[0079]

[0080] Among them, H k is a Hermitian matrix expressed in partitioned form, 0 k is the zero vector, ρ k is a scalar regularization term, and η k is the key component of the recursive update matrix, d yk is the error vector, the superscript T indicates transpose;

[0081] When an inverse matrix is ​​defined as When , formula (5) can be rewritten into a new form:

[0082]

[0083] in, represents the inverse matrix at k-1 steps, h k is the gradient vector, b k is the intermediate vector of the update direction.

[0084] It can be clearly seen from equation (5) that when the signal is reconstructed, the current matrix inversion can be expressed as a linear combination of the previous inversion results, thereby improving the efficiency of matrix inversion. Let the overcomplete dictionary be converted into the currently updated matrix D Δk =[D Δk- 1g yk ], where g yk is the best matching element after the current iteration update, so the matrix D Δk The update relationship is Θ Δk =[Θ Δk-1 d yk ], where Θ Δk =φψ Δk , d yk =φg yk , φ is the observation matrix, ψ Δk is the transformation matrix.

[0085] Therefore, let Constructed in a cascade manner, we finally get the following Hermitian matrix:

[0086]

[0087] According to formula (5), we get Thus, by k Substitute into equation (6) to calculate the inverse recursive formula, and then obtain the matrix used for signal reconstruction by the near-field frequency-space sparse decomposition algorithm proposed in this application.

[0088] S3. Perform L1 norm-singular value decomposition on the acoustic emission sub-bands of multiple snapshots to effectively reduce the number of background noise and redundant snapshots.

[0089] Assume that there are P given acoustic emission sources located at coordinates (x p ,y p ),p=1,2,…,P, the original acoustic emission signal is decomposed at point N0 by fast Fourier transform to obtain N l (f), where X l (f) is equivalent to the frequency snapshot after decomposition of each subband, where N0 is the number of subbands, Represents the center frequency of the N0th subband.

[0090] Therefore, the P acoustic emission source direction matrices with dimensions of m×P are expressed as follows:

[0091]

[0092] Among them, A p (x,y,f j ) is a column vector of m×1. The principle of the embodiment of the present application is to sample the original acoustic emission signal in the frequency domain to obtain the signals corresponding to different center frequencies f j (j=1,2,…,N0) multiple narrowband data sets.

[0093] At the same time, the narrowband characteristic acoustic emission signal X(f j ) to determine the location of the acoustic emission source. Based on the sparsity of the acoustic emission source, N grids are sampled in the entire near field area. Therefore, an overcomplete sparse array popular basic matrix A is constructed for the sampling points in the x-direction and y-direction N (x,y,f j ), the expression is as follows:

[0094]

[0095] Among them, α m (x N ,y N ,f j ) represents the amplitude function of the signal, x N Indicates the horizontal coordinate of the Nth spatial position, y N Represents the vertical coordinate of the Nth spatial position.

[0096] S4. Construct an array manifold matrix for each subband and establish a product function of the overcomplete dictionary array.

[0097] Solve the spatial spectrum of each subband signal in the entire spatial domain and construct a convex optimization model to process A N (x,y,f j ) to achieve the joint estimation of different frequency domain snapshots at a single frequency point, the expression is as follows:

[0098]

[0099] in,‖·‖ F represents the Frobenius norm, X sv (f j ) represents the center frequency f j The spatial domain signal observation matrix, S sv (f j ) represents the center frequency f j The sparse signal estimation coefficient matrix under , λ is the regularization parameter, Indicates the center frequency f j The regularized norm solution of .

[0100] S5. The spatial spectrum of the acoustic emission signal is obtained by summing the spatial spectra of all signal subbands, thereby realizing the location of the acoustic emission source. Subsequently, the non-sparse 2-norm is combined with the frequency domain multi-snap sampling, and the sparse 1-norm is used for spatial sampling. Therefore, the spatial spectrum of each subband signal in the entire spatial region is calculated by the following formula:

[0101] P(x,y,f j )=|S N (x,y,f j )| 2 ; (11)

[0102] All spatial spectra calculated by equation (11) are summed up and the spectrum peak is found to realize the location of the acoustic emission source, that is:

[0103]

[0104] Among them, S N (x,y,f j ) represents the position (x, y) and the center frequency f j The signal spectrum components below.

[0105] In a specific embodiment, the effectiveness of the method proposed in this application in locating the damage acoustic emission source is verified through the laser cladding mechanical structure damage acoustic emission experimental data set. The detailed coordinates of the acoustic emission sensor are summarized in Table 1:

[0106] Table 1 Coordinates of acoustic emission sensors (mm)

[0107]

[0108] In order to evaluate the performance of the method proposed in the embodiment of the present application, a 16-channel SH-II acoustic emission (AE) instrument was used to collect the damage acoustic emission signals generated during the laser cladding process of the mechanical structure. In the experiment, the laser cladding experiment was completed by the AFS-C1280 coaxial powder feeding laser material manufacturing equipment. To collect the acoustic emission signals, 6 PK15I piezoelectric acoustic emission sensors were used with a resonant frequency of 150 kHz, and CG-88 vaseline was used as an acoustic coupling agent to fix the sensors on the experimental mechanical structure. Figure 2The acquisition process of acoustic emission signals of laser cladding damage is demonstrated, which includes acoustic emission sensors 1-6, experimental mechanical structure (laser nozzle, sample) and computer (acquisition terminal). In addition, by setting the frequency range of the bandpass filter to 10kHz~500kHz and the threshold to 40dB according to Shannon sampling theorem, the environmental noise can be effectively eliminated. It is worth noting that the sampling frequency of the acoustic emission signal is set to 1MHz to ensure the integrity of the acoustic emission signal transmission. In addition, the experimental sample adopts a 316L stainless steel mechanical structure with a size of 400mm×400mm×8.5mm.

[0109] In order to ensure the uniformity between the medium and the substrate when the acoustic emission signal propagates in the laser cladding layer, 316L stainless steel powder similar to the experimental mechanical structure is used as the cladding material. The parameter configuration of the laser cladding experiment is shown in Table 2. In particular, in order to ensure the effectiveness and stability of the laser cladding additive manufacturing equipment and the acoustic emission acquisition system, the acoustic emission signal is first collected in advance under idling (i.e., no laser cladding powder is added). In addition, the laser cladding acoustic emission signals under the condition of changing process parameters are also collected to study the generalization ability of the proposed method. The technical parameters of the three groups of laser cladding experiments are shown in Table 3. Specifically, in this experiment, the damage acoustic emission data sets generated under different processing conditions were collected, and three groups of laser cladding acoustic emission signals were collected to obtain the laser cladding layer. Compared with the first and second groups of laser cladding layers (which showed insufficient cladding and slag inclusion characteristics), the third group of laser cladding layers are flat and plump, indicating that the third group of experimental process parameters are more suitable for actual industry.

[0110] Table 2 Parameter settings of laser cladding experiment

[0111]

[0112] Table 3 Process parameters of three groups of laser cladding experiments

[0113]

[0114] In order to verify the effectiveness of the near-field frequency space sparse decomposition (NFSSD) method proposed in this application, the data points in the first laser cladding layer are first selected to locate the damaged acoustic emission source. The acoustic emission characteristics of the original signal are extracted using NFSSD, and the frequency domain distribution is obtained. In order to conduct an intuitive analysis, taking an acoustic emission source as an example, the original acoustic emission time waveform and Fourier spectrum of six sensors are obtained by FFT analysis. The analysis results are shown in the figure. Figure 4 =3. Figure 3It can be seen that it is difficult to directly identify the characteristics of the acoustic emission signal by observing the time domain waveform and Fourier spectrum, because the acoustic emission data set of laser cladding has a certain self-similarity, which is not conducive to the identification of the signal. In addition, for the collected laser cladding acoustic emission signal, the acoustic emission events generated by all sensors must be separated first, and the acoustic emission events are usually directly extracted by setting the threshold voltage. However, due to the influence of various nonlinear factors (such as high energy, large amplitude and violent fluctuations), the laser cladding acoustic emission signals generated by mechanical metal structures are usually nonlinear and non-stationary, and their fault characteristics are easily submerged by strong background noise. Therefore, the effectiveness and accuracy of the threshold voltage selected by each sensor are difficult to determine. The NFSSD proposed in this application is an intelligent acoustic emission source positioning method that does not require prior knowledge, which can effectively extract the acoustic emission characteristics of laser cladding and reduce dependence on artificial factors.

[0115] Figure 4 The results of decomposing the acoustic emission signals of sensors 1 to 6 by the method S2 proposed in this application are shown. Figure 4 As shown, since the acoustic emission characteristic information of the mechanical metal structure is mainly concentrated in the high-frequency part, only the mode components of four different channels are extracted from each acquisition channel through the method proposed in this application. In addition, according to S3 of the method proposed in this application, Figure 5 The ring count history of the laser cladding acoustic emission signal collected by six acoustic emission sensors in the first laser cladding layer is shown. Figure 5 It can be seen that there is no significant correlation between the acoustic emission counts and the peak frequency of the acoustic emission signal, which indicates that the activity of the acoustic emission source is relatively independent of the type of acoustic emission waveform, thus revealing the necessity of extracting the frequency domain characteristics of the acoustic emission signal. Specifically, there is a certain lag between the acoustic emission signal and the laser cladding process, that is, there is a time mismatch between the acquisition of the acoustic emission signal and the laser trajectory. At the same time, the complexity and uncertainty of the laser cladding experiment make it difficult to accurately calibrate the peak frequency and depict the periodic nodes of the acoustic emission signal by analyzing only one set of signals, thereby reducing the recognition accuracy. But in general, the method proposed in this application is able to accurately record the changing trend of the acoustic emission counts of laser cladding. These results demonstrate the effectiveness of the method proposed in this application for acoustic emission feature extraction under different processing conditions. In addition, by executing S4 of the method proposed in this application, Figure 6 The reconstructed time domain waveforms and Fourier spectra of sensors 1 to 6 are shown. Figure 6 It can be seen that the frequency band of the laser cladding acoustic emission signal is mainly between 75kHz and 200kHz. Figure 7 The coordinate-based acoustic emission source localization result in (b) shows that after S5 processing of the method proposed in this application, the root mean square (RMS) localization error is 14.8 mm (see Figure 7Specifically, the positioning coordinates of the damage AE source are not completely close to the actual coordinates of the current laser cladding point, which indicates that the AE waveform may not be generated immediately after the laser stimulates the sample, but there is a certain time difference.

[0116] In addition, in order to study the influence of processing conditions on the positioning performance of the proposed method, the coordinate positioning of the laser cladding acoustic emission source collected under different process parameters was also analyzed in the experiment in this embodiment (see Figure 7 (c) and (d)). The results show that the root mean square (RMS) error of the proposed method in the positioning of the second and third layers of laser cladding acoustic emission sources is 21.3mm and 19.4mm respectively (see Figure 7 (e)). In summary, the experimental results preliminarily verify the effectiveness and accuracy of the proposed method in locating the acoustic emission source of complex mechanical structure damage under different laser cladding processing conditions.

[0117] In order to further verify the effectiveness and superiority of the method proposed in this application, this embodiment compares two traditional acoustic emission source positioning methods (i.e., PAC acoustic emission acquisition system and triangulation positioning method) and four representative methods for acoustic emission source positioning of complex mechanical structures (i.e., deep learning, Bayesian method, sparse decomposition (SD), sparse decomposition-singular value decomposition (SD-SVD)). All methods are compared and analyzed through the above experimental data, and the main details of the comparison results are summarized as follows: (1) In method 1 (PAC acquisition system), the original acoustic emission signal is directly used to calculate the time difference between sensors. Due to the influence of strong background noise, the acoustic emission source positioning accuracy is low. (2) In method 2 (triangulation positioning method), as the number of acoustic emission sensors increases, the computational complexity increases. Therefore, this method is limited in high-precision acoustic emission source positioning and has poor versatility. (3) In method 3 (deep learning), there are two disadvantages: A. When the number of layers or nodes is too small, the deep network is prone to underfitting; B. When the number of layers or nodes is too large, the training and calculation time increases, which easily leads to overfitting of the deep network. (4) In method 4 (Bayesian method), the waveform propagation speed and time difference are effectively corrected. However, the strong fluctuation of the AE energy in laser cladding leads to the expansion of the absolute error at the moment of impact, thereby reducing the accuracy of AE source positioning by this method. (5) In method 5 (SD), SD is directly applied to locate the AE source, which has two obvious shortcomings: A. When the distance between the AE source and the sensor increases, the source positioning accuracy decreases significantly; B. The computational complexity of AE source positioning is high, making it difficult to apply to situations with a large number of AE sources or grids. (6) In method 6 (SD-SVD), SVD is first applied to decompose the original AE signal to obtain the signal subspace, and then SD is applied to constrain the noise term and locate the AE source. However, for the AE source positioning calculation in the near field, the sparse manifold matrix of two parameters (direction angle and distance) needs to be jointly constructed, which leads to a large computational complexity of this method.

[0118] Figure 8 The results of locating the damage acoustic emission source using all analysis methods are shown. Among them, for the six comparison methods of (a) PAC acoustic emission acquisition system, (b) triangulation positioning method, (c) deep learning, (d) Bayesian method, (e) SD and (f) SD-SVD, the average acoustic emission source positioning RMS errors for the three laser cladding layers are 44.5mm, 52.9mm, 51.5mm, 56.6mm, 58.4mm and 41.8mm respectively. Although these six methods can roughly locate the laser cladding acoustic emission source, their RMS positioning errors are significantly higher than the method proposed in this application (18.5mm, see Figure 8 In (e), the method proposed in this application has the lowest mean absolute error compared with these methods.

[0119] More specifically, the RMS positioning error comparison results of the seven methods are shown in Fig. 9 In. From Fig. 9 It can be clearly seen that the method proposed in this application has the highest damage acoustic emission source location accuracy compared to other comparison methods. Therefore, these comparison results further prove the advantages of the proposed method in locating damage acoustic emission sources of complex mechanical structures (i.e., laser cladding mechanical structures).

[0120] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A mechanical structure damage location method based on near-field frequency space sparse decomposition, characterized in that: The following steps are involved: S1, processing the original acoustic emission signal frame by frame and performing fast Fourier transform, obtaining all frequency estimates of the original acoustic emission signal, and dividing the sub-bands according to the frequency estimates; S2, construct the objective function of the near-field frequency space sparse decomposition algorithm based on the particle swarm optimization algorithm; S3, performing L1 norm-singular value decomposition on the acoustic emission sub-bands of multiple snapshots to reduce the number of background noise and redundant snapshots; S4, constructing an array popularity matrix for each subband and establishing a product function of the overcomplete dictionary array; S5. Sum the spatial spectra of all signal subbands to obtain the spatial spectrum of the acoustic emission signal, thereby realizing the location of the acoustic emission source.

2. The mechanical structure damage location method based on near-field frequency space sparse decomposition according to claim 1 is characterized in that: The S1 comprises the following steps: The original acoustic emission signal is processed frame by frame to obtain an L-frame data set X with a length of T and a frame shift of T / 2. l (t)(l=1,2,…,L), where each frame can be regarded as a short-term stationary signal; The dataset X is obtained by fast Fourier transform approximation l (t), and dividing the sub-bands according to the frequency estimates, wherein each frequency point corresponds to the center frequency of the sub-band; The frame of the time domain signal obtained by the array acoustic emission sensor is converted into a frequency domain model, and the expression is as follows: X l (f)=A Nl (x,y,f)S Nl (f)+N l (f);(1) in, X l (f) and S Nl (f) is the array received data set after fast Fourier transform of each frame, N l (f) is the noise vector, A Nl (x, y, f) is the direction matrix of each discrete point on the N grid of each frame data set, x is X l (f) is the row index of the matrix, y is the number of sampling frequency points, that is, the number of discrete points of the signal spectrum in the frequency domain, f is the acoustic emission signal with noise, and m is X l (f) The number of rows in the matrix, N0 is the number of subbands.

3. The mechanical structure damage location method based on near-field frequency space sparse decomposition according to claim 2 is characterized in that: The S2 comprises the following steps: The particle swarm optimization algorithm regards individuals in the population as particles searching in a multidimensional space and characterizes these particles by position, velocity, and fitness functions; Let f = f s +f n , where f s is the original acoustic emission waveform, f n is the original acoustic emission noise, and the objective function is defined as: Among them, g i Indicates traversing the elements in the dictionary D; The Hermitian inversion method is used to reconstruct the signal, and the expression is as follows: Among them, H k is a Hermitian matrix expressed in partitioned form, 0 k is the zero vector, ρ k is a scalar regularization term, and η k is the key component of the recursive update matrix, d yk is the error vector, the superscript T indicates transpose; When an inverse matrix is ​​defined as When , equation (5) is rewritten as: in, represents the inverse matrix at k-1 steps, h k is the gradient vector, b k is the intermediate vector of the update direction; Let the overcomplete dictionary be converted to the currently updated matrix D Δk =[D Δk-1 g yk ], where g yk is the best matching element after the current iteration update, so the matrix D Δk The update relationship is Θ Δk =[Θ Δk-1 d yk ], where Θ Δk =φψ Δk , d yk =φg yk , φ is the observation matrix, ψ Δk is the transformation matrix; make Constructed in a cascade manner, we finally get the following Hermitian matrix: According to formula (5), we get Thus, by k Substitute into equation (6) to calculate the inverse recursive formula, and then obtain the matrix used for signal reconstruction by the near-field frequency-space sparse decomposition algorithm.

4. The mechanical structure damage location method based on near-field frequency space sparse decomposition according to claim 3 is characterized in that: The S3 comprises the following steps: Assume that there are P given acoustic emission sources located at coordinates (x p ,y p ),p=1,2,…,P, the original acoustic emission signal is decomposed at point N0 by fast Fourier transform to obtain N l (f), where X l (f) is equivalent to the frequency snapshot after decomposition of each sub-band, represents the center frequency of the N0th sub-band; Therefore, the P acoustic emission source direction matrices with dimensions of m×P are expressed as follows: Among them, A p (x,y,f j ) is a column vector of m×1 guiding quantity. By sampling the original acoustic emission signal in the frequency domain, the corresponding central frequencies f j (j=1,2,…,N0) multiple narrowband data sets; Introducing narrowband characteristic acoustic emission signal X(f j ) to determine the location of the acoustic emission source. Based on the sparsity of the acoustic emission source, N grids are sampled in the entire near field area, and an over-complete sparse array popular basic matrix A is constructed for the sampling points in the x-direction and y-direction. N (x,y,f j ), the expression is as follows: Among them, α m (x N ,y N ,f j ) represents the amplitude function of the signal, x N Indicates the horizontal coordinate of the Nth spatial position, y N Represents the vertical coordinate of the Nth spatial position.

5. The mechanical structure damage location method based on near-field frequency space sparse decomposition according to claim 4 is characterized in that: The S4 comprises the following steps: Solve the spatial spectrum of each subband signal in the entire spatial domain and construct a convex optimization model to process A N (x,y,f j ) to achieve the joint estimation of different frequency domain snapshots at a single frequency point, the expression is as follows: in,‖·‖ F represents the Frobenius norm, X sv (f j ) represents the center frequency f j The spatial domain signal observation matrix, S sv (f j ) represents the center frequency f j The sparse signal estimation coefficient matrix under , λ is the regularization parameter, Indicates the center frequency f j The regularized norm solution of .

6. The mechanical structure damage location method based on near-field frequency space sparse decomposition according to claim 5 is characterized in that: The S5 comprises the following steps: By combining non-sparse 2-norm with frequency domain multi-snap sampling and using sparse 1-norm for spatial sampling, the spatial spectrum of each subband signal in the entire spatial region is calculated by the following formula: P(x,y,f j )=|S N (x,y,f j )| 2 ;(11) All spatial spectra calculated by equation (11) are summed up and the spectrum peak is found to realize the location of the acoustic emission source, that is: Among them, S N (x,y,f j ) represents the position (x, y) and the center frequency f j The signal spectrum components below.

Images