A co-sampa pipeline defect positioning method based on NRBO adaptive sparsity
The CoSaMP method with NRBO adaptive sparsity solves the problem of decomposition result distortion in noisy environments by traditional algorithms, realizes adaptive sparsity selection and efficient defect localization, and is suitable for signal processing of metal pipes.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANTONG UNIV
- Filing Date
- 2026-03-23
- Publication Date
- 2026-07-14
AI Technical Summary
Traditional matching pursuit algorithms are prone to atomic misselection in noisy environments, leading to distorted decomposition results. Furthermore, the sparsity parameter selection of compressed sampling matching pursuit algorithms is highly dependent, affecting the detection performance under complex working conditions.
The CoSaMP method based on NRBO adaptive sparsity is adopted. By preprocessing the original echo signal, peak detection and window segmentation are performed to construct a sparse representation dictionary. The Newton-Raphson optimization algorithm is used to adaptively select the sparsity parameter. Combined with the compressed sampling matching pursuit algorithm, sparse decomposition and reconstruction are performed to realize defect localization.
It reduces the subjectivity of sparsity parameter selection, improves computational efficiency, enhances detection accuracy and stability under complex working conditions, and enables rapid and accurate defect location.
Smart Images
Figure CN122385773A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of signal processing and non-destructive testing, specifically relating to a CoSaMP pipeline defect localization method based on NRBO adaptive sparsity. Background Technology
[0002] With the development of science and technology, metallic materials are widely used in industrial production due to their excellent mechanical properties and processing characteristics, playing a crucial role in equipment manufacturing, transportation, and energy. However, most metallic materials have high chemical reactivity and are prone to oxidation and corrosion when exposed to air for extended periods, leading to the generation and propagation of defects such as cracks and pits. This not only weakens the mechanical properties of the material but also poses a potential threat to the overall safety of the system. Therefore, achieving rapid detection and accurate quantitative characterization of defects has become crucial for ensuring structural integrity.
[0003] Among various detection technologies, signal processing methods based on sparse representation and reconstruction have gained widespread attention in various detection scenarios (especially ultrasonic testing) due to their strong ability to extract weak echo features and good adaptability under complex conditions. However, the traditional matching pursuit (MP) algorithm is prone to atomic misselection in noisy environments, leading to distorted decomposition results. While the compressed sampling matching pursuit (CoSaMP) algorithm, an improvement on the traditional MP algorithm, offers some stability, its performance is highly dependent on the preset sparsity parameter. The rationality of the sparsity parameter selection directly affects the decomposition and reconstruction effects, limiting the practical application of this type of algorithm under complex conditions. Therefore, researching and developing a matching pursuit algorithm that can adaptively determine sparsity has significant theoretical value and application prospects. Summary of the Invention
[0004] This invention addresses the problems of existing technologies by providing a CoSaMP pipeline defect localization method based on NRBO adaptive sparsity. While ensuring the effective preservation of key information in the echo signal, it reduces the subjectivity of parameter selection and improves computational efficiency. The specific method includes: preprocessing the acquired raw echo signal, performing peak detection on the echo signal, and adaptively performing window segmentation; constructing a dictionary for sparse representation; and employing the Newton-Raphson optimization algorithm (NRBO) to adjust the sparsity parameters. Adaptive selection is performed; based on the optimal sparsity parameter. The Compressed Sampling Matched Pursuit (CoSaMP) algorithm is used to perform sparse decomposition and reconstruction of the echo signal, and the defect location is realized based on the reconstructed signal.
[0005] To solve the above technical problems, the present invention provides the following technical solution: a CoSaMP pipeline defect localization method based on NRBO adaptive sparsity, comprising the following steps:
[0006] S1. Preprocess the acquired raw echo signal and perform peak detection on the preprocessed signal to adaptively complete window segmentation;
[0007] S2. Construct a dictionary for sparse representation;
[0008] S3. The Newton-Raphson optimization algorithm (NRBO) is used to adaptively select the sparsity parameters to obtain the optimal sparsity parameters.
[0009] S4. Based on the optimal sparsity parameters, the echo signal is sparsely decomposed and reconstructed using the CoSaMP compressed sampling matching pursuit algorithm to obtain the reconstructed signal.
[0010] S5. Locate the pipeline defect based on the reconstructed signal.
[0011] Furthermore, the preprocessing step of the original echo signal in step S1 mentioned above includes the following sub-steps:
[0012] S1-A1, Excite SH guided waves on the surface of the pipe using an electromagnetic excitation device;
[0013] S1-A2: The guided wave signal is acquired using a piezoelectric ceramic sensor to obtain the original signal;
[0014] S1-A3 extracts effective signal segments from the original signal to obtain the preprocessed signal.
[0015] Furthermore, the peak detection and window segmentation in step S1 described above include the following sub-steps:
[0016] S1-B1: Identify wave packets that may correspond to echo signals in the preprocessed signal;
[0017] S1-B2. Construct a segmentation window for signal reconstruction centered on each wave packet. The length of the window is adaptively determined according to the distance between adjacent wave packets to avoid window overlap.
[0018] Furthermore, in the aforementioned steps S1-B2, to avoid window overlap, the length of the segmented window is determined by adjacent wave packets. The distance between them is determined by wave packets. When centered, the length of the segmented window is:
[0019] ,
[0020] in, The distance between two adjacent wave packets is the number of sampling points, and .
[0021] Furthermore, the aforementioned step S2 includes the following sub-steps:
[0022] S2-1. The sinusoidal carrier wave is windowed using the Hanning window to generate wave packet basis atoms;
[0023] S2-2. Apply discrete time delay, amplitude, and phase parameters to the base atoms to generate a wave packet atom set;
[0024] S2-3. Normalize the energy of the atoms and stack them in columns to form an overcomplete dictionary matrix.
[0025] Furthermore, the aforementioned step S3 includes the following sub-steps:
[0026] S3-1. Generating candidate sparsity parameters during the Newton-Raphson optimization algorithm's NRBO iteration process. ;
[0027] S3-2, For each candidate sparsity parameter The CoSaMP algorithm is called to perform sparse reconstruction of the signal to be processed, and the corresponding reconstruction error is used as the fitness feedback input to the Newton-Raphson optimization algorithm NRBO.
[0028] S3-3, the Newton-Raphson optimization algorithm (NRBO) updates candidate sparsity parameters based on fitness feedback and iteratively searches for the optimal sparsity parameters. ;
[0029] S3-4, Using optimal sparsity parameters The reconstructed signal is obtained by performing a final sparse reconstruction.
[0030] Furthermore, step S3-1 described above specifically involves: evaluating the quality of candidate sparsity using a fitness function, wherein the fitness function is as follows:
[0031] ,
[0032] in, Display window The error signal between the original signal and the reconstructed signal. Display window The original signal inside, As a penalty factor, It is a very small positive number. This represents the cumulative sparsity value used during the reconstruction process.
[0033] Furthermore, in the aforementioned step S4, the signal reconstruction step includes: based on the constructed dictionary The optimal solution of the obtained sparse coefficient vector The reconstructed signal is obtained. .
[0034] Furthermore, the aforementioned step S5 includes the following sub-steps:
[0035] S5-1. Extract the time points of the incident wave packet, defect echo wave packet and boundary echo wave packet from the reconstructed signal;
[0036] S5-2. Obtain the distance between the electromagnetic excitation device, the sensor, and the propagation boundary;
[0037] S5-3. Calculate the defect location based on the time difference and distance information, as follows:
[0038] ,
[0039] in, The distance for defect location. The distance between the electromagnetic excitation device and the sensor. The distance between the electromagnetic excitation device and the propagation boundary. The time of the incident wave packet, The time of the defect echo packet. The time of the boundary echo packet.
[0040] The present invention also provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the CoSaMP pipeline defect localization method based on NRBO adaptive sparsity of the present invention.
[0041] Compared with the prior art, the beneficial technical effects of the present invention using the above technical solution are as follows:
[0042] 1. This invention provides an adaptive sparsity search mechanism based on NRBO. The sparsity of CoSaMP is transformed from a fixed prior into an optimizable variable. NRBO combines the Newton-Raphson search rule (NRSR) with the trap avoidance operator (TAO). NRSR uses the Newton-Raphson method to improve the search capability of NRBO, increasing convergence speed to achieve an improved search space location. TAO helps NRBO avoid local optimum traps, thus achieving a trade-off between convergence speed and escaping local optima, enabling CoSaMP to more flexibly adapt to the sparsity characteristics of various guided wave signals.
[0043] 2. This invention proposes a peak-aware windowing mechanism. It first uses the signal envelope to detect information-containing "regions of interest," then constructs the observation matrix and performs CoSaMP reconstruction only for these local windows. This significantly reduces the dimensionality of the observation matrix, substantially improves computational efficiency, and naturally achieves "gated" filtering of background noise. Attached Figure Description
[0044] Figure 1 This is a flowchart of the method steps of a preferred embodiment of the present invention;
[0045] Figure 2 This is a flowchart of peak detection and window segmentation according to a preferred embodiment of the present invention;
[0046] Figure 3 This is a flowchart illustrating a preferred embodiment of the dictionary construction process of the present invention;
[0047] Figure 4 This is a flowchart of the NRBO-CoSaMP coupling iteration process according to a preferred embodiment of the present invention;
[0048] Figure 5 This is a diagram illustrating the peak detection and window segmentation effect of a preferred embodiment of the present invention;
[0049] Figure 6 Sparsity of a preferred embodiment of the present invention The selection process and the iterative convergence graph of the corresponding fitness function;
[0050] Figure 7 This is a schematic diagram of the reconstruction result of the original signal of the pipeline defect detection experimental system in a preferred embodiment of the present invention. Detailed Implementation
[0051] To better understand the technical content of the present invention, specific embodiments are described below in conjunction with the accompanying drawings.
[0052] In this invention, various aspects of the invention are described with reference to the accompanying drawings, in which numerous illustrative embodiments are shown. Embodiments of the invention are not limited to those depicted in the drawings. It should be understood that the invention is implemented through any of the various concepts and embodiments described above, as well as the concepts and embodiments described in detail below, because the concepts and embodiments disclosed herein are not limited to any particular implementation. Furthermore, some aspects of the invention disclosed may be used alone or in any suitable combination with other aspects of the invention disclosed.
[0053] Example 1
[0054] Reference Figure 1 - Figure 7A CoSaMP pipeline defect localization method based on NRBO adaptive sparsity includes the following steps:
[0055] S1. Preprocess the acquired raw echo signal and perform peak detection on the preprocessed signal to adaptively complete window segmentation.
[0056] As a preferred method, the acquired raw echo signal is preprocessed, specifically including the following sub-steps:
[0057] S1-A1. Using an electromagnetic excitation device capable of generating pure SH waves, SH guided waves are excited on the surface of the pipe;
[0058] S1-A2, The guided wave signal is acquired using a piezoelectric ceramic sensor to obtain the original signal. ,in, This represents the total length of the original signal.
[0059] S1-A3: Extract suitable signal segments to obtain the preprocessed signal. ,in, This represents the total length of the preprocessed signal.
[0060] Preferably, peak detection and window segmentation of the preprocessed signal refers to detecting the information-containing "region of interest" using the signal envelope, and constructing the observation matrix and performing CoSaMP reconstruction only for these local windows, such as... Figure 2 As shown. Specifically, it includes the following sub-steps:
[0061] S1-B1, Select wave packets that may correspond to echo signals from the preprocessed signal. ,in, The total number of wave packets;
[0062] S1-B2, Construct a window of appropriate size centered on the wave packet, such as... Figure 5 As shown, the length of the window ,in, The distance between two adjacent wave packets is the number of sampling points, and .
[0063] S2. Construct a dictionary for sparse representation. The dictionary construction is based on the Hann window function and is determined by three dimensions: delay, amplitude, and phase. Figure 3 As shown, preferably, step S2 includes the following sub-steps:
[0064] S2-1. A Hanning window is used to window the sinusoidal carrier wave to generate wave packet basis atoms, specifically:
[0065] A sinusoidal packet based on Hanning window modulation is constructed as the fundamental kernel function of the dictionary. This structure can well simulate the beam shape of ultrasonic guided waves after reflection from defects. Its fundamental window function... for: ,in, It is the atomic length, and ,in, The period number of the atom is determined by the adaptive window length. Sampling frequency, The signal center frequency is ; the corresponding sinusoidal carrier wave is ;
[0066] S2-2. Apply discrete time delay, amplitude, and phase parameters to the base atoms to generate a wave packet atom set;
[0067] Atoms are generated by discretizing sampling in three dimensions: time delay, amplitude, and phase. Specifically, the time delay set is: Amplitude set: Phase set: The final set of atoms after processing in three dimensions is as follows: ;
[0068] S2-3. Normalize the energy of the atoms and stack them column-wise to form an overcomplete dictionary matrix. Specifically, to ensure the comparability of each atom during sparse decomposition and avoid misselection due to differences in atomic energy, each generated atom needs to be normalized. Norm normalization: ,in, The total number of atoms, and , It is a very small positive number;
[0069] Finally, the final dictionary is generated. It is a matrix composed of all the above-mentioned normalized atoms as column vectors: ;
[0070] S3. The Newton-Raphson Optimization (NRBO) algorithm is used to adaptively select the sparsity parameters to obtain the optimal sparsity parameters. NRBO and CoSaMP are compared through sparsity... To perform coupling, NRBO performs... The optimal selection is determined, and the optimal sparsity is calculated using CoSaMP. Under the conditions of performing the final signal reconstruction, such as Figure 4 As shown.
[0071] S3-1. Generate candidate sparsity parameters : ,in, and These are the lower and upper limits of sparsity, respectively. This represents the total number of candidate sparsities.
[0072] S3-2, For each candidate sparsity parameter The CoSaMP (Cooperative Sagging and Matching Pursuit) algorithm is invoked to perform sparse reconstruction of the signal to be processed, and the corresponding reconstruction error is obtained and used as fitness feedback input to the Newton-Raphson optimization algorithm (NRBO). Specifically:
[0073] For each candidate sparsity parameter The CoSaMP algorithm is called to perform sparse reconstruction on the signal to be processed, and the NRBO is applied to the sparsity. When making a selection, the fitness function is used to evaluate the sparsity of the candidates, such as... Figure 6 As shown. The fitness function is: ,in, Display window The error signal between the original signal and the reconstructed signal. Display window The original signal inside, As a penalty factor, It is a very small positive number. This represents the cumulative sparsity value used during the reconstruction process. A smaller fitness value indicates a better reconstruction effect corresponding to the candidate sparsity.
[0074] S3-3, the Newton-Raphson optimization algorithm (NRBO) updates candidate sparsity parameters based on fitness feedback and iteratively searches for the optimal sparsity parameters. Specifically, the core strategy of the NBRO algorithm is the NRSR search rule. The NRSR search strategy guides the population search process to improve the ability to explore the solution space. The calculation formula for NRSR is: ,in, For random perturbations with positive and negative values, and Let these represent the sparsity of the worst and best individuals in the current population, respectively. The perturbation is the sparsity parameter sought in this paper. Since it is a one-dimensional variable, therefore take , This represents the sparsity value in the current iteration; the NRBO algorithm uses the TAO algorithm to avoid getting trapped in local optima. The solution update rule is: ,in, This represents the final position after the sparsity update. and They represent the first The particles in the current iteration step ( The original location and the pre-update location of ) This indicates the average position of a sparse population. It is a scaling factor, and , and To randomly scale the weights, and for Random numbers between for Random numbers between and These are random control parameters, and , In the formula, for Random numbers between for A random number between [a certain number of points].
[0075] S3-4, Using the aforementioned optimal sparsity parameters The final sparse reconstruction yields the reconstructed signal, such as... Figure 7 As shown.
[0076] S4. Based on the optimal sparsity parameters, the echo signal is sparsely decomposed and reconstructed using the CoSaMP compressed sampling matching pursuit algorithm to obtain the reconstructed signal.
[0077] S5. Locate the pipeline defect based on the reconstructed signal.
[0078] Preferably, the defect can be located by extracting the time points corresponding to the incident wave packet, the defect echo wave packet, and the boundary echo wave packet, and obtaining the distance between the electromagnetic excitation device, the sensor, and the propagation boundary. The propagation distance from the sensor to the defect is then calculated based on the time points. The specific steps are as follows:
[0079] First, determine the optimal sparsity. The CoSaMP signal was reconstructed, and the distance between the electromagnetic excitation device and the sensor was obtained. The distance between the electromagnetic excitation device and the propagation boundary Time of incident wave packet Time of defect echo packet and the time of the boundary echo packet .
[0080] Then, the defect location formula can be used: To calculate the distance from the electromagnetic excitation device to the defect. This allows for location tracking.
[0081] In this embodiment, a metal pipe with a circumference of 250 mm and a thickness of 5 mm is used to verify the performance of the above positioning method. A circumferential crack (length 20 mm × width 1.5 mm × depth 5 mm) is set at the center position of the pipe as the defect to be detected. Due to the fitting form of the electromagnetic excitation device and the curved pipe and the influence of magnetic field deflection, the original signal contains multi-modal clutter components. During the implementation process, the system parameters are configured as follows: reference distance 78 mm, sampling frequency 2 MHz, signal length 200 . The atomic dictionary is constructed based on 100 time-delay samples, 15 amplitude samples, and 6 phase samples, and its matrix dimension is 400×900. The search parameters of the NRBO algorithm are set as follows: population size 15, sparsity value range 6 to 100. It has been obtained =30mm,<00003This embodiment provides a non-transitory computer-readable storage medium. This storage medium stores computer-executable instructions, which, when read and executed by one or more processors, cause the processors to implement the steps of the CoSaMP pipeline defect localization method based on NRBO adaptive sparsity described in Embodiment 1.
[0085] Specifically, the computer-readable storage medium can be any device capable of storing program code, including integrated storage units within the terminal device (such as built-in hard disks, random access memory (RAM), and read-only memory (ROM), as well as external storage devices that can be separated from the terminal (such as pluggable hard disks, USB flash drives, smart memory cards, SD cards, flash memory cards, or optical discs). Furthermore, this storage medium can not only store the program code implementing the above methods, but can also be used to temporarily cache intermediate data generated during the computation process (such as the acquired raw echo signals, constructed atomic dictionaries, or reconstructed sparse coefficients).
[0086] Example 3:
[0087] This embodiment provides an intelligent signal processing terminal or computing device, whose hardware architecture includes a memory, a processor, and a bus interface connecting the various components. The memory stores a computer program, and the processor calls and runs the program to execute the various processes of the CoSaMP pipeline defect localization method based on NRBO adaptive sparsity described in Embodiment 1.
[0088] In this embodiment, the processor is not limited to a specific type; it can be a general-purpose central processing unit (CPU), or a digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device optimized for signal processing. It can also employ discrete hardware components (such as transistor logic circuits) or combinations of the above-mentioned chips. The memory includes, but is not limited to, high-speed random access memory (SRAM / DRAM) and non-volatile memory (such as hard disks and NVRAM), whose main function is to provide program instructions and data storage space for the processor (e.g., for storing acquired pipeline ultrasonic signal data).
[0089] Those skilled in the art will understand that the technical solutions disclosed in this application can be embodied as a method, a fully hardware system, or a computer program product combining software and hardware. Therefore, this invention can be implemented entirely in hardware, entirely in software, or in a combination of software and hardware.
[0090] Furthermore, each functional module shown in the flowcharts or block diagrams in the accompanying drawings can be understood as either a logical step controlled by computer program instructions or a dedicated hardware circuit module that implements a specific function. These computer program instructions can be loaded onto the processor of a general-purpose computer, a professional signal processor, or an embedded controller to generate a specific machine, enabling the functions specified in the flowcharts or block diagrams to be implemented through hardware entities.
[0091] It should be noted that the above embodiments are merely preferred embodiments of the present invention, intended to explain and illustrate the technical concept of the present invention, and are not intended to narrowly limit the scope of patent protection of the present invention.
[0092] Any equivalent transformations, simple modifications, feature combinations, or refinements made by those skilled in the art based on the technical logic and design concepts disclosed in this invention, according to actual application scenarios (such as different pipe materials, different waveguide modes, etc.), do not depart from the spirit of this invention and should therefore be covered within the patent protection scope of this invention.
[0093] While the present invention has been described above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.
Claims
1. A CoSaMP pipeline defect localization method based on NRBO adaptive sparsity, characterized in that, Includes the following steps: S1. Preprocess the acquired raw echo signal and perform peak detection on the preprocessed signal to adaptively complete window segmentation; S2. Construct a dictionary for sparse representation. The construction of the dictionary is based on the Hanning window function and is determined by three dimensions: time delay, amplitude, and phase. S3. The Newton-Raphson optimization algorithm (NRBO) is used to adaptively select the sparsity parameters to obtain the optimal sparsity parameters. S4. Based on the optimal sparsity parameters, the echo signal is sparsely decomposed and reconstructed using the CoSaMP compressed sampling matching pursuit algorithm to obtain the reconstructed signal. S5. Locate the pipeline defect based on the reconstructed signal.
2. The CoSaMP pipeline defect localization method based on NRBO adaptive sparsity according to claim 1, characterized in that, Step S1, the preprocessing of the original echo signal, includes the following sub-steps: S1-A1, Excite SH guided waves on the surface of the pipe using an electromagnetic excitation device; S1-A2: The guided wave signal is acquired using a piezoelectric ceramic sensor to obtain the original signal; S1-A3 extracts effective signal segments from the original signal to obtain the preprocessed signal.
3. The CoSaMP pipeline defect localization method based on NRBO adaptive sparsity according to claim 1, characterized in that, Step S1, which involves peak detection and window segmentation, includes the following sub-steps: S1-B1: Identify wave packets that may correspond to echo signals in the preprocessed signal; S1-B2. Construct a segmentation window for signal reconstruction centered on each wave packet. The length of the window is adaptively determined according to the distance between adjacent wave packets to avoid window overlap.
4. The CoSaMP pipeline defect localization method based on NRBO adaptive sparsity according to claim 3, characterized in that, In steps S1-B2, to avoid window overlap, the length of the segmented window is determined by the adjacent wave packets. The distance between them is determined by wave packets. When centered, the length of the segmented window ,in, The distance between two adjacent wave packets is the number of sampling points, and .
5. The CoSaMP pipeline defect localization method based on NRBO adaptive sparsity according to claim 1, characterized in that, Step S2 includes the following sub-steps: S2-1. The sinusoidal carrier wave is windowed using the Hanning window to generate wave packet basis atoms; S2-2. Apply discrete time delay, amplitude, and phase parameters to the base atoms to generate a wave packet atom set; S2-3. Normalize the energy of the atoms and stack them in columns to form an overcomplete dictionary matrix.
6. The CoSaMP pipeline defect localization method based on NRBO adaptive sparsity according to claim 1, characterized in that, Step S3 includes the following sub-steps: S3-1. Generating candidate sparsity parameters during the Newton-Raphson optimization algorithm's NRBO iteration process. ; S3-2, For each candidate sparsity parameter The CoSaMP algorithm is called to perform sparse reconstruction of the signal to be processed, and the corresponding reconstruction error is used as the fitness feedback input to the Newton-Raphson optimization algorithm NRBO. S3-3, the Newton-Raphson optimization algorithm (NRBO) updates candidate sparsity parameters based on fitness feedback and iteratively searches for the optimal sparsity parameters. ; S3-4, Using optimal sparsity parameters The final sparse reconstruction is performed to obtain the reconstructed signal.
7. The CoSaMP pipeline defect localization method based on NRBO adaptive sparsity according to claim 1, characterized in that, Step S3-1 specifically involves: evaluating the quality of candidate sparsity using a fitness function, wherein the fitness function is as follows: ， in, Display window The error signal between the original signal and the reconstructed signal. Display window The original signal inside, As a penalty factor, It is a very small positive number. This represents the cumulative sparsity value used during the reconstruction process.
8. The CoSaMP pipeline defect localization method based on NRBO adaptive sparsity according to claim 6, characterized in that, In step S4, the signal reconstruction step includes: based on the constructed dictionary The optimal solution of the obtained sparse coefficient vector The reconstructed signal is obtained. .
9. The CoSaMP pipeline defect localization method based on NRBO adaptive sparsity according to claim 1, characterized in that, Step S5 includes the following sub-steps: S5-1. Extract the time points of the incident wave packet, defect echo wave packet and boundary echo wave packet from the reconstructed signal; S5-2. Obtain the distance between the electromagnetic excitation device, the sensor, and the propagation boundary; S5-3. Calculate the defect location based on the time difference and distance information, as follows: ， in, The distance for defect location. The distance between the electromagnetic excitation device and the sensor. The distance between the electromagnetic excitation device and the propagation boundary. The time of the incident wave packet, The time of the defect echo packet. The time of the boundary echo packet.
10. A computer-readable storage medium, characterized in that, The storage medium stores computer program instructions, which, when executed by a processor, implement the CoSaMP pipeline defect location method based on NRBO adaptive sparsity as described in any one of claims 1-9.