Defect signal identification and positioning method based on dual-channel guided wave phase characteristics

Abstract

The invention discloses a defect signal identification and positioning method based on dual-channel guided wave phase characteristics. The method comprises the following steps: arranging a dual-channel excitation probe and a dual-channel receiving probe on the surface of a to-be-tested piece, exciting guided waves through the excitation probe and propagating the guided waves in the to-be-tested piece, obtaining dual-channel echo signals through detection of the receiving probe, and obtaining single-channel envelope signals and a wave packet window thereof after conversion processing; filtering the echo signals of the dual channels in each wave packet window according to the similarity to obtain defect signals so as to realize defect identification; and reconstructing the defect signal according to the wave packet window so as to determine the defect position on the to-be-tested piece, thereby realizing defect positioning. According to the method, the problems of frequency dispersion, multiple modes, mode conversion, non-thorough unidirectional direction of the transducer and the like of the guided waves are considered, compared with a conventional time flight method, the method has more advantages, real defect signals can be rapidly screened and positioned, the difficulty of guided wave signal processing is reduced, and the method has wide practical engineering application value.

Description

Technical Field

[0001] This invention relates to a method for defect signal identification and localization, and to the field of ultrasonic guided wave detection technology, specifically to a method for defect signal identification and localization based on the phase characteristics of dual-channel guided waves. Background Technology

[0002] In industry, magnetic particle testing, eddy current testing, and ultrasonic guided wave testing are commonly used non-destructive testing methods. In particular, ultrasonic guided wave testing is widely used in industrial fields for long-distance defect detection of pipes, plates, and other components due to its advantages such as long detection distance, high detection accuracy, wide range of applications, and low cost.

[0003] In ultrasonic guided wave testing, defect detection and localization are typically determined based on the number of wave packets and peak coordinates in the echo signal. Wave packets are usually not screened, as each packet corresponds to a unique characteristic. However, in actual testing, due to issues such as poor unidirectional performance, multimodal nature, and mode transitions, false information may exist in the wave packet data, significantly reducing the accuracy of the detection results. For example, in ultrasonic guided wave testing of a plate, when exciting a low-frequency lambda wave, both A0 and S0 modes are usually excited simultaneously. These two modes propagate independently in the waveguide, generating reflected echoes when they encounter defects. If the echo wave packets are not screened, the number of detected defects will be far higher than the actual number. Similarly, when exciting a low-frequency longitudinal mode L(0,2) guided wave in a pipeline, additional bending mode guided waves are generated when this mode passes through corrosion defects, bends, or other structures, which also reduces the accuracy of the detected signal. Furthermore, due to issues such as waveguide dispersion and wave packet superposition, the method of determining the defect location based on the wave packet peak coordinates is often not accurate enough, and better methods are needed to locate the defect structure.

[0004] In summary, this research on screening and identifying echo signals from ultrasonic guided wave detection, filtering out false feature information, and achieving accurate defect localization has profound practical significance. Summary of the Invention

[0005] To address the problems existing in the background technology, the present invention provides a method for defect signal identification and localization based on the phase characteristics of dual-channel guided waves.

[0006] The technical solution adopted in this invention is: The defect signal identification and localization method based on the phase characteristics of dual-channel guided waves of the present invention includes: Step 1) Place the dual-channel excitation probe and receiving probe on the surface of the test piece. Excite the guided wave through the excitation probe and let it propagate inside the test piece. Based on the principle of unidirectional guided wave excitation and reception, the echo signal of the dual channels is obtained by the receiving probe. After transformation and processing, the envelope signal of the single channel and its several wave packet windows are obtained.

[0007] Step 2) After filtering the echo signals of the two channels within each wave packet window according to similarity, the defect signal is obtained to achieve defect identification.

[0008] Step 3) Reconstruct the defect signal according to each wave packet window to obtain the time of the maximum peak point of the defect signal, and then determine the defect location on the test piece to achieve defect localization.

[0009] In step 1), both the excitation probe and the receiving probe consist of dual-channel coils. The two coils are installed as one unit but are not connected to each other. When exciting the guided wave, both channels of the excitation probe excite the same guided wave of a preset wavelength. The distance between the first channel coil and the second channel coil of the excitation probe is delayed by one-quarter of the preset wavelength in the axial direction, and the excitation time of the first channel coil of the excitation probe is advanced by one-quarter of the guided wave period. When receiving the guided wave, the distance between the first channel coil and the second channel coil of the receiving probe is delayed by one-quarter of the preset wavelength in the axial direction, and the reception time of the first channel coil of the receiving probe is delayed by one-quarter of the guided wave period. The determination of the period and wavelength depends on the preset guided wave frequency mode.

[0010] In step 1), one single-channel echo signal from the dual-channel echo signals detected by the receiving probe is arbitrarily selected, and Hilbert transform is performed to obtain the single-channel envelope signal. The N maxima in the envelope signal are determined as wave packet vertices. For each wave packet and its wave packet vertex, the minima are calculated for the other maxima on the left and right sides of the wave packet vertex in the wave packet region to obtain the left and right boundaries. The region within the left and right boundaries is used as the wave packet window, and finally N wave packet windows are obtained.

[0011] In step 2), for each wave packet window, the echo signals of the dual channels in the wave packet window are interpolated with cosine similarity. Specifically, the echo signal of the second channel is interpolated to the echo signal of the first channel to obtain the interpolated signal. Then, the cosine similarity between the interpolated signal and the echo signal of the first channel is calculated to obtain the similarity value. The interpolated signal with a similarity value higher than the preset similarity threshold is retained as the defect signal. If there is no defect signal, the test piece is defect-free. If there are one or more defect signals, the test piece has one or more defects.

[0012] In step 3), for each defect signal, the matching pursuit algorithm is used to reconstruct the signal, determine the time of the maximum peak point of the defect signal, and determine the defect location on the test piece based on the time of the maximum peak point and the propagation speed of the guided wave.

[0013] The reconstruction interval of the matching tracking algorithm is the wave packet window where the defect signal is located, and the matching atom is the generalized Gabor atom modulated by the Hanning window.

[0014] The beneficial effects of this invention are: This invention combines the phase characteristics of dual-channel guided waves to solve the cosine similarity of wave packets in the echo signal, sets a reasonable threshold to filter out false feature information in complex echo signals, thereby improving the authenticity of guided wave detection. In addition, it is combined with a matching tracking algorithm to accurately locate the detected features, which greatly reduces the difficulty of signal analysis.

[0015] The method of this invention takes into account the problems of waveguide dispersion, multimode, mode conversion, and transducer incomplete unidirectional operation. Compared with the conventional time-of-flight method, it has more advantages, can realize the rapid screening and location of real defect signals, reduce the difficulty of waveguide signal processing, and has broad practical engineering application value. Attached Figure Description

[0016] Figure 1 A velocity dispersion curve for a gas pipeline group; Figure 2 This is a schematic diagram of the experimental apparatus according to an embodiment of the present invention; Figure 3 Diagram of the received signal from the first channel of the receiving probe; Figure 4 This is a diagram of the received signal of the first channel after Hilbert transformation; Figure 5 The image shows the calculation results of the cosine similarity of wave packets in each window of the dual-channel signal from the receiving probe. Figure 6 This is a defect signal image after being processed and filtered by the algorithm; Figure 7 The result of reconstructing the defect signal using the matching pursuit algorithm; In the diagram: 1. Excitation probe, 2. Receiving probe. Detailed Implementation

[0017] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] The defect signal identification and localization method based on the phase characteristics of dual-channel guided waves of the present invention is as follows: Step 1) Arrange the dual-channel excitation probe 1 and receiving probe 2 on the surface of the test piece. Both excitation probe 1 and receiving probe 2 consist of dual-channel coils, which are installed as one unit but not connected to each other. When exciting the guided wave, both channels of the excitation probe 1 excite the same guided wave of the preset wavelength. The distance between the first channel coil of the excitation probe 1 and the second channel coil is delayed by one-quarter of the preset wavelength in the axial direction, and the excitation time of the first channel coil of the excitation probe 1 is advanced by one-quarter of the guided wave period. When receiving the guided wave, the distance between the first channel coil of the receiving probe 2 and the second channel coil is delayed by one-quarter of the preset wavelength in the axial direction, and the reception time of the first channel coil of the receiving probe 2 is delayed by one-quarter of the guided wave period. The determination of the period and wavelength depends on the preset guided wave frequency mode. In specific implementation, such as detecting defects in a pipe, the excitation probe 1 and receiving probe 2 can be fitted onto the outer wall of the pipe. If detecting defects in a flat plate, they can be directly attached to the surface of the flat plate.

[0019] Excitation probe 1 excites guided waves and propagates them inside the test piece. Based on the principle of unidirectional guided wave excitation and reception, receiving probe 2 detects and obtains dual-channel echo signals. One single-channel echo signal from the dual-channel echo signals detected by receiving probe 2 is arbitrarily selected and subjected to Hilbert transform to obtain a single-channel envelope signal. N maxima in the envelope signal are determined as wave packet vertices. For each wave packet and its vertices, the minima are calculated for the other maxima on the left and right sides of the wave packet vertices within the wave packet region to obtain the left and right boundaries. The region within the left and right boundaries is used as the wave packet window, ultimately obtaining N wave packet windows.

[0020] Step 2) After filtering the echo signals of the two channels within each wave packet window based on similarity, the defect signal is obtained. For each wave packet window, the echo signals of the two channels within the wave packet window are interpolated and cosine similarity is calculated. Specifically, the echo signal of the second channel is interpolated to the echo signal of the first channel to obtain the interpolated signal. Then, the cosine similarity is calculated between the interpolated signal and the echo signal of the first channel to obtain the similarity value. Alternatively, the echo signal of the first channel can be interpolated to the echo signal of the second channel to obtain the interpolated signal. Then, the cosine similarity is calculated between the interpolated signal and the echo signal of the second channel to obtain the similarity value. The interpolated signal with a similarity value higher than the preset similarity threshold is retained as the defect signal. If there is no defect signal, the test piece is defect-free. If there are one or more defect signals, the test piece has one or more defects, thus achieving defect identification.

[0021] Step 3) Reconstruct the defect signal according to each wave packet window. For each defect signal, use the matching pursuit algorithm to reconstruct it. The reconstruction interval of the matching pursuit algorithm is the wave packet window where the defect signal is located. The matching atom is the generalized Gabor atom modulated by the Hanning window. Determine the time of the maximum peak point of the defect signal. Determine the defect location on the test piece according to the time of the maximum peak point and the propagation speed of the guided wave, so as to achieve defect localization.

[0022] Specific embodiments of the present invention are as follows: Taking a gas pipeline with an outer diameter of 40mm, a wall thickness of 3mm, and made of Q235 steel as an example, due to its small diameter and the complexity of the working environment, it is difficult to generate a uniform circumferential magnetic field in the gas pipeline. Therefore, the longitudinal mode is usually used as the guided wave mode for defect detection in the gas pipeline. The group velocity dispersion curve of the longitudinal mode is plotted as follows: Figure 1 As shown in the figure, in order to excite the L(0,2) mode with weak dispersion characteristics in the gas pipe, the excitation frequency should be at least greater than 90kHz. At this time, the non-ideal L(0,1) mode will be excited in the gas pipe at the same time.

[0023] Setting up experimental setups, such as Figure 2 As shown, the gas pipe is 1.5m long. An excitation probe 1 is placed at one end of the pipe. A rectangular annular groove, 1.5mm deep and 3mm wide, is located 0.75m from the excitation probe 1. A receiving probe 2 is installed 0.15m from the right end of the excitation probe 1, exciting a 100kHz L(0,2) mode guided wave towards the defect side. The excitation signal is a five-cycle sinusoidal signal modulated by a Hanning window. The echo signal diagram of channel one in receiving probe 2 is shown in the figure. Figure 3 As shown in the diagram, the first wave packet in the echo signal diagram is the direct wave of L(0,2), the second wave packet is the direct wave of L(0,1), the third and fourth wave packets are the reflected echo of the L(0,2) mode after passing through the rectangular annular groove defect and the converted bending mode echo, and the fifth wave packet is the end face reflected echo after the L(0,2) mode reaches the end face. It is difficult to distinguish the above information from the echo signal alone. The interference of redundant wave packets makes the identification of pipe defects and end faces difficult. Therefore, a defect signal identification method based on the phase characteristics of dual-channel guided waves can be used to process the echo signal.

[0024] The channel 1 and channel 2 of the receiving probe 2 are separated by a quarter wavelength. The wavelength depends on the guided wave mode at the selected frequency. In this embodiment, the wavelength corresponds to the 100kHz L(0,2) mode guided wave wavelength, which is 53mm. Therefore, if the channel 2 signal is advanced by a quarter period relative to the channel 1 when receiving the signal, the dual-channel echo signals of the positive L(0,2) mode will be almost completely aligned in phase. The positive direction represents the right side of the receiving probe 2, and the negative direction represents the left side of the receiving probe 2. Other guided wave modes (including non-ideal modes during excitation and conversion modes caused by abnormal structures) and the reverse echo cannot be aligned in phase because of the mismatch between wavelength and period. This will form dual-channel superimposed signals with various phase differences. By calculating the cosine similarity of the dual-channel window signals, the positive L(0,2) mode guided wave that reflects the pipeline characteristic information can be directly screened out.

[0025] First, perform a Hilbert transform on the channel 1 signal of receiving probe 2, and obtain its envelope signal, such as... Figure 4 As shown, the maximum value in the envelope signal is found, the wave packet vertex is located, and then the minimum values ​​are calculated at the extreme points on its left and right sides to obtain the left and right boundaries of the wave packet window. After finding all windows, the cosine similarity is calculated for the received signals of channel 1 and channel 2 within the window range, as follows: When calculating cosine similarity within a specified time window, directly extracting signal amplitude using indexing may not be accurate enough. An interpolation method is needed to interpolate the channel 2 signal onto the time vector of the channel 1 signal before performing the similarity calculation. The calculated result is as follows: Figure 5 As shown, the cosine similarities obtained for windows 1-5 are -0.9331, 0.9117, 0.9998, -0.776, and 0.992, respectively. The closer the calculated result is to 1, the closer the phases of the dual-channel signals within the window are, and the higher the probability that it is a positive L(0,2) mode reflection echo. The reason for the high similarity of window 2 is that this wave packet is a reverse reflection echo and the wavelength of the L(0,1) mode is exactly close to half of the L(0,2) mode.

[0026] A similarity threshold of 0.98 is set. Window signals below this threshold are eliminated, while window signals above the threshold are retained after superposition and interpolation, resulting in a new echo signal. Figure 6 As shown, the wave packets in this echo signal diagram are reflected echoes from the forward propagation of the L(0,2) mode guided wave, eliminating interference from reverse echoes and non-ideal modes. Wave packet 1 represents the reflected echo of the L(0,2) mode after passing through the rectangular annular groove defect, and wave packet 2 represents the end-face reflected echo of the L(0,2) mode after reaching the end face. Each wave packet contains characteristic structural information of the pipeline.

[0027] After identifying the defect features, precise location is required. In engineering applications, this is typically achieved based on the peak points of wave packets. However, in reality, wave packet signals may exhibit deviations in key information due to issues such as waveguide dispersion, unstable signal reception, and overlapping wave packets, affecting defect feature location and subsequent in-depth research on wave packet information. This invention, based on the aforementioned wave packet window division, uses a matching pursuit algorithm to reconstruct the signal from the remaining window signal after defect identification. A predefined set of atoms best representing the local characteristics of the guided wave—generalized Gabor atoms modulated by the Hanning window—is used. Through a greedy iterative process, the strongest energy components are gradually matched, extracted, and separated from the original signal, ultimately sparsely reconstructing the original signal using linear combinations of atoms. In this embodiment, there are no overlapping wave packets, therefore the number of atoms is 1 and residual calculation is unnecessary. The final reconstructed signal is as follows: Figure 7 As shown, the signal reconstructed by the matched pursuit algorithm has higher time-frequency resolution and clarity, and can more accurately depict the arrival time of the guided wave modes, the center frequency variation, and the distribution of signal energy in the time-frequency plane, while also having a great ability to suppress noise.

[0028] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this invention.

Claims

1. A method for defect signal identification and location based on dual-channel guided-wave phase characteristics, characterized in that, The method comprises the following steps: Step 1) arranging a double-channel excitation probe (1) and a receiving probe (2) on the surface of a test piece, exciting a guided wave through the excitation probe (1) and propagating in the test piece, detecting a double-channel echo signal through the receiving probe (2), and obtaining a single-channel envelope signal and a plurality of wave packet windows after transformation and processing; Step 2) filtering the double-channel echo signal in each wave packet window according to similarity to obtain a defect signal, so as to realize defect identification; Step 3) reconstructing the defect signal according to each wave packet window to obtain the time of the maximum peak point of the defect signal, and determining the defect position on the test piece, so as to realize defect positioning.

2. The method according to claim 1, wherein the method is characterized in that: In the step 1), the excitation probe (1) and the receiving probe (2) are both composed of double-channel coils, when exciting the guided wave, the double-channel coils of the excitation probe (1) all excite the same guided wave of a preset wavelength, the distance between the first-channel coil and the second-channel coil of the excitation probe (1) is delayed by one quarter of the preset wavelength in the axial direction, and the excitation time of the first-channel coil relative to the second-channel coil of the excitation probe (1) is advanced by one quarter of the period of the guided wave; when receiving the guided wave, the distance between the first-channel coil and the second-channel coil of the receiving probe (2) is delayed by one quarter of the preset wavelength in the axial direction, and the receiving time of the first-channel coil relative to the second-channel coil of the receiving probe (2) is delayed by one quarter of the period of the guided wave.

3. The method of claim 1, wherein the method is based on dual-channel guided-wave phase characteristics. In the step 1), one of the single-channel echo signals in the double-channel echo signal detected by the receiving probe (2) is selected, and a single-channel envelope signal is obtained after Hilbert transformation processing, N maximum points in the envelope signal are determined as wave packet vertices, and for each wave packet and the wave packet vertex, the minimum values of other maximum points on the left and right sides of the wave packet vertex in the wave packet region are calculated, so as to obtain left and right limits, the region in the left and right limits is taken as a wave packet window, and finally N wave packet windows are obtained.

4. The dual-channel guided-wave phase characteristic based defect signal identification and location method according to claim 1, characterized in that: In the step 2), for each wave packet window, the interpolation cosine similarity of the double-channel echo signal in the wave packet window is calculated, specifically, the echo signal of the second channel is interpolated into the echo signal of the first channel to obtain an interpolation signal, and the interpolation signal and the echo signal of the first channel are calculated to obtain a similarity value, and the interpolation signal with a similarity value higher than a preset similarity threshold is reserved as a defect signal.

5. The dual-channel guided-wave phase characteristic based defect signal identification and location method according to claim 1, characterized in that: In the step 3), for each defect signal, a matching pursuit algorithm is used for reconstruction to determine the time of the maximum peak point of the defect signal, and the defect position on the test piece is determined according to the time of the maximum peak point and the propagation speed of the guided wave.

6. The dual-channel guided-wave phase characteristic based defect signal identification and location method according to claim 5, characterized in that: The reconstruction interval of the matching pursuit algorithm is the wave packet window where the defect signal is located, and the matching atom is a Hann window modulated generalized Gabor atom.

Images