Temperature compensation method of ultrasonic guided wave signal based on sliding window dynamic timing calibration

Abstract

This invention proposes a temperature compensation method for ultrasonic guided wave signals based on dynamic timing calibration of sliding windows. The present invention uses sliding windows to accelerate calculations and decompose signals into subsequences, and calculates the weighted impact energy parameter WSTE in the subsequence dimension to obtain the WSTE curve, and then determines the maximum amplitude of the WSTE curve to determine the sliding window corresponding to the weighted impact energy parameter WSTE of the maximum amplitude, and then calculates the location of defects in long and complex structural parts according to the sampling point sequence number of the normalized test signal corresponding to the starting point of the determined sliding window. The present invention monitors the defect information that may exist in the subsequence through the local perception concept of the sliding window, amplifies the echo characteristics of small defects on long components, increases the signal-to-noise ratio and identifiability of defects, effectively realizes the identification of defects by ultrasonic guided waves under different temperature conditions, and reduces the possibility of false alarms and missed alarms of defects.

Description

Technical Field

[0001] The invention relates to a temperature compensation method for ultrasonic guided wave monitoring signals, and in particular to a temperature compensation method for ultrasonic guided wave signals based on dynamic timing calibration of a sliding window. Background Art

[0002] Long and complex structural parts such as steel beams, turnout point rails, and railway base rails are key pressure-bearing components in various equipment and facilities in society. They are widely used in transportation and mechanical engineering. The health monitoring of these structural parts is of great significance to ensuring social and economic security and the safety of people’s property. Take the turnout point rail as an example. Once fatigue failure occurs in long-term work, it is very likely to cause extremely serious accidents such as train derailment and rollover.

[0003] In the non-destructive testing of complex structural parts, compared with magnetic flux leakage, X-ray and other detection schemes, ultrasonic guided waves are widely used in structural health monitoring due to their long propagation distance and full cross-section coverage. There have been a lot of research and practice on regular structural parts such as plates and tubes that do not undergo guided wave mode conversion. However, long and complex structural parts will cause a lot of ultrasonic guided wave dispersion and mode conversion, which will lead to complex ultrasonic guided wave echo signals and difficult feature extraction. More importantly, since the propagation of ultrasonic guided waves is affected by temperature, the increase in temperature will cause the ultrasonic wave velocity to change, which will cause the wave packet to shift, and even cause false alarms and missed alarms after defects occur in the monitoring system, which seriously threatens engineering safety.

[0004] Traditional temperature compensation algorithms, such as scale transformation and temperature library building methods, rely heavily on prior temperatures and a large amount of data, and are only applicable to regular structures with low frequency dispersion such as plates and tubes. They are powerless for long and complex structures that cause modal conversion of guided wave echoes. The dynamic timing calibration algorithm is a flexible and adaptive ultrasonic guided wave temperature compensation algorithm. It uses dynamic programming to adaptively find the best matching path between two ultrasonic sequences, which can effectively solve the problem of wave packet offset or stretching of ultrasonic signals due to temperature. However, this method has a large amount of calculation and will cause temperature over-compensation of defect signals, resulting in defects being missed. Summary of the invention

[0005] In order to solve the problem that ultrasonic guided wave transmission in long and complex structural parts is affected by temperature, which may cause false alarms or even missed alarms in the guided wave monitoring system, this invention proposes a temperature compensation method for ultrasonic guided wave signals based on dynamic timing calibration of sliding windows. The algorithm uses sliding windows to accelerate temperature compensation operations and decompose the signal into subsequences, and defines an abnormality indicator in the subsequence dimension. The local perception idea of ​​the sliding window is used to monitor the possible defect information in the subsequence. This method amplifies the echo characteristics of small defects on long components, increases the signal-to-noise ratio and identifiability of defects, effectively realizes the identification of defects by ultrasonic guided waves under different temperature conditions, reduces the possibility of false alarms and missed alarms of defects, overcomes the shortcomings of large computational complexity and defect over-compensation of dynamic timing calibration, and can effectively realize ultrasonic guided wave temperature compensation and defect monitoring of long components.

[0006] The present invention is achieved through the following technical solutions:

[0007] 1) Use ultrasonic monitoring instruments to collect guided wave reference signals and guided wave test signals of long and complex structural parts

[0008] 2) performing normalization processing on the waveguide reference signal and the waveguide test signal respectively to obtain a normalized reference signal and a normalized test signal;

[0009] 3) First, a Euclidean distance matrix is ​​constructed according to the sampling points of the normalized reference signal and the normalized test signal, and the (0,0) point of the Euclidean distance matrix is ​​used as the starting point of the initial sliding window. The sampling points of the normalized reference signal and the normalized test signal in the current sliding window are dynamically time-calibrated to obtain the optimal calibration path in the current sliding window. Then, the next sliding window is updated based on the optimal calibration path in each sliding window, and the sampling points of the normalized reference signal and the normalized test signal in the next sliding window are dynamically time-calibrated based on the Euclidean distance matrix to obtain the optimal calibration path in the next sliding window until the sliding window covers the boundary of the Euclidean distance matrix; then, the optimal calibration total path of the normalized reference signal and the normalized test signal is accumulated from the optimal calibration paths in each sliding window, and then the sampling points of the normalized reference signal and the normalized test signal are matched according to the optimal calibration total path, and finally, the normalized test signal is temperature compensated according to the matched sampling points to obtain the compensated test signal;

[0010] 4) According to the compensated test signal, the compensated test signal in each sliding window is used as the corresponding window subsequence Sw;

[0011] 5) Calculate the weighted shock energy parameter WSTE of the window subsequence Sw of each sliding window, and then connect each weighted shock energy parameter WSTE according to the sequence number of the sliding window to form a WSTE curve;

[0012] 6) Determine the maximum amplitude of the WSTE curve, thereby determining the sliding window corresponding to the weighted impact energy parameter WSTE of the maximum amplitude, and then calculate the location of the defect in the long complex structure according to the sampling point sequence number of the normalized test signal corresponding to the starting point of the determined sliding window.

[0013] When the path point of the optimal normalized path in the current sliding window meets the following conditions for the first time, the path point that meets the conditions is used as the starting point of the next sliding window. The formula is as follows:

[0014] (i,j)=(i-1,j-1),1 <i,j<n

[0015] Among them, (i, j) represents a path point of the optimal normalized path in the current sliding window, (i-1, j-1) represents the path point before the current path point of the optimal normalized path in the current sliding window; i represents the sampling point number of the normalized test signal, j represents the sampling point number of the normalized reference signal, and n represents the number of sampling points of the waveguide reference signal or the waveguide test signal.

[0016] The sliding window is a square window, and the side length is the length of the excitation signal.

[0017] The calculation formula of the weighted impact energy parameter WSTE of the window subsequence Sw is as follows:

[0018] ψ[x i ]=(x i ) 2 -x i+1 *x i-1

[0019]

[0020] Among them, WSTE μ represents the weighted impulse energy parameter WSTE of the window subsequence Sw of the μth sliding window, L represents the sliding window length, x ri represents the i-th sampling point of the normalized reference signal, x i 、x i+1 、x i-1 represent the i-th, i+1-th, and i-1-th sampling points of the normalized reference signal or the compensated test signal, respectively, ψ[x i ] represents the impact energy of the normalized reference signal or the compensated test signal at the i-th sampling point, t represents the sampling point number corresponding to the starting point of the μ-th sliding window, ψ[x ti ] represents the impact energy of the compensated test signal at the i-th sampling point, ψ[x ri ] represents the impact energy of the normalized reference signal at the i-th sampling point, and n represents the number of sampling points of the waveguide reference signal or the waveguide test signal.

[0021] In the step 6), the position of the defect in the long complex structural part is calculated according to the sampling point sequence number of the normalized test signal corresponding to the starting point of the determined sliding window. The specific calculation formula is as follows:

[0022]

[0023] Where, distance represents the distance of the defect relative to the installation position of the guided wave transducer, and f s represents the sampling frequency, N represents the sampling point number of the normalized test signal corresponding to the starting point of the sliding window, and v g represents the group velocity of the guided wave.

[0024] The present invention has the following beneficial effects:

[0025] The present invention performs temperature compensation on ultrasonic guided wave signals through dynamic timing calibration of sliding windows. While performing compensation, the local perception characteristics of the window can be used to cut the signal into subsequences, and then local perception of defects is achieved through the damage identification index WSTE on the subsequences, thereby overcoming the problem of dilution of defect echo information caused by too large a number of sampling points of ultrasonic guided waves on long groove parts.

[0026] At the same time, the present invention optimizes the algorithm time complexity through the sliding window, which can effectively reduce the algorithm running time, so that the defect alarm can be transmitted in time, and the location of the defect can also be determined through the WSTE indicator, so as to achieve the role of real-time monitoring of the structural health of the component. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a working diagram of the sliding window dynamic timing calibration algorithm;

[0028] Figure 2 This is the temperature compensation principle diagram of the dynamic timing calibration algorithm;

[0029] Figure 3 It is a schematic diagram of the sliding window updating and recording data segment subsequence in the sliding window dynamic time series return algorithm;

[0030] Figure 4 Schematic diagram of the steel rail measured in the embodiment;

[0031] Figure 5 It is a signal diagram before the sliding window dynamic timing normalization temperature compensation in the embodiment;

[0032] Figure 6 This is a diagram showing the effect of sliding window dynamic timing calibration temperature compensation in the embodiment;

[0033] Figure 7is a running time comparison diagram of the sliding window dynamic timing calibration and the dynamic timing calibration in the embodiment;

[0034] Figure 8 A defect recognition comparison diagram of the sliding window dynamic timing calibration and the dynamic timing calibration and scale transformation algorithm in the embodiment;

[0035] Fig. 9 It is a schematic diagram of the principle of positioning defects by dynamic timing calibration of sliding windows in the embodiment. DETAILED DESCRIPTION

[0036] The present invention will be further described below in conjunction with the accompanying drawings.

[0037] like Figure 1 As shown, the present invention comprises the following steps:

[0038] 1) Use ultrasonic monitoring equipment to collect a set of guided wave reference signals and guided wave test signals of long and complex structural parts, denoted as X R and X T , waveguide reference signal X R As the waveguide test signal X T As a reference for compensation, the waveguide reference signal refers to the waveguide signal collected from a long complex structure without defects, and the waveguide test signal refers to the waveguide signal collected from a long complex structure to be tested with defects.

[0039] 2) Normalizing the waveguide reference signal and the waveguide test signal respectively to obtain a normalized reference signal and a normalized test signal; in a specific implementation, the amplitude of the waveguide signal is between [-0.5V, 0.5V] after normalization, as shown in the following formula:

[0040]

[0041] Among them, X represents the normalized signal, x s represents the sth sampling point in the waveguide reference signal or the waveguide test signal, x k It represents the kth sampling point in the waveguide reference signal or the waveguide test signal, n represents the number of sampling points of the waveguide reference signal or the waveguide test signal, max() represents the maximum value operation, and min() represents the minimum value operation.

[0042] 3) First, a Euclidean distance matrix is ​​constructed according to the sampling points of the normalized reference signal and the normalized test signal, and the (0,0) point of the Euclidean distance matrix (i.e., the point composed of the first sampling points of the normalized reference signal and the normalized test signal) is used as the starting point of the initial sliding window. The sampling points of the normalized reference signal and the normalized test signal in the current sliding window are dynamically time-calibrated to obtain the optimal calibration path in the current sliding window. Then, the next sliding window is updated based on the optimal calibration path in each sliding window, and the sampling points of the normalized reference signal and the normalized test signal in the next sliding window are dynamically time-calibrated based on the Euclidean distance matrix to obtain The optimal normalization path in the next sliding window is obtained until the sliding window covers the boundary of the Euclidean distance matrix; then the optimal normalization total path of the normalized reference signal and the normalized test signal is accumulated from the optimal normalization paths in each sliding window, and the starting point of the optimal normalization path in each sliding window is used as the end point of the optimal normalization path in the previous sliding window, and then the sampling points of the normalized reference signal and the normalized test signal are matched according to the optimal normalization total path, and finally the normalized test signal is temperature compensated (i.e., reconstructed) according to the matching sampling points, so as to compensate the offset waveguide test signal and obtain the compensated test signal.

[0043] Dynamic timing calibration is performed on the sampling points of the normalized reference signal and the normalized test signal in the current sliding window to obtain the optimal calibration path in the current sliding window due to temperature changes, specifically:

[0044] Due to the change of temperature, the speed of the guided wave changes in the long and complex structure, which causes the wave packet to stretch and displace on the time axis. First, the Euclidean distance matrix D is constructed using the sampling points of the normalized reference signal and the normalized test signal. The formula is as follows:

[0045]

[0046] Among them, d i，j represents the Euclidean distance between the i-th sampling point of the normalized test signal and the j-th sampling point of the normalized reference signal. In the specific implementation, it is only necessary to construct the Euclidean distance matrix for the sampling points in the sliding window to reduce the amount of calculation.

[0047] Then, through the idea of ​​dynamic programming, a path is found on the Euclidean distance matrix D so that the total distance from the starting point (0, 0) to the end point (n, n) is the shortest. Due to the monotonicity of the path, the dynamic transfer equation is:

[0048]

[0049] Among them, D pm (i, j) represents the cumulative distance of the optimal normalized path from point (0, 0) to point (i, j), D pm(i-1, j) represents the cumulative distance of the optimal normalized path from point (0, 0) to point (i-1, j), D pm (i, j-1) represents the cumulative distance of the optimal normalized path from point (0, 0) to point (i, j-1), D pm (i-1, j-1) represents the cumulative distance of the optimal normalized path from point (0, 0) to point (i-1, j-1).

[0050] The optimal alignment path within each sliding window is determined according to the dynamic transfer equation.

[0051] Figure 2 The principle of the dynamic timing calibration algorithm for waveguide temperature compensation is explained. After calculating the distance matrix D between the two signals, the dynamic timing calibration algorithm uses the idea of ​​dynamic programming to find an optimal calibration path, such as Figure 2 As shown in (a), the point where the path passes is the matching point between the waveguide reference signal and the waveguide test signal, as shown in Figure 2 As shown by the gray line in (b), the waveguide test signal will then be remapped into a new signal according to this matching relationship (if the optimal normalization path passes through point (i, j), then the i-th point in the waveguide test signal will be remapped to the j-th point of the compensated waveguide test signal). The compensation result is shown in Figure 2 as shown in (c).

[0052] Figure 3 Improvement of dynamic timing calibration for sliding window dynamic timing calibration, Figure 3 (a) is a sliding window created from the starting point, and the local optimal calibration path is calculated. When the path exceeds the window, the optimal calibration path is recalculated in the latest updated window, and it is iterated in sequence until the end. Compared with dynamic timing calibration, the algorithm proposed in the present invention can effectively reduce the number of operations, because dynamic timing calibration requires calculating the distance of all points between two signals, and the time complexity is O(n*n), while the sliding window dynamic timing calibration only needs to calculate the Euclidean distance inside the window, and the time complexity is O(n*L). And by using the local receptive field of the sliding window, the original waveguide test signal can also be split into multiple overlapping data segments, such as Figure 3 As shown in (b), the impact energy is calculated within these data segments to identify defects.

[0053] When the path point of the optimal normalized path in the current sliding window meets the following conditions for the first time, the path point that meets the conditions will be used as the starting point of the next sliding window. The formula is as follows:

[0054] (i,j)=(i-1,j-1),1 <i,j<n

[0055] Among them, (i, j) represents a path point of the optimal normalized path in the current sliding window, (i-1, j-1) represents the path point before the current path point of the optimal normalized path in the current sliding window; i represents the sampling point number of the normalized test signal, j represents the sampling point number of the normalized reference signal, and n represents the number of sampling points of the waveguide reference signal or the waveguide test signal.

[0056] The sliding window is a square window with a side length equal to the length of the excitation signal. The excitation signal is the signal excited by the circuit board of the ultrasonic monitoring instrument. If the excitation signal is a five-cycle sine wave, the sliding window also covers the five-cycle sine wave to better capture the defect echo.

[0057] 4) According to the compensated test signal, the compensated test signal in each sliding window is used as the corresponding window subsequence Sw, and the length of the window subsequence is equal to the sliding window length L;

[0058] 5) Calculate the weighted impulse energy parameter WSTE of the window subsequence Sw of each sliding window, and then connect each weighted impulse energy parameter WSTE according to the sequence number of the sliding window to form a WSTE curve, and use the WSTE curve to characterize the waveguide signal;

[0059] The calculation formula of the weighted impact energy parameter WSTE of the window subsequence Sw is as follows:

[0060] ψ[x i ]=(x i ) 2 -x i+1 *x i-1

[0061]

[0062] Among them, WSTE μ represents the weighted impulse energy parameter WSTE of the window subsequence Sw of the μth sliding window, L represents the sliding window length, x ri represents the i-th sampling point of the normalized reference signal, x i 、x i+1 、x i-1 represent the i-th, i+1-th, and i-1-th sampling points of the normalized reference signal or the compensated test signal, respectively, ψ[x i ] represents the impact energy of the normalized reference signal or the compensated test signal at the i-th sampling point, t represents the sampling point number corresponding to the starting point of the μ-th sliding window, ψ[x ti ] represents the impact energy of the compensated test signal at the i-th sampling point, ψ[x ri ] represents the impact energy of the normalized reference signal at the i-th sampling point, and n represents the number of sampling points of the waveguide reference signal or the waveguide test signal.

[0063] 6) In order to locate the defect, the location of the defect can be inferred by the maximum peak value of the amplitude in the WSTE curve. Determine the maximum amplitude of the WSTE curve, thereby determining the sliding window corresponding to the weighted impact energy parameter WSTE with the maximum amplitude, and then calculate the location of the defect in the long complex structure according to the sampling point number of the normalized test signal corresponding to the starting point of the determined sliding window.

[0064] In step 6), the position of the defect in the long complex structural part is calculated according to the sampling point sequence number of the normalized test signal corresponding to the starting point of the determined sliding window. The specific calculation formula is as follows:

[0065]

[0066] Where, distance represents the distance of the defect relative to the installation position of the guided wave transducer, and f s represents the sampling frequency, N represents the sampling point number of the normalized test signal corresponding to the starting point of the sliding window, and v g represents the group velocity of the guided wave.

[0067] Taking the shear wave velocity v=3250m / s as an example, after knowing the sampling point number N of the normalized test signal corresponding to the starting point of the determined sliding window, the actual position of the defect can be calculated according to the sampling frequency fs.

[0068] Embodiments of the present invention are as follows:

[0069] Railway stock rail, as a typical long and complex structural component, will be used to illustrate the proposed algorithm.

[0070] Step 1: Select a 12m long railway basic rail, install a transducer at the head end to collect ultrasonic guided wave signals at different temperatures, first collect a set of guided wave reference signals without introduced defects at any temperature, then after a period of time, use an electric saw to make a 3mm*12mm cut on the bottom of the rail at a distance of 5m from the transducer, and collect a set of guided wave test signals. The rails and defects used are as follows: Figure 4 shown.

[0071] Step 2: normalize the two sets of signals and compare them, as Figure 5 As shown, it can be seen that the temperature causes the phase shift of the ultrasonic guided wave. The fundamental reason is that the temperature increase affects the material properties of the rail and thus leads to a decrease in the guided wave velocity.

[0072] Step 3, in order to reflect the compensation effect of the method of the present invention on the waveguide signal, the signal is compensated using the algorithm described above to obtain a signal such as Figure 6 As shown, Figure 6This is a partial enlarged view of the compensated waveguide test signal at 10m. From the comparison between the local test signal and the reference signal, it can be seen that the phase shift phenomenon is perfectly solved, that is, the influence of temperature is compensated by the algorithm.

[0073] In order to reflect the improvement of the invention on dynamic timing calibration, Python3 is used to perform sliding window dynamic timing calibration and temperature compensation of dynamic timing calibration algorithm on the waveguide test signal. Figure 7 It can be seen from the running time that the single running time of the algorithm is 0.3 seconds, while the dynamic timing reduction criterion takes 15.9 seconds to execute. Therefore, the method proposed in the present invention can significantly improve the alarm real-time performance of the long component guided wave monitoring system.

[0074] Step 4: extract the subsequences in the sliding window to form a new waveguide test signal matrix:

[0075] X WT ={S w1 , S w2 , ... S wi ,……,S wp}

[0076] S wμ =[x μ*l x μ*l+1 ……x μ*(l+1) ]

[0077] Where S wμ represents the μth subsequence.

[0078] Step 5, in order to reflect the results of the damage monitoring of the rail by the method of the present invention, the WSTE value of each window is calculated to form a WSTE curve, and compared with the traditional temperature compensation algorithm: scale transformation and dynamic time series calibration. The traditional method uses the residual amplitude value as the defect measurement index, and the algorithm of the present invention uses WSTE (short-time subsequence impact energy) as the measurement index, and the signal-to-noise ratio is used as the basis for defect monitoring. The signal-to-noise ratio SNR is defined as:

[0079]

[0080] V N Refers to the amplitude signal of the noise, V D Refers to the amplitude signal of the defect.

[0081] from Figure 8 The comparison chart can be obtained in Table 1, where Figure 8 (a) is the defect amplitude without treatment, Figure 8 (b) is the defect amplitude after scale transformation. Figure 8 (c) is the defect amplitude under dynamic time series calibration processing, Figure 8 (d) is the defect energy under the sliding window dynamic timing calibration processing.

[0082] Table 1: Comparison of signal-to-noise ratio (SNR) between the existing method and the present invention

[0083] Algorithm Name none Scale Transformation Dynamic Timing Correction Sliding window dynamic timing correction SNR 0.13 0.9 1.1 1.67

[0084] from Figure 8 As shown in Table 1, the defect signal clarity, defect recognizability and signal-to-noise ratio obtained by the algorithm proposed in the present invention are higher than those of the traditional scale transformation and dynamic time series calibration methods.

[0085] Step 6: Finally, the defect location is determined based on the peak value of the WSTE indicator of each signal. Since the optimal normalization path can be mapped to the number of sampling points, the location of the defect can be inferred, such as Fig. 9 As shown, the defect located by the algorithm is 4.93m, with an error of less than 2% from the actual defect.

[0086] It can be seen that the present invention slices and temperature compensates the signal through the idea of ​​sliding window, thereby realizing the rapid calculation of temperature compensation for long structural parts. By calculating the WSTE index, defects can also be quickly monitored and located. The sliding window ensures that the data segment where the defect is located can be scanned and captured. The algorithm still has excellent sensitivity to defects at a long sampling distance and will not be affected by temperature changes. It provides ideas for ultrasonic guided wave nondestructive testing and structural health monitoring of long and complex structural parts, and has outstanding and significant technical effects.

[0087] The above specific implementation modes are used to explain the present invention rather than to limit the present invention. Any modification and change made to the present invention within the spirit of the present invention and the protection scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A temperature compensation method for ultrasonic guided wave signals based on sliding window dynamic timing calibration. It is characterized in that The following steps are involved: 1) Use ultrasonic monitoring instruments to collect guided wave reference signals and guided wave test signals of long and complex structural parts 2) performing normalization processing on the waveguide reference signal and the waveguide test signal respectively to obtain a normalized reference signal and a normalized test signal; 3) First, a Euclidean distance matrix is ​​constructed according to the sampling points of the normalized reference signal and the normalized test signal, and the (0,0) point of the Euclidean distance matrix is ​​used as the starting point of the initial sliding window. The sampling points of the normalized reference signal and the normalized test signal in the current sliding window are dynamically time-calibrated to obtain the optimal calibration path in the current sliding window. Then, the next sliding window is updated based on the optimal calibration path in each sliding window, and the sampling points of the normalized reference signal and the normalized test signal in the next sliding window are dynamically time-calibrated based on the Euclidean distance matrix to obtain the optimal calibration path in the next sliding window until the sliding window covers the boundary of the Euclidean distance matrix; then, the optimal calibration total path of the normalized reference signal and the normalized test signal is accumulated from the optimal calibration paths in each sliding window, and then the sampling points of the normalized reference signal and the normalized test signal are matched according to the optimal calibration total path, and finally, the normalized test signal is temperature compensated according to the matched sampling points to obtain the compensated test signal; 4) According to the compensated test signal, the compensated test signal in each sliding window is used as the corresponding window subsequence Sw; 5) Calculate the weighted shock energy parameter WSTE of the window subsequence Sw of each sliding window, and then connect each weighted shock energy parameter WSTE according to the sequence number of the sliding window to form a WSTE curve; 6) Determine the maximum amplitude of the WSTE curve, thereby determining the sliding window corresponding to the weighted impact energy parameter WSTE of the maximum amplitude, and then calculate the position of the defect in the long complex structure according to the sampling point sequence number of the normalized test signal corresponding to the starting point of the determined sliding window; When the path point of the optimal normalized path in the current sliding window meets the following conditions for the first time, the path point that meets the conditions is used as the starting point of the next sliding window. The formula is as follows: (i,j)=(i-1,j-1),1 <i,j<n Wherein, (i, j) represents a path point of the optimal normalized path in the current sliding window, (i-1, j-1) represents the path point before the current path point of the optimal normalized path in the current sliding window; i represents the sampling point number of the normalized test signal, j represents the sampling point number of the normalized reference signal, and n represents the number of sampling points of the waveguide reference signal or the waveguide test signal; The calculation formula of the weighted impact energy parameter WSTE of the window subsequence Sw is as follows: ψ[x i ]=(x i ) 2 -x i+1 *x i-1 Among them, WSTE μ represents the weighted impulse energy parameter WSTE of the window subsequence Sw of the μth sliding window, L represents the sliding window length, x ri represents the i-th sampling point of the normalized reference signal, x i 、x i+1 、x i-1 represent the i-th, i+1-th, and i-1-th sampling points of the normalized reference signal or the compensated test signal, respectively, ψ[x i ] represents the impact energy of the normalized reference signal or the compensated test signal at the i-th sampling point, t represents the sampling point number corresponding to the starting point of the μ-th sliding window, ψ[x ti ] represents the impact energy of the compensated test signal at the i-th sampling point, ψ[x ri ] represents the impact energy of the normalized reference signal at the i-th sampling point, and n represents the number of sampling points of the waveguide reference signal or the waveguide test signal; In the step 6), the position of the defect in the long complex structural part is calculated according to the sampling point sequence number of the normalized test signal corresponding to the starting point of the determined sliding window. The specific calculation formula is as follows: Where, distance represents the distance of the defect relative to the installation position of the waveguide transducer, and f s represents the sampling frequency, N represents the sampling point number of the normalized test signal corresponding to the starting point of the sliding window, and v g represents the group velocity of the guided wave.

2. According to claim 1, a method for temperature compensation of ultrasonic guided wave signals based on sliding window dynamic timing calibration, It is characterized in that The sliding window is a square window, and the side length is the length of the excitation signal.

Images