Ultrasonic thickness extraction method based on wavelet compression energy estimation

Abstract

The invention discloses an ultrasonic thickness extraction method based on wavelet compression energy estimation, and relates to the technical field of ultrasonic thickness measurement, and the method comprises the following steps: S1, obtaining a first ultrasonic signal of a time domain ultrasonic wave, and calculating the power spectral density of the first ultrasonic signal; s2, calculating total energy of the first ultrasonic signal and accumulated energy distribution under different cut-off frequencies, and finding out a filtering frequency band; s3, wavelet transformation is carried out, and a first time-frequency graph is established; s4, filtering the first time-frequency diagram according to the filtering frequency band to obtain a second time-frequency diagram, and reconstructing the first ultrasonic signal to obtain a second ultrasonic signal; and S5, segmenting the second time-frequency diagram according to energy, calculating an ultrasonic echo position, and finally calculating the thickness. When facing a larger wave packet, the method can quickly find the center frequency and effectively process filtering, and solves the problem of wave packet dispersion caused by the change of the roughness of the reflecting surface through wave packet energy cutting, finding out a small specific envelope and combining self-correlation or cross-correlation with a single wave packet.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ultrasonic thickness measurement, in particular to an ultrasonic thickness extraction method based on wavelet compression energy estimation. BACKGROUND

[0002] As an important means of metal corrosion monitoring, ultrasonic thickness measurement is widely used in industry, especially in the current industrial equipment life, its corrosion monitoring is of great significance. The traditional ultrasonic thickness measurement device absorbs and attenuates the ultrasonic oscillation through the backing, which can significantly accelerate the attenuation of the ultrasonic pulse wave packet and reduce the wave packet width. However, in recent years, in order to reduce the thickness of the sensor, the monitoring unit is placed under the paint layer, and the wave packet attenuation rate is significantly reduced, and the wave packet width is significantly increased.

[0003] The waveform diagram of the existing reflection surface corrosion causing roughness reduction is shown in Figure 2 During the long-period monitoring process, the reflection surface corrosion will cause the roughness to decrease as the coupling state changes. The existing technology shows that if the roughness of the reflection surface changes, the wave packet will be scattered, which will cause the signal quality to deteriorate and make it difficult to extract the thickness. Therefore, how to combine the waveform to realize accurate thickness extraction has become a difficult problem. SUMMARY

[0004] The present application aims to overcome the shortcomings of the prior art and provide an ultrasonic thickness extraction method based on wavelet compression energy estimation.

[0005] The purpose of the present application is achieved by the following technical solutions: The present application discloses an ultrasonic thickness extraction method based on wavelet compression energy estimation, comprising the following steps: S1, obtaining a first ultrasonic signal of a time-domain ultrasonic wave, and calculating the power spectral density of the first ultrasonic signal; S2, calculating the total energy of the first ultrasonic signal, and then calculating the cumulative energy distribution under different cutoff frequencies to find the filter frequency band; S3, wavelet transforming the first ultrasonic signal to establish a first time-frequency diagram; S4, filtering the first time-frequency diagram according to the filter frequency band to obtain a second time-frequency diagram, and reconstructing the first ultrasonic signal based on the second time-frequency diagram to obtain a second ultrasonic signal; S5, dividing the second time-frequency diagram according to the energy to find the time period of different echoes, calculating the ultrasonic echo position, and finally calculating the thickness.

[0006] Further, step S1 specifically comprises: the obtained first ultrasonic signal is x[n], the total length of which is N, then the first ultrasonic signal x[n] is divided into K segments, each segment has a length of L, and there are M overlapping points between adjacent segments, wherein Then the value of the nth point in the i-th segment is , ,in, , Then, through the formula Calculate power spectral density ,in The factor for calculating the power spectral density is represented by f, where f represents the frequency and j represents the imaginary number. This represents the value of the nth point in the j'th segment.

[0007] Preferably, step S2 specifically includes: using the formula Calculate the total energy of the first ultrasound signal ;in The frequency resolution is expressed by the following formula: , This represents the sampling rate; then, it is expressed using the formula... Calculate the cumulative energy distribution at different cutoff frequencies Where m corresponds to different frequencies, as shown by the formula and Determine the energy percentage minimum frequency band Energy percentage The calculation formula is .

[0008] Preferably, step S3 specifically includes: selecting a discrete wavelet basis. Perform wavelet transform on the first ultrasonic signal x[n] to obtain the first time-frequency diagram. Where v represents the scale parameter and k represents the translation parameter. This represents the mother wavelet.

[0009] Preferably, step S4 specifically includes: based on the minimum frequency band obtained in step S2 For the first time-frequency diagram Filtering is performed, setting values ​​outside the frequency band to 0, while retaining the original amplitude and phase of values ​​within the frequency band, to obtain the second time-frequency diagram. Based on the second time-frequency diagram The first ultrasonic signal x[n] is reconstructed to obtain the second ultrasonic signal. , The second ultrasonic signal The system segments the data by window and calculates the average energy. It then finds five consecutive segments with energy below the average energy and performs wave packet segmentation. Preferably, step S5 includes the following steps: S501, transfer the second ultrasonic signal The first envelope is deleted, generating the third ultrasound signal x. ’ [n], through the formula calculating a third ultrasonic signal x ’ autocorrelation of [n], obtaining autocorrelation result ; S502, performing envelope extraction on the autocorrelation result , constructing envelope curve a(m), and performing moving average filtering on the envelope ; The smoothed envelope curve is a(m). S503, calculating the first peak position P of the smoothed envelope curve , and finally calculating the thickness D through the formula , wherein v represents the speed of sound.

[0010] Preferably, step S5 further comprises the following steps: S511, deleting the first envelope in the second ultrasonic signal , generating a third ultrasonic signal x ’ [n] ’ , extracting the second wave packet x2(n) of the third ultrasonic signal x ’ [n] and its position n2 and half wave width t2, and extracting the third wave packet of the third ultrasonic signal x [n] ’ S512, cross-correlating the third ultrasonic signal x [n] and the second wave packet x2(n), , obtaining cross-correlation result , then performing envelope extraction on the cross-correlation result, constructing envelope curve a(m), and performing moving average filtering on the envelope ; The smoothed envelope curve is a(m). S513, based on the smoothed envelope curve , finding the first maximum point in the first range , finding the second maximum point in the second range , calculating the position difference P' of the first maximum point and the second maximum point , and finally calculating the thickness D through the formula , wherein v represents the speed of sound.

[0011] The beneficial effects of the present application are: 1) The present application can quickly find the center frequency and effectively process filtering when facing large wave packets, thereby realizing accurate measurement of thickness under conditions of uncertain center frequency, large excitation wave packets, frequency dispersion, and high roughness of the reflecting surface.

[0012] 2) The application finds small specific envelopes by wave packet energy cutting, and solves the problem of wave packet dispersion due to the change of roughness of the reflecting surface by combining autocorrelation or cross-correlation with a single wave packet. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 Waveform of the prior art; Figure 2 Waveform of the prior art caused by the roughness reduction of the corroded reflecting surface; Figure 3 The step of directly performing data autocorrelation in the wavelet compression energy estimation-based ultrasonic thickness extraction method of the embodiment of the application is shown in Fig. 1. Figure 4 The step of performing cross-correlation on a single wave packet in the wavelet compression energy estimation-based ultrasonic thickness extraction method of the embodiment of the application is shown in Fig. 2. DETAILED DESCRIPTION

[0014] The technical solutions of the application will be described clearly and completely in combination with the embodiments. Obviously, the described embodiments are only some of the embodiments of the application, but not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0015] The application discloses a wavelet compression energy estimation-based ultrasonic thickness extraction method. In some ultrasonic monitoring scenarios, center frequency migration (such as wireless coupling system dispersion, piezoelectric ceramic and multiple inductive coil coupling resonance frequency change, and ultrasonic transmission process dispersion) may occur, resulting in uncertain center frequency. For single-pulse ultrasonic detection with fast attenuation, the influence of the center frequency migration does not need to be considered. However, for the case of large wave packet, it is crucial to find the center frequency and effectively filter. The existing waveforms are shown in Figs. 1 and 2. Figure 1 The application can quickly find the center frequency and effectively filter in the case of large wave packet by using specific algorithms, so as to accurately measure the thickness under the conditions of uncertain center frequency, large excitation wave packet, dispersion and high roughness of the reflecting surface.

[0016] In the embodiment, data autocorrelation is directly performed in the process of data extraction after obtaining the ultrasonic thickness waveform. Although the operation amount is larger, the reliability is higher. The step of directly performing data autocorrelation is shown in Fig. 1, and specifically includes the following steps. Figure 3 S1, obtaining a first ultrasonic signal of a time-domain ultrasonic wave, and calculating the power spectral density of the first ultrasonic signal; S2, calculating the total energy of the first ultrasonic signal, and calculating the cumulative energy distribution under different cutoff frequencies to find the filter frequency band;​ S3. Perform wavelet transform on the first ultrasonic signal to establish the first time-frequency diagram; S4. Filter the first time-frequency diagram according to the filtering frequency band to obtain the second time-frequency diagram, and reconstruct the first ultrasonic signal based on the second time-frequency diagram to obtain the second ultrasonic signal; S5. Divide the second time-frequency diagram according to energy, find the time periods of different echoes, calculate the ultrasonic echo positions, and finally calculate the thickness.

[0017] For example, step S1 specifically includes: acquiring a first ultrasound signal x[n] with a total length of N, and then dividing the first ultrasound signal x[n] into... K Each segment is of length L, and there are M overlapping points between adjacent segments. Then the value of the nth point in the i-th segment is , ,in, , Then, through the formula Calculate power spectral density ,in The factor for calculating the power spectral density is represented by f, where f represents the frequency and j represents the imaginary number. This represents the value of the nth point in the j'th segment.

[0018] For example, step S2 specifically includes: using the formula Calculate the total energy of the first ultrasound signal ;in The frequency resolution is expressed by the following formula: , This represents the sampling rate; then, it is expressed using the formula... Calculate the cumulative energy distribution at different cutoff frequencies Where m corresponds to different frequencies, as shown by the formula and Determine the energy percentage minimum frequency band Energy percentage The calculation formula is Energy percentage The value is generally between 0.4 and 0.95.

[0019] For example, step S3 specifically includes: selecting a discrete wavelet basis. Perform wavelet transform on the first ultrasonic signal x[n] to obtain the first time-frequency diagram. Where v represents the scale parameter and k represents the translation parameter. Representing the mother wavelet, such as the Haar wavelet, when When = 1, 0≤t<1 / 2); when -1, 1 / 2≤t<1. Exemplarily, step S4 specifically comprises: based on the minimum frequency band t obtained in step S2 , filtering the first time-frequency graph , setting the values not in the frequency band to 0, and keeping the original amplitude and phase of the values in the frequency band, to obtain a second time-frequency graph , reconstructing the first ultrasonic signal x[n] based on the second time-frequency graph , to obtain a second ultrasonic signal , , segmenting the second ultrasonic signal by window and calculating the average energy, finding 5 continuous segments lower than the average energy, and then performing wave packet segmentation. Exemplarily, step S5 comprises the following steps: S501, deleting the first envelope in the second ultrasonic signal to generate a third ultrasonic signal x ’ [n], calculating the autocorrelation of the third ultrasonic signal x [n] by the formula ’ , and obtaining an autocorrelation result ; S502, performing envelope extraction on the autocorrelation result , constructing an envelope curve a(m), and performing moving average filtering on the envelope ; , wherein the envelope curve a(m) is a smoothed envelope curve; S503, calculating the first peak position P of the smoothed envelope curve , and finally calculating the thickness D by the formula , wherein v represents the sound speed.

[0020] When it is required to reduce the amount of calculation, the embodiment directly extracts a single wave packet for cross-correlation in the data extraction process after obtaining the ultrasonic thickness measurement waveform, but the accurate extraction of a single wave packet may be wrong when the noise is large, causing a large error in thickness calculation. The step diagram of cross-correlation of a single wave packet is shown in Figure 4 , which specifically comprises the following steps: S1, obtaining a first ultrasonic signal of a time-domain ultrasonic wave, and calculating the power spectral density of the first ultrasonic signal; S2, calculating the total energy of the first ultrasonic signal, and then calculating the cumulative energy distribution under different cutoff frequencies to find the filtering frequency band; S3, performing wavelet transformation on the first ultrasonic signal to establish a first time-frequency graph; S4, filtering the first time-frequency map according to a filtering frequency band to obtain a second time-frequency map, reconstructing the first ultrasonic signal based on the second time-frequency map to obtain a second ultrasonic signal; S5, segmenting the second time-frequency map according to energy to find time periods of different echoes, calculating ultrasonic echo positions, and finally calculating thickness.

[0021] Exemplarily, step S1 specifically comprises: the obtained first ultrasonic signal is x[n], and the total length thereof is N; then the first ultrasonic signal x[n] is divided into segments, each segment has a length of L, and there are M overlapping points between adjacent segments, wherein K , then the value of the nth point in the ith segment is , wherein , ; then the power spectral density is calculated by formula wherein denotes a factor for calculating the power spectral density, f denotes frequency, and j denotes an imaginary number, denotes the value of the nth point in the j'th segment.

[0022] Exemplarily, step S2 specifically comprises: the total energy of the first ultrasonic signal is calculated by formula ; wherein denotes frequency resolution, and the calculation formula thereof is , denotes sampling rate; then the cumulative energy distribution under different cutoff frequencies is calculated by formula wherein m corresponds to different frequencies, the energy proportion is determined by formula and the minimum frequency band of the energy proportion is determined by formula The calculation formula of the energy proportion is , and the value of the energy proportion is generally between 0.4 and 0.95.

[0023] Exemplarily, step S3 specifically comprises: selecting a discrete wavelet basis to perform wavelet transform on the first ultrasonic signal x[n] to obtain a first time-frequency map wherein v denotes a scale parameter, k denotes a translation parameter, denotes a mother wavelet.

[0024] Exemplarily, step S4 specifically comprises: based on the minimum frequency band obtained in step S2, the first time-frequency map ​​​Filtering is performed, setting values ​​outside the frequency band to 0, while retaining the original amplitude and phase of values ​​within the frequency band, to obtain the second time-frequency diagram. Based on the second time-frequency diagram The first ultrasonic signal x[n] is reconstructed to obtain the second ultrasonic signal. , The second ultrasonic signal Segment by window and calculate average energy, find 5 consecutive segments with energy below the average energy, and then perform wave packet segmentation.

[0025] For example, step S5 further includes the following steps: S511, transfer the second ultrasonic signal The first envelope is deleted, generating the third ultrasound signal x. ’ [n], extract the third ultrasound signal x ’ The position n2 and half-wave width t2 of the second wave packet x2(n) of [n] are then used to extract the third ultrasound signal x. ’ The position n3 of the third wave packet of [n] and the half-wave width t3; S512, transfer the third ultrasonic signal x ’ Cross-correlation is performed between [n] and the second wave packet x2(n). Obtain cross-correlation results Then, the envelope of the cross-correlation results is extracted to construct the envelope curve a(m), and a moving average filter is applied to the envelope. ; The envelope curve after smoothing filtering; S513, Envelope Curve Based on Smoothing Filtering In the first range Find the first maximum value point In the second range Find the second maximum point Calculate the first maximum point Second maximum point The positional difference P' is finally determined by the formula. Calculate the thickness D, where v represents the speed of sound.

[0026] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. An ultrasonic thickness extraction method based on wavelet compression energy estimation, characterized by, The method comprises the following steps: S1, acquiring a first ultrasonic signal of a time domain ultrasonic wave, and calculating a power spectral density of the first ultrasonic signal; S2, calculating a total energy of the first ultrasonic signal, and then calculating an accumulated energy distribution under different cutoff frequencies to find a filtering frequency band; S3, performing wavelet transformation on the first ultrasonic signal to establish a first time-frequency graph; S4, filtering the first time-frequency graph according to the filtering frequency band to obtain a second time-frequency graph, reconstructing the first ultrasonic signal based on the second time-frequency graph, and obtaining a second ultrasonic signal; S5, segmenting the second time-frequency graph according to energy to find time periods of different echoes, calculating ultrasonic echo positions, and finally calculating thickness.

2. The method of claim 1, wherein, Step S1 specifically comprises: the acquired first ultrasonic signal is x[n], the total length of which is N, then the first ultrasonic signal x[n] is divided into K segments, each segment has a length of L, and there are M points of overlap between adjacent segments, wherein then the value of the nth point in the ith segment is , wherein , then the power spectral density is calculated by the formula wherein denotes a factor for calculating the power spectral density, f denotes frequency, j denotes an imaginary number, and denotes the value of the nth point in the j'th segment.​ 3. The ultrasonic thickness extraction method based on wavelet compression energy estimation according to claim 2, characterized in that, Step S2 specifically includes: using the formula Calculate the total energy of the first ultrasound signal ;in The frequency resolution is expressed by the following formula: , This represents the sampling rate; then, it is expressed using the formula... Calculate the cumulative energy distribution at different cutoff frequencies Where m corresponds to different frequencies, as shown by the formula and Determine the energy percentage minimum frequency band Energy percentage The calculation formula is .

4. The method of claim 3, wherein the wavelet compression energy estimation based ultrasonic thickness extraction method is characterized by, Step S3 specifically comprises: selecting a discrete wavelet base Wavelet transform is performed on the first ultrasonic signal x[n] to obtain a first time-frequency graph wherein v represents a scale parameter, and k represents a time domain translation parameter, denotes a mother wavelet.

5. The method of claim 4, wherein the wavelet compression energy estimation based ultrasonic thickness extraction method is characterized by, Step S4 specifically comprises: based on the minimum frequency band obtained in step S2 , filtering the first time-frequency map , setting the values not in the frequency band to 0, and keeping the original amplitude and phase of the values in the frequency band, to obtain a second time-frequency map , reconstructing the first ultrasonic signal x[n] based on the second time-frequency map , to obtain a second ultrasonic signal , , segmenting the second ultrasonic signal by window and calculating the average energy, and finding 5 consecutive segments lower than the average energy, and then performing wave packet segmentation.

6. The ultrasonic thickness extraction method based on wavelet compression energy estimation according to claim 5, characterized in that, Step S5 comprises the following steps: S501, transfer the second ultrasonic signal The first envelope is deleted, generating the third ultrasound signal x. ’ [n], through the formula Calculate the third ultrasound signal x ’ The autocorrelation of [n] is calculated to obtain the autocorrelation result. ; S502, to the autocorrelation result Envelop extraction is performed to construct an envelope curve a(m) and moving average filtering is performed on the envelope is a smoothed envelope curve​ S503、Calculate the envelope curve after smoothing filtering the first peak position P of the envelope curve, and finally calculate the thickness D by the formula where v represents the sound speed.

7. The method of claim 5, wherein the wavelet compression energy estimation based ultrasonic thickness extraction method is characterized by, Step S5 further comprises the following steps: S511, deleting the first envelope in the second ultrasound signal x[n] to generate a third ultrasound signal x ’ [n] to extract a second wave packet x2(n) and its position n2 and half wave width t2 of the third ultrasound signal x ’ [n] to extract a third wave packet and its position n3 and half wave width t3 of the third ultrasound signal x ’ [n] to extract a third wave packet and its position n3 and half wave width t3 of the third ultrasound signal x S512, the third ultrasonic signal x ’ [n] and the second wave packet x2(n) are cross-correlated, , to obtain a cross-correlation result , and then the cross-correlation result is envelope extracted to construct an envelope curve a(m), and the envelope is moving average filtered ; is a smoothed envelope curve; S513、based on the envelope curve after smoothing filtering , the first maximum point is found in the first range , the second maximum point is found in the second range , the position difference P' of the first maximum point and the second maximum point is calculated, and finally the thickness D is calculated by the formula , wherein v represents the speed of sound.​​

Images