Non-linear frequency modulation carbon fiber composite material full-focus detection method and system

Abstract

The invention discloses a non-linear frequency modulation carbon fiber composite material full-focus detection method, which comprises the following steps: designing a non-linear frequency modulation excitation waveform, generating a modulation coding signal by using the non-linear frequency modulation excitation waveform, exciting a single unit in an ultrasonic transducer array by the modulation coding signal to emit ultrasonic waves, after the defects in the sample to be detected are reflected, echo signals are received by all array elements on the transducer, and decoded echo signals are obtained through pulse compression in a signal receiving system; imaging the decoded echo signal through a full focusing method to obtain an original noise detection image; inputting the original noise detection image into the second-order total generalized total variation image denoising model to obtain a denoised image matrix; and inputting the de-noised image matrix into an improved Butterworth high-pass filter, and carrying out filtering processing to obtain a final detection image data matrix. According to the invention, the problems of insufficient detection depth and imaging signal-to-noise ratio of carbon fiber composite material full-focusing detection can be solved.

Description

Technical Field

[0001] The invention belongs to the technical field of ultrasonic nondestructive testing, and in particular relates to a nonlinear frequency-modulated carbon fiber composite material full-focusing detection method and system. Background Art

[0002] Carbon fiber composite materials have a strong attenuation effect on sound waves, and the useful signal received by the system is relatively weak, resulting in a low signal-to-noise ratio, low depth resolution, and poor imaging quality; this limits the application of full-focus imaging in carbon fiber materials.

[0003] In order to improve the detection depth and signal-to-noise ratio of the full-focus imaging method in carbon fiber composites, the traditional method is to reduce the frequency of the excitation signal and increase the width of the time domain signal. However, this will lead to a decrease in the axial resolution of the detection results. Therefore, there is a trade-off between axial resolution and signal-to-noise ratio. In the current research, the traditional coded signal is used as the excitation signal for phased array ultrasonic detection. Although the detection quality has been improved, there are still problems such as high sidelobe level and limited improvement in the signal-to-noise ratio of the detection results. At the same time, carbon fiber composites have a complex structure of internal fiber layer stacking, and the acoustic impedance of the matrix material is different from that of the carbon fiber material, forming serious structural noise, which affects the resolution of deep defects. Another new challenge facing carbon fiber is that the high-frequency part of the image signal contains noise and edge texture details; in the denoising process, how to effectively solve the problem of removing noise and retaining the edge texture details of the image.

[0004] Therefore, it is urgent to solve the above problems. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a nonlinear frequency modulation carbon fiber composite material full focusing detection method, the present invention can solve the deficiencies of carbon fiber composite material full focusing detection in terms of detection depth and imaging signal-to-noise ratio.

[0006] The second object of the present invention is to provide a nonlinear frequency modulated carbon fiber composite material full focusing detection system.

[0007] Technical solution: To achieve the above purpose, the present invention discloses a nonlinear frequency modulated carbon fiber composite material full focusing detection method, comprising the following steps:

[0008] (1) Design a nonlinear frequency modulation excitation waveform, use the nonlinear frequency modulation excitation waveform to generate a modulation coding signal, a single unit in the ultrasonic transducer array is excited by the modulation coding signal to emit ultrasound, after reflecting the defects in the sample to be tested, the echo signal is received by all array elements on the transducer, and the decoded echo signal is obtained by pulse compression in the signal receiving system;

[0009] (2) imaging the decoded echo signal through the full focusing method to obtain the original noise detection image F;

[0010] (3) Input the original noise detection image F into the second-order total generalized total variation image denoising model to obtain the denoised image matrix;

[0011] (4) The denoised image matrix is ​​input into the improved Butterworth high-pass filter for filtering to obtain the final detection image data matrix.

[0012] Optionally, the designing of the nonlinear frequency modulation excitation waveform in step (1) specifically includes the following steps:

[0013] To evaluate the detection capability of the proposed waveform, the two waveforms are based on the conventional linear frequency modulation excitation waveform defined by the same center frequency. In this work, a center frequency of 2.25 MHz is used. The pulse widths of the two waveforms are: T p =3μs;

[0014] The linear frequency modulation excitation waveform is: v LFM (t) = cos[2πf LFM (t)t]

[0015] The instantaneous frequency is expressed as:

[0016]

[0017] The center frequency can be defined as:

[0018]

[0019] The excitation duration and frequency range of the nonlinear FM waveform are defined in the same way as the linear FM waveform. The nonlinear FM excitation waveform is:

[0020] v NLFM (t) = cos[2πf 非线性调频 (t)·t-π]

[0021] The instantaneous frequency of the nonlinear FM waveform is a quadratic function of the excitation duration and is defined as:

[0022]

[0023] Optionally, the imaging by the total focusing method in step (2) specifically includes the following steps:

[0024] (2.1) The first array element of the transducer is excited, and all array elements receive the echo signal. The received echo signal is defined as: P(1,1)(t)~P(1,n)(t); The second array element of the transducer is excited, and all array elements are used to receive independently, and so on, until all array elements independently transmit a spherical wave and receive an echo signal N times, and a full matrix data set is obtained, which is expressed as:

[0025]

[0026] A represents the transmitting array element number, and B represents the receiving array element number;

[0027] (2.2) Divide the imaging area and determine the position of each pixel. The pixel will become the virtual focus object. Set an imaging point in the imaging area as point P. The coordinates of point P are (x P , z P ); By calculating the wave path difference of the sound wave from the transmitting array element to point P and then to the receiving array element, the focusing delay time of all ultrasonic echo signals at point P is calculated. The calculation formula is:

[0028]

[0029] Where, d AP is the distance from the imaging point P to the transmitting array element, d BP is the distance from the imaging point P to the receiving array element, x A is the horizontal coordinate of the transmitting array element, x B is the horizontal coordinate of the receiving array element, c is the sound velocity of the material;

[0030] (2.3) Based on the delay time, the full matrix data set is processed, and the amplitude of the focused pixel point P is obtained as the imaging matrix value by phase shifting and superimposing the echo signal data points, which is expressed as:

[0031]

[0032] (2.4) The imaging area is divided into M×N pixels, and the pixel amplitudes of all pixels are calculated according to the method of calculating the amplitude of point P to obtain the imaging matrix value data set I(T), that is, the original noise detection image F;

[0033]

[0034] Among them, I(X) 11 ~I(X) mn Represents the amplitude from the first imaged pixel to the imaged pixel.

[0035] Optionally, the denoising by the second-order generalized total variation image denoising model in step (3) specifically includes the following steps:

[0036] The second-order total generalized total variation denoising model can be simplified to the following minimization problem:

[0037]

[0038] In the formula, T represents the denoised image; F represents the original noise detection image; is the regularization term of the second-order variational model; (λ / 2)∫ Ω (TF) 2 dx dy is the data fidelity term of the second-order variational model; Ω is the image domain, λ is the Lagrange constant factor, ω∈BD(Ω) is the function defined in the image domain Ω, BD(Ω) is the set of all differentiable functions on a bounded open set, is the gradient of the image, D 2 T is the second-order derivative of the image, ‖.‖ 1 represents the L1 norm, a 1 and a 0 is the regularization parameter;

[0039] The Chambole-Lock iterative method is used to solve the second-order total generalized total variation image denoising model and output the denoised image matrix T 1 (N x ,N y ), N x ,N y is a matrix element.

[0040] Optionally, the filtering process performed by the improved Butterworth high-pass filter in step (4) specifically includes the following steps:

[0041] The transfer function expression of the improved Butterworth high-pass filter is:

[0042] H(u,v)=(R h -R 1 ){1 / [1+D 0 / cD(u,v)]}+R 1

[0043] Where D 0 is the cut-off frequency, R h is the high frequency gain, R 1 is the low frequency gain, c is a constant;

[0044] Read the image data matrix T after second-order total generalized total variation denoising 1 (N x ,N y ), and perform a fast two-dimensional Fourier transform on the image data matrix P(u,v)=ff2(T 1 );

[0045] Set the parameters of the Butterworth high-pass filter, calculate the Butterworth high-pass filter, and multiply the filter by the Fourier transformed image data matrix G(t,v)=P(t,v)·H(u,v);

[0046] Perform a two-dimensional inverse fast Fourier transform on the filtered data matrix to obtain the image data matrix T after feature sharpening. 2 (N x ,N y )=ifft2(G), which is the final detection image data matrix.

[0047] Based on the same inventive concept, the present invention discloses a nonlinear frequency modulated carbon fiber composite material full focusing detection system, comprising:

[0048] The signal acquisition unit is used to design a nonlinear frequency modulation excitation waveform, and use the nonlinear frequency modulation excitation waveform to generate a modulation coding signal. A single unit in the ultrasonic transducer array is excited by the modulation coding signal to emit ultrasound. After reflecting the defects in the sample to be tested, the echo signal is received by all array elements on the transducer, and the decoded echo signal is obtained by pulse compression in the signal receiving system;

[0049] An original imaging unit, used to image the decoded echo signal through a full focusing method to obtain an original noise detection image F;

[0050] An image denoising unit is used to input the original noise detection image F into a second-order total generalized total variation image denoising model to obtain a denoised image matrix;

[0051] The filtering processing unit is used to input the denoised image matrix into the improved Butterworth high-pass filter to perform filtering processing to obtain the final detection image data matrix.

[0052] Optionally, designing a nonlinear frequency modulation excitation waveform in the signal acquisition unit refers to:

[0053] To evaluate the detection capability of the proposed waveform, the two waveforms are based on the conventional linear frequency modulation excitation waveform defined by the same center frequency. In this work, a center frequency of 2.25 MHz is used. The pulse widths of the two waveforms are: T p =3μs;

[0054] The linear frequency modulation excitation waveform is: v LFM (t) = cos[2πf LFM (t)t]

[0055] The instantaneous frequency is expressed as:

[0056]

[0057] The center frequency can be defined as:

[0058]

[0059] The excitation duration and frequency range of the nonlinear FM waveform are defined in the same way as the linear FM waveform. The nonlinear FM excitation waveform is:

[0060] v NLFM (t) = cos[2πf 非线性调频 (t)·t-π]

[0061] The instantaneous frequency of the nonlinear FM waveform is a quadratic function of the excitation duration and is defined as:

[0062]

[0063] Optionally, the imaging by the total focusing method in the original imaging unit is specifically performed as follows:

[0064] The first array element of the transducer is excited, and all array elements receive the echo signal. The received echo signal is defined as: P(1,1)(t)~P(1,n)(t); the second array element of the transducer is excited, and all array elements are used to receive independently, and so on, until all array elements independently transmit a spherical wave and receive an echo signal N times, and a full matrix data set is obtained, which is expressed as:

[0065]

[0066] A represents the transmitting array element number, and B represents the receiving array element number;

[0067] Divide the imaging area and determine the position of each pixel. The pixel will become the virtual focus object. Set an imaging point in the imaging area as point P. The coordinates of point P are (x P , z P ); By calculating the wave path difference of the sound wave from the transmitting array element to point P and then to the receiving array element, the focusing delay time of all ultrasonic echo signals at point P is calculated. The calculation formula is:

[0068]

[0069] Where, d AP is the distance from the imaging point P to the transmitting array element, d BP is the distance from the imaging point P to the receiving array element, x A is the horizontal coordinate of the transmitting array element, x B is the horizontal coordinate of the receiving array element, c is the sound velocity of the material;

[0070] The full matrix data set is processed based on the delay time, and the amplitude of the focused pixel point P is obtained as the imaging matrix value by phase shifting and superimposing the echo signal data points, which is expressed as:

[0071]

[0072] The imaging area is divided into M×N pixels, and the pixel amplitudes of all pixels are calculated according to the method of calculating the amplitude of point P to obtain the imaging matrix value data set I(T), that is, the original noise detection image F;

[0073]

[0074] Among them, I(X) 11 ~I(X) mn Represents the amplitude from the first imaged pixel to the imaged pixel.

[0075] Optionally, the second-order total generalized total variation image denoising model in the image denoising unit performs denoising specifically as follows:

[0076] The second-order total generalized total variation denoising model can be simplified to the following minimization problem:

[0077]

[0078] In the formula, T represents the denoised image; F represents the original noise detection image; is the regularization term of the second-order variational model; (λ / 2)∫ Ω (TF) 2 dx dy is the data fidelity term of the second-order variational model; Ω is the image domain, λ is the Lagrange constant factor, ω∈BD(Ω) is the function defined in the image domain Ω, BD(Ω) is the set of all differentiable functions on a bounded open set, is the gradient of the image, D 2 T is the second-order derivative of the image, ‖.‖ 1 represents the L1 norm, a 1 and a 0 is the regularization parameter;

[0079] The Chambole-Lock iterative method is used to solve the second-order total generalized total variation image denoising model and output the denoised image matrix T 1 (N x ,N y ), N x ,N y is a matrix element.

[0080] Optionally, the improved Butterworth high-pass filter in the filtering processing unit performs filtering processing specifically as follows:

[0081] The transfer function expression of the improved Butterworth high-pass filter is:

[0082] H(u,v)=(R h -R 1 ){1 / [1+D0 / cD(u,v)]}+R 1

[0083] Where D 0 is the cut-off frequency, R h is the high frequency gain, R 1 is the low frequency gain, c is a constant;

[0084] Read the image data matrix T after second-order total generalized total variation denoising 1 (N x ,N y ), perform a fast two-dimensional Fourier transform on the image data matrix P(u,v)=fft2(T 1 );

[0085] Set the parameters of the Butterworth high-pass filter, calculate the Butterworth high-pass filter, and multiply the filter by the Fourier transformed image data matrix G(u,v)=P(u,v)·H(u,v);

[0086] Perform a two-dimensional inverse fast Fourier transform on the filtered data matrix to obtain the image data matrix T after feature sharpening. 2 (N x ,N y )=ifft2(G), which is the final detection image data matrix.

[0087] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: the present invention utilizes the characteristic that nonlinear frequency modulation signals contain more low-frequency energy, which makes up for the shortcomings of full-focus detection of composite materials in detection depth and imaging signal-to-noise ratio; in order to further eliminate the strong background noise in the detection image of carbon fiber composite materials, the present invention introduces the second-order total generalized total variation and Butterworth high-pass filter, which solves the technical problem of denoising and sharpening the defect features in the detection image. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 It is a schematic diagram of the coded excitation full focusing method in the present invention;

[0089] Figure 2 It is a schematic diagram of the time domain representation of the nonlinear frequency modulation waveform and the linear frequency modulation waveform in the present invention;

[0090] Figure 3 It is a schematic diagram of frequency domain representation of a nonlinear frequency modulation waveform and a linear frequency modulation waveform in the present invention;

[0091] Figure 4 It is a schematic diagram comparing the power spectrum density of linear frequency modulation and nonlinear frequency modulation waveforms in the present invention;

[0092] Figure 5It is a schematic diagram of denoising and sharpening an image using a second-order generalized total variation and a Butterworth high-pass filter in the present invention. DETAILED DESCRIPTION

[0093] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0094] Example 1

[0095] The present invention discloses a nonlinear frequency modulation carbon fiber composite material full focusing detection method, comprising the following steps:

[0096] (1) Design a nonlinear frequency modulation excitation waveform, use the nonlinear frequency modulation excitation waveform to generate a modulation coding signal, a single unit in the ultrasonic transducer array is excited by the modulation coding signal to emit ultrasound, after reflecting the defects in the sample to be tested, the echo signal is received by all array elements on the transducer, and the decoded echo signal is obtained by pulse compression in the signal receiving system;

[0097] The present invention utilizes the characteristic that nonlinear frequency modulation signals contain more low-frequency energy, and makes up for the deficiencies of carbon fiber reinforced polymer carbon fiber composite material full focusing detection in terms of detection depth and imaging signal-to-noise ratio.

[0098] like Figure 1 As shown, the energy of the echo signal is first increased through coded transmission and pulse compression, and then the image resolution is improved through the full focusing method to form an ultrasonic image with improved signal-to-noise ratio and resolution; a single unit in the ultrasonic transducer array is excited by the modulated coded signal to emit ultrasound, and the center frequency and bandwidth of the carrier are consistent with the transducer parameters; after reflecting the defects in the sample to be tested, the echo signal is received by all array elements on the transducer; in the signal receiving system, the decoded echo signal is obtained through pulse compression, and finally the decoded echo signal is output to the full focusing imaging program for imaging; after coded excitation and pulse compression processing, the signal-to-noise ratio of the echo signal will be effectively improved, thereby improving the signal-to-noise ratio of the image.

[0099] like Figure 2 , Figure 3 and Figure 4 As shown in the figure, the linear frequency modulation coded signal commonly used in the field of ultrasonic testing is used as the excitation signal for high attenuation material detection, but the signal-to-noise ratio is limited. Based on the relationship between acoustic attenuation and ultrasonic signal frequency, the higher the frequency, the greater the attenuation. A nonlinear frequency modulation waveform is designed whose instantaneous frequency is a quadratic function. This waveform enhances the energy within the signal frequency range and realizes phased array ultrasonic imaging detection of deep-buried defects in carbon fiber composite materials.

[0100] Designing the nonlinear FM excitation waveform specifically includes the following steps:

[0101] To evaluate the detection capability of the proposed waveform, the two waveforms are based on the conventional linear frequency modulation excitation waveform defined by the same center frequency. In this work, a center frequency of 2.25 MHz is used. The pulse widths of the two waveforms are: T p =3μs;

[0102] The linear frequency modulation excitation waveform is: v LFM (t) = cos[2πf LFM (t)t]

[0103] The instantaneous frequency is expressed as:

[0104]

[0105] The center frequency can be defined as:

[0106]

[0107] The excitation duration and frequency range of the nonlinear FM waveform are defined in the same way as the linear FM waveform. The nonlinear FM excitation waveform is:

[0108] v NLFM (t) = cos[2πf 非线性调频 (t)·t-π]

[0109] The instantaneous frequency of the nonlinear FM waveform is a quadratic function of the excitation duration and is defined as:

[0110]

[0111] (2) The decoded echo signal is imaged by the full focusing method to obtain the original noise detection image F(x, y, t); the phased array ultrasound full focusing imaging method uses the excitation and receiving element information contained in the full matrix data set to superimpose the echo on all imaging points in the measured area through post-processing to obtain the pixel amplitude information of all imaging points in the measured area.

[0112] Imaging by the total focusing method specifically includes the following steps:

[0113] (2.1) The first array element of the transducer is excited, and all array elements receive the echo signal. The received echo signal is defined as: P(1,1)(t)~P(1,n)(t); The second array element of the transducer is excited, and all array elements are used to receive independently, and so on, until all array elements independently transmit a spherical wave and receive an echo signal N times, and a full matrix data set is obtained, which is expressed as:

[0114]

[0115] A represents the transmitting array element number, and B represents the receiving array element number;

[0116] (2.2) Divide the imaging area and determine the position of each pixel. The pixel will become the virtual focus object. Set an imaging point in the imaging area as point P. The coordinates of point P are (x P , z P ); By calculating the wave path difference of the sound wave from the transmitting array element to point P and then to the receiving array element, the focusing delay time of all ultrasonic echo signals at point P is calculated. The calculation formula is:

[0117]

[0118] Where, d AP is the distance from the imaging point P to the transmitting array element, d BP is the distance from the imaging point P to the receiving array element, x A is the horizontal coordinate of the transmitting array element, x B is the horizontal coordinate of the receiving array element, c is the sound velocity of the material;

[0119] (2.3) Based on the delay time, the full matrix data set is processed, and the amplitude of the focused pixel point P is obtained as the imaging matrix value by phase shifting and superimposing the echo signal data points, which is expressed as:

[0120]

[0121] (2.4) The imaging area is divided into M×N pixels, and the pixel amplitudes of all pixels are calculated according to the method of calculating the amplitude of point P to obtain the imaging matrix value data set I(T), that is, the original noise detection image F;

[0122]

[0123] Among them, I(X) 11 ~I(X) mn Represents the amplitude from the first imaged pixel to the imaged pixel.

[0124] (3) The original noise detection image F(x, y, t) is converted into a two-dimensional data matrix F(x, y), which is input into the second-order total generalized total variation image denoising model to obtain the denoised image T;

[0125] The second-order total generalized total variation image denoising method is an image denoising method based on total variation regularization. It can penalize the non-smooth areas in the image, effectively remove the noise in the image, and use the second-order derivative information to protect the details of the image, ultimately improving the clarity and contrast of the image.

[0126] The establishment of the second-order generalized total variation image denoising model specifically includes the following steps:

[0127] The second-order total generalized total variation denoising model can be simplified to the following minimization problem:

[0128]

[0129] In the formula, T represents the denoised image; F represents the original noise detection image; is the regularization term of the second-order variational model; (λ / 2)∫ Ω (TF) 2 dxdy is the data fidelity term of the second-order variational model; Ω is the image domain, λ is the Lagrange constant factor, ω∈BD(Ω) is the function defined in the image domain Ω, BD(Ω) is the set of all differentiable functions on a bounded open set, is the gradient of the image, D 2 T is the second-order derivative of the image, ‖.‖ 1 represents the L1 norm, a 1 and a 0 is the regularization parameter;

[0130] The Chambole-Lock iterative method is used to solve the second-order total generalized total variation image denoising model and output the denoised image matrix T 1 (N x ,N y ), N x ,N y is a matrix element;

[0131] Initialization: Select T 1 (0) , K,τ,α,β;

[0132] Repeat the following steps until convergence or the maximum number of iterations K is reached;

[0133]

[0134] When the stopping criteria are met, the iteration process stops and the final denoised image T is output. 1 (N x ,N y ), otherwise let K = K + 1. (4) The denoised image matrix T 1 (N x ,N y ) is input into the improved Butterworth high-pass filter, and the final detection image data matrix T is obtained by filtering. 2 (N x ,N y );

[0135] The frequency component of the Fourier transform of an image is directly related to the spatial characteristics of the image. The low frequency corresponds to the area where the grayscale changes slowly in the image, and the high frequency corresponds to the part where the grayscale changes rapidly in the image. The establishment of the improved Butterworth high-pass filter specifically includes the following steps:

[0136] The defect features in the carbon fiber detection image belong to a part of the image mutation and are related to the high-frequency components. In the frequency-domain processing of the image, a high-pass filter is used to enhance the prominent defect feature information, realizing the sharpening processing of the detection image and further enhancing the defect feature information of the detection image.

[0137] The transfer function expression of the improved Butterworth high-pass filter is:

[0138] H(u,v) = (R h -R 1 ){1 / [1 + D 0 / cD(u,v)]} + R 1

[0139] In the formula, D 0 is the cut-off frequency, R h is the high-frequency gain, R 1 is the low-frequency gain, and c is a constant;

[0140] Read the image data matrix T 1 (N x , N y ) after the second-order total generalized variational denoising, and perform a fast two-dimensional Fourier transform on the image data matrix P(u,v) = fft2(T 1 );

[0141] Set the parameters of the Butterworth high-pass filter, calculate the Butterworth high-pass filtering, and multiply the filter by the Fourier-transformed image data matrix G(u,v) = P(u,v)·H(u,v);

[0142] Perform an inverse fast two-dimensional Fourier transform on the filtered data matrix to obtain the image data matrix T 2 (N x , N y ) = ifft2(G), which is the final detection image data matrix.

[0143] The second-order generalized total variation and Butterworth high-pass filtering processing of the present invention are used to further overcome the problem of strong background noise in the imaging detection structure of carbon fiber composites, denoise and sharpen the defect features in the detection image respectively, and further improve the clarity and contrast of the image defect information.

[0144] As Figure 5 shown, the image denoising model based on the second-order generalized total variation can remove the structural noise in the carbon fiber detection image, while the improved Butterworth high-pass filter can enhance the defect feature information of the detection image, obtaining a detection image with denoised and sharpened defect feature information.

[0145] Example 2

[0146] Based on the same inventive concept, the present invention discloses a nonlinear frequency modulated carbon fiber composite material full focusing detection system, comprising:

[0147] The signal acquisition unit is used to design a nonlinear frequency modulation excitation waveform, and use the nonlinear frequency modulation excitation waveform to generate a modulation coding signal. A single unit in the ultrasonic transducer array is excited by the modulation coding signal to emit ultrasound. After reflecting the defects in the sample to be tested, the echo signal is received by all array elements on the transducer, and the decoded echo signal is obtained by pulse compression in the signal receiving system;

[0148] Designing a nonlinear frequency modulation excitation waveform in the signal acquisition unit means:

[0149] To evaluate the detection capability of the proposed waveform, the two waveforms are based on the conventional linear frequency modulation excitation waveform defined by the same center frequency. In this work, a center frequency of 2.25 MHz is used. The pulse widths of the two waveforms are: T p =3μs;

[0150] The linear frequency modulation excitation waveform is: v LFM (t) = cos[2πf LFM (t)t]

[0151] The instantaneous frequency is expressed as:

[0152]

[0153] The center frequency can be defined as:

[0154]

[0155] The excitation duration and frequency range of the nonlinear FM waveform are defined in the same way as the linear FM waveform. The nonlinear FM excitation waveform is:

[0156] v NLFM (t) = cos[2πf 非线性调频 (t)·t-π]

[0157] The instantaneous frequency of the nonlinear FM waveform is a quadratic function of the excitation duration and is defined as:

[0158]

[0159] An original imaging unit, used to image the decoded echo signal through a full focusing method to obtain an original noise detection image F;

[0160] The imaging by the full focusing method in the original imaging unit is specifically as follows:

[0161] The first array element of the transducer is excited, and all array elements receive the echo signal. The received echo signal is defined as: P(1,1)(t)~P(1,n)(t); the second array element of the transducer is excited, and all array elements are used to receive independently, and so on, until all array elements independently transmit a spherical wave and receive an echo signal N times, and a full matrix data set is obtained, which is expressed as:

[0162]

[0163] A represents the transmitting array element number, and B represents the receiving array element number;

[0164] Divide the imaging area and determine the position of each pixel. The pixel will become the virtual focus object. Set an imaging point in the imaging area as point P. The coordinates of point P are (x P , z P ); By calculating the wave path difference of the sound wave from the transmitting array element to point P and then to the receiving array element, the focusing delay time of all ultrasonic echo signals at point P is calculated. The calculation formula is:

[0165]

[0166] Where, d AP is the distance from the imaging point P to the transmitting array element, d BP is the distance from the imaging point P to the receiving array element, x A is the horizontal coordinate of the transmitting array element, x B is the horizontal coordinate of the receiving array element, c is the sound velocity of the material;

[0167] The full matrix data set is processed based on the delay time, and the amplitude of the focused pixel point P is obtained as the imaging matrix value by phase shifting and superimposing the echo signal data points, which is expressed as:

[0168]

[0169] The imaging area is divided into M×N pixels, and the pixel amplitudes of all pixels are calculated according to the method of calculating the amplitude of point P to obtain the imaging matrix value data set I(T), that is, the original noise detection image F;

[0170]

[0171] Among them, I(X) 11 ~I(X) mn Represents the amplitude from the first imaged pixel to the imaged pixel.

[0172] An image denoising unit is used to input the original noise detection image F into a second-order total generalized total variation image denoising model to obtain a denoised image matrix;

[0173] The second-order total generalized total variation image denoising model in the image denoising unit performs denoising as follows:

[0174] The second-order total generalized total variation denoising model can be simplified to the following minimization problem:

[0175]

[0176] In the formula, T represents the denoised image; F represents the original noise detection image; is the regularization term of the second-order variational model; (λ / 2)∫ Ω (TF) 2 dx dy is the data fidelity term of the second-order variational model; Ω is the image domain, λ is the Lagrange constant factor, ω∈BD(Ω) is the function defined in the image domain Ω, BD(Ω) is the set of all differentiable functions on a bounded open set, is the gradient of the image, D 2 T is the second-order derivative of the image, ‖.‖ 1 represents the L1 norm, a 1 and a 0 is the regularization parameter;

[0177] The Chambole-Lock iterative method is used to solve the second-order total generalized total variation image denoising model and output the denoised image matrix T 1 (N x ,N y ), N x ,N y is a matrix element.

[0178] The filtering processing unit is used to input the denoised image matrix into the improved Butterworth high-pass filter to perform filtering processing to obtain the final detection image data matrix.

[0179] The improved Butterworth high-pass filter in the filtering processing unit performs filtering processing specifically as follows:

[0180] The transfer function expression of the improved Butterworth high-pass filter is:

[0181] H(u,v)=(R h -R 1 ){1 / [1+D 0 / cD(u,v)]}+R 1

[0182] Where D 0 is the cut-off frequency, R h is the high frequency gain, R 1 is the low frequency gain, c is a constant;

[0183] Read the image data matrix T after second-order total generalized total variation denoising1 (N x ,N y ), perform a fast two-dimensional Fourier transform on the image data matrix P(u,v)=fft2(T 1 );

[0184] Set the parameters of the Butterworth high-pass filter, calculate the Butterworth high-pass filter, and multiply the filter by the Fourier transformed image data matrix G(u,v)=P(u,v)·H(u,v);

[0185] Perform a two-dimensional inverse fast Fourier transform on the filtered data matrix to obtain the image data matrix T after feature sharpening. 2 (N x ,N y )=ifft2(G), which is the final detection image data matrix.

Claims

1. A nonlinear frequency modulated carbon fiber composite material full focusing detection method, characterized in that: The steps include: (1) Design a nonlinear frequency modulation excitation waveform, use the nonlinear frequency modulation excitation waveform to generate a modulation coding signal, a single unit in the ultrasonic transducer array is excited by the modulation coding signal to emit ultrasound, after reflecting the defects in the sample to be tested, the echo signal is received by all array elements on the transducer, and the decoded echo signal is obtained by pulse compression in the signal receiving system; (2) imaging the decoded echo signal through the full focusing method to obtain the original noise detection image F; (3) Input the original noise detection image F into the second-order total generalized total variation image denoising model to obtain the denoised image matrix; (4) The denoised image matrix is ​​input into the improved Butterworth high-pass filter for filtering to obtain the final detection image data matrix.

2. A nonlinear frequency modulated carbon fiber composite material full focusing detection method according to claim 1, characterized in that: The design of the nonlinear frequency modulation excitation waveform in step (1) specifically includes the following steps: To evaluate the detection capability of the proposed waveform, the two waveforms are based on the conventional linear frequency modulation excitation waveform defined by the same center frequency. In this work, a center frequency of 2.25 MHz is used. The pulse widths of the two waveforms are: T p =3μs; The linear frequency modulation excitation waveform is: v LFM (t) = cos[2πf LFM (t)t] The instantaneous frequency is expressed as: The center frequency can be defined as: The excitation duration and frequency range of the nonlinear FM waveform are defined in the same way as the linear FM waveform. The nonlinear FM excitation waveform is: v NLFM (t)=cos[2πf 非线性调频 (t)·t-π] The instantaneous frequency of the nonlinear FM waveform is a quadratic function of the excitation duration and is defined as:

3. A nonlinear frequency modulated carbon fiber composite material full focusing detection method according to claim 2, characterized in that: The imaging by the total focusing method in step (2) specifically comprises the following steps: (2.1) The first array element of the transducer is excited, and all array elements receive the echo signal. The received echo signal is defined as: P(1,1)(t)~P(1,n)(t); The second array element of the transducer is excited, and all array elements are used to receive independently, and so on, until all array elements independently transmit a spherical wave and receive an echo signal N times, and a full matrix data set is obtained, which is expressed as: A represents the transmitting array element number, and B represents the receiving array element number; (2.2) Divide the imaging area and determine the position of each pixel. The pixel will become the virtual focus object. Set an imaging point in the imaging area as point P. The coordinates of point P are (x P , z P ); By calculating the wave path difference of the sound wave from the transmitting array element to point P and then to the receiving array element, the focusing delay time of all ultrasonic echo signals at point P is calculated. The calculation formula is: Where, d AP is the distance from the imaging point P to the transmitting array element, d BP is the distance from the imaging point P to the receiving array element, x A is the horizontal coordinate of the transmitting array element, x B is the horizontal coordinate of the receiving array element, c is the sound velocity of the material; (2.3) Based on the delay time, the full matrix data set is processed, and the amplitude of the focused pixel point P is obtained as the imaging matrix value by phase shifting and superimposing the echo signal data points, which is expressed as: (2.4) The imaging area is divided into M×N pixels, and the pixel amplitudes of all pixels are calculated according to the method of calculating the amplitude of point P to obtain the imaging matrix value data set I(T), that is, the original noise detection image F; Among them, I(X) 11 ~I(X) mn Represents the amplitude from the first imaged pixel to the imaged pixel.

4. The method for full focusing detection of nonlinear frequency modulated carbon fiber composite materials according to claim 3, characterized in that: In step (3), the second-order generalized total variation image denoising model performs denoising, which specifically includes the following steps: The second-order total generalized total variation denoising model can be simplified to the following minimization problem: In the formula, T represents the denoised image; F represents the original noise detection image; is the regularization term of the second-order variational model; (λ / 2)∫ Ω (TF) 2 dx dy is the data fidelity term of the second-order variational model; Ω is the image domain, λ is the Lagrange constant factor, ω∈BD(Ω) is the function defined in the image domain Ω, BD(Ω) is the set of all differentiable functions on a bounded open set, is the gradient of the image, D 2 T is the second-order derivative of the image, ‖.‖1 represents the L1 norm, and a1 and a0 are regularization parameters; The Chambole-Lock iterative method is used to solve the second-order total generalized total variation image denoising model and output the denoised image matrix T1(N x ,N y ), N x ,N y is a matrix element.

5. The method for full focusing detection of nonlinear frequency modulated carbon fiber composite materials according to claim 4, characterized in that: The improved Butterworth high-pass filter in step (4) performs filtering processing, which specifically includes the following steps: The transfer function expression of the improved Butterworth high-pass filter is: H(u,v)=(R h -R1){1 / [1+D0 / cD(u,v)]}+R1 Where D0 is the cutoff frequency, R h is the high frequency gain, R1 is the low frequency gain, and c is a constant; Read the image data matrix T1(N after second-order total generalized total variation denoising) x ,N y ), perform fast two-dimensional Fourier transform P(u,v)=fft2(T1) on the image data matrix; Set the parameters of the Butterworth high-pass filter, calculate the Butterworth high-pass filter, and multiply the filter by the Fourier transformed image data matrix G(u,v)=P(u,v)·H(u,v); Perform a two-dimensional inverse fast Fourier transform on the filtered data matrix to obtain the image data matrix T2 (N x ,N y )=ifft2(G), which is the final detection image data matrix.

6. A nonlinear frequency modulated carbon fiber composite material full focusing detection system, characterized in that: include: The signal acquisition unit is used to design a nonlinear frequency modulation excitation waveform, and use the nonlinear frequency modulation excitation waveform to generate a modulation coding signal. A single unit in the ultrasonic transducer array is excited by the modulation coding signal to emit ultrasound. After reflecting the defects in the sample to be tested, the echo signal is received by all array elements on the transducer, and the decoded echo signal is obtained by pulse compression in the signal receiving system; An original imaging unit, used to image the decoded echo signal through a full focusing method to obtain an original noise detection image F; An image denoising unit is used to input the original noise detection image F into a second-order total generalized total variation image denoising model to obtain a denoised image matrix; The filtering processing unit is used to input the denoised image matrix into the improved Butterworth high-pass filter to perform filtering processing to obtain the final detection image data matrix.

7. The nonlinear frequency modulated carbon fiber composite material total focusing detection system according to claim 6, characterized in that: Designing a nonlinear frequency modulation excitation waveform in the signal acquisition unit refers to: To evaluate the detection capability of the proposed waveform, the two waveforms are based on the conventional linear frequency modulation excitation waveform defined by the same center frequency. In this work, a center frequency of 2.25 MHz is used. The pulse widths of the two waveforms are: T p =3μs; The linear frequency modulation excitation waveform is: v LFM (t) = cos[2πf LFM (t)t] The instantaneous frequency is expressed as: The center frequency can be defined as: The excitation duration and frequency range of the nonlinear FM waveform are defined in the same way as the linear FM waveform. The nonlinear FM excitation waveform is: v NLFM (t)=cos[2πf 非线性调频 (t)·t-π] The instantaneous frequency of the nonlinear FM waveform is a quadratic function of the excitation duration and is defined as:

8. The nonlinear frequency modulated carbon fiber composite material total focusing detection system according to claim 7, characterized in that: The imaging by the full focusing method in the original imaging unit is specifically performed as follows: The first array element of the transducer is excited, and all array elements receive the echo signal. The received echo signal is defined as: P(1,1)(t)~P(1,n)(t); the second array element of the transducer is excited, and all array elements are used to receive independently, and so on, until all array elements independently transmit a spherical wave and receive an echo signal N times, and a full matrix data set is obtained, which is expressed as: A represents the transmitting array element number, and B represents the receiving array element number; Divide the imaging area and determine the position of each pixel. The pixel will become the virtual focus object. Set an imaging point in the imaging area as point P. The coordinates of point P are (x P , z P ); By calculating the wave path difference of the sound wave from the transmitting array element to point P and then to the receiving array element, the focusing delay time of all ultrasonic echo signals at point P is calculated. The calculation formula is: Where, d AP is the distance from the imaging point P to the transmitting array element, d BP is the distance from the imaging point P to the receiving array element, x A is the horizontal coordinate of the transmitting array element, x B is the horizontal coordinate of the receiving array element, c is the sound velocity of the material; The full matrix data set is processed based on the delay time, and the amplitude of the focused pixel point P is obtained as the imaging matrix value by phase shifting and superimposing the echo signal data points, which is expressed as: The imaging area is divided into M×N pixels, and the pixel amplitudes of all pixels are calculated according to the method of calculating the amplitude of point P to obtain the imaging matrix value data set I(T), that is, the original noise detection image F; Among them, I(X) 11 ~I(X) mn Represents the amplitude from the first imaged pixel to the imaged pixel.

9. The nonlinear frequency modulated carbon fiber composite material total focusing detection system according to claim 8, characterized in that: The second-order total generalized total variation image denoising model in the image denoising unit performs denoising specifically as follows: The second-order total generalized total variation denoising model can be simplified to the following minimization problem: In the formula, T represents the denoised image; F represents the original noise detection image; is the regularization term of the second-order variational model; (λ / 2)∫ Ω (TF) 2 dx dy is the data fidelity term of the second-order variational model; Ω is the image domain, λ is the Lagrange constant factor, ω∈BD(Ω) is the function defined in the image domain Ω, BD(Ω) is the set of all differentiable functions on a bounded open set, is the gradient of the image, D 2 T is the second-order derivative of the image, ‖.‖1 represents the L1 norm, and a1 and a0 are regularization parameters; The Chambole-Lock iterative method is used to solve the second-order total generalized total variation image denoising model and output the denoised image matrix T1(N x ,N y ), N x ,N y is a matrix element.

10. The nonlinear frequency modulated carbon fiber composite material total focusing detection system according to claim 9, characterized in that: The improved Butterworth high-pass filter in the filtering processing unit performs filtering processing specifically as follows: The transfer function expression of the improved Butterworth high-pass filter is: H(u,v)=(R h -R1){1 / [1+D0 / cD(u,v)]}+R1 Where D0 is the cutoff frequency, R h is the high frequency gain, R1 is the low frequency gain, and c is a constant; Read the image data matrix T1(N after second-order total generalized total variation denoising) x ,N y ), perform fast two-dimensional Fourier transform P(u,v)=ff2(T1) on the image data matrix; Set the parameters of the Butterworth high-pass filter, calculate the Butterworth high-pass filter, and multiply the filter by the Fourier transformed image data matrix G(t,v)=P(t,v)·H(u,v); Perform a two-dimensional inverse fast Fourier transform on the filtered data matrix to obtain the image data matrix T2 (N x ,N y )=ifft2(G), which is the final detection image data matrix.

Images