Continuous terahertz wave imaging quantitative detection system and method for internal defects of multi-layer non-metal structure

Abstract

The invention discloses a continuous terahertz wave imaging quantitative detection system and method for internal defects of a multilayer nonmetal structure, and belongs to the technical field of terahertz nondestructive testing. The detection system comprises a continuous wave linear frequency modulation signal source, a spread spectrum module, a terahertz transceiving probe, a focusing lens, a system control and processing computer, a four-dimensional scanning frame, a scanning frame controller, a detected test piece and a lifting platform, the detection method comprises the steps of determining detection parameters of the detection system; collecting a terahertz detection signal; performing wavelet transform on the terahertz detection signal to obtain a continuous wavelet coefficient; using empirical mode decomposition to decompose the continuous wavelet coefficient into an intrinsic mode function and a residual function; and positioning and imaging the internal defects of the non-metallic material by using the first-order intrinsic mode function as an imaging parameter. According to the method, high-resolution visual detection and accurate positioning can be effectively carried out on the internal defects of the multi-layer non-metal structure, and the method has a relatively high application value for ensuring the safe operation of a major structure.

Description

Technical Field

[0001] The present invention relates to the technical field of terahertz imaging and detection, and in particular to a continuous terahertz wave imaging quantitative detection system and method for internal defects of a multi-layer non-metallic structure. Background Art

[0002] As a non-contact, non-invasive detection method, terahertz imaging technology has broad application prospects in the field of internal defect detection of materials.

[0003] In terahertz non-destructive testing applications, defects inside non-metallic materials can be identified on the upper surface and internal interface of non-metallic materials, that is, there is information about internal defects within the range of the upper and lower surfaces of non-metallic materials, making it difficult to distinguish the exact interface position of the internal defects. The main reason for this result is that the wavelength of the terahertz detection signal is relatively long, resulting in the merging of the lower surface signal of the non-metallic material and the upper surface signal of the non-metallic material that are too close to each other to produce a "peak-merging" problem, thereby reducing the longitudinal resolution of the detection image and making it impossible to accurately judge the internal defects of non-metallic materials. Summary of the invention

[0004] In view of the deficiencies in the above-mentioned prior art, the purpose of the present invention is to provide a continuous terahertz wave imaging quantitative detection system and method for internal defects of multi-layer non-metallic structures, which solves the problem of insufficient longitudinal resolution in the existing terahertz imaging non-destructive testing field. The processed images are significantly improved in longitudinal resolution and clarity, which not only makes the positioning of non-metallic internal defects more accurate, but also provides a new method for terahertz tomography imaging and terahertz non-destructive testing application research.

[0005] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0006] In a first aspect, a continuous terahertz wave imaging quantitative detection system for internal defects of a multilayer non-metallic structure is provided, the system comprising a continuous wave linear frequency modulation signal source, a system control and processing computer, a scanning frame controller, a four-dimensional scanning frame, a spectrum expansion module, a terahertz transceiver probe, a focusing lens, a test piece to be inspected and a lifting platform;

[0007] The continuous wave linear frequency modulation signal source is electrically connected to the system control and processing computer and the spectrum expansion module, and is used to transmit the linear frequency modulation continuous wave to the spectrum expansion module, and collect the terahertz detection signal of the test piece under test, and transmit it to the system control and processing computer;

[0008] The scanning frame controller is electrically connected to the system control and processing computer and the four-dimensional scanning frame, and controls the movement of the four-dimensional scanning frame through the system control and processing computer;

[0009] The four-dimensional scanning frame includes an x-axis scanning axis, a y-axis scanning axis, a z-axis scanning axis and a P-axis rotating axis arranged on the z-axis scanning axis, which are connected in sequence. The four-dimensional scanning frame is controlled by a system control and processing computer to move, and the terahertz transceiver probe arranged on the spectrum expansion module is moved row by row or column by column at equal intervals to form a scanning surface.

[0010] The spectrum expansion module is connected to the terahertz transceiver probe to multiply the signal frequency output by the continuous wave linear frequency modulation signal source to a higher frequency required for detection;

[0011] The terahertz transceiver probe consists of a self-transmitting and self-receiving horn antenna, which is used to transmit and receive terahertz detection signals to detect the test piece under test;

[0012] The focusing lens is mechanically connected to the terahertz transceiver probe and is used to focus the terahertz transmission signal beam to the focal plane;

[0013] The lifting platform is mechanically connected to the test piece to be tested, and is used to fix the test piece to be tested, while providing a height adjustment function for the test piece to be tested.

[0014] Furthermore, uniform spatial sampling of two-dimensional plane data is ensured by controlling the moving speed of the terahertz transceiver module and the sampling frequency of data acquisition.

[0015] Furthermore, the detection frequency band is determined according to the material characteristics of the test piece under detection.

[0016] Furthermore, the optimal distance between the test piece under inspection and the terahertz transceiver module is determined by the focal length of the focusing lens.

[0017] In a second aspect, a detection method of a continuous terahertz wave imaging quantitative detection system for internal defects of a multilayer non-metallic structure is provided, comprising the following steps:

[0018] Step 1: Determine the detection parameters: The detection parameters include the x-direction scanning step length Δx, the y-direction scanning step length Δy, and the x-direction scanning range L x ,,y-direction scanning step length L y and a lift-off distance R0, and determining a detection frequency band according to material properties of the test piece (8) to be detected;

[0019] Step 2: experimentally testing the test piece: collecting the detection signal of the test piece (8) through a continuous terahertz wave imaging quantitative detection system; controlling the four-dimensional scanning frame (4), determining the x-direction scanning step length Δx and the y-direction scanning step length Δy, and realizing two-dimensional spatial sampling of the detection signal in a row-by-row or column-by-column scanning manner, thereby completing data collection of the test piece (8);

[0020] Step 3: Collecting terahertz detection signals: The continuous wave linear frequency modulation signal source (1) of the continuous terahertz wave imaging quantitative detection system generates a fast sawtooth low-frequency scanning signal, which is amplified by a frequency multiplier as a transmission signal, wherein a part of the transmission signal is radiated into the air through the terahertz transceiver probe (6), and the other part directly enters the spectrum spreading module as a reference signal. The transmission signal radiated into the air is converged on the focal plane through a focusing lens (7). When the transmission signal radiated into the air encounters an interface between different media, it returns along the original path and then enters the spectrum spreading module (5) as a receiving signal. The transmission signal and the receiving signal are frequency-converted in the spectrum spreading module (5), and finally, an intermediate frequency signal is output;

[0021] The transmitted signal and received signal are expressed as:

[0022]

[0023]

[0024] Among them, S ts (t) is the transmitted signal, S rs (t) is the received signal, A t is the amplitude of the transmitted signal, A r is the amplitude of the received signal, f0 is the initial frequency, K is the frequency modulation slope, expressed as B / t s , B is the bandwidth of the system, t s is the frequency sweep time, Δt is the delay time; t is time, j is the imaginary unit;

[0025] The output intermediate frequency signal is expressed as:

[0026]

[0027] Among them, S if (t) is the output intermediate frequency signal; when the continuous terahertz wave imaging quantitative detection system detects the test piece, when the detection signal interacts with the internal defects of the test piece (8), the positions of different layers of the internal defects correspond to different delay times Δt, and the delay time Δt increases as the thickness of the internal defects increases, and at the same time, the amplitude of the reflected signal changes as the thickness of the internal defects increases;

[0028] The data obtained by the continuous terahertz wave imaging quantitative detection system is the sum of cosine functions at different frequencies. Each frequency component corresponds to a different defect interface. The obtained data is subjected to Fourier transformation to obtain the relationship between the frequency modulation bandwidth and the internal defect position of the test piece (8) under inspection, which is expressed as follows:

[0029]

[0030] Wherein, d is the position distance of the test piece (8) under test, c is the speed of light, Δf is the frequency modulation bandwidth, and n is the refractive index of the test piece under test within the working frequency band of the continuous terahertz wave imaging quantitative detection system. According to the above principle, the continuous terahertz wave imaging quantitative detection system is used to scan the test piece (8) under test point by point, and finally a terahertz detection signal obtained by sampling the entire test piece (8) under test is obtained;

[0031] The terahertz detection signal obtained by two-dimensional scanning plane sampling is represented by s(x, y, k), which contains three variables. x, y, and k represent the modulus of the wave number vector corresponding to the sampling point on the x-axis coordinate, the y-axis coordinate, and the sampling signal. The modulus of the wave number vector corresponding to the sampling signal is proportional to the frequency f of the sampling signal. The relationship between the two is expressed as:

[0032]

[0033] Where, f is the frequency of the sampling signal;

[0034] Step 4: performing wavelet transform on the obtained terahertz detection signal: in order to extract the local characteristic information of each interface of the test piece (8) in the detection signal, enhance the signal characteristic and discrimination degree at the defect, and realize the accurate positioning of the defect in the depth direction, the obtained terahertz detection signal is subjected to continuous wavelet transform processing;

[0035] The time domain signal S(t) in the z direction at each scanning point on the two-dimensional scanning plane is expressed as:

[0036] S(t)=Aexp(j2πft)

[0037] Where S(t) is the time domain signal in the z direction at each scanning point on the two-dimensional scanning plane xy, and A is the signal amplitude;

[0038] The wavelet transform of the time domain signal S(t) is expressed as:

[0039]

[0040] Among them, W(a,b) is the continuous wavelet coefficient, a is the scale factor, ψ(t) is the conjugate function of the continuous wavelet basis function, and b is the time factor;

[0041] The conjugate function ψ(t) of the continuous wavelet basis function must satisfy the following conditions:

[0042]

[0043] in, is the inverse Fourier transform of ψ(t), and ω is the angular frequency;

[0044] For continuous wavelet transform, S(t) is projected into a two-dimensional time-frequency space to form a new time moment with a frequency of The component information W(a,b) at the time, the scale factor a and the time translation factor b respectively determine the center and time domain width of the continuous wavelet basis function, and under the premise of the selection of the wavelet basis function, the window shape of the wavelet window function is related to the scale factor a. The larger a is, the smaller the width of the frequency window will be, and the window shape is also adaptive;

[0045] Step 5: Perform empirical mode decomposition on the continuous wavelet coefficients: In order to maximize the characteristic of the detection signal and improve the longitudinal resolution of the detection signal, perform empirical mode decomposition on the obtained continuous wavelet coefficients W(a,b);

[0046] Empirical mode decomposition is an adaptive time-frequency analysis method that can decompose the continuous wavelet coefficients W(a,b) according to the scale characteristics of the signal into a series of intrinsic mode functions IMF and residual functions containing different oscillation modes. Different orders of eigenmode functions contain local characteristic signals of different time scales. Lower eigenmode functions contain more energy and frequency information in the signal before decomposition. Each order of eigenmode functions needs to meet the following two basic conditions:

[0047] (1) The number of local extreme points and zero crossing points of the eigenmode function must be equal or differ by at most one in the entire time range;

[0048] (2) At any time, the average value of the envelope of the local maximum and the envelope of the local minimum must be zero;

[0049] The continuous wavelet coefficient W(a,b) is expressed as follows after empirical mode decomposition:

[0050]

[0051] Among them, IMF k (t) is the kth order eigenmode function component, r N (t) is the residual function;

[0052] After the empirical mode analysis is completed, the first-order intrinsic mode function IMF contains the most original signal energy information and frequency information, and the time resolution Δt of the terahertz time domain detection signal is improved compared with the original signal. After the whole wave packet corresponding to the original terahertz echo time domain signal of the internal defect of the test piece (8) is processed by wavelet transform and empirical mode analysis, two peaks corresponding to the upper and lower surfaces of the defect are decomposed. The time information corresponding to the first peak of the defect is obtained by the first-order intrinsic mode function IMF corresponding to the defect at different positions in the internal direction of the test piece (8). The time information corresponding to the first peak of the defect corresponding to the first-order intrinsic mode function IMF is converted into distance information, so as to locate the defect. The conversion between the two is expressed as:

[0053]

[0054] Wherein, d1 is the depth of the internal defect of the test piece (8) under test, Δt1 is the time corresponding to the first peak value of the defect corresponding to the first-order intrinsic mode function IMF; the upper and lower surfaces of the internal defect of the test piece (8) under test corresponding to the first-order intrinsic mode function IMF are characterized by the peak values ​​of two wave packets decomposed from one wave packet at the internal defect of the original data. With respect to one wave packet of the original data, the above processing can decompose the upper and lower surfaces of the internal defect of the test piece (8) under test, thereby greatly improving the longitudinal resolution of the terahertz detection signal;

[0055] Step 6: Visualization of 3D imaging: The task of 3D imaging is to reconstruct the scattering coefficient distribution δ(x, y, k) of the target based on the measured terahertz detection signal s(x, y, k), and give the 3D scattering distribution map of the target;

[0056] Based on the first-order intrinsic mode function IMF in the lift-off direction z direction at each scanning point on the obtained two-dimensional scanning plane, the frequency domain signal s'(x, y, k) of the two-dimensional scanning plane is obtained by Fourier transform;

[0057] Based on the converted two-dimensional scanning plane frequency domain signal s'(x, y, k), the main steps of three-dimensional imaging algorithm data processing are:

[0058] (1) Two-dimensional Fourier transform in the spatial domain, the processed frequency domain signal s'(x, y, k) is expanded along the transverse measurement plane to obtain the component value S(k) of the plane wave propagating in any direction on the measurement plane z = R0 x ,k y ,k), the frequency domain signal s'(x,y,k) is transformed into two-dimensional Fourier transform in the spatial domain with respect to the azimuth xy:

[0059]

[0060] Among them, S(kx ,k y , k) is the component of the terahertz plane wave in the z direction, L x and L y are the scanning range, k x and k y are the wave number components in the x and y directions;

[0061] (2) Frequency domain filtering. Since there is a certain distance z = R0 between the scanning plane of the self-transmitting and self-receiving horn antenna and the target, it is necessary to perform frequency domain filtering on the spatial domain transformation data to obtain the component S' (k x ,k y ,k), the processing process is expressed as:

[0062]

[0063] Wherein, the spatial wave number k = 4πf / c, f is the transmission signal frequency, c is the electromagnetic wave propagation speed in the medium, R0 represents the vertical distance between the array plane and the plane of the test piece (8) under test, i.e., the lift-off, k x , k y , k z are the wave number components in the x, y, and z directions respectively;

[0064] (3) Stolt interpolation, by replacing the variables, we can obtain the complete spectrum S'(k x ,k y ,k z ), the processing process is expressed as:

[0065]

[0066] (4) Perform three-dimensional inverse Fourier transform on the data, and superimpose the plane waves in all propagation directions to obtain the three-dimensional scattering intensity distribution δ(x, y, z) of the target, and obtain a three-dimensional image. The processing process is expressed as:

[0067]

[0068] Preferably, when performing continuous wavelet transform processing on the terahertz detection signal in step 4, the second-order derivative of the Gaussian wavelet basis is used as the wavelet basis function and a wavelet combination with a scale of 2.

[0069] Preferably, the specific process of empirical mode decomposition of continuous wavelet coefficients in step 5 is as follows:

[0070] (1) Find all local maximum and local minimum values ​​of the continuous wavelet coefficients obtained after the detection signal is transformed by wavelet in the whole time range;

[0071] (2) Connect all local maxima as the upper envelope, connect all local minima as the lower envelope, and calculate the average value of the upper and lower envelopes, which is denoted as m1(t);

[0072] (3) Extract the new data sequence and calculate it by the following formula:

[0073] h1(t)=W(a,b)-m1(t)

[0074] Among them, h1(t) is the new data sequence;

[0075] (4) Iterate the new data sequence h1(t) and repeat steps (2), (3) and (4) until h1(t) satisfies the two basic conditions of the intrinsic mode function. At this time, the first-order intrinsic mode function IMF1(t) = h1(t);

[0076] (5) Calculate the first-order residual function. The calculation of the first-order residual function is expressed as:

[0077] r1(t)=W(a,b)-IMF1(t)

[0078] Among them, r1(t) is the first-order residual function;

[0079] Repeat steps (1), (2), (3) and (4) to obtain the eigenmode functions of the remaining orders;

[0080] When the residual function r N When (t) becomes a monotonic function or a constant, the above iterative solution process ends.

[0081] The beneficial effects of the present invention are:

[0082] The present invention provides a continuous terahertz wave imaging quantitative detection system and method for internal defects of multi-layer non-metallic structures, which utilizes terahertz detection signals to realize visual non-destructive detection of the tested test piece, thereby ensuring the integrity of the equipment structure and service safety. The system and method described in the present invention have the advantages of high efficiency and high longitudinal resolution. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 It is a continuous terahertz wave imaging quantitative detection system for internal defects of multi-layer non-metallic structures.

[0084] Among them, 1. Continuous wave linear frequency modulation signal source; 2. System control and processing computer; 3. Scanning frame controller; 4. Four-dimensional scanning frame; 5. Spread spectrum module; 6. Terahertz transceiver probe; 7. Focusing lens; 8. Test piece under inspection; 9. Lifting platform.

[0085] Figure 2 This is a schematic diagram of the four-dimensional scanning frame structure.

[0086] Figure 3 Schematic diagram of the two-dimensional plane scanning principle of the terahertz imaging nondestructive testing system. Among them, 6 is the terahertz transceiver probe; 8 is the test piece under test; Δx and Δy are the scanning step lengths, L x , L y is the scanning range, and R0 is the lift-off distance.

[0087] Figure 4 This is the defect design diagram of the test piece under inspection, where the black rectangular area represents the internal defects.

[0088] Figure 5 It is the original time domain signal at the defect of the tested piece and the first-order and second-order intrinsic mode functions IMF.

[0089] Figure 6 It is the original time domain signal of the defect of the tested piece.

[0090] Figure 7 is the first order intrinsic mode function IMF.

[0091] Figure 8 It is a three-dimensional slice diagram of the original time domain signal of the defect position in the depth direction of the test piece.

[0092] Fig. 9 It is a three-dimensional slice diagram of the first-order intrinsic mode function IMF of the defect position in the depth direction of the test piece. DETAILED DESCRIPTION

[0093] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0094] like Figure 1 As shown, the present invention provides a continuous terahertz wave imaging quantitative detection system for internal defects of a multi-layer non-metallic structure, characterized in that it includes a continuous wave linear frequency modulation signal source 1, a system control and processing computer 2, a scanning frame controller 3, a four-dimensional scanning frame 4, a spectrum expansion module 5, a terahertz transceiver probe 6, a focusing lens 7, a test piece to be detected 8 and a lifting platform 9;

[0095] The continuous wave linear frequency modulation signal source 1 is electrically connected to the system control and processing computer 2 and the spectrum expansion module 5, and is used to transmit the linear frequency modulation continuous wave to the spectrum expansion module 5, and collect the terahertz detection signal of the test piece under test, and transmit it to the system control and processing computer 2; the scanning frame controller 3 is electrically connected to the system control and processing computer 2 and the four-dimensional scanning frame 4, and controls the movement of the four-dimensional scanning frame 4 through the system control and processing computer 2, such as Figure 2 As shown, the four-dimensional scanning frame 4 is composed of an x-axis scanning axis 10, a y-axis scanning axis 11, a z-axis scanning axis 12 and a P-axis rotating axis 13, the P-axis rotating axis 13 is arranged on the z-axis scanning axis 12, and the spectrum expansion module 5 is arranged on the P-axis rotating axis 13; the four-dimensional scanning frame 4 is mechanically connected with the spectrum expansion module 5 and the terahertz transceiver probe 6, and the four-dimensional scanning frame 4 is controlled to move by the system control and processing computer 2, and the terahertz transceiver probe 6 is moved row by row or column by column at equal intervals to form a scanning surface; the spectrum expansion module 5 is connected to the terahertz transceiver probe 6, and the signal frequency output by the continuous wave linear frequency modulation signal source 1 is multiplied to a higher frequency required for detection; as shown Figure 3 As shown, the terahertz transceiver probe 6 is composed of a self-transmitting and self-receiving horn antenna, which is used to transmit and receive terahertz detection signals to detect the test piece 8 under test; the focusing lens 7 is mechanically connected to the terahertz transceiver probe 6, and is used to focus the terahertz transmission signal beam to the focal plane; the lifting platform 9 is mechanically connected to the test piece 8 under test, and is used to fix the test piece 8 under test, and at the same time provide adjustment of the height of the test piece 8 under test.

[0096] Uniform spatial sampling of two-dimensional plane data is ensured by controlling the moving speed of the terahertz transceiver module 6 and the sampling frequency of the spectrum spreading module 5 .

[0097] The optimal distance between the test piece and the terahertz transceiver module is determined by the focal length of the focusing lens.

[0098] The present invention also provides a continuous terahertz wave imaging quantitative detection method for internal defects of a multilayer non-metallic structure, comprising the steps of:

[0099] Step 1: Determine the detection parameters: The detection parameters include the x-direction scanning step length Δx and the y-direction scanning step length Δy, the x-direction scanning range L x and the y-scanning step length L y and a lift-off distance R0, and a detection frequency band determined according to the material characteristics of the test piece 8 to be detected;

[0100] Step 2: Experimentally test the test piece: collect the test signal of the test piece 8 through the continuous terahertz wave imaging quantitative detection system; control the four-dimensional scanning frame 4 to determine the x-direction scanning step length Δx and the y-direction scanning step length Δy, and implement two-dimensional spatial sampling of the test signal in a row-by-row or column-by-column scanning manner to complete the data collection of the test piece 8;

[0101] In this example, the terahertz frequency band used is 0.2-0.3 THz, and the reflection method is used for scanning detection, such as Figure 3 As shown, the scanning step lengths are Δx = 1 mm and Δy = 1 mm. The scanning ranges are selected according to the size of the test piece. x =120mm and L y =20mm. The lift-off distance R0 is the focal length 100mm, and the signal is collected by continuous column-by-column scanning.

[0102] Step 3: Collecting terahertz detection signals: The continuous wave linear frequency modulation signal source 1 of the continuous terahertz wave imaging quantitative detection system generates a fast sawtooth low-frequency scanning signal, which is amplified by a frequency multiplier as a transmission signal, wherein a part of the transmission signal is radiated into the air through the terahertz transceiver probe 6, and the other part directly enters the spectrum spreading module as a reference signal. The transmission signal radiated into the air is converged on the focal plane through the focusing lens 7. When the transmission signal radiated into the air encounters an interface between different media, it returns along the original path, and then enters the spectrum spreading module 5 as a receiving signal. The transmission signal and the receiving signal are frequency-converted in the spectrum spreading module 5, and finally, an intermediate frequency signal is output;

[0103] The transmitted signal and received signal are expressed as:

[0104]

[0105] Among them, S ts (t) is the transmitted signal, S rs (t) is the received signal, A t is the amplitude of the transmitted signal, A r is the amplitude of the received signal, f0 is the initial frequency, and K is the frequency modulation slope, which can be expressed as B / t s , B is the bandwidth of the system, t s is the frequency sweep time, Δt is the delay time; t is time, j is the imaginary unit;

[0106] The output intermediate frequency signal is expressed as:

[0107]

[0108] Among them, S if (t) is the output intermediate frequency signal; when the continuous terahertz wave imaging quantitative detection system detects the test piece, when the detection signal interacts with the internal defects of the test piece 8, the positions of different layers of the internal defects correspond to different delay times Δt, and the delay time Δt increases with the thickness of the internal defects.

[0109] Increases with the increase of the internal defect thickness, and at the same time, the amplitude of the reflected signal changes with the increase of the internal defect thickness;

[0110] The data obtained by the continuous terahertz wave imaging quantitative detection system is the sum of cosine functions at different frequencies. Each frequency component corresponds to a different defect interface. After Fourier transforming the obtained data, the relationship between the frequency modulation bandwidth and the internal defect position of the test piece 8 can be obtained, which is expressed as follows:

[0111]

[0112] Wherein, d is the position distance of the test piece 8 under test, c is the speed of light, Δf is the frequency modulation bandwidth, and n is the refractive index of the test piece under test within the working frequency band of the continuous terahertz wave imaging quantitative detection system. According to the above principle, the continuous terahertz wave imaging quantitative detection system can be used to scan the test piece 8 under test point by point, and finally obtain the terahertz detection signal obtained by sampling the entire test piece under test (8);

[0113] The terahertz detection signal obtained by two-dimensional scanning plane sampling is represented by s(x, y, k), which contains three variables x, y, and k, which represent the modulus of the wave number vector corresponding to the sampling point on the x-axis coordinate, the y-axis coordinate, and the sampling signal. The modulus of the wave number vector corresponding to the sampling signal is proportional to the frequency f of the sampling signal. The relationship between the two is expressed as:

[0114]

[0115] Where, f is the frequency of the sampling signal;

[0116] Step 4: Perform wavelet transform on the obtained terahertz detection signal: In order to extract the local feature information of each interface of the test piece 8 in the detection signal, enhance the signal characteristic and discrimination at the defect, and realize the accurate positioning of the defect in the depth direction, the obtained terahertz detection signal is processed by continuous wavelet transform. When analyzing non-stationary signals, wavelet transform can have good local feature characterization ability in both time domain and frequency domain. Wavelet transform refers to decomposing the original signal into different scales under the action of a set of wavelet basis functions, observing the general appearance of the signal in a large-scale space, and observing the details of the signal in a small-scale space, which can better observe the instantaneous characteristics of the signal;

[0117] The time domain signal S(t) in the z direction at each scanning point on the two-dimensional scanning plane is expressed as:

[0118] S(t)=Aexp(j2πft)

[0119] Where S(t) is the time domain signal in the z direction at each scanning point on the two-dimensional scanning plane xy, and A is the signal amplitude;

[0120] The wavelet transform of the time domain signal S(t) is expressed as:

[0121]

[0122] Among them, W(a,b) is the continuous wavelet coefficient, a is the scale factor, ψ(t) is the conjugate function of the continuous wavelet basis function, and b is the time factor;

[0123] The conjugate function ψ(t) of the continuous wavelet basis function must satisfy the following conditions:

[0124]

[0125] in, is the inverse Fourier transform of ψ(t), and ω is the angular frequency;

[0126] For continuous wavelet transform, S(t) is projected into a two-dimensional time-frequency space to form a new time moment with a frequency of The component information W(a,b) at the time, the scale factor a and the time translation factor b respectively determine the center and time domain width of the continuous wavelet basis function, and under the premise of the selection of the wavelet basis function, the window shape of the wavelet window function is related to the scale factor a. The larger a is, the smaller the width of the frequency window will be, and the window shape is also adaptive. When studying high-frequency signals, the width of the time window will decrease, thereby improving the time resolution; when studying low-frequency signals, the width of the time window will increase, thereby improving the frequency resolution, so as to achieve the purpose of accurately extracting detailed information in the signal;

[0127] When the terahertz detection signal is processed by continuous wavelet transform, the second-order derivative of the Gaussian wavelet basis is used as the wavelet basis function and the wavelet combination with a scale of 2. The scale of 2 is selected to obtain a smaller time window width and improve the time resolution. The advantage of this is that the signal characteristic peak can be well maintained, and on the other hand, the interference of noise can be suppressed;

[0128] Step 5: Perform empirical mode decomposition on the continuous wavelet coefficients. In order to maximize the characteristic of the detection signal and improve the longitudinal resolution of the detection signal, perform empirical mode decomposition on the obtained continuous wavelet coefficients W(a,b);

[0129] Empirical mode decomposition is an adaptive time-frequency analysis method that can decompose the continuous wavelet coefficients W(a, b) according to the scale characteristics of the signal into a series of intrinsic mode functions (IMFs) and residual functions containing different oscillation modes. Different orders of eigenmode functions contain local characteristic signals of different time scales. Lower eigenmode functions contain more energy and frequency information in the signal before decomposition. Each order of eigenmode functions needs to meet the following two basic conditions:

[0130] (1) The number of local extreme points and zero crossing points of the eigenmode function must be equal or differ by at most one in the entire time range;

[0131] (2) At any time, the average value of the envelope of the local maximum and the envelope of the local minimum must be zero;

[0132] The continuous wavelet coefficient W(a,b) is expressed as follows after empirical mode decomposition:

[0133]

[0134] Among them, IMF k (t) is the kth order eigenmode function component, r N (t) is the residual function;

[0135] The specific process of empirical mode decomposition of continuous wavelet coefficients is as follows:

[0136] (1) Find all local maximum and local minimum values ​​of the continuous wavelet coefficients obtained after the detection signal is transformed by wavelet in the whole time range;

[0137] (2) Connect all local maxima as the upper envelope, connect all local minima as the lower envelope, and calculate the average value of the upper and lower envelopes, which is denoted as m1(t);

[0138] (3) Extract the new data sequence and calculate it by the following formula:

[0139] h1(t)=W(a,b)-m1(t)

[0140] Among them, h1(t) is the new data sequence;

[0141] (4) Iterate the new data sequence h1(t) and repeat steps (2), (3) and (4) until h1(t) satisfies the two basic conditions of the intrinsic mode function. At this time, the first-order intrinsic mode function IMF1(t) = h1(t);

[0142] (5) Calculate the first-order residual function. The calculation of the first-order residual function is expressed as:

[0143] r1(t)=W(a,b)-IMF1(t)

[0144] Among them, r1(t) is the first-order residual function;

[0145] Repeat steps (1), (2), (3) and (4) to obtain the eigenmode functions of the remaining orders;

[0146] When the residual function r N When (t) becomes a monotonic function or a constant, the above iterative solution process ends.

[0147] After the above-mentioned empirical mode analysis is completed, the first-order intrinsic mode function IMF contains the most original signal energy information and frequency information, and the time resolution Δt of the terahertz time-domain detection signal is improved compared with the original signal. The entire wave packet corresponding to the original terahertz echo time-domain signal of the internal defect of the test piece 8 is decomposed into two peaks corresponding to the upper and lower surfaces of the defect after wavelet transformation and empirical mode analysis. The time information corresponding to the first peak of the defect is obtained through the first-order intrinsic mode function IMF corresponding to the defects at different positions in the internal direction of the test piece 8. The time information corresponding to the first peak of the defect corresponding to the first-order intrinsic mode function IMF is converted into distance information to locate the defect. The conversion of the two is expressed as:

[0148]

[0149] Wherein, d1 is the depth of the internal defect of the test piece 8 under test, Δt1 is the time corresponding to the first peak value of the defect corresponding to the first-order intrinsic mode function IMF; the characteristics of the upper and lower surfaces of the internal defect of the test piece 8 under test corresponding to the first-order intrinsic mode function IMF are the peak values ​​of two wave packets decomposed from one wave packet at the internal defect of the original data. Relative to one wave packet of the original data, the above processing can decompose the upper and lower surfaces of the internal defect of the test piece 8 under test, which greatly improves the longitudinal resolution of the terahertz detection signal;

[0150] Step 6: Visualization of 3D imaging: The task of 3D imaging is to reconstruct the scattering coefficient distribution δ(x, y, k) of the target based on the measured terahertz detection signal s(x, y, k), and give the 3D scattering distribution map of the target;

[0151] Based on the above processing, the first-order intrinsic mode function IMF in the z direction at each scanning point on the two-dimensional scanning plane is transformed by Fourier transform to obtain the two-dimensional scanning plane frequency domain signal s'(x, y, k);

[0152] Based on the converted two-dimensional scanning plane frequency domain signal s'(x, y, k), the main steps of three-dimensional imaging algorithm data processing are:

[0153] (1) Two-dimensional Fourier transform in the spatial domain, the processed frequency domain signal s'(x, y, k) is expanded along the transverse measurement plane to obtain the component value S(k) of the plane wave propagating in any direction on the measurement plane z = R0 x ,k y ,k), the frequency domain signal s'(x,y,k) is transformed into two-dimensional Fourier transform in the spatial domain with respect to the azimuth xy:

[0154]

[0155] Among them, S(kx ,k y , k) is the component of the terahertz plane wave in the z direction, L x and L y are the scanning range, k x and k y are the wave number components in the x and y directions;

[0156] (2) Frequency domain filtering. Since there is a certain distance z = R0 between the scanning plane of the self-transmitting and self-receiving horn antenna and the target, it is necessary to perform frequency domain filtering on the spatial domain transformation data to obtain the component S' (k x ,k y ,k), the processing process is expressed as:

[0157]

[0158] Wherein, the spatial wave number k = 4πf / c, f is the transmission signal frequency, c is the propagation speed of electromagnetic waves in the medium, R0 represents the vertical distance between the array plane and the plane of the test piece 8 under test, i.e., the lift-off, k x , k y , k z are the wave number components in the x, y, and z directions respectively;

[0159] (3) Stolt interpolation, by replacing the variables, we can obtain the complete spectrum S'(k x ,k y ,k z ), the processing process is expressed as:

[0160]

[0161] (4) Perform three-dimensional inverse Fourier transform on the data, and superimpose the plane waves in all propagation directions to obtain the three-dimensional scattering intensity distribution δ(x, y, z) of the target, and obtain a three-dimensional image. The processing process is expressed as:

[0162]

[0163] The processing results and imaging are as follows Figure 5-Figure 9 As shown, Figure 5 It is the original time domain signal at the defect of the tested piece and the first-order and second-order intrinsic mode functions IMF. It can be seen from the first-order and second-order intrinsic mode functions IMF that the first-order intrinsic mode function IMF contains the most original signal energy information and frequency information. Figure 6 It is the original time domain signal at the defect of the tested piece. There is a “combined peak” problem in the original time domain signal. Figure 7It is the first-order intrinsic mode function IMF, which contains the most original signal energy information and frequency information. Therefore, the first-order intrinsic mode function IMF is used as the imaging parameter and three-dimensional image reconstruction is performed. Figure 8 It is a three-dimensional slice image of the original time domain signal of the defect position in the depth direction of the test piece, and the defect positioning and imaging effects are poor. Fig. 9 This is a three-dimensional slice image of the first-order intrinsic mode function IMF of the defect position in the depth direction of the test piece. It can be seen that the method of the present invention has significantly improved the longitudinal resolution and clarity of terahertz detection of non-metallic internal defects, and also makes the positioning of non-metallic internal defects more accurate.

[0164] The continuous terahertz wave imaging quantitative detection system and method for internal defects of multi-layer non-metallic structures of the present invention solves the problem of insufficient longitudinal resolution in the existing terahertz imaging non-destructive testing field. The processed images are significantly improved in longitudinal resolution and clarity, which not only makes the positioning of non-metallic internal defects more accurate, but also provides a new method for terahertz tomography imaging and terahertz non-destructive testing application research.

Claims

1. A continuous terahertz wave imaging quantitative detection system for internal defects of multi-layer non-metallic structures, characterized in that: The system comprises a continuous wave linear frequency modulation signal source (1), a system control and processing computer (2), a scanning frame controller (3), a four-dimensional scanning frame (4), a spectrum expansion module (5), a terahertz transceiver probe (6), a focusing lens (7), a test piece to be tested (8) and a lifting platform (9); The continuous wave linear frequency modulation signal source (1) is electrically connected to the system control and processing computer (2) and the spectrum expansion module (5), and is used to transmit a linear frequency modulation continuous wave to the spectrum expansion module (5), and collect the terahertz detection signal of the test piece under test, and transmit it to the system control and processing computer (2); The scanning frame controller (3) is electrically connected to the system control and processing computer (2) and the four-dimensional scanning frame (4), and controls the movement of the four-dimensional scanning frame (4) through the system control and processing computer (2); The four-dimensional scanning frame (4) comprises an x-axis scanning axis (10), a y-axis scanning axis (11), a z-axis scanning axis (12) connected in sequence, and a P-axis rotating axis (13) arranged on the z-axis scanning axis (12), and the spectrum expansion module (5) is arranged on the P-axis rotating axis (13); the four-dimensional scanning frame (4) is controlled to move by a system control and processing computer (2), and the terahertz transceiver probe (6) arranged on the spectrum expansion module (5) is moved row by row or column by column at equal intervals to form a scanning surface; The spectrum expansion module (5) is connected to the terahertz transceiver probe (6) to multiply the frequency of the signal output by the continuous wave linear frequency modulation signal source (1) to a higher frequency required for detection; The terahertz transceiver probe (6) is composed of a self-transmitting and self-receiving horn antenna, and is used to transmit and receive terahertz detection signals to detect the test piece (8); The focusing lens (7) is mechanically connected to the terahertz transceiver probe (6) and is used to focus the terahertz transmission signal beam onto a focal plane; The lifting platform (9) is mechanically connected to the test piece (8) to fix the test piece (8) and to adjust the height of the test piece (8).

2. The continuous terahertz wave imaging quantitative detection system for internal defects of a multi-layer non-metallic structure according to claim 1, characterized in that: Uniform sampling of sampling plane data is ensured by controlling the moving speed of the terahertz transceiver probe (6) and controlling the sampling frequency of the spectrum expansion module (5).

3. The continuous terahertz wave imaging quantitative detection system for internal defects of a multi-layer non-metallic structure according to claim 1, characterized in that: The optimal distance between the tested test piece (8) and the terahertz transceiver probe (6) is determined by the focal length of the focusing lens (7).

4. A detection method of a continuous terahertz wave imaging quantitative detection system for internal defects of a multi-layer non-metallic structure according to any one of claims 1 to 3, characterized in that: The steps include: Step 1: Determine the detection parameters: The detection parameters include the x-direction scanning step length Δx, the y-direction scanning step length Δy, and the x-direction scanning range L x ,,y-direction scanning step length L y and a lift-off distance R0, and determining a detection frequency band according to material properties of the test piece (8) to be detected; Step 2: experimentally testing the test piece: collecting the detection signal of the test piece (8) through a continuous terahertz wave imaging quantitative detection system; controlling the four-dimensional scanning frame (4), determining the x-direction scanning step length Δx and the y-direction scanning step length Δy, and realizing two-dimensional spatial sampling of the detection signal in a row-by-row or column-by-column scanning manner, thereby completing data collection of the test piece (8); Step 3: Collecting terahertz detection signals: The continuous wave linear frequency modulation signal source (1) of the continuous terahertz wave imaging quantitative detection system generates a fast sawtooth low-frequency scanning signal, which is amplified by a frequency multiplier as a transmission signal, wherein a part of the transmission signal is radiated into the air through the terahertz transceiver probe (6), and the other part directly enters the spectrum spreading module as a reference signal. The transmission signal radiated into the air is converged on the focal plane through a focusing lens (7). When the transmission signal radiated into the air encounters an interface between different media, it returns along the original path and then enters the spectrum spreading module (5) as a receiving signal. The transmission signal and the receiving signal are frequency-converted in the spectrum spreading module (5), and finally, an intermediate frequency signal is output; The transmitted signal and received signal are expressed as: Among them, S ts (t) is the transmitted signal, S rs (t) is the received signal, A t is the amplitude of the transmitted signal, A r is the amplitude of the received signal, f0 is the initial frequency, K is the frequency modulation slope, expressed as B / t s , B is the bandwidth of the system, t s is the frequency sweep time, Δt is the delay time; t is the time, j is the imaginary unit; The output intermediate frequency signal is expressed as: Among them, S if (t) is the output intermediate frequency signal; when the continuous terahertz wave imaging quantitative detection system detects the test piece (8), when the detection signal interacts with the internal defects of the test piece (8), the positions of different layers of the internal defects correspond to different delay times Δt, and the delay time Δt increases as the thickness of the internal defects increases, and at the same time, the amplitude of the reflected signal changes as the thickness of the internal defects increases; The data obtained by the continuous terahertz wave imaging quantitative detection system is the sum of cosine functions at different frequencies. Each frequency component corresponds to a different defect interface. The obtained data is subjected to Fourier transformation to obtain the relationship between the frequency modulation bandwidth and the internal defect position of the test piece (8) under inspection, which is expressed as follows: Wherein, d is the position distance of the test piece (8) under test, c is the speed of light, Δf is the frequency modulation bandwidth, and n is the refractive index of the test piece (8) under test within the working frequency band of the continuous terahertz wave imaging quantitative detection system. According to the above principle, the continuous terahertz wave imaging quantitative detection system is used to scan the test piece (8) under test point by point, and finally a terahertz detection signal obtained by sampling the entire test piece (8) under test is obtained; The terahertz detection signal obtained by two-dimensional scanning plane sampling is represented by s(x, y, k), which contains three variables. x, y, and k represent the modulus of the wave number vector corresponding to the sampling point on the x-axis coordinate, the y-axis coordinate, and the sampling signal. The modulus of the wave number vector corresponding to the sampling signal is proportional to the frequency f of the sampling signal. The relationship between the two is expressed as: Where, f is the frequency of the sampling signal; Step 4: performing wavelet transform on the obtained terahertz detection signal: in order to extract the local characteristic information of each interface of the test piece (8) in the detection signal, enhance the signal characteristic and discrimination degree at the defect, and realize the accurate positioning of the defect in the depth direction, the obtained terahertz detection signal is subjected to continuous wavelet transform processing; The time domain signal S(t) in the z direction at each scanning point on the two-dimensional scanning plane is expressed as: S(t)=Aexp(j2πft) Where S(t) is the time domain signal in the z direction at each scanning point on the two-dimensional scanning plane xy, and A is the signal amplitude; The wavelet transform of the time domain signal S(t) is expressed as: Among them, W(a,b) is the continuous wavelet coefficient, a is the scale factor, ψ(t) is the conjugate function of the continuous wavelet basis function, and b is the time factor; The conjugate function ψ(t) of the continuous wavelet basis function must satisfy the following conditions: in, is the inverse Fourier transform of ψ(t), and ω is the angular frequency; For continuous wavelet transform, S(t) is projected into a two-dimensional time-frequency space to form a new time moment with a frequency of The component information W(a,b) at the time, the scale factor a and the time translation factor b respectively determine the center and time domain width of the continuous wavelet basis function, and under the premise of the selection of the wavelet basis function, the window shape of the wavelet window function is related to the scale factor a. The larger a is, the smaller the width of the frequency window will be, and the window shape is also adaptive; Step 5: Perform empirical mode decomposition on the continuous wavelet coefficients: In order to maximize the characteristic of the detection signal and improve the longitudinal resolution of the detection signal, perform empirical mode decomposition on the obtained continuous wavelet coefficients W(a,b); Empirical mode decomposition is an adaptive time-frequency analysis method that can decompose the continuous wavelet coefficients W(a,b) according to the scale characteristics of the signal into a series of intrinsic mode functions IMF and residual functions containing different oscillation modes. Different orders of eigenmode functions contain local characteristic signals of different time scales. Lower eigenmode functions contain more energy and frequency information in the signal before decomposition. Each order of eigenmode functions needs to meet the following two basic conditions: (1) The number of local extreme points and zero crossing points of the eigenmode function must be equal or differ by at most one in the entire time range; (2) At any time, the average value of the envelope of the local maximum and the envelope of the local minimum must be zero; The continuous wavelet coefficient W(a,b) is expressed as follows after empirical mode decomposition: Among them, IMF k (t) is the kth order eigenmode function component, r N (t) is the residual function; After the empirical mode analysis is completed, the first-order intrinsic mode function IMF contains the most original signal energy information and frequency information, and the time resolution Δt of the terahertz time domain detection signal is improved compared with the original signal. After the whole wave packet corresponding to the original terahertz echo time domain signal of the internal defect of the test piece (8) is processed by wavelet transform and empirical mode analysis, two peaks corresponding to the upper and lower surfaces of the defect are decomposed. The time information corresponding to the first peak of the defect is obtained by the first-order intrinsic mode function IMF corresponding to the defect at different positions in the internal direction of the test piece (8). The time information corresponding to the first peak of the defect corresponding to the first-order intrinsic mode function IMF is converted into distance information, so as to locate the defect. The conversion between the two is expressed as: Wherein, d1 is the depth of the internal defect of the test piece (8) under test, Δt1 is the time corresponding to the first peak value of the defect corresponding to the first-order intrinsic mode function IMF; the upper and lower surfaces of the internal defect of the test piece (8) under test corresponding to the first-order intrinsic mode function IMF are characterized by the peak values ​​of two wave packets decomposed from one wave packet at the internal defect of the original data. With respect to one wave packet of the original data, the above processing can decompose the upper and lower surfaces of the internal defect of the test piece (8) under test, thereby greatly improving the longitudinal resolution of the terahertz detection signal; Step 6: Visualization of 3D imaging: The task of 3D imaging is to reconstruct the scattering coefficient distribution δ(x, y, k) of the target based on the measured terahertz detection signal s(x, y, k), and give the 3D scattering distribution map of the target; Based on the first-order intrinsic mode function IMF in the z direction at each scanning point on the obtained two-dimensional scanning plane, the frequency domain signal s'(x, y, k) of the two-dimensional scanning plane is obtained by Fourier transformation; Based on the converted two-dimensional scanning plane frequency domain signal s'(x, y, k), the main steps of three-dimensional imaging algorithm data processing are: (1) Two-dimensional Fourier transform in the spatial domain, the processed frequency domain signal s'(x, y, k) is expanded along the transverse measurement plane to obtain the component value S(k) of the plane wave propagating in any direction on the measurement plane z = R0 x ,k y ,k), the frequency domain signal s'(x,y,k) is transformed into two-dimensional Fourier transform in the spatial domain with respect to the azimuth xy: Among them, S(k x ,k y , k) is the component of the terahertz plane wave in the z direction, L x and L y are the scanning range, k x and k y are the wave number components in the x and y directions; (2) Frequency domain filtering. Since there is a certain distance z = R0 between the scanning plane of the self-transmitting and self-receiving horn antenna and the target, it is necessary to perform frequency domain filtering on the spatial domain transformation data to obtain the component S' (k x ,k y ,k), the processing process is expressed as: Wherein, the spatial wave number k = 4πf / c, f is the transmission signal frequency, c is the electromagnetic wave propagation speed in the medium, R0 represents the vertical distance between the array plane and the plane of the test piece (8) under test, i.e., the lift-off, k x , k y , k z are the wave number components in the x, y, and z directions respectively; (3) Stolt interpolation, by replacing the variables, we can obtain the complete spectrum S'(k x ,k y ,k z ), the processing process is expressed as: (4) Perform three-dimensional inverse Fourier transform on the data, and superimpose the plane waves in all propagation directions to obtain the three-dimensional scattering intensity distribution δ(x, y, z) of the target, and obtain a three-dimensional image. The processing process is expressed as:

5. The detection method according to claim 4, characterized in that: When the terahertz detection signal is processed by continuous wavelet transform in step 4, the second-order derivative of the Gaussian wavelet basis is used as the wavelet basis function and the wavelet combination with a scale of 2.

6. The detection method according to claim 4, characterized in that: The specific process of empirical mode decomposition continuous wavelet coefficients in step 5 is as follows: (1) Find all local maximum and local minimum values ​​of the continuous wavelet coefficients obtained after the detection signal is transformed by wavelet in the whole time range; (2) Connect all local maxima as the upper envelope, connect all local minima as the lower envelope, and calculate the average value of the upper and lower envelopes, which is denoted as m1(t); (3) Extract the new data sequence and calculate it by the following formula: h1(t)=W(a,b)-m1(t) Among them, h1(t) is the new data sequence; (4) Iterate the new data sequence h1(t) and repeat steps (2), (3) and (4) until h1(t) satisfies the two basic conditions of the intrinsic mode function. At this time, the first-order intrinsic mode function IMF1(t) = h1(t); (5) Calculate the first-order residual function. The calculation of the first-order residual function is expressed as: r1(t)=W(a,b)-IMF1(t) Among them, r1(t) is the first-order residual function; Repeat steps (1), (2), (3) and (4) to obtain the eigenmode functions of the remaining orders; When the residual function r N When (t) becomes a monotonic function or a constant, the above iterative solution process ends.

Images