A Full-Matrix Ultrasonic Imaging Method for Defects in Curved Surface Parts Based on Frequency Domain Elastic Reverse Time Migration
By reconstructing the sound source and received wave field using the frequency domain elastic reverse time migration method, the problem of high precision and high efficiency in detecting internal defects of complex surface parts was solved, and high-precision imaging of curved surface parts was realized.
Patent Information
- Application Number
- CN202310715304.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-06-16
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-06-16
AI Technical Summary
Existing technologies suffer from high computational complexity, low accuracy, and long computation time when detecting internal defects in parts with complex surface shapes. In particular, traditional ray tracing methods and phase-shift-based methods are not accurate enough when imaging curved parts, while time-domain reverse time-shifting methods take too long to compute.
A frequency-domain elastic reverse time migration method is adopted to reconstruct the sound source and receiver wave fields through the fully elastic wave equation in the frequency domain. Combined with Helmholtz decomposition and LU decomposition, efficient imaging of P-wave and S-wave mode conversion is achieved.
It achieves high-precision imaging of internal defects in curved parts, improving imaging accuracy by 2-36 times and reducing computation time to 1/14 of ERTM, significantly improving detection efficiency.
Smart Images

Figure SMS_1 
Figure SMS_4 
Figure SMS_5
Abstract
Description
Technical Field
[0001] This invention belongs to the field of detection method technology, specifically relating to an ultrasonic full-matrix imaging method for defects in curved parts based on frequency domain elastic reverse time offset. Background Technology
[0002] Critical components in the aerospace, nuclear, and energy industries often have complex surface shapes. Due to complex manufacturing processes and harsh operating conditions, they are prone to internal cracks and defects, which significantly weaken their mechanical properties and lead to premature failure. Information on the location, shape, and size of these defects can be used to assess product failure modes and remaining service life; therefore, high-precision non-destructive testing (NDT) of internal defects is of great importance. Ultrasonic testing, with its advantages of deep penetration and ease of use, has been widely applied to the measurement of internal defects in various components.
[0003] Ultrasonic phased array technology is a mature technology with advantages such as portability, reliability, and high precision. Phased array probes can perform full matrix acquisition (FMC), and FMC datasets contain complete information about the measurement area, offering higher imaging capabilities and resolution than other data acquisition techniques. Imaging methods for FMC datasets have recently become a research hotspot, with extensive studies conducted on the detection of internal defects in planar parts. With the increasing demand for inspecting curved products, corresponding shaped wedges have been manufactured to assist phased array probes in measuring products during their use. Here, an acoustic parameter model of the stacked shape is formed, which poses a challenge to high-precision imaging of FMC datasets.
[0004] Imaging methods for the FMC dataset can be categorized into three types: time-delay-based methods, phase-shift-based methods, and wave equation-based methods. Time-delay-based methods use rays to simulate ultrasonic propagation, reconstructing the wavefield by time-delaying the FMC data, and finally obtaining the imaging result through superposition. Total Focusing Method (TFM) is widely accepted as the standard method for ultrasonic imaging. Introducing ray tracing into TFM allows it to be adapted for the inspection of curved products. Ray tracing uses Fermat's principle to determine the ultrasonic propagation path in layered structures. However, the computational complexity of ray-tracing-based TFM increases exponentially with the number of layers, while the imaging accuracy decreases with increasing layers. For phase-shift-based methods, the wavefield in the measurement region is reconstructed through phase shift offset (PSM) in the frequency-wavenumber (fk) domain. This method can easily adapt to horizontally layered structures and shows significant advantages in computational efficiency. However, when imaging curved parts, PSM-based methods use a first-order approximation of the wave equation in the fk domain to extrapolate the wavefield, which leads to a significant decrease in accuracy when the horizontal direction of the sound velocity model changes significantly. Therefore, phase-shift-based methods are more suitable for detecting non-metallic bent products, and their measurement accuracy may not meet the requirements for hypersonic metallic products.
[0005] Methods based on time delay and phase shift simplify the second-order wave equation to a first-order unidirectional wave equation. This equation fails to account for multiple echoes at the scatterer and reflector interfaces, leading to background artifacts and a reduced signal-to-noise ratio. Methods based on the wave equation directly reconstruct the wave field of the FMC dataset using the full wave equation, enabling high-precision modeling of multiple scattering and reflection during sound propagation. Reverse-time migration (RTM) reconstructs the propagation of longitudinal waves in solid media using the acoustic wave equation and is not limited to sound velocity models, making it well-suited for detecting curved products. Grohmann used RTM to image the interface shape of building structures, and Liu introduced RTM to detect voids in concrete and steel-concrete composite structures. However, traditional RTM methods reconstruct the wave field in the time domain, typically requiring long computation times. To address this, Ji proposed a frequency-domain reverse-time migration method to achieve efficient and high-precision measurement of internal defects in complex surface products. In fact, RTM uses an acoustic propagation model to approximate the propagation of ultrasonic signals in solid media, but it cannot model the propagation of transverse waves or the conversion between transverse and longitudinal waves. The elastic wave equation can accurately simulate wave propagation in solids, leading geophysicists to propose elastic reverse time migration (ERTM) for more precise imaging results. Rao introduced ERTM to process the FMC dataset for imaging irregularly shaped notches and further combined it with point cloud reconstruction to detect internal defects in parts with vertical surfaces with high accuracy; Liu used ERTM to inspect grouting splice sleeves in precast concrete structures. However, ERTM is implemented in the time domain and requires iterative calculations at each time point, where the time step should be short enough to ensure computational accuracy. Therefore, the computation time of ERTM is typically very long, severely limiting its applications. Summary of the Invention
[0006] To address the problems existing in the prior art, this invention provides an ultrasonic full-matrix imaging method for defects in curved parts based on frequency domain elastic reverse time migration. This method considers multiple echoes and mode conversion between longitudinal and transverse waves, achieving higher accuracy compared to traditional methods. At the same time, it significantly reduces computation time compared to the time domain elastic reverse time migration method, making it a new full-matrix imaging method that combines accuracy and efficiency.
[0007] An ultrasonic full-matrix imaging method for defects in curved parts based on frequency-domain elastic reverse-time migration includes the following steps:
[0008] (1) Input the sound source signal, the parameter model of the curved part, and the measured full matrix data;
[0009] (2) Discretize the parameter model to obtain the impedance matrix; preprocess the source signal and the full matrix data to obtain the source matrix and the receiver matrix respectively;
[0010] (3) Take any untraversed frequency as the current frequency, and use the LU decomposition of the impedance matrix to process the source matrix and receiver matrix of the current frequency respectively to obtain the source wave field and receiver wave field of the current frequency.
[0011] (4) After reshaping the sound source wavefield and the received wavefield at the current frequency, apply imaging conditions to obtain the imaging result corresponding to the current frequency.
[0012] (5) Traverse all discrete frequencies and superimpose the imaging results corresponding to all discrete frequencies to obtain the imaging results of internal defects of curved parts.
[0013] Preferably, in step (1), the parameter model includes Lame parameters λ(x,z) and μ(x,z), and density parameter ρ(x,z).
[0014] Preferably, in step (2), the preprocessing of the sound source signal is as follows:
[0015] The sound source signal is converted to the frequency domain through Fourier transform and then distributed to each discretized node to obtain the initial sound source matrix.
[0016] Preferably, in step (2), the preprocessing procedure for the full matrix data is as follows:
[0017] After removing the first arrival wave from the full matrix data using a window function, the data is transformed to the frequency domain using a Fourier transform and the conjugate is taken to obtain the inverse-time data in the frequency domain. The inverse-time data is then distributed to each discretized node to obtain the initial receiving matrix.
[0018] Preferably, in step (4), the current frequency ω is... j The corresponding imaging conditions for the reconstructed source wavefield and receiver wavefield are as follows:
[0019]
[0020] in, Represents the current frequency ω j The corresponding imaging results; j∈[1, N] ω ], N ω Indicates the number of discrete frequencies; i∈[1,M], where M represents the number of oscillators; * indicates taking the complex conjugate; These represent the current frequency ω. j The corresponding scalar wavefield is extracted from the reconstructed source and receiver wavefields using Helmholtz decomposition, and the specific expression is as follows:
[0021]
[0022]
[0023] In the formula, • Represents the divergence operator; Represents the current frequency ω j The corresponding reshaped sound source wave field; Represents the current frequency ω j The corresponding reshaped receiving wave field.
[0024] Preferably, in step (5), the calculation formula for superimposing the imaging results corresponding to all discrete frequencies is as follows:
[0025]
[0026] In the formula, Im(x,z) represents the imaging result of internal defects in the curved surface part.
[0027] In step (3), the theoretical formulas for processing the source matrix and receiver matrix using the LU decomposition of the impedance matrix are derived as follows:
[0028] Materials in a solid medium are simultaneously subjected to normal stress and shear stress, generating longitudinal and transverse waves. For an isotropic solid medium, the two-dimensional elastic wave equation in the frequency domain is:
[0029]
[0030] Where i represents the imaginary unit; x and z are the horizontal and vertical coordinates of the two-dimensional spatial field, respectively; ω is the frequency; u = (u x ,u z ) represents the vibration velocity vector; σ xx σ zz and σ xz There are three stress components; f = (f x ,f z ) is the sound source vector; λ and μ are the Lame parameters of the solid medium; ρ is the density of the solid medium; where u and f are functions of x, z and ω.
[0031] After introducing perfectly matched (PML) boundary conditions in the x and z directions, equation (1) can be written as:
[0032]
[0033] in:
[0034]
[0035] Where, x i and z iThese represent the horizontal and vertical distances between nodes inside the Perfect Match Layer (PML) and the inside of the PML boundary, respectively; a0 is a scalar, typically set to 1.79; L PML f0 is the thickness of the PML layer; f0 is the center frequency of the sound source.
[0036] In engineering practice, the velocity wave field u can be calculated using the finite difference method, where the wave field is discretized as follows: Figure 2 As shown in (a), the entire measurement area is discretized into N = N x ×N z The staggered leapfrog structure of discrete meshes with a certain number of nodes has been well developed in the finite-difference time domain, ensuring accurate derivative estimation and good wave propagation dispersion performance. Therefore, a staggered mesh strategy is proposed to solve the velocity wave field, with structural details as follows: Figure 2 As shown in (b), two coordinate systems are established in space, with (x′, z′) rotated by 45° relative to the original coordinate system (x, z). The finite difference operator in space... Viewed as the superposition of corresponding finite difference operators in two coordinate systems, it can be written as:
[0037]
[0038] in, and , respectively, are operators in the two coordinate systems (x,z) and (x′,z′); a is an average coefficient used to account for the coupling between nodes on the Cartesian mesh and the rotated mesh.
[0039] The velocity wave field is discretized at integer nodes, and the parameters are discretized at intermediate nodes. The discretized form of the wave equation can be written as:
[0040] A m-1,n-1 u m-1,n-1 (ω)+B m-1,n u m-1,n (ω)+C m-1,n+1 u m-1,n+1 (ω)+
[0041] D m,n-1 u m,n-1 (ω)+E m,n (ω)u m,n (ω)+F m,n+1 u m,n+1 (ω)+
[0042] G m+1,n-1 u m+1,n-1 +H m+1,n u m+1,n (ω)+J m+1,n+1 u m+1,n+1 (ω)+
[0043] f m,n (ω)=0 (5)
[0044] Among them, u m,n (ω) is a 2×1 velocity vector at node (m,n) and frequency ω, A m-1,n-1 ~J m+1,n+1 It is a series of 2×2 matrices.
[0045] Therefore, each discrete point has two equations, and a large sparse system of equations across the entire computational grid can be written as:
[0046]
[0047] Where K(ω) is a 2N-order square matrix, and Given a 2N-dimensional vector, K(ω) contains the horizontal and vertical components of the vibration velocities at all nodes. Therefore, the elastic wave field can be modeled by inverting the LU decomposition of K(ω) frequency-by-frequency.
[0048] When a phased array probe is used to inspect a part, each element is excited sequentially, and all elements simultaneously receive ultrasonic signals, resulting in a full matrix dataset. An element can be viewed as a piston model vibrating only in the z-direction; therefore, the element signal is considered a measurement of the vibration velocity in the z-direction at the contact point on the surface.
[0049] Therefore, forward modeling of full matrix data (FMC) can be written as:
[0050]
[0051] in, Let be the sound source wave field when the i-th vibration element is excited; Let K(ω) be the received wave field in the measurement area after the excitation of the i-th element; M is the number of elements in the phased array. Therefore, all M vibration velocity fields at frequency ω can be solved simultaneously by the inversion of matrix K(ω). In practical applications, formula (7) can be efficiently solved by LU decomposition of matrix K(ω).
[0052] In the ultrasonic full-matrix imaging method for defects in curved parts based on frequency domain elastic reverse time migration (FD-ERTM) of the present invention, the sound source signal s(x r ,x s The signal s(x,z) can be approximated by a suitable wavelet, whose spectral distribution should closely approximate the measured signal. Input parameters include Lamé parameters λ(x,z) and μ(x,z), and density parameter ρ(x,z), which are discretized to obtain the impedance matrix K. The source signal and FMC data can be considered as measurements of vibration velocity in the z-direction at surface nodes.r ,x s ,t) is transformed into the frequency domain by Fast Fourier Transform to obtain s(x) r ,x s ,ω j ), s(x r ,x s ,ω j The sound source matrix [S1(ω)] is obtained by assigning it to the corresponding node. j ),...,S M (ω j )], where S i (ω j ) represents the sound source vector at the i-th excitation (excitation of the i-th vibration element). During preprocessing of the full matrix data, the first arrival wave is first removed in the time domain using a window function. Then, the dataset is transformed to the frequency domain and its conjugate is taken to obtain the inverse-time data d(x) in the frequency domain. r ,x s ,ω j Then the signal d(x) r ,x s ,ω j ) is assigned to a specific node to obtain the receive matrix [D1(ω j ),...,D M (ω j )], where D i (ω j ) represents the received vector during the i-th excitation.
[0053] For discrete frequency ω j Impedance matrix K(ω) j The efficiency of inversion is enhanced by performing LU decomposition. Since the source matrix and receiver matrix correspond to the same impedance matrix, all excited source wave fields [U1(ω j ),...,U M (ω j )] and received wave field [V1(ω j ),...,V M (ω j The ability to reconstruct simultaneously in one go greatly improves the efficiency of the imaging method of this invention compared to the time-domain reverse migration method.
[0054] Reconstructed wave field U i (ω j ) and V i (ω j The values are 2N×1, which contain the vertical and horizontal components of the vibration velocity at all nodes, and can be reconstructed into N. x ×N z Wavefield matrix ×2 and and The direct cross-correlation is affected by mode conversion and sensitivity to errors in the parametric model. Therefore, the pressure wave component is extracted from the vector wave field through Helmholtz decomposition, written as:
[0055]
[0056]
[0057] in, • For divergence operators, and It is a scalar wave field.
[0058] Imaging conditions are defined as follows:
[0059]
[0060] Where * denotes the complex conjugate, N ω The number of discrete frequencies.
[0061] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0062] This invention presents a full-matrix ultrasonic imaging method for defects in curved parts based on frequency-domain elastic reverse time migration (FD-ERTM), used for high-precision imaging of internal defects in curved products. This imaging method reconstructs the source and receiver wavefields at each frequency using the fully elastic wave equation in the frequency domain to accurately simulate the propagation of ultrasound waves within an elastic body. Then, Holmholtz decomposition imaging conditions are used to process the wavefield, obtaining the longitudinal wave component and avoiding aliasing between longitudinal and transverse waves, thus extracting defects with high precision. Compared with traditional imaging methods, this method has significant advantages. Compared to TFM, this method recovers the full waveform and models mode transitions in solid media, improving imaging accuracy by 2-36 times. Furthermore, frequency selection and parallel computing reduce the computational complexity of the proposed method and reduce the time cost to 1 / 14 of ERTM. Attached Figure Description
[0063] Figure 1 This is a flowchart of an embodiment of the present invention;
[0064] Figure 2 In the middle: (a) is the discretized velocity field; (b) is a magnified view of the staggered mesh.
[0065] Figure 3(a) FMC data acquisition image of internal hole defects in part; (b) Imaging result of internal hole defects in part using the ray-based TFM imaging method; (c) Imaging result of internal hole defects in part using the TD-ERTM imaging method; (d) Imaging result of internal hole defects in part using the FD-ERTM imaging method; (e) FMC data acquisition image of internal crack defects in part; (f) Imaging result of internal crack defects in part using the ray-based TFM imaging method; (g) Imaging result of internal crack defects in part using the TD-ERTM imaging method; (h) Imaging result of internal crack defects in part using the FD-ERTM imaging method. Detailed Implementation
[0066] like Figure 1 As shown, a full-matrix ultrasonic imaging method for defects in curved parts based on frequency-domain elastic reverse-time migration includes the following steps:
[0067] 1. Input sound source signal s(x) r ,x s The parametric model of the curved surface part with internal defects (including Lame parameters λ(x,z) and μ(x,z), and density parameter ρ(x,z)) and the measured full matrix data d(x) are given. r ,x s ,t);
[0068] Where, x r and x s These correspond to the r-th receiving element and the s-th transmitting element, respectively, where s∈[1,M], r∈[1,M], M is the number of elements, and t represents time.
[0069] 2. Input data processing:
[0070] 2.1 Discretize the parameter model to obtain the impedance matrix K;
[0071] 2.2 Convert the sound source signal s(x) r ,x s The source matrix [S1,...,S] is transformed to the frequency domain using Fourier transform and then distributed to each discretized node to obtain the initial source matrix [S1,...,S]. M ], where S i Let represent the sound source vector at the i-th excitation, i∈[1,M].
[0072] 2.3 Convert the full matrix data d(x) r ,x s After removing the first arrival wave using a window function, the signal is transformed to the frequency domain using a Fourier transform and its conjugate is taken to obtain the inverse-time data in the frequency domain. This inverse-time data is then distributed to each discretized node to obtain the initial receiver matrix [D1,...,D]. M], where D i Let represent the received vector during the i-th excitation.
[0073] 3. Take any untraversed frequency as the current frequency ω. j Using the impedance matrix K(ω) j The LU decomposition of the source matrix [S1(ω)] j ),...,S M (ω j )] and the receiver matrix [D1(ω j ),...,D M (ω j The sound source wavefield [U1(ω]] is processed separately to obtain the current frequency sound source wavefield. j ),...,U M (ω j )] and received wave field [V1(ω j ),...,V M (ω j )).
[0074] 4. The obtained sound source wave field U i (ω j ) and received wave field V i (ω j After reshaping, apply imaging conditions to obtain the imaging result corresponding to the current frequency;
[0075] The imaging conditions are:
[0076]
[0077] in, Represents the current frequency ω j The corresponding imaging results; j∈[1, N] ω ], N ω Indicates the number of discrete frequencies; i∈[1,M], where M represents the number of oscillators; * indicates taking the complex conjugate; These represent the current frequency ω. j The corresponding scalar wave field is obtained from the reconstructed sound source wave field through Helmholtz decomposition. and receiving wave field The extracted value is as follows:
[0078]
[0079]
[0080] In the formula, • Represents the divergence operator.
[0081] 5. Traverse all discrete frequencies and superimpose the imaging results corresponding to all discrete frequencies to obtain the imaging results of internal defects of the curved surface part:
[0082]
[0083] In the formula, Im(x,z) represents the imaging result of internal defects in the curved surface part.
[0084] Figure 1 middle,
[0085] Detection experiment
[0086] Experiments were conducted to verify the effectiveness of the method of the present invention in imaging defects in curved components.
[0087] like Figure 3 As shown in (a) and (e), a wedge made of acrylonitrile-butadiene-styrene (ABS) was used to assist in the measurement of an aluminum component with a radius of 80 mm. The component contained defects such as a hole (a) and a crack (e). A 64-element phased array probe (Shantou Ultrasonic) with a center frequency of 2.5 MHz and an element spacing of 0.75 mm was used in the experiment, along with a 64 / 64OEM-PA data acquisition card (AOS. Ltd, America) to achieve FMC data acquisition. The sampling frequency was 50 MHz, and the time range was 80 μs. A two-layer model was formed in the measurement area (ABS: L-wave velocity = 2200 m / s, S-wave velocity = 1100 m / s, density = 1150 kg / m³). 3 Aluminum: L-wave velocity = 63000 m / s, S-wave velocity = 3120 m / s, density = 2700 kg / m³ 3 ).
[0088] Three imaging methods were used to process the FMC dataset, including ray-based TFM, time-domain elastic wave reverse time migration (TD-ERTM), and the imaging method of this embodiment (FD-ERTM). Ray-based TFM uses a ray tracing method to calculate the time delay of each image pixel. For TD-ERTM and FD-ERTM, the discretization step size of the measurement region in both the x and z directions is 0.075 mm, with 13 grid points for each shortest wavelength. The time step size of TD-ERTM is 9.52 × 10⁻⁶. -9 The frequency step size of FD-ERTM is 12.5 kHz, determined by the FMC dataset itself, and the frequency band is selected as [1.5, 3] MHz.
[0089] The imaging results of the porous and cracked components using the three imaging methods are as follows: Figure 3 (b)-(d) and Figure 3As shown in (f)-(h), the enlarged envelope of the intermediate defect image is shown in the lower right corner of the corresponding figure.
[0090] For Experiment 1 (the part with the hole defect), the ray-based TFM method and FD-ERTM can image all three defects, such as Figure 3 (b) and (d). TD-ERTM failed to image the bottom defect (see...) Figure 3 (c) This is likely because TD-ERTM is more susceptible to numerical dispersion and accumulates too many errors in the deep region during wavefield extrapolation. Furthermore, the enlarged envelope plot shows the imaging results of FD-ERTM (see [image]). Figure 3 The morphology of pore defects in (d) is superior to that of other imaging methods.
[0091] For Experiment 2 (cracked component), the imaging results of TFM and FD-ERTM based on rays (corresponding to respectively) Figure 3 The crack defects in (f) and (h) are compared with the imaging results of TD-ERTM (corresponding to respectively) Figure 3 The middle (g) is much clearer.
[0092] The coordinates of the pore defects and the size and location information of the crack defects are also marked on the imaging results. The measurement errors are shown in Table 1.
[0093] Table 1. Measurement error, average API, and imaging time for different methods in the experiment.
[0094]
[0095]
[0096] In Table 1, for Experiment 1, the positional errors in the horizontal and vertical directions are expressed as E, respectively. x and E z As can be seen from Table 1, ERTM effectively improves positioning accuracy, while FD-ERTM has the smallest positioning errors in both the x and z directions, which are almost 1 / 5 and 1 / 2 of those of ray-based TFM, respectively.
[0097] To quantitatively analyze the imaging results, a dimensionless parameter API (array performance indicator) is introduced, as follows:
[0098] API = A -6dB / λ 2
[0099] Among them, A -6dB The area whose peak value exceeds -6 dB relative to the defect region image is defined as λ, where λ is the wavenumber corresponding to the center frequency. Clearly, a smaller API value implies a higher resolution.
[0100] Table 1 shows that the average API of the imaging results in Experiment 1 demonstrates that ERTM can effectively improve resolution. Although the API of TD-ERTM is smaller than that of FD-ERTM, the former only detected two defects, so only two defects are included in the API calculation.
[0101] For Experiment 2, E in Table 1 L E H E D and E F Corresponding to Figure 3 Dimensions L, H, D, and F in (f)-(h). E of FD-ERTM L E H E D and E F Further evidence demonstrates its high imaging accuracy, which is 1 / 26, 1 / 16, 1 / 24, and 1 / 35 of that of the X-ray-based TFM, and 1 / 4, 1 / 6, 1 / 30, and 1 / 1.25 of that of TD-ERTM. Furthermore, Table 1 lists the time consumption of the three methods in the two experiments. TD-ERTM has a time cost as high as 14 hours, while FD-ERTM reduces the computation time to approximately 1 hour. Therefore, FD-ERTM has significant advantages in both imaging accuracy and efficiency, improving imaging accuracy by 2-35 times compared to the X-ray-based TFM method and imaging efficiency by 14 times compared to the TD-ERTM method.
Claims
1. A method for ultrasonic full-matrix imaging of defects in curved parts based on frequency-domain elastic reverse-time migration, characterized in that, Includes the following steps: (1) Input the sound source signal, the parametric model of the curved part, and the measured full matrix data; (2) Discretize the parameter model to obtain the impedance matrix; preprocess the source signal and the full matrix data to obtain the source matrix and the receiver matrix respectively; (3) Take any untraversed frequency as the current frequency, and use the LU decomposition of the impedance matrix to process the source matrix and receiver matrix of the current frequency respectively to obtain the source wave field and receiver wave field of the current frequency. (4) After reshaping the sound source wave field and the received wave field at the current frequency, apply imaging conditions to obtain the imaging result corresponding to the current frequency; (5) Traverse all discrete frequencies and superimpose the imaging results corresponding to all discrete frequencies to obtain the imaging results of internal defects of curved parts; In step (1), the parameter model includes Lame parameters. and and density parameters ; In step (4), the current frequency is... The imaging conditions under the influence of the reconstructed source wavefield and the receiver wavefield are as follows: ; in, Indicates the current frequency The corresponding imaging results; j∈[1, ], Indicates the number of discrete frequencies; ∈[1, M], where M represents the number of oscillating elements; * indicates taking the complex conjugate; , These represent the current frequency. The corresponding scalar wavefield is extracted from the reconstructed source and receiver wavefields through Helmholtz decomposition, and the specific expression is as follows: ; ; In the formula, Represents the divergence operator; Indicates the current frequency The corresponding reshaped sound source wave field; Indicates the current frequency The corresponding reshaped receiving wave field.
2. The ultrasonic full-matrix imaging method for defects in curved parts based on frequency-domain elastic reverse-time migration according to claim 1, characterized in that, In step (2), the preprocessing of the sound source signal is as follows: The sound source signal is converted to the frequency domain through Fourier transform and then distributed to each discretized node to obtain the initial sound source matrix.
3. The ultrasonic full-matrix imaging method for defects in curved parts based on frequency-domain elastic reverse-time migration according to claim 1, characterized in that, In step (2), the preprocessing procedure for the full matrix data is as follows: After removing the first arrival wave from the full matrix data using a window function, the data is transformed to the frequency domain using a Fourier transform and the conjugate is taken to obtain the inverse-time data in the frequency domain. The inverse-time data is then distributed to each discretized node to obtain the initial receiving matrix.
Citation Information
Patent Citations
Full-matrix data high-quality imaging method based on frequency domain reverse time migration
CN114858926A
