Method for dynamic characterization of internal deformation of materials based on double half-range ct scanning and stress relaxation evolution

Abstract

This invention discloses a method for dynamically characterizing the internal deformation of materials based on dual half-span CT scanning and stress relaxation evolution. Utilizing the physical property that "rotation angle and absolute time are linearly mapped" during continuous CT scanning at a constant speed, this invention performs two uninterrupted half-span continuous scans throughout the entire stress relaxation cycle triggered by constant displacement loading of the material. A pseudo-full-angle image is reconstructed through time mapping, and an invariant reference domain is introduced for rigid body displacement targeted registration to eliminate mechanical drift errors. Finally, the dynamic radial deformation of the entire cycle is analyzed by extracting the image difference under the conjugate mapping angle. This invention completely abandons the digital volume correlation algorithm, which is extremely dependent on high-resolution images, and overcomes the interference of artifacts from finite-angle scanning. It achieves high-precision, low-cost quantitative coupling characterization of microscopic damage and macroscopic stress evolution within materials in a single, uninterrupted experiment.

Description

Technical Field

[0001] This invention belongs to the field of material mechanical property testing and non-destructive testing technology, specifically involving a method for dynamic characterization of internal deformation of materials based on dual half-path CT scanning and stress relaxation evolution. Background Technology

[0002] Studying the time-dependent deformation behavior of materials under constant loads or displacements (such as creep and stress relaxation) is crucial for assessing their long-term mechanical properties. Traditional methods mainly rely on external extensometers or strain gauges to measure macroscopic deformation on the material surface, which cannot reveal localized non-uniform deformations caused by microcrack evolution and damage accumulation within the material.

[0003] While X-ray computed tomography (CT) technology can obtain internal structures non-destructively, it faces significant bottlenecks in characterizing continuous dynamic mechanical processes: on the one hand, in order to obtain transient responses, the scanning time must be shortened; on the other hand, if a limited angle (such as 180° half-scan) is used to shorten the time, the reconstructed image will inevitably have severe artifacts and edge blurring.

[0004] Existing technologies typically employ digital volumetric image correlation (DVC) for deformation analysis; however, DVC algorithms are extremely sensitive to image quality. When reconstructing CT data from the first half-scan (0°~180°) and the second half-scan (180°~360°), the DVC calculation generates significant noise due to image blurring and indistinct internal feature points, leading to distorted results or even algorithm failure. Furthermore, during macroscopic stress relaxation, while the material maintains a constant axial macroscopic displacement, complex stress redistribution occurs internally, accompanied by radial microscopic deformation evolution. Current detection methods cannot effectively couple macroscopic constant displacement with microscopic dynamic radial deformation. Summary of the Invention

[0005] The purpose of this invention is to provide a dynamic characterization method for internal deformation of materials based on dual half-span CT scanning and stress relaxation evolution. This method breaks away from the traditional dependence on high-resolution images and complex DVC algorithms in deformation analysis. It cleverly utilizes the inherent "time-sequence and angle mapping" physical property of continuous rotational CT scanning, and through image subtraction of limited angular data and rigid body motion compensation techniques, it can accurately extract the radial deformation rate of the sample in a single, uninterrupted constant displacement loading experiment, thus achieving dynamic characterization of internal deformation of the material.

[0006] The technical solution to achieve the purpose of this invention is: a method for dynamic characterization of internal deformation of materials based on dual half-span CT scanning and stress relaxation evolution, comprising the following steps:

[0007] S1: Apply axial displacement to the test sample to a predetermined value and keep it constant to trigger stress relaxation, and record the macroscopic axial stress relaxation curve σ(t); Simultaneously execute two continuous CT scans of 180° half-range each to complete the full-range projection acquisition from 0° to 360° and obtain the initial image;

[0008] S2: Establish a time mapping relationship, perform mirror symmetric augmentation on the projection data of the two halves respectively, and reconstruct the pseudo full-angle image;

[0009] S3: A rigid body displacement calibration mechanism is introduced to register the image to obtain a radial deformation grayscale image; based on the early pseudo full-angle image A1 and the late pseudo full-angle image A2 reconstructed by the equilateral mirror logic, the grayscale difference of the slices under the conjugate physical angle is calculated to obtain the true dynamic radial physical deformation of the material.

[0010] S4: Obtain the discrete relationship of "radial deformation amount - time" throughout the entire cycle;

[0011] S5: Introduce a sub-pixel edge positioning algorithm to subdivide the difference boundary, and perform mathematical fitting on the discrete "radial deformation amount - time" scatter points obtained in step S4 to obtain the dynamic radial deformation curve and radial deformation rate.

[0012] S6: Simultaneously map the macroscopic axial stress relaxation curve σ(t) recorded in step S1 and the dynamic radial deformation curve obtained in step S5 to the same time axis to construct a coupled curve of "radial deformation amount - stress relaxation change relationship"; convert the fitted dynamic radial deformation into strain, and calculate the dynamic Poisson's ratio evolution curve ν(t) by combining it with the constant initial macroscopic axial strain.

[0013] Furthermore, the CT scan in step S1 specifically involves: initiating the CT scan during the displacement holding period, with the sample rotary stage moving in set angular steps. The system continuously rotates in steps and records the absolute scanning time corresponding to each angle to complete the full-range projection acquisition from 0° to 360°. 0° to 180° is defined as the first half-range, and 180° to 360° is defined as the second half-range.

[0014] Furthermore, step S2 specifically involves:

[0015] S21, based on the real data of the first half of the process, mirrors the data of angle θ to (180° + θ) to fill in the fake data of the second half of the process, and reconstructs a pseudo full-angle image A1 representing "relaxation in the early stage".

[0016] S22, based on the real data of the second half of the process, mirrors the data of the angle (180° + θ) to θ, fills in the fake first half of the data, and reconstructs a pseudo full-angle image A2 representing the "relaxation-biased later stage".

[0017] Furthermore, step S3 specifically includes:

[0018] S31 Extract the undeformed hardware regions from the first and second half pseudo-full-angle images of step S2 as invariant reference domains.

[0019] S32 calculates the spatial offset vector of the invariant reference domain between the first half-image and the corresponding second half-image;

[0020] S33 subtracts the spatial offset vector from the pseudo full-angle image A1 and pseudo full-angle image A2 to obtain the first half-registered and aligned image B1 and the second half-registered and aligned image B2, thereby eliminating artifacts caused by rigid body motion.

[0021] S34 performs grayscale subtraction on the image B1 after registration and alignment of the first half and the image B2 after registration and alignment of the second half to obtain a radially deformed grayscale image.

[0022] Furthermore, step S4 specifically involves:

[0023] S41 uses the difference in grayscale values ​​of two volume data slices of the first half-registered and aligned image B1 and the second half-registered and aligned image B2 generated in step S3 at the same physical angle θ to trace the sample contour, where θ∈[0°,180°].

[0024] S42 extracts the radial deformation ΔR at angle θ based on the mirror mapping law;

[0025] S43 continuously extracts the radial deformation at all slice angles from 0° to 180°, calculates the corresponding continuous time span Δt(θ), and thus obtains the discrete relationship of "radial deformation - time" for the entire cycle.

[0026] Furthermore, the determination logic for radial deformation ΔR is as follows: in the image after grayscale subtraction, the pixel displacement difference of the outer contour boundary of the sample is extracted; if the pixel displacement difference is greater than the effective spatial resolution of the CT system, it is recorded as the effective radial deformation; if the pixel displacement difference is less than or equal to the effective spatial resolution, it is determined that the material stress relaxation tends to be stable during this time period.

[0027] Furthermore, the radial deformation rate is calculated as follows: Let Δt be the actual time span corresponding to any paired conjugate image, and let ΔR(t) be the dynamic radial deformation amount extracted within this time span. Then, the radial deformation rate for the corresponding period is calculated as follows: .

[0028] Furthermore, the specific process for calculating the dynamic Poisson's ratio evolution curve ν(t) in step S6 is as follows:

[0029] S61 divides the extracted dynamic radial deformation ΔR(t) by the initial radius R0 of the specimen to obtain the dynamic radial strain of the material. ;

[0030] S62 records the initial macroscopic axial strain during the dynamic radial strain and constant displacement loading phase. By comparing the absolute values, a nonlinear function curve of the material's dynamic Poisson's ratio over time is constructed:

[0031]

[0032] Compared with the prior art, the significant advantages of this invention are:

[0033] The innovative "mirror mapping continuous time measurement method" breaks through the limitation that conventional subtraction can only obtain a few isolated data points. By deriving the continuous function of angle and time difference through the principle of symmetrical mirroring, the continuous deformation time curve covering the entire relaxation cycle can be completely analyzed with a single CT reconstruction.

[0034] Deep decoupling of multidimensional mechanical properties: For the first time, the "macroscopic axial stress signal" of the in-situ loading device and the "microscopic radial strain signal" extracted by CT are deeply coupled on the time axis. In addition to obtaining Poisson's ratio, it can also directly reveal the stiffness degradation and micro-damage accumulation mechanism during the relaxation process.

[0035] This method overcomes the bottleneck of DVC algorithm application in low-quality images: abandoning the traditional digital image correlation (DVC) algorithm that highly relies on image contrast and feature points, it directly uses half-projection or reconstructed slices to perform gray-level difference tracking of the sample contour. This method has strong robustness to stripe artifacts and feature blurring caused by finite-angle reconstruction and significantly reduces computational noise.

[0036] A pioneering multi-timescale deformation extraction method based on "spatiotemporal mapping" was developed: it cleverly utilizes the physical property that "rotation angle is equivalent to absolute time" during uniform CT scanning. Using two consecutive half-range data points, only one continuous 360° rotation is needed to analyze the dynamic deformation rate of the material at different relaxation stages.

[0037] Eliminating the interference of mechanical errors, this invention boasts high precision and strong practicality: It introduces a rigid body motion calibration mechanism based on the invariant reference domain of the testing machine (such as the loading plate). During the image calculation stage, it effectively eliminates eccentricity and thermal drift errors generated during long-term equipment operation, ensuring the absolute accuracy of extracting extremely small radial deformations, making it highly compatible with conventional commercial industrial CT equipment. Attached Figure Description

[0038] Figure 1 This invention is based on reconstructed slices from the first half of CT data.

[0039] Figure 2 This invention is based on reconstructed slices from the second half of CT data.

[0040] Figure 3 The difference between the front and back slices of this invention and the corresponding radial displacement distribution.

[0041] Figure 4 The relationship curve between radial displacement and axial stress relaxation time obtained by this invention.

[0042] Figure 5 The relationship curve between radial deformation and axial stress relaxation in this invention. Detailed Implementation

[0043] The present invention will now be described in further detail with reference to the accompanying drawings.

[0044] A method for dynamic characterization of internal deformation of materials based on dual half-span CT scanning and stress relaxation evolution includes the following steps:

[0045] S1. Apply an axial displacement to the test sample to a predetermined value and lock it to maintain a constant value to trigger stress relaxation. Simultaneously, use a mechanical sensor to continuously record the relaxation curve σ(t) of the macroscopic axial stress decreasing over time. During the displacement holding period, start a CT scan. The sample rotary stage rotates continuously in small step increments, and the absolute scan time corresponding to each angle is recorded to complete the full-range projection acquisition from 0° to 360°. 0°~180° is defined as the first half-range, and 180°~360° is defined as the second half-range.

[0046] S2. To eliminate artifacts in the reconstruction of limited angles and establish a time mapping relationship, the projection data is subjected to mirror symmetry expansion processing: Based on the real data of the first half, the data of angle θ is mirrored to (180° + θ) to fill in the fake second half data and reconstruct the first three-dimensional volume data representing "early relaxation"; Based on the real data of the second half, the data of angle (180° + θ) is mirrored to θ to fill in the fake first half data and reconstruct the second three-dimensional volume data representing "late relaxation".

[0047] S3. Before performing image subtraction, extract the undeformed hardware region (such as the upper and lower loading plates of the testing machine) in the image field of view as an invariant reference domain. Calculate the spatial offset vector of the invariant reference domain between the corresponding images in the first and second halves of the image. This offset vector represents the mechanical error and rigid body displacement of the equipment during the long-term scanning process. Subtract this spatial offset vector from the global image (i.e., perform image registration and alignment) to eliminate artifacts caused by rigid body motion. Then, perform grayscale subtraction on the sample area to ensure that the extracted edge displacement originates purely from the physical deformation of the material itself.

[0048] S4. Track the sample contour by subtracting the grayscale values ​​of two volume data slices at the same physical angle θ (θ∈[0°,180°]). According to the mirror mapping law, the radial deformation ΔR extracted at this angle actually reflects the deformation of the sample between absolute times t1(θ) and t2(180°+θ). Since the scanning is continuous and uniformly stepped, by continuously extracting the radial deformation at all slice angles from 0° to 180°, the corresponding continuous time span Δt(θ) can be calculated, thus constructing a discrete correspondence between "radial deformation - time" for the entire cycle.

[0049] S5. To address the physical quantity jumps (discrete step-like data) caused by the effective resolution of the CT system (e.g., 10μm), a subpixel edge localization algorithm is introduced to subdivide the difference boundary; mathematical fitting is performed on the obtained discrete "deformation amount-time" scatter points to obtain the dynamic radial deformation curve.

[0050] S6. The fitted dynamic radial deformation is converted into strain, and combined with the constant initial macroscopic axial strain, the smooth dynamic Poisson's ratio evolution curve ν(t) is calculated.

[0051]

[0052] The macroscopic axial stress relaxation curve σ(t) synchronously recorded in S1 is mapped synchronously to the same time axis as the microscopic radial deformation curve obtained in this step. A coupled curve of "macroscopic stress reduction rate - microscopic radial expansion rate" is constructed to determine the critical point of microcrack propagation and damage evolution law of the material at different stages of relaxation.

[0053] In step S3, the image registration and subtraction process specifically includes: the image is a two-dimensional slice image processed by the CT reconstruction algorithm or a two-dimensional projection image obtained directly; the grayscale subtraction process is to perform pixel-level subtraction on the paired images at the conjugate angle (i.e., 180° difference) to highlight the displacement change of the sample contour boundary within the corresponding time span.

[0054] In step S4, the specific determination logic for extracting the dynamic radial deformation ΔR(t) is as follows: in the image after grayscale subtraction, extract the pixel displacement difference of the outer contour boundary of the sample; if the pixel displacement difference is greater than the effective spatial resolution of the CT system, it is recorded as the effective radial deformation; if the pixel displacement difference is less than or equal to the effective spatial resolution, it is determined that the material stress relaxation tends to be stable within this time period, and the radial deformation increment within the corresponding time span tends to be zero.

[0055] In step S5, the radial deformation rate is calculated as follows: Let Δt be the actual time span corresponding to any paired conjugate image, and ΔR(t) be the dynamic radial deformation amount extracted within this time span. Then, the radial deformation rate for the corresponding period is calculated as follows: .

[0056] In step S6, the specific process of calculating the material's time-dependent dynamic Poisson's ratio is as follows: divide the extracted dynamic radial deformation ΔR(t) by the initial radius R0 of the specimen to obtain the material's dynamic radial strain. The dynamic radial strain and the initial macroscopic axial strain recorded during the constant displacement loading stage. By comparing the absolute value of the material with the time-varying nonlinear function curve of the material's dynamic Poisson's ratio, a nonlinear function curve of the material's dynamic Poisson's ratio is constructed.

[0057] Example

[0058] The following example is a stress relaxation experiment of ultra-high performance concrete (UHPC) under uniaxial constant displacement compression.

[0059] 1. A cylindrical UHPC specimen with a diameter of Φ=10mm and a height of H=15mm was used, and the effective spatial resolution of CT was 10μm.

[0060] 2. After applying compressive stress to 100 MPa, the axial displacement was locked, triggering stress relaxation. Over the next 42 minutes, the mechanical sensor continuously recorded the curve of the axial stress σ(t) gradually decreasing from 100 MPa.

[0061] 3. During the 42-minute relaxation period, the sample is rotated from 0° to 360° (each angle pause is recorded for 42 / 1080 minutes). Mirror mapping is performed: taking an 85° slice angle as an example. In the first half of the reconstruction, the 85° slice uses real data of t1 = (85° / 360°) × 42 = 9.9 minutes (see...). Figure 1 In the second half of the reconstruction, the 85° slice uses mirrored data with the actual angle (180° + 85°) = 275° (see...). Figure 2 The corresponding real time t2 = (275° / 360°) × 42 = 32.1 minutes.

[0062] 4. By subtracting the two registered volume data at an 85° slice, the radial displacement difference of the UHPC within the time interval of 9.9 minutes to 32.1 minutes (span Δt = 22.2 minutes) can be directly obtained (see...). Figure 3 By iterating through slices corresponding to θ continuously from 0° to 180°, the time span Δt covers the entire interval from 42 minutes to 0 minutes. This yields a stepped discrete scatter plot with 10μm intervals, thus generating a continuous and differentiable radial deformation-axial stress relaxation time curve. (See Figure 4 ).

[0063] 5. Couple the fitted continuous radial expansion curve with the axial stress relaxation curve σ(t) measured in step 2 by time axis alignment (see...). Figure 5 By analyzing the radial expansion increment caused by a unit stress reduction, the active period (first 8 minutes) of internal micro-damage propagation in the UHPC under constant displacement and the critical time point for entering quasi-static stability were determined.

Claims

1. A method for dynamic characterization of internal deformation of materials based on dual half-span CT scanning and stress relaxation evolution, characterized in that, Includes the following steps: S1: Apply axial displacement to the test sample to a predetermined value and keep it constant to trigger stress relaxation, and record the macroscopic axial stress relaxation curve σ(t); Simultaneously execute two continuous CT scans of 180° half-range each to complete the full-range projection acquisition from 0° to 360° and obtain the initial image; S2: Establish a time mapping relationship, perform mirror symmetric augmentation on the projection data of the two halves respectively, and reconstruct the pseudo full-angle image; S3: Introduce a rigid body displacement calibration mechanism to register and subtract images to obtain a radial deformation grayscale image; S4: Obtain the discrete relationship of "radial deformation amount - time" throughout the entire cycle; S5: Introduce a subpixel edge positioning algorithm to subdivide the difference boundary, and perform mathematical fitting and smoothing on the discrete "radial deformation amount - time" scatter points obtained in step S4 to obtain a smooth and continuous dynamic radial deformation curve and radial deformation rate. S6: Simultaneously map the macroscopic axial stress relaxation curve σ(t) recorded in step S1 and the dynamic radial deformation curve obtained in step S5 to the same time axis to construct a coupled curve of "radial deformation amount - stress relaxation change relationship"; convert the fitted dynamic radial deformation into strain, and combine it with constant initial macroscopic axial strain to calculate the dynamic Poisson's ratio evolution curve ν(t).

2. The method according to claim 1, characterized in that, The CT scan in step S1 specifically involves initiating the CT scan during the displacement holding period, with the sample rotary stage moving in set angular steps. The system continuously rotates in steps and records the absolute scanning time corresponding to each angle to complete the full-range projection acquisition from 0° to 360°. 0° to 180° is defined as the first half-range, and 180° to 360° is defined as the second half-range.

3. The method according to claim 2, characterized in that, Step S2 is as follows: S21: Based on the real data of the first half, by mirroring the data of angle θ to (180°+θ), the fake data of the second half is filled in, and the pseudo full-angle image A1 representing "relaxation in the early stage" is reconstructed. S22: Based on the real data of the second half, by mirroring the data of the angle (180°+θ) to θ, the fake first half data is filled in, and the pseudo full-angle image A2 representing the "relaxation-biased later stage" is reconstructed.

4. The method according to claim 3, characterized in that, Step S3 is as follows: S31: Extract the undeformed hardware regions from the first and second half pseudo-full-angle images in step S2 as invariant reference domains. S32: Calculate the spatial offset vector of the invariant reference domain between the first half-image and the corresponding second half-image; S33: Subtract the spatial offset vector from the pseudo full-angle image A1 and pseudo full-angle image A2 to obtain the first half-registered and aligned image B1 and the second half-registered and aligned image B2, thereby eliminating artifacts caused by rigid body motion; S34: Perform grayscale difference between the image B1 after registration and alignment of the first half and the image B2 after registration and alignment of the second half to obtain a radially deformed grayscale image.

5. The method according to claim 4, characterized in that, Step S4 is as follows: S41: Using the two volume data slices of the first half-registered and aligned image B1 and the second half-registered and aligned image B2 generated in step S3, the grayscale difference at the same physical angle θ is used to trace the sample contour, where θ∈[0°, 180°]. S42: Based on the mirror mapping law, extract the radial deformation ΔR at angle θ; S43: By continuously extracting the radial deformation at all slice angles from 0° to 180°, the corresponding continuous time span Δt(θ) is calculated, thus obtaining the discrete relationship of "radial deformation - time" for the entire cycle.

6. The method according to claim 5, characterized in that, The determination logic for radial deformation ΔR is as follows: In the image after grayscale subtraction, extract the pixel displacement difference of the outer contour boundary of the sample; if the pixel displacement difference is greater than the effective spatial resolution of the CT system, it is recorded as the effective radial deformation; if the pixel displacement difference is less than or equal to the effective spatial resolution, it is determined that the material stress relaxation tends to be stable during this time period.

7. The method according to claim 6, characterized in that, The radial deformation rate is calculated as follows: Let Δt be the actual time span corresponding to any paired conjugate image, and let ΔR(t) be the dynamic radial deformation amount extracted within this time span. Then, the radial deformation rate for the corresponding period is calculated as follows: .

8. The method according to claim 7, characterized in that, The specific process for calculating the radial deformation evolution curve ν(t) in step S6 is as follows: S61: Divide the extracted dynamic radial deformation ΔR(t) by the initial radius R0 of the specimen to obtain the dynamic radial strain of the material. ; S62: The initial macroscopic axial strain recorded during the dynamic radial strain and constant displacement loading stage. By comparing the absolute values, a nonlinear function curve of the material's dynamic Poisson's ratio over time is constructed: 。

Images