Tensile test error compensation and elastic deformation detection method for universal testing machine
By setting feature points on the surface of the specimen and using image processing algorithms to track the displacement of the feature points, and combining force data to construct stress-strain curves, the problems of fixture slippage error and lack of elastic deformation monitoring in universal testing machines are solved, and more accurate material performance evaluation is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-11
- Publication Date
- 2026-03-27
AI Technical Summary
Traditional universal testing machines cannot accurately distinguish between the actual deformation of the specimen body and the displacement caused by the slippage of the fixture in tensile testing, resulting in measurement data containing errors and neglecting the elastic deformation part, which affects the evaluation of material properties.
Multiple feature points are set on the surface of the specimen. Continuous image data is acquired by a high frame rate industrial camera. The displacement of the feature points is calculated by a feature point tracking algorithm. The stress-strain curve is constructed by combining the force data to eliminate fixture slippage error and monitor elastic deformation.
It improves the accuracy and precision of tensile length measurement, comprehensively describes the mechanical behavior of materials in the elastic and plastic deformation stages, and provides a more comprehensive assessment of material properties.
Smart Images
Figure CN121740592A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of tensile testing technology for universal testing machines, and more specifically, to a method for error compensation and elastic deformation detection in tensile testing of universal testing machines. Background Technology
[0002] In mechanics laboratories, universal testing machines are widely used for testing the mechanical properties of materials.
[0003] In tensile testing, traditional methods rely on the displacement of the testing machine beam or the fixture to measure the tensile length of the material. This method cannot distinguish between the actual deformation of the specimen and the displacement caused by fixture slippage, leading to errors in the measurement data. These systematic errors affect the reliability and accuracy of the test results.
[0004] Existing universal testing machines typically test the tensile length of materials after the specimen has returned to rest following the end of the tensile process. This only monitors the plastic deformation of the material, neglecting the elastic deformation. In many traditional testing methods, the experiment ends when the material enters the plastic deformation stage, ignoring the tensile characteristics of the elastic stage. This results in an inability to fully understand the deformation characteristics of the material throughout the tensile process, especially the mechanical behavior at small strains and in the elastic stage, thus affecting the overall assessment of the material's properties. Summary of the Invention
[0005] This invention provides a method for compensating for tensile testing errors and detecting elastic deformation in a universal testing machine, solving the technical problems in related technologies where clamp slippage leads to inaccurate tensile length measurement and inability to accurately monitor elastic deformation.
[0006] This invention discloses a method for compensating for tensile testing errors and detecting elastic deformation using a universal testing machine, comprising the following steps: Multiple feature points are set in the non-clamping area on the surface of the specimen; The specimen is stretched, and continuous frame image data from the start to the end of the stretching is acquired from the front of the universal testing machine, and the force value data is recorded simultaneously. The continuous frame image data is preprocessed to generate a standard image sequence; A feature point tracking algorithm is used to track the position changes of each feature point in the standard image sequence in different frames to obtain the displacement trajectory of each feature point. Select any two feature points to form a feature point pair, calculate the difference between the spatial distance of the feature point pair in the current frame and the initial distance in the initial frame, and obtain the relative displacement of the feature point pair; The relative displacements of all feature point pairs are averaged to calculate the true tensile length variation sequence of the specimen. Based on the actual tensile length change sequence and the force data, the stress and strain values at each moment are calculated to construct the stress-strain curve of elastic deformation. The continuous frame image data and the force value data are synchronized and correspond to each other through timestamp marking.
[0007] Furthermore, the number of feature points is at least three, and the multiple feature points are evenly distributed in the non-clamping area of the specimen body.
[0008] Furthermore, the preprocessing includes filtering, grayscale normalization, distortion correction, and registration.
[0009] Furthermore, a feature point tracking algorithm is used to track the positional changes of each feature point in the standard image sequence across different frames, obtaining the displacement trajectory of each feature point, including: Extract the initial frame of the standard image sequence and identify the initial position coordinates of each feature point in the initial frame; The KLT tracking algorithm is used to track the position coordinates of each feature point in the standard image sequence, excluding the initial frame. The position coordinates of each feature point in the standard image sequence are integrated into a time series to generate the displacement trajectory of each feature point.
[0010] Further, calculating the spatial distance between the feature point pairs in the current frame includes: Obtain the position coordinates of the first and second feature points in the feature point pair in the current frame, wherein the position coordinates include horizontal and vertical coordinates; Calculate the difference in horizontal coordinates and the difference in vertical coordinates between the first feature point and the second feature point, respectively; The spatial distance between the feature point pairs in the current frame is obtained by calculating the square root of the sum of the squares of the horizontal coordinate differences and the squares of the vertical coordinate differences.
[0011] Further, the relative displacement of the feature point pair is obtained, including: The initial distance of the feature point pair in the initial frame is obtained using the same calculation method; The difference between the spatial distance of the feature point pair in the current frame and the initial distance in the initial frame is calculated to obtain the relative displacement of the feature point pair.
[0012] Furthermore, the relative displacements of all feature point pairs are averaged to calculate the true tensile length variation sequence of the specimen, including: Count the total number of all selected feature point pairs; Sum the relative displacements of all feature point pairs; Divide the summation result by the total number of feature point pairs to obtain the actual tensile length change value of the specimen corresponding to the current frame; Perform the above calculations on all frames in the standard image sequence to generate the true tensile length change sequence of the specimen.
[0013] Furthermore, the continuous frame image data and the force value data are synchronized and correspond to each other through timestamp marking, including: While triggering the camera to acquire images, the force value data output by the universal testing machine at the current moment is recorded simultaneously; Each frame of image and its corresponding force value data is marked with the same timestamp; A one-to-one correspondence is established between the consecutive frame image data and the force value data based on the timestamp.
[0014] Furthermore, based on the actual tensile length change sequence and the force data, the stress and strain values at each moment are calculated to construct the stress-strain curve of elastic deformation, including: Based on the timestamp, extract the force value data and the actual tensile length change data corresponding to each frame; The stress value corresponding to each frame is calculated based on the force data and the cross-sectional area of the specimen. The strain value corresponding to each frame is calculated based on the actual tensile length change data and the initial gauge length of the specimen; The stress and strain values corresponding to each timestamp are combined into point-to-point data, which are then plotted into a two-dimensional curve to generate the stress-strain curve of elastic deformation.
[0015] This invention discloses a universal testing machine tensile testing error compensation and elastic deformation detection system, comprising: A universal testing machine is used to apply tensile force to a specimen and output force value data. An industrial camera, fixed to the front of the universal testing machine, is used to acquire continuous frame image data of the surface of the specimen; A data acquisition system is used to synchronously record the continuous frame image data and the force value data, and to establish the correspondence between the two by using timestamps; Computer equipment is used to preprocess the continuous frame image data, track feature points, calculate relative displacement, calculate the actual tensile length change, and construct stress-strain curves.
[0016] This invention sets feature points in the non-clamping area of the specimen body, uses image processing algorithms to track the displacement trajectory of the feature points during the tensile process, calculates the change in distance between different feature point pairs to obtain relative displacement data based on the deformation of the specimen body, and calculates the true tensile length change of the specimen by averaging the measurement results of multiple feature point pairs. This solves the technical problem of inaccurate tensile length measurement caused by clamp slippage, and achieves the technical effects of eliminating systematic errors, improving test accuracy and data reliability.
[0017] This invention continuously acquires image and force data during the stretching process, establishes a correspondence between image and force data through timestamps, and achieves continuous monitoring of the material stretching process. It obtains minute deformation data of the material in the elastic deformation stage, and constructs a complete stress-strain curve based on the real stretching length change data and force data throughout the entire process. This solves the technical problem of missing elastic deformation monitoring and achieves the technical effect of comprehensively describing the mechanical behavior of the material in the elastic and plastic deformation stages and improving the measurement accuracy of key mechanical parameters.
[0018] The feature point pair relative displacement comprehensive calculation mechanism adopted in this invention selects multiple feature points and calculates the relative displacement of all possible feature point pairs, and then averages these relative displacements. This reduces the impact of single feature point tracking error on the final result, solves the technical problem of large single-point measurement error, and achieves the technical effect of improving the calculation accuracy of the actual tensile length change. Attached Figure Description
[0019] Figure 1 This is a technical flowchart of steps 100 to 500 in specific embodiment 1 of the present invention; Figure 2 This is a technical flowchart of steps 600 to 1000 in specific embodiment 2 of the present invention. Detailed Implementation Detailed Implementation Method 1 In mechanics laboratories, universal testing machines are widely used for testing the mechanical properties of materials. Two main technical problems exist in tensile testing: First, the difference between the actual deformation during tensile testing and the slippage of the clamps leads to inaccurate measured tensile lengths. This systematic error affects the reliability and accuracy of the test results. Traditional methods rely on measuring the displacement of the testing machine's beam or clamps, which cannot distinguish between the actual deformation of the specimen and the displacement caused by clamp slippage, resulting in erroneous measurement data. Second, existing universal testing machines typically test the tensile length of materials after the tensile test has ended and the specimen has returned to rest. This only monitors the plastic deformation of the material, ignoring the elastic deformation. In many traditional testing methods, the experiment ends when the material enters the plastic deformation stage, neglecting the tensile characteristics of the elastic stage. This results in an inability to fully understand the deformation characteristics of the material throughout the tensile process, especially the mechanical behavior at small strains and in the elastic stage.
[0021] This embodiment addresses the technical problems of fixture slippage error and lack of elastic deformation monitoring in tensile testing of universal testing machines by providing a method for tensile testing error compensation and elastic deformation detection. This embodiment is performed in a universal testing machine environment, which includes: the universal testing machine and its associated force sensor, a high frame rate industrial camera fixed to the front of the universal testing machine, a data acquisition system for timestamp synchronization, and computer equipment for image processing and data analysis.
[0022] The method of this embodiment includes the following steps. Step 100: Acquire feature point image data of the non-clamping area on the specimen surface, continuous frame image data of the tensile process, and corresponding force value data. Multiple feature points are set on the non-clamping area of the specimen surface. These feature points are set on the specimen surface using laser etching or spraying dot marks and are used for subsequent image tracking analysis. The specimen with the feature points set is then mounted onto the clamps of the universal testing machine, ensuring a secure clamping and alignment of both ends. The universal testing machine is started, and a tensile force is applied to the specimen at the set loading rate. During the tensile process, continuous frame image data is acquired from the camera on the front of the universal testing machine, from the start of the tensile process to its end. The force value data output by the universal testing machine is recorded simultaneously with the image acquisition. A timestamp is used to correlate the continuous frame image data with the force value data.
[0023] It should be noted that the number of feature points is at least three, evenly distributed in the non-clamping area of the specimen body. The more feature points there are, the higher the accuracy of the subsequent calculation of the actual tensile length change. In a specific example of this embodiment, more than 10 feature points are used.
[0024] It should be noted that the correspondence between the continuous frame image data and the force value data through timestamp marking means that when the camera is triggered to acquire images, the data acquisition system simultaneously records the force value data of the universal testing machine at the current moment, and marks each frame image and the corresponding force value data with the same timestamp, thereby establishing a one-to-one correspondence between image data and force value data.
[0025] Step 200: Process the consecutive frame image data to generate a standard image sequence. The continuous frame image data is preprocessed to obtain a standard image sequence. The preprocessing includes filtering, grayscale normalization, distortion correction, and registration. This preprocessing eliminates the influence of image noise, uneven illumination, and lens distortion on subsequent feature point tracking, generating a standard image sequence suitable for feature point tracking analysis.
[0026] Step 300: Trace the displacement trajectories of feature points in the standard image sequence and calculate the relative displacement of feature point pairs. Extract the initial frame of the standard image sequence and identify the location of feature points in the initial frame. The first [frame name missing] in the initial frame... The initial coordinates of the feature points are: The feature point tracking algorithm is used to track the position of each feature point in the standard image sequence, excluding the initial frame. Then, the position of the first feature point is determined. The first frame The coordinates of the feature points are ,in , Let be the total number of frames in the standard image sequence. The positions of feature points in the standard image sequence are integrated into displacement trajectories in a time series, then the th... The displacement trajectory of the feature points is .
[0027] Select any two feature points and Composition of feature point pairs Calculate the feature point pairs In the Frame spatial distance The calculation formula is: in, For feature points In the The frame's position coordinates. When hour, For feature point pairs At the initial distance in the initial frame, the feature point pair... relative displacement for: It should be noted that the feature point tracking algorithm can be the KLT tracking algorithm.
[0028] The input to the aforementioned KLT tracking algorithm is a standard image sequence and a set of feature point coordinates in the initial frame. The output is a set of position coordinates for each feature point in each frame of a standard image sequence. ,in The KLT tracking algorithm tracks the position of feature points in consecutive frames by iteratively solving for the displacement of feature points by minimizing the brightness difference of image blocks between adjacent frames.
[0029] Step 400: Combine the relative displacements of multiple feature point pairs to generate a sequence of actual tensile length changes in the specimen. The relative displacements of all selected feature point pairs are averaged to calculate the first... The actual tensile length change of the specimen corresponding to the frame The calculation formula is: in, The total number of selected feature point pairs, Iterate through all selected feature point pairs. Perform the above calculation on all frames in the standard image sequence to generate the true tensile length variation sequence of the specimen. .
[0030] Step 500: Based on the actual tensile length variation sequence and force value data, generate the stress-strain curve of elastic deformation. Based on the timestamp, extract the force value data and the actual tensile length change data corresponding to each frame. Calculate the stress value for each frame based on the force value data and the strain value for each frame based on the actual tensile length change data. Combine the stress and strain values corresponding to each timestamp into point-to-point data, plot them into a two-dimensional curve, and generate a stress-strain curve for elastic deformation.
[0031] It should be noted that the formula for calculating the stress value is as follows: ,in This is the stress value. For the force value data at the corresponding time, Let be the cross-sectional area of the specimen. The formula for calculating the strain value is: ,in The strain value, This represents the actual change in tensile length at the corresponding moment. This is the initial gauge length of the specimen.
[0032] In this embodiment, a universal testing machine was used in a mechanics laboratory to perform a tensile test on a specimen with an original length of 100 mm. Table 1 is derived from a method for error compensation and elastic deformation detection using a universal testing machine in this embodiment. Table 1 provides an example of the point-to-point data used to plot the two-dimensional curve.
[0033] Table 1 Examples of Actual Tensile Length-Stress-Strain Point Pairs In Table 1, the force values are not used as data for plotting the curves, but rather as data values to aid understanding. When plotting two-dimensional curves, a dual-axis format can be used; for example, strain values can be used as the X-axis, stress data as the left Y-axis, and the actual tensile length data as the right Y-axis. This is because in tensile testing, strain represents the degree of material deformation and is usually used as the abscissa to describe the relative proportion of the change in length of the specimen material from its original length to its deformed length. Strain can clearly describe how the material exhibits different mechanical properties at different stages of tensile testing. For example, in the initial stage of tensile testing, the strain change of the specimen material is relatively small, while the strain of the material increases rapidly after reaching the yield point. The increase in strain value is usually accompanied by a change in stress in the material; therefore, using strain as the abscissa can intuitively reflect the elastic, yielding, and strengthening stages of the material.
[0034] In a two-dimensional graph, two curves will be displayed: one is the stress-strain curve, and the other shows the relationship between the actual tensile length and strain. In this embodiment, to more comprehensively demonstrate the deformation characteristics of the material during tensile testing, a curve showing the relationship between the actual tensile length and strain is generated alongside the stress-strain curve of elastic deformation. Specifically, a two-dimensional graph is plotted using a dual-axis format, with strain as the horizontal axis, stress as the left vertical axis, and the actual tensile length as the right vertical axis. This two-dimensional graph presents two curves: the first is the stress-strain curve, reflecting the stress response characteristics of the material under different strains; the second is the curve showing the relationship between the actual tensile length and strain, visually demonstrating the change in length of the material from its initial length to its deformed length. This hyperbolic format can simultaneously display the mechanical properties and geometric deformation characteristics of the material.
[0035] This embodiment sets feature points in the non-clamping area of the specimen body and uses image processing algorithms to track the displacement trajectory of the feature points during the tensile process, calculating the change in distance between different feature point pairs to obtain relative displacement data based on the deformation of the specimen body. This relative displacement data integrates the measurement results of multiple feature point pairs and calculates the true tensile length change of the specimen through averaging. Since the feature points are set in the non-clamping area of the specimen body, the displacement trajectory of the feature points only reflects the true deformation of the specimen body and does not include the displacement caused by clamp slippage. Therefore, the true tensile length change calculated based on the relative displacement of feature points eliminates the clamp slippage error, a factor that leads to inaccurate measurement, and solves the technical problem of inaccurate tensile length measurement caused by clamp slippage.
[0036] This implementation method continuously monitors the material's tensile process by acquiring image and force data during continuous stretching and establishing a correspondence between the image and force data using timestamps. At each moment of the stretching process, the actual stretching length change calculated based on the current frame image reflects the total deformation of the material at that moment. Combined with the corresponding force data, the stress and strain values at that moment can be calculated. Since the image acquisition time covers the entire process from the start to the end of stretching, including the elastic and plastic deformation stages of the material, it is possible to acquire minute deformation data of the material during the elastic deformation stage. The stress-strain curve constructed based on the actual stretching length change data and force data throughout the entire process fully describes the mechanical behavior of the material during both the elastic and plastic deformation stages. This method overcomes the limitations of traditional methods that only measure deformation after stretching, resulting in the inability to monitor elastic deformation, and solves the technical problem of missing elastic deformation monitoring.
[0037] The feature point pair relative displacement comprehensive calculation mechanism employed in this embodiment further improves measurement accuracy. By selecting multiple feature points and calculating the relative displacement of all possible feature point pairs, and then averaging these relative displacements, the impact of individual feature point tracking errors on the final result can be reduced, improving the accuracy of calculating the true tensile length change. This mechanism utilizes the redundant information provided by multiple spatially distributed feature points and reduces measurement noise through statistical averaging. Detailed Implementation Method 2 In the tensile testing process of a universal testing machine, multiple feature points are placed on the surface of the specimen, and the displacement trajectory of each feature point is obtained using the KLT tracking algorithm. The relative displacement change between feature point pairs is then calculated to obtain the true tensile length of the specimen. However, in actual tensile testing, the degree of deformation in different areas of the specimen surface is not completely uniform. Material defects, differences in grain orientation, or stress concentration can lead to abnormal deformation gradients in local areas.
[0039] Characterizing the global tensile length using the average relative displacement of discrete feature point pairs presents the following technical problems: First, it ignores the spatial non-uniformity of deformation on the specimen surface, causing the deformation characteristics of local stress concentration areas to be masked by averaging. Second, it cannot identify potential material defects or crack initiation locations, affecting the comprehensiveness of the material's mechanical property assessment. Third, it can only obtain displacement information of discrete points, failing to reconstruct the continuous deformation field across the entire specimen surface, thus limiting a deeper understanding of the material deformation mechanism.
[0040] This embodiment addresses the technical problem that discrete feature point tracking cannot capture the non-uniformity of spatial deformation in tensile testing using a universal testing machine, and provides a multi-scale deformation analysis method based on feature point drift field reconstruction. This embodiment is executed in a universal testing machine environment, which includes: the universal testing machine and its associated force sensor, a high frame rate industrial camera fixed to the front of the universal testing machine, a data acquisition system for timestamp synchronization, and computer equipment for image processing and data analysis.
[0041] The method of this embodiment includes the following steps: Step 600: Obtain the position coordinate data of feature points during the stretching process. Obtain the first image from the standard image sequence during the stretching process. The set of coordinates of all feature points in a frame and the set of feature point coordinates of the initial frame The coordinates of the feature points are obtained through a feature point tracking algorithm. Each feature point... In the The position coordinates of the frame are represented as .
[0042] It should be noted that the standard image sequence is obtained by preprocessing consecutive frame image data acquired during the stretching process. Preprocessing includes filtering, grayscale normalization, distortion correction, and registration. The feature point tracking algorithm can employ the KLT tracking algorithm or other optical flow tracking algorithms.
[0043] Step 700: Calculate the discrete displacement vector field and reconstruct the continuous drift field Calculate the displacement vector of each feature point This generates a discrete displacement vector field. The displacement vector reflects the displacement of the feature point from its initial position to its current position, and includes horizontal and vertical components.
[0044] Using the radial basis function interpolation algorithm, the discrete displacement vector field is reconstructed into a continuous drift field on the specimen surface. ,in The coordinates are arbitrary positions on the specimen surface. The radial basis function interpolation algorithm constructs radial basis functions centered at each feature point, and performs weighted interpolation on the displacements at arbitrary positions on the specimen surface, thereby extending the displacement information of discrete points to the entire specimen surface and generating a continuous displacement field covering the entire domain.
[0045] It should be noted that the input to the radial basis function interpolation algorithm is a set of discrete feature point coordinates. and its corresponding set of displacement vectors The output is any position on the surface of the specimen. continuous displacement field Radial basis function interpolation spatially interpolates the displacement of the specimen surface by constructing radial basis functions centered at each feature point. The mathematical expression is: ,in These are the weighting coefficients. For radial basis functions, This represents the total number of feature points.
[0046] The aforementioned radial basis functions can be Gaussian functions, multiple quadratic functions, or thin-plate spline functions.
[0047] Step 800: Perform multi-scale wavelet decomposition on the continuous drift field. Multi-scale wavelet decomposition is performed on the continuous drift field to separate the global stretching component. With local disturbance components The multi-scale wavelet decomposition employs a two-dimensional discrete wavelet transform to decompose the continuous drift field into approximate components and detail components at different scales.
[0048] The global stretching component The low-frequency approximation component corresponding to wavelet decomposition reflects the uniform tensile deformation of the entire specimen. The local perturbation component... The high-frequency detail components corresponding to wavelet decomposition reflect the local non-uniform deformation of the specimen surface, including local features such as stress concentration, defect areas, and crack initiation.
[0049] It should be noted that the input to the multi-scale wavelet decomposition is a continuous drift field. The output is a global stretch component. and local disturbance components Multiscale wavelet decomposition decomposes a continuous drift field into approximate and detail components at different scales through two-dimensional discrete wavelet transform. The low-frequency approximate components correspond to the global stretching components, while the high-frequency detail components correspond to the local perturbation components.
[0050] The aforementioned two-dimensional discrete wavelet transform can employ Haar wavelet, Daubechies wavelet, or Symlet wavelet.
[0051] Step 900: Calculate the strain gradient distribution and identify stress concentration regions. Calculate the spatial gradient tensor of a continuous drift field The strain gradient distribution at various locations on the specimen surface is obtained. The spatial gradient tensor is obtained by spatially differentiating the continuous drift field, and its expression is: in, and These are the horizontal and vertical components of the continuous drift field, respectively.
[0052] Based on the strain gradient distribution, peak regions of the strain gradient are identified, generating spatial distribution maps of local stress concentration regions. Specifically, the Frobenius norm of the strain gradient tensor is calculated. As an index of strain gradient intensity, among which Traverse all row and column indices of the gradient tensor, and when the index exceeds a set threshold, identify the corresponding location as a stress concentration region.
[0053] It should be noted that the strain gradient peak regions correspond to the locations of stress concentration within the material, which are typically potential sites for material defects, crack initiation, or fracture. Spatial distribution maps can visually demonstrate the distribution characteristics of stress concentration regions on the specimen surface.
[0054] Step 1000: Generate a multi-scale deformation analysis report By combining the global stretching component and the local perturbation component, a multi-scale deformation analysis report is generated. The multi-scale deformation analysis report includes the following: (1) Visual representation of global tensile components, showing the uniform tensile deformation field of the specimen as a whole; (2) Visual representation of local disturbance components, showing the non-uniform deformation distribution on the surface of the specimen; (3) A pseudo-color map of strain gradient distribution, which uses color mapping to show the spatial distribution of strain gradient intensity; (4) Annotation of stress concentration areas: Mark the locations of the identified stress concentration areas on the specimen image; (5) Characteristic parameter statistics table, including quantitative parameters such as global tensile amount, maximum strain gradient value, and area ratio of stress concentration region.
[0055] It should be noted that the multi-scale deformation analysis report not only provides overall tensile deformation information of the specimen, but also reveals deformation anomalies in local areas, providing richer data support for a comprehensive evaluation of the material's mechanical properties. This report can identify weak points in the material and predict potential failure locations.
[0056] This implementation reconstructs the displacement of discrete feature points into a continuous drift field using a radial basis function interpolation algorithm. This expands the data, which originally only characterized the displacement of discrete points, to deformation information covering the entire surface of the specimen, overcoming the problem of insufficient spatial resolution caused by discrete sampling. The continuous drift field provides displacement information at any position on the specimen surface, eliminating the spatial blind spots caused by the limitation of feature point density and solving the technical problem of not being able to obtain full-domain deformation information.
[0057] This implementation separates global and local deformation components through multi-scale wavelet decomposition, avoiding the masking effect of averaging operations on local anomalous deformations. The global tensile component reflects the overall uniform deformation of the specimen, while the local perturbation component reveals non-uniform deformation caused by material defects and stress concentration. This multi-scale separation mechanism can accurately locate stress concentration areas, overcoming the inability of traditional average relative displacement methods to capture spatial deformation non-uniformity, and solving the technical problem of deformation characteristics being masked in local stress concentration areas.
[0058] This implementation converts displacement information into a physical quantity that directly reflects the internal stress state of the material by calculating the strain gradient tensor. The strain gradient distribution shows the spatial rate of change of strain on the specimen surface, and the peak regions of the strain gradient correspond to the locations of stress concentration, which are often the initiation points of material failure. The spatial distribution map can predict potential crack initiation locations, solving the problem that traditional methods cannot identify potential defect locations and improving the comprehensiveness and predictive ability of material mechanical property evaluation.
Claims
1. A method for compensating for tensile testing errors and detecting elastic deformation using a universal testing machine, characterized in that, Includes the following steps: Multiple feature points are set in the non-clamping area on the surface of the specimen; The specimen is stretched, and continuous frame image data from the start to the end of the stretching is acquired from the front of the universal testing machine, and the force value data is recorded simultaneously. The continuous frame image data is preprocessed to generate a standard image sequence; A feature point tracking algorithm is used to track the position changes of each feature point in the standard image sequence in different frames to obtain the displacement trajectory of each feature point. Select any two feature points to form a feature point pair, calculate the difference between the spatial distance of the feature point pair in the current frame and the initial distance in the initial frame, and obtain the relative displacement of the feature point pair; The relative displacements of all feature point pairs are averaged to calculate the true tensile length variation sequence of the specimen. Based on the actual tensile length change sequence and the force data, the stress and strain values at each moment are calculated to construct the stress-strain curve of elastic deformation. The continuous frame image data and the force value data are synchronized and correspond to each other through timestamp marking.
2. The method for error compensation and elastic deformation detection in tensile testing of a universal testing machine according to claim 1, characterized in that, The number of feature points is at least three, and the multiple feature points are evenly distributed in the non-clamping area of the specimen body.
3. The method for error compensation and elastic deformation detection in tensile testing of a universal testing machine according to claim 1, characterized in that, The preprocessing includes filtering, grayscale normalization, distortion correction, and registration.
4. The method for error compensation and elastic deformation detection in tensile testing of a universal testing machine according to claim 1, characterized in that, The feature point tracking algorithm is used to track the position changes of each feature point in the standard image sequence in different frames, and to obtain the displacement trajectory of each feature point, including: Extract the initial frame of the standard image sequence and identify the initial position coordinates of each feature point in the initial frame; The KLT tracking algorithm is used to track the position coordinates of each feature point in the standard image sequence, excluding the initial frame. The position coordinates of each feature point in the standard image sequence are integrated into a time series to generate the displacement trajectory of each feature point.
5. The method for error compensation and elastic deformation detection in tensile testing of a universal testing machine according to claim 1, characterized in that, Calculating the spatial distance between the feature point pairs in the current frame includes: Obtain the position coordinates of the first and second feature points in the feature point pair in the current frame, wherein the position coordinates include horizontal and vertical coordinates; Calculate the difference in horizontal coordinates and the difference in vertical coordinates between the first feature point and the second feature point, respectively; The spatial distance between the feature point pairs in the current frame is obtained by calculating the square root of the sum of the squares of the horizontal coordinate differences and the squares of the vertical coordinate differences.
6. The method for error compensation and elastic deformation detection in tensile testing of a universal testing machine according to claim 5, characterized in that, Obtaining the relative displacement of the feature point pair includes: The initial distance of the feature point pair in the initial frame is obtained using the same calculation method; The difference between the spatial distance of the feature point pair in the current frame and the initial distance in the initial frame is calculated to obtain the relative displacement of the feature point pair.
7. The method for error compensation and elastic deformation detection in tensile testing of a universal testing machine according to claim 1, characterized in that, The relative displacements of all feature point pairs are averaged to calculate the true tensile length variation sequence of the specimen, including: Count the total number of all selected feature point pairs; Sum the relative displacements of all feature point pairs; Divide the summation result by the total number of feature point pairs to obtain the actual tensile length change value of the specimen corresponding to the current frame; Perform the above calculations on all frames in the standard image sequence to generate the true tensile length change sequence of the specimen.
8. The method for error compensation and elastic deformation detection in tensile testing of a universal testing machine according to claim 1, characterized in that, The continuous frame image data and the force value data are synchronized and correspond to each other through timestamp marking, including: While triggering the camera to acquire images, the force value data output by the universal testing machine at the current moment is recorded simultaneously; Each frame of image and its corresponding force value data is marked with the same timestamp; A one-to-one correspondence is established between the consecutive frame image data and the force value data based on the timestamp.
9. The method for error compensation and elastic deformation detection in tensile testing of a universal testing machine according to claim 1, characterized in that, Based on the actual tensile length change sequence and the force data, the stress and strain values at each moment are calculated to construct the stress-strain curve of elastic deformation, including: Based on the timestamp, extract the force value data and the actual tensile length change data corresponding to each frame; The stress value corresponding to each frame is calculated based on the force data and the cross-sectional area of the specimen. The strain value corresponding to each frame is calculated based on the actual tensile length change data and the initial gauge length of the specimen; The stress and strain values corresponding to each timestamp are combined into point-to-point data, which are then plotted into a two-dimensional curve to generate the stress-strain curve of elastic deformation.
10. A universal testing machine tensile testing error compensation and elastic deformation detection system, used to execute the universal testing machine tensile testing error compensation and elastic deformation detection method according to any one of claims 1 to 9, characterized in that, include: A universal testing machine is used to apply tensile force to a specimen and output force value data. An industrial camera, fixed to the front of the universal testing machine, is used to acquire continuous frame image data of the surface of the specimen; A data acquisition system is used to synchronously record the continuous frame image data and the force value data, and to establish the correspondence between the two by using timestamps; Computer equipment is used to preprocess the continuous frame image data, track feature points, calculate relative displacement, calculate the actual tensile length change, and construct stress-strain curves.