Deformation tracking method for aluminum material in stamping oil-free stamping process

Abstract

The invention provides a deformation tracking method for an aluminum material in a stamping oil-free stamping process, and the method comprises the steps: calculating the change rate of a direction angle between adjacent images according to an initial direction angle sequence of a texture direction, recognizing a potential jump region, and determining a jump position set and a corresponding space coordinate; obtaining deformation path data of a corresponding area from the jump position set and the corresponding space coordinates, and simulating a shear band evolution trend to obtain a risk change trend chart in a local shear set; for the local shear concentration risk change trend chart, identifying and marking a high-risk area with local reverse rotation, and determining a corrected flow track according to a flow path deviation analysis result; according to the corrected flow track, material flow instability critical state data in the deformation increasing stage are obtained, microcrack initiation signals are continuously monitored, and initiation probability distribution is obtained; and judging an area of which the probability is higher than a preset threshold value from the initiation probability distribution, generating an instability initial position map, and determining extension path prediction.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for tracking deformation during the non-stamping process of aluminum-based materials. Background Technology

[0002] The forming process of aluminum alloy sheets under oil-free stamping conditions is widely used in the production of lightweight structural components for automobiles and aerospace. This field is crucial for improving material utilization, reducing environmental pollution, and ensuring the dimensional accuracy of parts. Accurate tracking of material flow behavior during stamping directly determines forming quality and defect control capabilities, and is a core requirement for achieving high-precision oil-free stamping. Current methods relying on surface texture orientation angles to characterize material flow often exhibit judgment biases as deformation increases. The principal orientation angle of the surface texture should smoothly deflect with deformation to reflect the true flow direction, but in actual observation, abnormal jumps or local reversals frequently occur, causing the extracted orientation angles to fail to reliably correspond to the actual material flow path. This bias makes it difficult for process engineers to detect flow anomalies in a timely manner using conventional morphological images, thus affecting early intervention for defects. The root cause of this bias lies in the complex relationship between the principal axis of surface texture directionality and material flow behavior. When the deformation path is complex, the gradual evolution of local shear bands significantly alters the texture formation mechanism, and anisotropic hardening further exacerbates this effect. Under the combined effect of these two factors, once the deformation exceeds certain critical states or triggers the initiation of microcracks, the main texture axis may undergo discontinuous jumps or local reverse rotations. These jumps cause the direction angles extracted based on single or few morphological observations to lose continuity, failing to accurately reflect the true flow trajectory of the material throughout the deformation process. For example, in the middle of deep drawing, a region that should exhibit a uniform radial flow texture direction may suddenly experience a local reversal of the main texture axis, leading to the region being misjudged as having normal flow, when in fact shear concentration has begun to form, ultimately developing into surface orange peel or cracking defects in subsequent deformation. Therefore, how to continuously and accurately capture the true evolution trajectory of the main surface texture direction angle with the amount of deformation during unlubricated stamping, thereby reliably identifying the starting position and propagation path of local flow instability, becomes a key issue for achieving early warning and dynamic process control. Summary of the Invention

[0003] This invention provides a method for tracking deformation during the non-stamping process of aluminum-based materials, mainly including: This process involves acquiring surface texture image sequences at multiple time points during the non-stamping process of aluminum-based materials, extracting the principal direction angle distribution of the texture in each image to obtain the initial direction angle sequence of the texture direction, calculating the direction angle change rate between adjacent images based on the initial direction angle sequence, identifying potential jump regions, and determining the jump location set and corresponding spatial coordinates, obtaining deformation path data for the corresponding regions from the jump location set and corresponding spatial coordinates, simulating the shear band evolution trend, and obtaining a local shear concentration risk change trend map, identifying and marking high-risk areas with local reverse rotation based on the local shear concentration risk change trend map, and determining the corrected flow trajectory through flow path deviation analysis results, obtaining material flow instability critical state data during the deformation increase stage based on the corrected flow trajectory, continuously monitoring microcrack initiation signals, and obtaining the initiation probability distribution, identifying regions with probabilities higher than a preset threshold from the initiation probability distribution, generating an instability initiation location map, and determining the propagation path prediction, and using the propagation path prediction to fuse texture direction and flow trajectory data from all time points to generate a complete flow behavior report, identifying flow anomaly regions caused by texture principal axis reversal in the report to achieve early warning. Furthermore, the step of acquiring a sequence of surface texture images at multiple time points during the stamping process of aluminum-based materials without stamping oil, extracting the principal direction angle distribution of the texture in each image, and obtaining the initial direction angle sequence of the texture direction includes: During the stamping and deformation process of aluminum alloy sheets, surface images are acquired at preset time intervals, and the corresponding stamping stroke position and forming force value are recorded. A correspondence between timestamps and deformation amounts is established to obtain an original image sequence marked with deformation amounts. For each image in the original image sequence, grayscale processing and histogram equalization are performed. Texture response values ​​in each direction are extracted at set intervals within a preset angle range. The main texture direction of each pixel is determined by comparing the response intensity in each direction, and a texture direction field matrix is ​​generated. Based on the texture direction field matrix, the frequency distribution of the main direction angle is statistically analyzed within each preset size pixel region. The main direction angles of each region are arranged in the order of image coordinates to obtain the initial direction angle sequence of the texture direction.

[0004] Furthermore, the step of calculating the rate of change of orientation angle between adjacent images based on the initial orientation angle sequence of the texture direction, identifying potential transition regions, and determining the set of transition positions and their corresponding spatial coordinates includes: Extract the orientation angle data of the same spatial position at adjacent time points from the initial orientation angle sequence of the texture direction, calculate the orientation angle difference of corresponding pixel regions in two consecutive frames of images, and record their time index, spatial coordinates, and angle difference to obtain a set of jump candidate points; based on the spatial coordinates of each point in the set of jump candidate points, use the DBSCAN algorithm to identify the degree of aggregation of jump points on the board surface and obtain the boundary range of the region; for potential jump regions, calculate the actual physical coordinates of each jump point on the board through the pre-calibrated transformation relationship between the image coordinate system and the board physical coordinate system, arrange them in the order of jump occurrence time, and determine the set of jump positions and their corresponding spatial coordinates.

[0005] Furthermore, the step of obtaining deformation path data for the corresponding region from the set of jump locations and their corresponding spatial coordinates, simulating the evolution trend of the shear band, and obtaining a local shear concentration risk change trend map includes: The coordinate sequence of each jump point at consecutive time steps is extracted from the set of jump locations. The ratio of displacement increment to time interval between adjacent time steps is calculated to obtain the velocity field. The spatial derivative of velocity is calculated using the finite difference method to obtain the strain rate component. The strain rate is accumulated over time to obtain the local strain tensor distribution. The principal strain value and direction are extracted based on the local strain tensor distribution. Combined with the aluminum alloy yield strength, hardening index, and anisotropy coefficient, the equivalent stress is calculated using the Hill yield criterion. For the strain data within the local strain tensor distribution, the local strain tensor and material parameters are input into the anisotropic hardening model. The initiation probability value is calculated using the shear band formation criterion. The shear band development direction is determined based on the direction of the maximum shear strain increment, obtaining the shear band evolution characteristic data for each location point. Based on the shear band evolution characteristic data, a gridded evaluation matrix is ​​established on the plate surface. Spatially continuous distribution is achieved by weighted averaging of risk values ​​of adjacent grid points, and a local shear concentration risk change trend map is plotted.

[0006] Furthermore, the process of identifying and marking high-risk areas with local reverse rotation based on the local shear concentration risk trend map, and determining the corrected flow trajectory through flow path deviation analysis results, includes: For each grid point in the local shearing concentrated risk change trend map, the texture principal axis angle of the corresponding position in the image at adjacent time points is extracted, the angle change between consecutive frames is calculated, and the average value of the angle change in the surrounding preset neighborhood is counted as the mainstream change direction. If the sign of the angle change is opposite to the mainstream change direction and the absolute value exceeds a preset threshold, it is determined that a local reverse rotation has occurred at that position, and the set of coordinates of the reverse rotation points is recorded. Based on the set of coordinates of the reverse rotation points, a preset rotation anomaly weight value is added to the original risk level value of the corresponding position in the risk change trend map to generate a comprehensive risk distribution map. Based on the comprehensive risk distribution map, the original flow trajectory is corrected point by point to determine the corrected flow trajectory.

[0007] Furthermore, based on the comprehensive risk distribution map, the actual flow direction of each point in the area is obtained, the deviation angle between the actual flow direction and the ideal flow direction is calculated, the spatial distribution function of the deviation angle is obtained by least squares fitting, and a flow path deviation correction matrix is ​​constructed; the flow path deviation correction matrix is ​​used to correct the original flow trajectory point by point.

[0008] Furthermore, after correcting the original flow trajectory point by point using the flow path deviation correction matrix, the corrected displacement field is obtained by vector synthesis of the direction correction amount in the correction matrix and the original displacement vector, and the material point motion trajectory is reconstructed by time integration of the corrected displacement field.

[0009] Furthermore, the step of obtaining critical state data of material flow instability during the deformation increase stage based on the corrected flow trajectory, continuously monitoring microcrack initiation signals, and obtaining the initiation probability distribution includes: Based on the corrected flow trajectory, the cumulative equivalent strain value of each material point is extracted, and the strain rate is calculated. When the strain rate reaches its peak and then shows a continuous downward trend, the material is determined to have entered a critical instability state. The strain value, stress state, and deformation rate at this moment are recorded as the critical instability state data of the material flow. For the high strain region in the critical instability state data of the material flow, where the strain value exceeds a preset threshold, the surface morphology change signal and internal elastic wave signal of the region are collected, and the number of germination signals per unit time is counted. Based on the number of germination signals, the microcrack germination frequency of each monitoring point within a preset time window is calculated. The discrete frequency data is spatially convolved and smoothed using a Gaussian kernel function using a kernel density estimation method to obtain a continuous germination probability distribution.

[0010] Furthermore, the step of identifying regions with probabilities higher than a preset threshold from the germination probability distribution, generating an instability initiation location map, and determining the predicted expansion path includes: The coordinates of grid points with probability values ​​exceeding a preset threshold are extracted from the initiation probability distribution. The material plastic strain limit value and current cumulative strain value corresponding to the grid points are obtained, and the remaining plastic deformation capacity index is calculated to obtain a set of instability candidate points. Based on the set of instability candidate points, a connected component labeling algorithm is used to identify spatially adjacent groups of instability points. The geometric center coordinates of each group are calculated as the instability initiation position. The dominant instability direction is determined by combining the direction information of the corrected flow trajectory, and an instability initiation position map containing position coordinates and direction information is generated. Based on the instability initiation position map, the strain gradient distribution is calculated along the dominant instability direction. The direction with the maximum gradient is taken as the crack preferential propagation direction. The stress distribution state in this direction is obtained through finite element stress analysis to obtain the propagation path prediction.

[0011] Furthermore, by extending path prediction and fusing texture direction and flow trajectory data from all time points, a complete flow behavior report is generated. This report identifies abnormal flow areas caused by texture axis reversal to achieve early warning, including: By extending the path prediction results, the texture direction data and flow trajectory data at all time points are arranged in chronological order. Data at the same time and location are correlated and matched to form a flow behavior report containing the complete deformation process. Based on the flow behavior report, the records of texture principal axis angle changes in the report are scanned. If a region with reverse rotation of the angle exceeding a preset threshold is detected, it is determined to be a flow abnormal region.

[0012] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects: This invention discloses a method for tracking deformation during the stamping process of aluminum-based materials without stamping oil. This method addresses the material instability problems caused by abnormal surface texture, shear band concentration, and microcrack initiation in aluminum alloy sheet stamping under unlubricated conditions. It acquires surface texture image sequences at multiple time points, uses a continuous tracking algorithm to extract the distribution of the main direction angle of the texture and form an initial direction angle sequence, and then calculates the rate of change of direction angle between adjacent images to identify potential jump regions and determine the jump location and its spatial coordinates. Subsequently, based on the deformation path data of the jump regions, and combined with an anisotropic hardening model to simulate the probability and development direction of shear band initiation, a local shear concentration risk change trend map is generated. The difference in the main axis angle of the texture between consecutive frames is used to identify high-risk areas of reverse rotation, and the flow trajectory is corrected to obtain the probability distribution of microcrack initiation under the critical state of instability. High-probability areas are identified, and combined with the material's plastic deformation capacity assessment, the instability initiation location and propagation path prediction are generated. Finally, the texture direction and flow trajectory data from all time points are integrated to generate a complete flow behavior report, achieving early warning of abnormal flow areas caused by texture main axis reversal. This invention provides a non-contact real-time monitoring method by integrating dynamic surface texture tracking and multi-step simulation, effectively predicting and preventing the risk of local shear instability and crack propagation in dry friction stamping, thereby improving the forming quality and reliability of aluminum materials. Attached Figure Description

[0013] Figure 1 This is a flowchart of a method for tracking deformation during the non-stamping process of aluminum-based materials according to the present invention.

[0014] Figure 2 This is a schematic diagram of a method for tracking deformation during the non-stamping process of aluminum-based materials according to the present invention.

[0015] Figure 3 This is another schematic diagram of a method for tracking deformation during the non-stamping process of aluminum-based materials according to the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] like Figures 1-3 This embodiment of a method for tracking deformation during the non-stamping process of aluminum-based materials may specifically include: Step S101: Obtain surface texture image sequences at multiple time points during the stamping process of aluminum materials without stamping oil, extract the main direction angle distribution of texture in each image, and obtain the initial direction angle sequence of texture direction.

[0018] Surface images of aluminum alloy sheets are acquired at preset time intervals using a high-speed camera during the stamping and deformation process. Simultaneously, the corresponding stamping stroke position and forming force value are recorded. A correspondence between timestamps and deformation amounts is established based on the stroke position, resulting in an original image sequence marked with deformation amounts. Each image in the original image sequence undergoes grayscale conversion and histogram equalization. A Gabor filter bank is used to extract texture response values ​​in each direction at set intervals within a preset angle range. The principal direction of the texture at each pixel is determined by comparing the response intensity in each direction, generating a texture direction field matrix. Based on the texture direction field matrix, the frequency distribution of the principal direction angle is statistically analyzed within each preset-sized pixel region. If the frequency within a certain angle range exceeds a preset threshold, the principal direction angle of that region is determined to be the median of that angle. The principal direction angles of each region are arranged in image coordinate order to obtain an initial direction angle sequence for the texture direction.

[0019] In one embodiment, a high-speed camera acquisition system is arranged above the stamping machine, with the lens vertically aligned with the surface of the aluminum alloy sheet. The system is synchronized with the stamping machine control system via a trigger signal to achieve precise correspondence between image acquisition and stamping stroke.

[0020] Specifically, as the stamping punch begins its descent, a high-speed camera continuously acquires images of the sheet metal surface at preset time intervals. The timestamp of each image is calibrated using the stamping stroke position read by the encoder. Simultaneously, the forming force value recorded by the pressure sensor is also stored in association with this timestamp, forming an original image sequence containing three-dimensional information of time, position, and forming force. The Gabor filter bank is constructed based on a two-dimensional Gabor function, generating multiple filter kernels in different directions within a preset angle range by adjusting the direction parameter θ.

[0021] In one possible implementation, each image in the original image sequence is first converted to grayscale to eliminate the interference of color information on texture analysis. Then, histogram equalization is performed to enhance image contrast and make texture features more prominent. Each Gabor filter is convolved with the processed image to obtain a texture response map for the corresponding direction. By comparing the response intensity values ​​of the same pixel location across all directional response maps, the direction with the largest response is selected as the principal texture direction for that pixel. The principal texture directions of all pixels are aggregated to form a texture direction field matrix, where each element stores the principal direction angle value for the corresponding pixel location.

[0022] For example, when performing region statistical analysis based on the texture orientation field matrix, a fixed-size sliding window is used to move pixel by pixel on the matrix.

[0023] It is understandable that the main direction angle values ​​of all pixels contained within the window constitute a local directional distribution. The frequency of occurrence of each angle range is calculated by the histogram statistical method. When the frequency of a certain angle range exceeds a preset threshold, the region is considered to have a consistent texture directionality, and the median value of that angle range is taken as the main direction angle of the entire window region.

[0024] For example, in the flange area of ​​aluminum alloy sheet deep drawing, the texture direction usually exhibits radial distribution characteristics. The above statistical method can accurately identify the dominant flow direction in a local area.

[0025] In one embodiment, the principal orientation angles obtained from each window area are arranged according to their spatial coordinate positions in the image to form an initial angle sequence of texture direction corresponding to the deformation moment. This sequence reflects the material flow direction distribution characteristics of the plate surface at a specific deformation stage.

[0026] Step S102: Based on the initial orientation angle sequence of the texture direction, calculate the orientation angle change rate between adjacent images, identify potential jump regions, and determine the set of jump positions and their corresponding spatial coordinates.

[0027] The orientation angle data of the same spatial position at adjacent time points are extracted from the initial orientation angle sequence of the texture direction. The orientation angle difference of corresponding pixel regions in two consecutive frames is calculated. If the absolute value of the difference is greater than a preset angle threshold, the position is marked as a candidate jump point, and its time index, spatial coordinates, and angle difference are recorded to obtain a jump candidate point set. Based on the spatial coordinates of each point in the jump candidate point set, the DBSCAN algorithm is used to identify the degree of aggregation of jump points on the board surface. When the density of jump points in a certain area exceeds a preset density threshold, it is determined as a potential jump region, and the boundary range of the region is obtained. For the potential jump regions, the actual physical coordinates of each jump point on the board are calculated through the pre-calibrated transformation relationship between the image coordinate system and the board physical coordinate system. The jump points are arranged in the order of jump occurrence time to determine the jump position set and its corresponding spatial coordinates.

[0028] In one implementation, when performing temporal comparison analysis on the initial angle sequence of the texture direction, the frame-by-frame difference method is used to calculate the rate of change of the direction angle.

[0029] Specifically, for the corresponding spatial positions of the i-th frame and the (i+1)-th frame in the sequence, the orientation angle values ​​of the same coordinate points are extracted. and Calculate the angle difference When the periodicity of the angle needs to be considered for this difference, take... The actual change is then divided by the inter-frame time interval to obtain the rate of change. When identifying spatial clustering features of transition points, the DBSCAN algorithm requires setting two key parameters: the neighborhood radius ε and the minimum number of points. .

[0030] Preferably, the neighborhood radius ε is determined based on the plate size and imaging resolution, and is typically set to 1 / 50 to 1 / 100 of the plate width; minimum number of points The number of points is typically set to 3 to 5 based on the expected density of the transition points within a unit area.

[0031] In one possible implementation, the algorithm starts with any unvisited jump candidate point and searches for all candidate points within its ε-neighborhood. If the number of points in the neighborhood is greater than or equal to... If the point is marked as the core point, a new cluster is created, and the search continues to expand until no more new points can be included, forming multiple transition point clusters, each cluster corresponding to a potential transition region.

[0032] For example, during the deep drawing process of aluminum alloy sheets, abrupt changes in texture direction often occur at the junction of the flange area and the rounded corner area. The above cluster analysis can accurately locate these key deformation areas.

[0033] For example, when more than five candidate transition points are detected clustered within a radius of 10 pixels in a certain region, the region is identified as a potential transition region, and its boundary is determined by calculating the minimum bounding rectangle of all points within the cluster. The conversion between the image coordinate system and the physical coordinate system of the board material needs to be pre-calibrated.

[0034] In one embodiment, a calibration plate of known size is placed on the surface of the plate, and calibration images are captured using a high-speed camera. The 3×3 homography transformation matrix H is calculated using the correspondence between feature points on the calibration plate in the image coordinate system and the physical coordinate system. Further, for any image coordinate point within the potential transition region... Through matrix operations Obtain the corresponding physical coordinates This allows for accurate positioning of the jump position within the actual space of the board.

[0035] Step S103: Obtain deformation path data of the corresponding region from the set of jump positions and corresponding spatial coordinates, simulate the evolution trend of shear zone, and obtain a local shear concentration risk change trend map.

[0036] The coordinate sequence of each jump point in consecutive time steps is extracted from the set of jump locations. The ratio of displacement increment to time interval between adjacent time steps is calculated to obtain the velocity field. The spatial derivative of the velocity is calculated using the finite difference method to obtain the strain rate component. The strain rate is accumulated over time to obtain the local strain tensor distribution. The principal strain values ​​and directions are extracted based on the local strain tensor distribution. Combined with the aluminum alloy yield strength, hardening index, and anisotropy coefficient, the equivalent stress is calculated using the Hill yield criterion. If the equivalent stress exceeds a preset multiple of the current yield stress of the material, it is determined that the material has entered a plastic instability state, and the range of the instability region is determined. For the strain data within the range of the instability region, the local strain tensor and material parameters are input into the anisotropic hardening model. The initiation probability value is calculated using the shear band formation criterion. The shear band development direction is determined based on the direction of the maximum shear strain increment, and the shear band evolution characteristic data of each location point is obtained. Based on the shear band evolution characteristic data, a gridded evaluation matrix is ​​established on the surface of the plate. A risk level value is assigned according to the initiation probability of each grid point. The spatial continuous distribution is achieved by weighted averaging of the risk values ​​of adjacent grid points, and a trend map of local shear concentration risk changes is plotted.

[0037] In one implementation, when extracting deformation path data from the set of jump positions, it is necessary to establish a complete kinematic description system.

[0038] Specifically, for each transition point, its complete trajectory from the initial time t0 to the current time t is recorded by tracking its position change in the time series image. The displacement increment Δu between adjacent time points is obtained through an image matching algorithm, and the velocity field... It reflects the instantaneous flow state of the material. The strain rate tensor is calculated using the finite difference method, and the velocity gradient tensor is obtained by spatially differentiating the velocity field. , where x represents the spatial coordinate.

[0039] Preferably, the velocity gradient tensor is decomposed into a symmetric part and an antisymmetric part, the symmetric part being the strain rate tensor. By integrating the strain rate in the time domain Obtain the cumulative strain tensor ε, which contains complete information about the material deformation history. Its main diagonal elements represent the normal strain components, and the off-diagonal elements represent the shear strain components.

[0040] For example, the application of the Hill yield criterion in determining the plastic instability of aluminum alloy sheets takes into account the anisotropic properties of the material.

[0041] In one possible implementation, the Hill yield function is expressed as: Where F, G, H, L, M, and N are anisotropic parameters, determined by the ratio of yield stress measured in the rolling direction, transverse direction, and 45° direction through uniaxial tensile tests. When the equivalent stress... Exceeding the material's current yield stress When the strain is 1.1 times that of the material, the region is considered to have entered a state of plastic instability, at which point strain concentration begins to occur inside the material.

[0042] Understandably, anisotropic hardening models are a core tool for predicting shear band evolution.

[0043] Specifically, the model receives the local strain tensor ε and material parameters as inputs, including the initial yield strength σ0, hardening exponent n, anisotropy coefficient r, and strain rate sensitivity coefficient m.

[0044] In one embodiment, the model first calculates the hardening modulus under the current strain state. When the hardening modulus decreases to a certain critical value, the material loses its load-bearing capacity, and shear bands begin to emerge. The probability P of shear band emergence is determined by the instability criterion. Calculate, where W is the strain energy density function, Its second derivative, when When the probability of initiation changes from positive to negative, the initiation probability increases sharply. The development direction of the shear band is determined by analyzing the maximum shear strain increment. It is confirmed that this direction forms a 45° angle with the principal stress direction, and the material is most prone to shear failure along this direction.

[0045] For example, in the deep drawing of cup-shaped parts from 5052 aluminum alloy sheet, the rounded corner area of ​​the punch is subjected to a complex stress state, with radial tensile stress and circumferential compressive stress acting together to reduce the material thickness. Furthermore, when the local equivalent strain reaches 0.3, the probability of shear band initiation in this area exceeds 60%, indicating a high risk of fracture. In one embodiment, the shear band evolution characteristic data includes three key parameters: the initiation probability value, the development direction angle, and the propagation rate at each evaluation point. These parameters collectively describe the spatial distribution and temporal evolution of shear instability. The gridded evaluation matrix is ​​constructed by covering the entire sheet surface with a regular quadrilateral grid. The grid size is determined based on the sheet size and the required resolution, typically taking 1 / 100 of the sheet's characteristic size.

[0046] Preferably, the risk level value R of each grid node is assigned in a graded manner according to the occurrence probability P. When P < 0.2, R = 1 indicates low risk; when 0.2 ≤ P < 0.5, R = 2 indicates medium risk; and when P ≥ 0.5, R = 3 indicates high risk.

[0047] For example, the risk values ​​between adjacent grid points are spatially continuous through bilinear interpolation. For any location (x, y), the risk value is obtained by weighted averaging of the risk values ​​of its four surrounding grid nodes. Where α and β are normalized local coordinates, , , , Using the risk values ​​of four adjacent nodes, this interpolation method ensures the continuity and smoothness of the risk distribution. The resulting local shear concentration risk change trend diagram can intuitively reflect the instability risk distribution in each region of the plate.

[0048] Step S104: Based on the local shear concentration risk change trend map, identify and mark the high-risk areas with local reverse rotation, and determine the corrected flow trajectory through the flow path deviation analysis results.

[0049] For each grid point in the local shear concentration risk change trend map, the texture principal axis angle of the corresponding position in the image at adjacent time points is extracted, the angle change Δθ between consecutive frames is calculated, and the average value of the angle change in the surrounding preset neighborhood is counted as the mainstream change direction. If the sign of Δθ is opposite to the mainstream change direction and the absolute value exceeds a preset threshold, it is determined that a local reverse rotation has occurred at that position, and the set of coordinates of the reverse rotation points is recorded. Based on the set of coordinates of the reverse rotation points, a preset rotation anomaly weight value is added to the original risk level value of the corresponding position in the risk change trend map. If the accumulated comprehensive risk value exceeds the high risk threshold, the area is marked as a high-risk area, and a comprehensive risk distribution map is generated. Based on the high-risk area in the comprehensive risk distribution map, the actual flow direction of each point in the area is obtained. The radial flow determined according to the sheet metal forming theory is the ideal flow direction. The deviation angle between the actual and ideal flow directions is calculated, and the spatial distribution function of the deviation angle is obtained through least squares fitting. Based on this, the flow direction correction amount of each point is calculated, and a flow path deviation correction matrix is ​​constructed. The original flow trajectory is corrected point by point using the flow path deviation correction matrix. The corrected displacement field is obtained by vector synthesis of the direction correction amount in the correction matrix and the original displacement vector. The material point motion trajectory is reconstructed by time integration of the corrected displacement field to determine the corrected flow trajectory.

[0050] In one implementation, for texture angle analysis of the risk change trend map of local shear concentration, it is necessary to establish a time-series comparison mechanism to identify abnormal rotation phenomena.

[0051] Specifically, for each grid point in the risk trend map, the sequence of texture principal axis angles for that point across multiple consecutive frames is extracted. The rotation change is quantified by calculating the angle difference Δθ between adjacent time points. The calculation of the angle difference needs to consider the periodicity of the angles. When two angles are near 0° and 180° respectively, direct subtraction will produce incorrect results. Therefore, a different approach is adopted. The actual angle change is calculated using this method.

[0052] Preferably, the mainstream direction of change is determined using statistical methods. A neighborhood window of a preset size is selected around the target point, usually a 5×5 or 7×7 grid. The angle change values ​​of all points within the window are statistically analyzed, and their mean and standard deviation are calculated.

[0053] For example, when more than 80% of the points within the window have positive angle changes, the mainstream rotation direction is considered positive; otherwise, it is considered negative rotation. If the angle change of the target point is opposite to the mainstream direction and its absolute value exceeds a preset threshold, it is determined to be a local reverse rotation. In the deep drawing process of aluminum alloy sheets, this reverse rotation often occurs in areas where material flow is obstructed, indicating a potential risk of wrinkling or cracking.

[0054] Understandably, the construction of a comprehensive risk distribution map requires the integration of multi-dimensional risk information.

[0055] In one possible implementation, the original risk level value The probability of shear band emergence is used to determine the value, which ranges from 0 to 1. When reverse rotation is detected, rotational anomaly weights are introduced. This weight is dynamically adjusted based on the magnitude of the reverse rotation angle. When the reverse angle is between 15° and 30°, =0.3; when the reversal angle is between 30° and 45°, When the angle exceeds 45°, =0.7. Overall Risk Value = + ,like If the risk level exceeds a preset high-risk threshold of 0.8, the area is marked as a high-risk zone. This dual-risk assessment mechanism can more accurately identify potential failure areas and avoid missed detections caused by a single indicator.

[0056] For example, in the forming process of automotive body panels, complex stress states often occur at the corners of the window frames of the inner door panels, where the material must withstand tensile stress while also adapting to changes in curvature. Furthermore, the determination of the ideal flow direction is based on the principle of minimum resistance in sheet metal forming theory.

[0057] Specifically, in axisymmetric deep-drawn parts, the ideal flow direction is radial from the flange edge to the die center; in non-axisymmetric parts, the ideal flow direction is obtained by solving the gradient direction of the plastic flow potential function. The deviation angle φ between the actual flow direction and the ideal flow direction reflects the degree of non-uniformity of material flow; the larger the deviation angle, the worse the deformation compatibility in that region.

[0058] It should be noted that the application of the least squares method in fitting the spatial distribution of the deviation angle aims to obtain a smooth and continuous correction field.

[0059] In one embodiment, a quadratic polynomial is chosen as the fitting function. Where x and y are the coordinates of the plate surface. to These are undetermined coefficients. The coefficients are solved by minimizing the sum of squared errors, yielding the spatial distribution function of the deviation angle. The flow direction correction for each grid point is then calculated based on this function. , where k is the correction coefficient, usually taken as 0.5 to 0.8, construct the flow path deviation correction matrix M, in which each element contains the direction correction amount and correction intensity at the corresponding position.

[0060] For example, the trajectory correction process of vector synthesis requires converting the correction amount into actual displacement adjustment.

[0061] Preferably, the original displacement vector Decomposed into two components: magnitude |V| and direction θ, with the corrected direction... The corrected displacement vector The corrected material point trajectory is reconstructed by integrating the corrected displacement field in the time domain. In one embodiment, a fourth-order Runge-Kutta method is used for numerical integration, with a time step of 1 / 100 of the total forming time to ensure the accuracy of the trajectory reconstruction. The corrected flow trajectory can more accurately reflect the actual deformation path of the material.

[0062] Step S105: Based on the corrected flow trajectory, obtain the critical state data of material flow instability during the deformation increase stage, continuously monitor the microcrack initiation signal, and obtain the initiation probability distribution.

[0063] The cumulative equivalent strain value of each material point is extracted based on the corrected flow trajectory. The strain rate, i.e., the ratio of strain increment to time increment, is calculated. When the strain rate reaches its peak and then shows a continuous downward trend, the material is determined to have entered a critical instability state. The strain value, stress state, and deformation rate at this moment are recorded as the critical instability state data of the material flow. For the high-strain region in the critical instability state data of the material flow, where the strain value exceeds a preset threshold, surface morphology change signals and internal elastic wave signals of this region are collected. When the signal amplitude exceeds a preset multiple of the background noise and the signal characteristics meet the preset crack initiation criteria, it is identified as a microcrack initiation signal. The number of initiation signals per unit time is counted. Based on the number of initiation signals, the microcrack initiation frequency of each monitoring point within a preset time window is calculated. The discrete frequency data is spatially convolved and smoothed using a kernel density estimation method through a Gaussian kernel function to obtain a continuous initiation probability distribution.

[0064] In one implementation, when identifying instability based on the corrected flow trajectory, it is necessary to track the deformation history of the material points throughout the entire process.

[0065] Specifically, by extracting the complete path of each material point from its initial position to its current position, the cumulative equivalent strain at each time step along the path is calculated. ,in For the strain tensor components, the peak value of the strain rate corresponds to the most intense stage of plastic flow in the material. When the strain rate continues to decrease from the peak value and maintains a downward trend for more than a preset time threshold, it indicates that the material has lost its ability to deform uniformly and has entered a critical state of instability.

[0066] Preferably, the identification of microcrack initiation signals adopts a multi-source information fusion method.

[0067] For example, the surface morphology change signal is obtained by laser scanning to acquire surface roughness evolution data. When the roughness growth rate suddenly accelerates, it indicates that microcracks may occur. The internal elastic wave signal is acquired by a piezoelectric sensor with a frequency range set between 100 kHz and 1 MHz, which corresponds to the microcrack initiation characteristic frequency of aluminum alloy materials.

[0068] In one possible implementation, the crack initiation criteria include: the signal amplitude exceeding three times the background noise; the signal duration being in the range of 10 to 100 microseconds; and spectral analysis showing that the dominant frequency is concentrated in a characteristic frequency band. When these criteria are met simultaneously, the event is identified as a valid microcrack initiation event.

[0069] Understandably, kernel density estimation methods have unique advantages when dealing with discrete germination frequency data.

[0070] For example, for n germination events within a monitoring area, where the germination frequency of each event at location xi is fi, the kernel density estimation is performed using a Gaussian kernel function. Perform spatial smoothing to obtain the probability density at any position x. , where h is the bandwidth parameter, which controls the degree of smoothing. Furthermore, the choice of the bandwidth parameter h directly affects the smoothing effect of the probability distribution; too small a value will result in an overly discrete distribution, while too large a value will lead to the loss of local features.

[0071] In one embodiment, a cross-validation method is used to determine the optimal bandwidth, and a suitable h value is obtained by minimizing the integral mean square error.

[0072] Specifically, the obtained crack initiation probability distribution map can intuitively reflect the crack initiation risk in each region of the plate, with high-probability areas usually concentrated near the strain concentration zone.

[0073] Step S106: From the germination probability distribution, identify regions with probabilities higher than a preset threshold, generate an instability initiation location map, and determine the predicted expansion path.

[0074] Extract the coordinates of grid points whose probability values ​​exceed a preset threshold from the germination probability distribution, and obtain the material plastic strain limit value corresponding to the grid points. and current cumulative strain value Calculate the residual plastic deformation capacity index If R is lower than a preset safety margin, it is marked as a potential instability point, resulting in a set of instability candidate points. Based on this set of candidate points, a connected component labeling algorithm is used to identify spatially adjacent groups of instability points. The geometric center coordinates of each group are calculated as the instability initiation position. Combined with the direction information of the previously corrected flow trajectory, the dominant instability direction is determined, generating an instability initiation position map containing position coordinates and direction information. Based on the instability initiation position map, the strain gradient distribution is calculated along the dominant instability direction. The direction with the largest gradient is taken as the crack preferential propagation direction. The stress distribution state in this direction is obtained through finite element stress analysis. The continuous path of the high-stress region is tracked to obtain the propagation path prediction.

[0075] In one implementation, when screening high-risk areas from the germination probability distribution, it is necessary to take into account both the probability value and the residual deformation capacity of the material.

[0076] Specifically, the initiation probability threshold is typically set between 0.6 and 0.8, a range corresponding to high-risk standards in engineering practice. The calculation of the residual plastic deformation capacity index R is based on the material's constitutive relation, where the plastic strain limit value... Obtained through uniaxial tensile testing, this value is typically in the range of 0.25 to 0.35 for 5052-O aluminum alloy; current cumulative strain value. Extracted from real-time monitoring data, the safety margin is generally taken as 0.2, that is, when R<0.2, the material is considered to be close to failure.

[0077] Preferably, the connected component labeling algorithm uses the eight-neighbor connectivity criterion when identifying unstable point groups.

[0078] For example, for each point in the set of unstable candidate points, check whether its eight neighboring grid points are also unstable candidate points. If they are adjacent, they are grouped into the same group. The labeling of all connected components is completed by recursive scanning.

[0079] In one possible implementation, the centroid formula is used to calculate the geometric center of each group, which involves taking the arithmetic mean of the coordinates of all points within the group. Determining the dominant instability direction requires considering the material's flow history. By analyzing the main flow direction of the corrected flow trajectory in the region, the direction with the highest frequency of occurrence is selected as the dominant instability direction. Furthermore, the extended path prediction is based on the strain energy release rate theory.

[0080] For example, in the deep drawing of aluminum alloy sheets, cracks tend to propagate along the direction of the greatest strain gradient because the strain energy release rate is highest in that direction.

[0081] Specifically, the stress field distribution around the instability initiation position is calculated using the finite element method, and stress components in each direction are extracted. Calculate the stress intensity factor , where a is the equivalent crack length, and a continuous path is formed by tracing along the direction with the largest K value.

[0082] In one embodiment, the identification of high-stress regions employs a contour tracing method, setting a stress threshold of 0.9 times the material's yield strength. Continuous regions exceeding this threshold are connected in series to form a predicted propagation path, which can indicate the possible evolution direction of material failure.

[0083] Step S107: By extending path prediction, the texture direction and flow trajectory data of all time points are integrated to generate a complete flow behavior report, and the abnormal flow areas caused by the reversal of the texture main axis in the report are identified to achieve early warning.

[0084] By extending the path prediction results, texture direction data and flow trajectory data at all time points are arranged in chronological order. Data at the same time and location are correlated and matched to form a flow behavior report containing the complete deformation process. Based on the flow behavior report, the records of texture principal axis angle changes in the report are scanned. If a region with a reverse rotation angle exceeding a preset threshold is detected, it is determined to be a flow anomaly region and an early warning is triggered.

[0085] In one implementation, when integrating multi-source data through extended path prediction results, it is necessary to establish a unified spatiotemporal coordinate system.

[0086] Specifically, the texture direction data at each time point includes angle values ​​and position coordinates, while the flow trajectory data includes displacement vectors and timestamps. By aligning timestamps and matching spatial coordinates, data association of the same material point at different times is achieved.

[0087] Preferably, the flow behavior report is stored in a structured format, including five fields: time series index, spatial grid coordinates, texture principal axis angle, flow velocity vector, and strain state parameters.

[0088] It should be noted that each record in the report corresponds to the complete deformation information of a monitoring point on the surface of the plate at a specific moment.

[0089] For example, the determination of texture axis reversal is based on the angle change trend of more than three consecutive frames. When the angle change direction is reversed and the reversal angle exceeds 15°, it is considered that an abnormal rotation has occurred.

[0090] In one possible implementation, the early warning mechanism sets multiple thresholds. A yellow warning is triggered when the reversal angle is between 15° and 30°, and a red warning is triggered when it exceeds 30°, reminding process engineers to adjust the stamping parameters in time to avoid serious forming defects in the future.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for tracking deformation during the non-stamping process of aluminum-based materials, characterized in that, The method includes: This process involves acquiring surface texture image sequences at multiple time points during the non-stamping process of aluminum-based materials, extracting the principal direction angle distribution of the texture in each image to obtain the initial direction angle sequence of the texture direction, calculating the direction angle change rate between adjacent images based on the initial direction angle sequence, identifying potential jump regions, and determining the jump location set and corresponding spatial coordinates, obtaining deformation path data for the corresponding regions from the jump location set and corresponding spatial coordinates, simulating the shear band evolution trend, and obtaining a local shear concentration risk change trend map, identifying and marking high-risk areas with local reverse rotation based on the local shear concentration risk change trend map, and determining the corrected flow trajectory through flow path deviation analysis results, obtaining material flow instability critical state data during the deformation increase stage based on the corrected flow trajectory, continuously monitoring microcrack initiation signals, and obtaining the initiation probability distribution, identifying regions with probabilities higher than a preset threshold from the initiation probability distribution, generating an instability initiation location map, and determining the propagation path prediction, and using the propagation path prediction to fuse texture direction and flow trajectory data from all time points to generate a complete flow behavior report, identifying flow anomaly regions caused by texture principal axis reversal in the report to achieve early warning.

2. The method for tracking deformation during the non-stamping process of aluminum-based materials according to claim 1, characterized in that, The process of acquiring surface texture image sequences at multiple time points during the non-stamping process of aluminum-based materials, extracting the principal direction angle distribution of the texture in each image, and obtaining the initial direction angle sequence of the texture direction includes: During the stamping and deformation process of aluminum alloy sheets, surface images are acquired at preset time intervals, and the corresponding stamping stroke position and forming force value are recorded. A correspondence between timestamps and deformation amounts is established to obtain an original image sequence marked with deformation amounts. For each image in the original image sequence, grayscale processing and histogram equalization are performed. Texture response values ​​in each direction are extracted at set intervals within a preset angle range. The main texture direction of each pixel is determined by comparing the response intensity in each direction, and a texture direction field matrix is ​​generated. Based on the texture direction field matrix, the frequency distribution of the main direction angle is statistically analyzed within each preset size pixel region. The main direction angles of each region are arranged in the order of image coordinates to obtain the initial direction angle sequence of the texture direction.

3. The method for tracking deformation during the non-stamping process of aluminum-based materials according to claim 1, characterized in that, The step of calculating the rate of change of orientation angle between adjacent images based on the initial orientation angle sequence of the texture direction, identifying potential transition regions, and determining the set of transition positions and their corresponding spatial coordinates includes: Extract the orientation angle data of the same spatial position at adjacent time points from the initial orientation angle sequence of the texture direction, calculate the orientation angle difference of corresponding pixel regions in two consecutive frames of images, and record their time index, spatial coordinates, and angle difference to obtain a set of jump candidate points; based on the spatial coordinates of each point in the set of jump candidate points, use the DBSCAN algorithm to identify the degree of aggregation of jump points on the board surface and obtain the boundary range of the region; for potential jump regions, calculate the actual physical coordinates of each jump point on the board through the pre-calibrated transformation relationship between the image coordinate system and the board physical coordinate system, arrange them in the order of jump occurrence time, and determine the set of jump positions and their corresponding spatial coordinates.

4. The method for tracking deformation during the non-stamping oil stamping process of aluminum-based materials according to claim 1, characterized in that, The process of obtaining deformation path data for the corresponding region from the set of jump locations and their corresponding spatial coordinates, simulating the evolution trend of the shear zone, and obtaining a local shear concentration risk change trend map includes: The coordinate sequence of each jump point at consecutive time steps is extracted from the set of jump locations. The ratio of displacement increment to time interval between adjacent time steps is calculated to obtain the velocity field. The spatial derivative of velocity is calculated using the finite difference method to obtain the strain rate component. The strain rate is accumulated over time to obtain the local strain tensor distribution. The principal strain value and direction are extracted based on the local strain tensor distribution. Combined with the aluminum alloy yield strength, hardening index, and anisotropy coefficient, the equivalent stress is calculated using the Hill yield criterion. For the strain data within the local strain tensor distribution, the local strain tensor and material parameters are input into the anisotropic hardening model. The initiation probability value is calculated using the shear band formation criterion. The shear band development direction is determined based on the direction of the maximum shear strain increment, obtaining the shear band evolution characteristic data for each location point. Based on the shear band evolution characteristic data, a gridded evaluation matrix is ​​established on the plate surface. Spatially continuous distribution is achieved by weighted averaging of risk values ​​of adjacent grid points, and a local shear concentration risk change trend map is plotted.

5. The method for tracking deformation during the non-stamping process of aluminum-based materials according to claim 1, characterized in that, The method for analyzing the risk trend of local shear concentration identifies and marks high-risk areas with local reverse rotation, and determines the corrected flow trajectory based on the flow path deviation analysis results, including: For each grid point in the local shearing concentrated risk change trend map, the texture principal axis angle of the corresponding position in the image at adjacent time points is extracted, the angle change between consecutive frames is calculated, and the average value of the angle change in the surrounding preset neighborhood is counted as the mainstream change direction. If the sign of the angle change is opposite to the mainstream change direction and the absolute value exceeds a preset threshold, it is determined that a local reverse rotation has occurred at that position, and the set of coordinates of the reverse rotation points is recorded. Based on the set of coordinates of the reverse rotation points, a preset rotation anomaly weight value is added to the original risk level value of the corresponding position in the risk change trend map to generate a comprehensive risk distribution map. Based on the comprehensive risk distribution map, the original flow trajectory is corrected point by point to determine the corrected flow trajectory.

6. The method for tracking deformation during the non-stamping process of aluminum-based materials according to claim 5, characterized in that, The actual flow direction at each point within the area is obtained based on the comprehensive risk distribution map. The deviation angle between the actual flow direction and the ideal flow direction is calculated. The spatial distribution function of the deviation angle is obtained through least squares fitting, and a flow path deviation correction matrix is ​​constructed. The original flow trajectory is corrected point by point using the flow path deviation correction matrix.

7. The method for tracking deformation during the non-stamping process of aluminum-based materials according to claim 6, characterized in that, After the original flow trajectory is corrected point by point using the flow path deviation correction matrix, the corrected displacement field is obtained by vector synthesis of the direction correction amount in the correction matrix and the original displacement vector. The material point motion trajectory is reconstructed by time integration of the corrected displacement field.

8. The method for tracking deformation during the non-stamping process of aluminum-based materials according to claim 1, characterized in that, The process of obtaining critical state data of material flow instability during the deformation increase stage based on the corrected flow trajectory, continuously monitoring microcrack initiation signals, and obtaining the initiation probability distribution includes: Based on the corrected flow trajectory, the cumulative equivalent strain value of each material point is extracted, and the strain rate is calculated. When the strain rate reaches its peak and then shows a continuous downward trend, the material is determined to have entered a critical instability state. The strain value, stress state, and deformation rate at this moment are recorded as the critical instability state data of the material flow. For the high strain region in the critical instability state data of the material flow, where the strain value exceeds a preset threshold, the surface morphology change signal and internal elastic wave signal of the region are collected, and the number of germination signals per unit time is counted. Based on the number of germination signals, the microcrack germination frequency of each monitoring point within a preset time window is calculated. The discrete frequency data is spatially convolved and smoothed using a Gaussian kernel function using a kernel density estimation method to obtain a continuous germination probability distribution.

9. The method for tracking deformation during the non-stamping oil stamping process of aluminum-based materials according to claim 1, characterized in that, The step of identifying regions with probabilities higher than a preset threshold from the germination probability distribution, generating an instability initiation location map, and determining the propagation path prediction includes: The coordinates of grid points with probability values ​​exceeding a preset threshold are extracted from the initiation probability distribution. The material plastic strain limit value and current cumulative strain value corresponding to the grid points are obtained, and the remaining plastic deformation capacity index is calculated to obtain a set of instability candidate points. Based on the set of instability candidate points, a connected component labeling algorithm is used to identify spatially adjacent groups of instability points. The geometric center coordinates of each group are calculated as the instability initiation position. The dominant instability direction is determined by combining the direction information of the corrected flow trajectory, and an instability initiation position map containing position coordinates and direction information is generated. Based on the instability initiation position map, the strain gradient distribution is calculated along the dominant instability direction. The direction with the maximum gradient is taken as the crack preferential propagation direction. The stress distribution state in this direction is obtained through finite element stress analysis to obtain the propagation path prediction.

10. The method for tracking deformation during the non-stamping process of aluminum-based materials according to claim 1, characterized in that, The process involves expanding path prediction, fusing texture direction and flow trajectory data from all time points to generate a complete flow behavior report, and identifying abnormal flow areas caused by texture axis reversal in the report to achieve early warning. This includes: By extending the path prediction results, the texture direction data and flow trajectory data at all time points are arranged in chronological order. Data at the same time and location are correlated and matched to form a flow behavior report containing the complete deformation process. Based on the flow behavior report, the records of texture principal axis angle changes in the report are scanned. If a region with reverse rotation of the angle exceeding a preset threshold is detected, it is determined to be a flow abnormal region.

Images