A Forming Quality Analysis Method for Hot Stamped Parts with Complex Shapes

Abstract

The present invention provides a method for analyzing the forming quality of hot stamping parts with complex shapes, including establishing a three-dimensional model; extracting the position information of all points and the coordinate information including the x, y, and z axis directions to establish a three-dimensional point set information database; dividing the part cross-section slices and constructing a slice combination database; sequentially extracting the coordinate information of the slice combinations to construct a slice coordinate information database; correlating the relationship between the feature information and the position information of a single slice; constructing a slice feature database; determining whether the traversal of the slice coordinate information database is completed; performing machine self-learning and forming a feature recognition evaluation model; establishing a database for the feature data that conforms to the feature recognition evaluation model; determining whether the traversal of the slice combination database is completed, dividing the original hot stamping part with complex shape into a combination of basic shape stamping parts and outputting. This system links the geometric features of hot stamping parts with complex shapes and the stamping forming process parameters, improving the parameter optimization efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of metal plastic forming, and particularly relates to a method for analyzing the forming quality of hot stamping parts with complex shapes. Background Art

[0002] The design of the hot stamping forming process based on traditional methods generally includes the following processes: qualitative analysis, physical simulation tests, and actual production verification, etc. The relevant process design mainly depends on the design experience of the staff. After the initial design of the hot stamping forming die is completed and trial-produced, it is necessary to repeatedly modify the design parameters and process parameters through the "trial-and-error method" to finally complete the entire design. In actual production, the parts formed by hot stamping often have complex shapes and complex contour dimensions. When such parts are formed, the deformation states of each part of the blank are complex and the stress distribution is uneven. Such parts are more likely to appear defects such as wrinkling and cracking during the forming process. It is very difficult to directly carry out the process design for them. With the development of simulation technology, using CAE software for finite element analysis to simulate the forming process of hot stamping parts can shorten the product development cycle and reduce production costs, and plays a very important role in the hot stamping process design of parts.

[0003] The existing simulation analysis technology for the hot stamping forming of parts first establishes a three-dimensional model of the part, then establishes a three-dimensional model of the simplified die, performs mesh division in the CAE software, inputs relevant stamping process parameters for simulation analysis, and obtains the forming quality results of the part after stamping, which can effectively save design time and reduce the die design cost.

[0004] Due to market demand, the geometric shapes of stamping finished products are becoming more and more complex, such as automotive fenders, automotive A-pillars, etc. How to ensure the mechanical property requirements of parts after stamping also puts forward requirements for the improvement of stamping processes. During the stamping process of metal sheet materials, the materials undergo elastoplastic deformation, and the relationship between stamping forming process parameters and forming quality is often highly nonlinear. Therefore, how to find the nonlinear relationship between sheet metal forming process parameters and forming quality and control and optimize multiple forming defects is extremely crucial. If conventional analysis methods are used, the requirements for the computer are extremely high. Not only is the speed of the simulation analysis process unable to meet the production requirements, but also the forming quality and accuracy of the simulated parts are relatively poor. Summary of the Invention

[0005] In view of this, the present invention aims to propose a method for analyzing the forming quality of hot stamping parts with complex shapes, which can solve the above technical problems.

[0006] To achieve the above object, the technical solution of the present invention is realized as follows:

[0007] A method for analyzing the forming quality of hot stamping parts with complex shapes includes the following steps:

[0008] Step 1: Establish a 3D model of the hot stamping part with complex shape;

[0009] Step 2: Extract the position information of all points of the hot stamping part with complex shape and the coordinate information including the x, y, and z axis directions, and establish a 3D point set information database;

[0010] Step 3: Divide the part cross-section slices according to the position information of all points of the hot stamping part with complex shape, and extract three adjacent part cross-section slices to form a slice combination, and construct a slice combination database;

[0011] Step 4: Extract the coordinate information of the slice combination in sequence for traversal operation;

[0012] Step 5: Construct a slice coordinate information database;

[0013] Step 6: Associate the relationship between the feature information and position information of a single slice and perform a traversal operation;

[0014] Step 7: Construct a slice feature database and store the associated slice feature information and position information data;

[0015] Step 8: Judge whether the traversal of the slice coordinate information database is completed. If it is, enter Step 9; if the traversal is not completed, return to Step 6;

[0016] Step 9: Use the data in the feature database for machine self-learning and form a feature recognition evaluation model;

[0017] Step 10: Judge whether each feature data in the feature database conforms to the feature recognition evaluation model. If it conforms, enter Step 11; if it does not conform, input the feature data into Step 9 for machine self-learning;

[0018] Step 11: Establish a feature recognition model database for the feature data that conforms to the feature recognition evaluation model;

[0019] Step 12: Judge whether the traversal of the slice combination database is completed. If the traversal is completed, enter Step 13; if it is not completed, return to Step 4;

[0020] Step 13: Perform feature evaluation and integration on the feature recognition model;

[0021] Step 14: Divide the original hot stamping part with complex shape into a combination of basic shape stamping parts;

[0022] Step 15: Output the combination of basic shape stamping parts and perform subsequent finite element simulation analysis.

[0023] Further, in step 3, the method of dividing the part cross-section slices according to the position information of all points of the complex-shaped hot stamping part and extracting a slice combination composed of three adjacent part cross-section slices is to select one of the x, y, and z coordinate axes as the cross-section direction. The set of three-dimensional points corresponding to each coordinate value on this coordinate axis is a cross-section slice. Along this cross-section direction, three adjacent cross-section slices with equal intervals are selected as a slice combination. The cross-section slices in this slice combination are denoted as the i-th slice, the (i + 1)-th slice, and the (i + 2)-th slice, where i is an integer greater than or equal to 1.

[0024] Further, in step 6, the method of correlating the feature information and position information of a single slice is to extract three consecutive feature coordinates in each single slice in the order of position, identify the connection feature of the three coordinate points as a line segment or an arc and use it as the feature information, and extract the position information of the three consecutive feature coordinates, and then correlate the feature information with the position information.

[0025] Further, in step 9, the specific method of using the data in the feature database for machine self-learning and forming a feature recognition evaluation model is to select a specific number of feature information in the feature database to construct a data set with the geometric features of the basic stamping part, and use the feature recognition algorithm of machine learning for feature recognition training to form a feature recognition evaluation model.

[0026] Further, the geometric features of the data set with the geometric features of the basic stamping part include line segment features and arc features.

[0027] Further, the specific method of using the feature recognition algorithm of machine learning for feature recognition training and forming a feature recognition evaluation model is that in each slice combination, within the same slice, the line segments or arcs are correlated and recognized, and the line segments or arcs with the same features are spliced into larger-scale line segments or arcs; the line segment or arc features between every two adjacent slices are correlated and recognized, and the line segments or arcs with the same features are spliced into a plane or an arc surface.

[0028] Further, in step 10, the method of judging whether the feature data conforms to the feature recognition evaluation model is to consider those belonging to the same feature as positive examples and those not having the same feature as negative examples, and use a formula for evaluation

[0029]

[0030] Where:

[0031] TP (True Positive): Represents the number of times of being judged as a positive example and being judged correctly;

[0032] FP (False Positive): Represents the number of times of being judged as a positive example but being judged wrongly;

[0033] TN (True Negative): indicates the number of times a negative example is judged to be correct;

[0034] FN (False Negative): Indicates the number of times a negative example is judged incorrectly.

[0035] Furthermore, the method for feature evaluation and integration of the feature recognition model in step 13 is to splice the same features in the current slice combination and the adjacent slice combination in the feature recognition model database, and compare them with the set error after splicing. If it is within the set error range, it is retained; if it is outside the set error range, the slice combination adjacent to the current slice combination is eliminated.

[0036] The present invention mentions a forming quality analysis method for complex-shaped hot stamping parts. By extracting geometric features, the features of complex-shaped hot stamping parts are identified, and the geometric features of complex-shaped hot stamping parts are linked to the stamping forming process parameters, thereby improving the parameter optimization efficiency. A relatively simple mathematical model can be found to approximate the nonlinear relationship between the forming parameters and the forming quality objective function. The number of sheet metal stamping simulations can be reduced and optimized parameters can be quickly obtained, thereby improving simulation efficiency, effectively controlling forming defects and improving forming quality, which is of great significance to the hot stamping forming process design of complex parts. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0038] Figure 1 This is a step diagram of a forming quality analysis method for hot stamping parts of complex shapes according to an embodiment of the present invention;

[0039] Figure 2 The figure is a schematic diagram of a specific process of the forming quality analysis method for hot stamping parts with complex shapes according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0041] The technical solutions between the various embodiments can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the technical solutions between the embodiments can be combined, they are all within the protection scope required by the present invention.

[0042] The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0043] A method for analyzing the forming quality of hot stamping parts with complex shapes, as Figure 1 and Figure 2 shown below:

[0044] Step 1: Establish a three-dimensional model of the hot stamping part with complex shape;

[0045] Step 2: Extract the position information of all points of the hot stamping part with complex shape and the coordinate information including the x, y, and z axis directions, and establish a three-dimensional point set information database;

[0046] Step 3: Divide the part cross-section slices according to the position information of all points of the hot stamping part with complex shape, and extract three adjacent part cross-section slices to form a slice combination, and construct a slice combination database;

[0047] Select one of the x, y, and z axis directions as the cross-section direction. The set of three-dimensional points corresponding to each coordinate value on this axis is a cross-section slice. Select three adjacent cross-section slices with equal intervals along this cross-section direction as a slice combination. The cross-section slices in this slice combination are denoted as the i-th slice, the (i + 1)-th slice, and the (i + 2)-th slice, where i is an integer greater than or equal to 1.

[0048] Step 4: Extract the coordinate information of the slice combination in sequence for traversal operations; when the current slice is the i-th one, i, i + 1, and i + 2 form a slice combination, denoted as A. When the current slice is the (i + 1)-th one, i + 1, i + 2, and i + 3 are the slice combination adjacent to slice combination A; when extracting the coordinate information of the slice combination, also extract the position information of each slice in the slice combination for subsequent analysis;

[0049] Step 5: Construct a slice coordinate information database;

[0050] Step 6: Associate the relationship between the feature information and the position information of a single slice and perform traversal operations;

[0051] Extract the coordinate information of three consecutive three-dimensional points in each single slice in the order of position, identify the connection feature of the three coordinates as a line segment or an arc as the feature information, and extract the position information of the three consecutive three-dimensional points, and associate the feature information with the position information. This slice coordinate information database includes the coordinate information of all three-dimensional points within the current group of slice combinations, and the slice coordinate information database stores the coordinate information of a group of slice combinations each time;

[0052] Step 7: Construct a slice feature database and store the associated slice feature information and position information data;

[0053] Step 8: Determine whether the slicing coordinate information database has been traversed completely. If so, proceed to Step 9; if not, return to Step 6. The slicing coordinate information database is considered traversed completely when all points from the starting point of the first slice to the ending point of the last slice in the current slice combination have been traversed.

[0054] Step 9: Use the data in the feature database for machine self-learning to form a feature recognition evaluation model.

[0055] Select a specific number of feature information from the slicing feature database to construct a dataset with the geometric features of basic stampings. Use the feature recognition algorithm of machine learning for feature recognition training, and the learning method uses deep learning to complete the automatic recognition of features. Through learning a large amount of feature, coordinate data, and stamping types, the training set is 80% and the prediction set is 20%. And form a feature recognition evaluation model. The geometric features of the dataset with the geometric features of basic stampings include line segment features and arc features. The specific method of using the feature recognition algorithm of machine learning for feature recognition training and forming a feature recognition evaluation model is that in each slice combination, within the same slice, the line segments or arcs are associated and recognized, and the line segments or arcs with the same features are spliced into larger-scale line segments or arcs; the line segment or arc features between every two adjacent slices are associated and recognized, and the line segments or arcs with the same features are spliced into planes or arc surfaces, and then the planes or arc surfaces with the same features in the slice combination are spliced into larger planes or arc surfaces as the feature recognition evaluation model. When the feature is a line segment or a plane, the same slope is considered to have the same feature; when the feature is an arc or an arc surface, the same center and curvature are considered to have the same feature.

[0056] Step 10: Determine whether each feature data in the feature database conforms to the feature recognition evaluation model. If so, proceed to Step 11; if not, input the feature data into Step 9 for machine self-learning. Consider those belonging to the same feature as positive examples and those not having the same feature as negative examples, and use the formula for evaluation:

[0057]

[0058] Where:

[0059] TP (True Positive): Represents the number of times determined as a positive example and determined correctly.

[0060] FP (False Positive): Represents the number of times determined as a positive example but judged wrongly.

[0061] TN (True Negative): Represents the number of times determined as a negative example and determined correctly.

[0062] FN (False Negative): The number of times that a negative example is misjudged as a negative example.

[0063] Use Accuracy (which can be set to 0.99) as the evaluation criterion to evaluate the quality of the model.

[0064] Step 11: Establish a feature recognition model database for the feature data that meets the feature recognition evaluation model.

[0065] Step 12: Determine whether the slice combination database has been traversed. If it has been traversed, go to Step 13; if not, return to Step 4. The slice combination database has been traversed when all the slice combinations extracted from the three-dimensional point set information database, from the first to the last, have been traversed.

[0066] Step 13: Conduct feature evaluation and integration on the feature recognition model.

[0067] Stitch the same features in the current slice combination and the adjacent slice combination in the feature recognition model database. Compare the geometric shape of the stitched plane or surface with the geometric shape at the same position of the original three-dimensional model, and there is a certain actual error. Compare this actual error with the set error. If it is within the set error range, keep it; if it is outside the set error range, remove the slice combination adjacent to the current slice combination, and then stitch the same features of the current slice combination and the next slice combination adjacent to it.

[0068] Step 14: Divide the original complex-shaped hot stamping part into a combination of basic-shaped stamping parts.

[0069] Step 15: Output the combination of basic-shaped stamping parts and conduct subsequent finite element simulation analysis.

[0070] In a specific embodiment, a three-dimensional model is established for a hot stamping part with a complex shape. Using 3D drawing software such as Meshlab, the position information and coordinate information of all points of the complex-shaped part are exported to establish a three-dimensional point information database containing x, y, and z coordinates. Taking the x-axis direction as the cross-section direction, along the x direction, the cross-section slice where the starting x coordinate point is located is selected as the first cross-section slice, and the spacing r between adjacent cross-section slices is set. The value of r is determined according to the calculation accuracy, and the unit can be centimeters, millimeters, or micrometers. According to the position information of all points of the hot stamping part with a complex shape, the part cross-section slices are divided, and a slice combination composed of three adjacent part cross-section slices i, i + 1, and i + 2 is extracted to construct a slice combination database. The coordinate information of the slice combination is sequentially extracted for traversal operations. In the currently extracted slice combination, a slice coordinate information database is constructed, and all point sets on the i-th slice are loaded. Starting from the first point of this point set, it is scanned point by point to the last point on the other side. During the scanning process, the feature information between three adjacent points is extracted to determine whether the feature is a line segment or an arc. The feature information is associated with the previously extracted slice position information, a slice feature database is constructed, and the associated slice feature information and position information data are stored. It is judged whether the traversal of the slice coordinate information database is completed. If so, it enters the machine self-learning stage; if the traversal is not completed, the point set of the (i + 1)-th slice along the x direction is continuously loaded, and the traversal process of the i-th slice is repeated. Subsequently, the (i + 2)-th slice is loaded along the x direction for traversal recognition. A specific number of feature data, including line segment features and arc features, are selected from the slice feature database, and a feature recognition algorithm of machine learning is used for feature recognition training. The specific method for forming a feature recognition evaluation model is that in each slice combination and within the same slice, the line segments or arcs are associated and recognized, and the line segments or arcs with the same features are spliced into line segments or arcs with a larger scale;Associate and identify the line segment or arc features between every two adjacent slices. The line segments or arcs with the same features are spliced into a plane or an arc surface, and then the planes or arc surfaces with the same features in the slice combination are spliced into a larger plane or arc surface as the feature recognition evaluation model. Identify and evaluate according to the feature recognition evaluation model in the slice feature database. Those belonging to the same features are regarded as positive examples, and those without the same features are regarded as negative examples. Then, judge according to the formula in step 10, establish a feature recognition model database based on the recognized feature data, and judge whether the current slice combination database has been traversed. If not, extract the adjacent slice combination of the current slice combination for traversal. If so, perform feature evaluation and integration on the feature recognition model. The same features in the current slice combination and the adjacent slice combination are spliced. There is a certain actual error when comparing the geometric shape of the spliced plane or curved surface with the same position of the original three-dimensional model. Compare this actual error with the set error. If it is within the set error range, it is retained. If it is outside the set error range, the slice combination adjacent to the current slice combination is removed. Then, the current slice combination and the next slice combination of the adjacent slice combination are spliced with the same features. Thus, the original complex-shaped hot stamping part is divided into a combination of basic-shaped hot stamping parts. The part shape gradually transitions from box-shaped parts, spherical parts, and cup-shaped parts to U-shaped parts, V-shaped parts, and sheet stretching parts. Finally, the three-dimensional point set is labeled with many simple-shaped three-dimensional features to realize the "differential" processing of complex-shaped parts, and the combination of basic-shaped hot stamping parts is output for subsequent finite element analysis of the forming quality of the part.;

[0071] Divide the original complex-shaped hot stamping part into a combination of basic-shaped hot stamping parts, and quantitatively evaluate the relative position of each feature to the whole part and the interaction between each feature and the surrounding features, so that the forming prediction distribution of simple three-dimensional features can more accurately reflect the real situation.

[0072] Although the present invention has been described in detail with general descriptions and specific embodiments above, based on the present invention, some modifications or improvements can be made, which are obvious to those skilled in the art. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of the present invention claimed.

Claims

1. A method for analyzing the forming quality of hot stamping parts with complex shapes, characterized in that: Step 1: Establish a three-dimensional model of the hot stamping part with complex shape; Step 2: Extract the position information of all points of the hot stamping part with complex shape and the coordinate information including the x, y, and z axis directions to establish a three-dimensional point set information database; Step 3: Divide the part cross-section slices according to the position information of all points of the hot stamping part with complex shape, and extract three adjacent part cross-section slices to form a slice combination, and construct a slice combination database; Step 4: Extract the coordinate information of the slice combination in sequence for traversal operation; Step 5: Construct a slice coordinate information database; Step 6: Associate the relationship between the feature information and the position information of a single slice and perform a traversal operation; Step 7: Construct a slice feature database and store the associated slice feature information and position information data; Step 8: Determine whether the traversal of the slice coordinate information database is completed. If so, go to Step 9; if the traversal is not completed, return to Step 6; Step 9: Use the data in the feature database for machine self-learning and form a feature recognition evaluation model; Step 10: Determine whether each feature data in the feature database conforms to the feature recognition evaluation model. If it conforms, go to Step 11; if it does not conform, input the feature data into Step 9 for machine self-learning; Step 11: Establish a feature recognition model database for the feature data that conforms to the feature recognition evaluation model; Step 12: Determine whether the traversal of the slice combination database is completed. If the traversal is completed, go to Step 13; if it is not completed, return to Step 4; Step 13: Perform feature evaluation and integration on the feature recognition model; Step 14: Divide the original hot stamping part with complex shape into a combination of basic shape stamping parts; Step 15: Output the combination of basic shape stamping parts and perform subsequent finite element simulation analysis; The method for associating the feature information and the position information of a single slice in Step 6 is to extract the three consecutive three-dimensional point features coordinates in each single slice according to the position order, identify the connection feature of the three coordinate points as a line segment or an arc as the feature information, and extract the position information of the three consecutive feature coordinates, and associate the feature information with the position information; The method for performing feature evaluation and integration on the feature recognition model in Step 13 is to splice the same features in the current slice combination and the adjacent slice combination in the feature recognition model database, and compare with the set error after splicing. If it is within the set error range, it is retained; if it is outside the set error range, the slice combination adjacent to the current slice combination is excluded.

2. The forming quality analysis method of the hot stamping part with complex shape according to claim 1, characterized in that, In step 3, the method of dividing the part cross-section slices according to the position information of all points of the complex-shaped hot stamping part and extracting three adjacent part cross-section slices to form a slice combination is to select one of the x, y, and z coordinate axes as the cross-section direction. The set of three-dimensional points corresponding to each coordinate value on this coordinate axis is a cross-section slice. Three adjacent cross-section slices with equal intervals are selected along this cross-section direction as a slice combination. The cross-section slices in this slice combination are denoted as the i-th slice, the (i + 1)-th slice, and the (i + 2)-th slice, where i is an integer greater than or equal to 1.

3. The method for analyzing the forming quality of hot stamping parts with complex shapes according to claim 1, characterized in that In step 9, the specific method of using the data in the feature database for machine self-learning and forming a feature recognition evaluation model is to select a specific number of feature information in the feature database to construct a data set with the geometric features of the basic stamping part, and use the feature recognition algorithm of machine learning for feature recognition training to form a feature recognition evaluation model.

4. The forming quality analysis method of the hot stamping part with complex shape according to claim 3, characterized in that, The geometric features for constructing the data set with the geometric features of the basic stamping part include line segment features and arc features.

5. The forming quality analysis method for hot stamping parts with complex shapes according to claim 4, characterized in that, The specific method of using the feature recognition algorithm of machine learning for feature recognition training and forming a feature recognition evaluation model is that in each slice combination, within the same slice, the line segments or arcs are associated and recognized, and the line segments or arcs with the same features are spliced into larger-scale line segments or arcs; the line segment or arc features between every two adjacent slices are associated and recognized, and the line segments or arcs with the same features are spliced into a plane or an arc surface.

6. The forming quality analysis method for hot stamping parts with complex shapes according to claim 1, characterized in that, In step 10, the method of determining whether the feature data conforms to the feature recognition evaluation model is to consider those belonging to the same features as positive examples and those not having the same features as negative examples, and use the formula for evaluation: Where: TP (True Positive): represents the number of times determined as a positive example and determined correctly; FP (False Positive): represents the number of times determined as a positive example but judged wrongly; TN (True Negative): represents the number of times determined as a negative example and determined correctly; FN (False Negative): represents the number of times determined as a negative example but judged wrongly.

Images