Chain mold forming springback deviation prediction method based on sequential synchronous springback coupling

Abstract

The invention discloses a chain die forming springback deviation prediction method based on sequential synchronous springback coupling, and the method comprises the steps: decoupling the springback generated by a part in chain die forming into sequential springback of time sequence accumulation and global collaborative synchronous springback through a physical mechanism; and for sequential springback, calculating a local cumulative springback amount of the sequential springback by adopting a segmented equivalent beam theory, for synchronous springback, predicting a global coupling springback amount of the synchronous springback by adopting a point sequence neural network, and finally realizing dynamic coupling of the local cumulative springback and the global coupling springback through an exponential attenuation weight distribution function. Therefore, accurate characterization of the rebound nonlinear characteristic is realized. According to the springback deviation prediction method, the prediction precision and the calculation efficiency are greatly improved, precision control over chain die forming parts can be efficiently supported, and the calculation method for different springback modes is high in universality, easy to expand and suitable for industrial production of ultrahigh-strength steel beam parts.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ultra-high strength steel sheet forming manufacturing, and particularly relates to a chain die forming springback deviation prediction method based on sequential synchronous springback coupling. BACKGROUND

[0002] In recent years, with the increasing requirements of the automobile industry for lightweight and safety, ultra-high strength steel (UHSS) is increasingly widely used in the manufacture of vehicle body structures. Although UHSS has high strength, its obvious Bauschinger effect, significant cyclic hardening characteristics and relatively low ductility bring a series of challenges to its forming process. The traditional processing method is easy to cause defects such as rupture and wrinkling of the component, and the springback problems such as web buckling, flange edge wavy deformation and part end opening caused by the release of residual stress also seriously reduce the dimensional accuracy of the component, limiting its application in large-scale production. The stamping process relies on large pressure equipment, and the die loss is large; the roll forming is prone to longitudinal redundant strain due to strong geometric constraints and high local contact stress in the deformation zone, thereby increasing the complexity of springback control. Therefore, the chain die forming technology has been developed in the industry to solve this problem, which effectively suppresses springback and improves forming quality through process innovation. Previous studies have shown that this technology has significant advantages, but its feasibility and stability in industrial production still need to be further verified through systematic research and multi-dimensional engineering cases.

[0003] Chain die forming is a new type of incremental forming process that combines the constraints of stamping and the gradual nature of roll forming, which makes the sheet metal produce plastic deformation by sequentially pressing the discrete die blocks. It is mainly applied to the forming of beam parts such as automobile seat beams. Chain die forming has the advantages of small forming load and small redundant deformation, so it is good for the forming of ultra-high strength steel. However, due to the process characteristics of chain die forming, the forming process is coupled by multiple dies, and the sequential pressing and springback overlap. Specifically, the springback of the previous die overlaps with the pressing of the next die, and its springback deformation directly affects the loading of the next die, resulting in complex part springback mechanism under chain die forming process, making prediction and control difficult. The existing die surface compensation and process parameter optimization methods for springback control can reduce the springback phenomenon to some extent, but they have not fundamentally revealed the coupling mechanism between local effects and global responses; although the full-process finite element simulation performs well in terms of prediction accuracy, it has a heavy computational burden and low efficiency, making it difficult to provide effective support for the rapid iteration and optimization of process parameters. Therefore, it is important to analyze the multi-die coupling characteristics in chain die forming, establish an accurate springback theory and deviation analysis model, and predict the surface of beam parts and control the accuracy. SUMMARY

[0004] The application is to solve the problem of complex part springback mechanism and insufficient surface prediction accuracy in chain die forming, and proposes a chain die forming springback deviation prediction method based on sequential synchronous springback coupling, which realizes efficient prediction through decoupling and dynamic coupling algorithm, and effectively solves the above technical problems.

[0005] According to the first aspect of the application, the application provides a chain die forming springback deviation prediction method based on sequential synchronous springback coupling, characterized in that it comprises the following steps: Step 1: decoupling the springback of the part generated in the chain die forming into sequential springback and synchronous springback; Step 2: based on the measured part, respectively calculating the local cumulative springback amount generated by the part to obtain the deviation representation of sequential springback, and the global coupled springback amount to obtain the deviation representation of synchronous springback; Step 3: dynamically coupling the deviation representation of the sequential springback of the measured part with the deviation representation of the synchronous springback to obtain the final prediction result of the measured part.

[0006] Further, in the step 2, the sequential springback calculates the local cumulative springback amount by the segmented equivalent beam theory, and the synchronous springback calculates the global coupled springback amount by the point sequence neural network.

[0007] Further, the process of obtaining the local cumulative springback amount of the measured part is as follows: The segmented equivalent beam theory is used to decompose the longitudinal axis deformation of the measured part into several characteristic regions; The local springback amount of each characteristic region is calculated by the deflection curve equation; The local springback amount of each characteristic region is added to obtain the local cumulative springback amount of the sequential springback.

[0008] Further, the process of obtaining the global coupled springback amount of the measured part is as follows: The surface of the measured part is meshed to generate a point sequence containing node coordinates; A deviation prediction model of the synchronous springback is constructed; The spatial geometric features of the point sequence are taken as input, and the global coupled springback amount of the synchronous springback is output.

[0009] Further, the deviation prediction model of the synchronous springback is constructed based on point sequence and LSTM neural network, and the LSTM neural network includes a spatial feature extraction layer and a fully connected network layer.

[0010] Further, when performing the dynamic coupling, based on the mold sequence in the forming stage, an exponential decay weight function is used to assign the weights of the sequential springback and the synchronous springback: And through the least square method fitting actual data, the optimal and : In the formula, The mold sequence number is, The actual rebound section is; The predicted section based on the sequential synchronous rebound coupling is: In the formula, The sequential rebound weight is, The sequential rebound section is, The synchronous rebound section is, The correction error term is.

[0011] Further, the chain die forming rebound deviation prediction method based on the sequential synchronous rebound coupling further comprises: Step 4, a chain die forming finite element simulation model is established, and the prediction accuracy of the sequential synchronous rebound coupling is verified through simulation data, the finite element simulation model comprises a forming stage and a rebound stage, the forming stage adopts a dynamic explicit algorithm, and the rebound stage adopts a static implicit algorithm.

[0012] According to the second aspect of the present application, the present application further provides a chain die forming rebound deviation prediction device based on the sequential synchronous rebound coupling, which is used for predicting the rebound of a part in a chain die forming stage, characterized in that the rebound is decoupled into sequential rebound and synchronous rebound, and the device comprises: A sequential rebound calculation module is used for calculating the local cumulative rebound amount generated by a to-be-tested part based on the to-be-tested part to obtain a deviation representation of the sequential rebound; A synchronous rebound calculation module is used for calculating the global coupling rebound amount generated by the to-be-tested part based on the to-be-tested part to obtain a deviation representation of the synchronous rebound; A dynamic coupling prediction module is used for dynamically coupling the deviation representation of the sequential rebound of the to-be-tested part with the deviation representation of the synchronous rebound, so as to obtain a final rebound prediction result of the to-be-tested part.

[0013] According to the third aspect of the present application, the present application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor implements the steps of any chain die forming rebound deviation prediction method based on the sequential synchronous rebound coupling when executing the program.

[0014] According to a fourth aspect of the present application, the present application also provides a storage medium, in which a plurality of instructions are stored, and the instructions are suitable for being loaded by a processor to execute the steps of any one of the sequential synchronization rebound coupling-based chain die forming rebound deviation prediction methods.

[0015] By one or more of the above embodiments of the present application, at least the following technical effects can be achieved: the sequential synchronization rebound coupling-based chain die forming rebound deviation prediction method of the present application decouples the rebound of the part produced in the chain die forming into a sequential rebound accumulated in time sequence and a synchronous rebound globally coordinated through a physical mechanism, calculates the local cumulative rebound amount of the sequential rebound by using the segmented equivalent beam theory, predicts the global coupled rebound amount of the synchronous rebound by using the point sequence neural network, and finally realizes the dynamic coupling of the local cumulative rebound and the global coupled rebound through an exponential decay weight distribution function, thereby realizing the accurate characterization of the rebound nonlinear characteristics. The rebound deviation prediction method of the present application greatly improves the prediction accuracy, can efficiently support the precision control of the chain die forming part, and is suitable for the industrialized production of ultrahigh-strength steel beam parts. BRIEF DESCRIPTION OF DRAWINGS

[0016] The technical solutions of the present application and other beneficial effects will become apparent through the following detailed description of the specific embodiments of the present application in combination with the accompanying drawings.

[0017] Figure 1 is a flowchart of the sequential synchronization rebound coupling-based chain die forming rebound deviation prediction method provided by the embodiments of the present application; Figure 2 is a finite element simulation model schematic diagram provided by the embodiments of the present application; Figure 3 is a feature region division schematic diagram based on the segmented equivalent beam theory provided by the embodiments of the present application; Figure 4 is a schematic diagram for calculating the sequential rebound based on the segmented equivalent beam theory provided by the embodiments of the present application; Figure 5 is a point sequence-based LSTM neural network construction schematic diagram provided by the embodiments of the present application; Figure 6 is a schematic diagram for calculating the synchronous rebound based on the LSTM neural network provided by the embodiments of the present application; Figure 7 is a schematic diagram of the cross section selection position of each die provided by the embodiments of the present application; Figure 8 is a schematic diagram of the half cross section of each die under three rebound modes provided by the embodiments of the present application; Figure 9 is a sequential synchronization rebound dynamic coupling calculation schematic diagram provided by the embodiments of the present application; Figure 10 is a structural schematic diagram of a chain die forming springback deviation prediction device based on sequential synchronous springback coupling provided by an embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person skilled in the art without creative work fall within the scope of protection of the present application.

[0019] In the description of the present application, it should be noted that, unless otherwise explicitly specified and limited, the term “and / or” in this paper is only to describe the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B can represent the three cases of A alone, A and B together, and B alone. In addition, the character “ / ” in this paper generally represents an “or” relationship between the front and rear associated objects without special explanation.

[0020] Based on the above technical problems, the present application provides a chain die forming springback deviation prediction method and system based on sequential synchronous springback coupling. First, the method will be introduced below with reference to the drawings.

[0021] As shown in Figure 1 The chain die forming springback deviation prediction method based on sequential synchronous springback coupling of the present embodiment includes the following steps: Step S1, decoupling the springback of the part generated in the chain die forming into sequential springback and synchronous springback.

[0022] The existing die surface compensation and process parameter optimization methods for springback control can reduce the springback phenomenon to a certain extent, but have not fundamentally revealed the coupling mechanism between local effect and global response; although the full-process finite element simulation performs well in terms of prediction accuracy, it has heavy calculation burden and low efficiency, and it is difficult to provide effective support for the rapid iteration and optimization of process parameters. The traditional springback analysis method, whether it is to deal with completely independent single-step springback or to deal with overall springback of all dies unloaded at the same time, cannot accurately describe the characteristics of chain die forming, which is both step-by-step and overlapping, both locally independent and globally related. Therefore, it is necessary to decouple the chain die forming springback behavior into sequential springback and synchronous springback from the mechanism level, and describe the interaction between them by means of stiffness matrix.

[0023] Therefore, based on stiffness decomposition theory, this application first decouples chain mold forming springback into sequential springback and synchronous springback, corresponding to two physical mechanisms: local accumulation and global coordination, respectively. The following embodiments of this application model synchronous springback, sequential springback, and chain mold forming springback to obtain corresponding deviation characteristics in order to illustrate the above decoupling mechanism.

[0024] In synchronous springback, multiple molds simultaneously apply constraints and loads to the component. The overall stiffness matrix includes off-diagonal coupling terms to characterize the inter-mold interactions. This stiffness matrix can be expressed as: (1) In the formula, For the total number of molds, For the first The local stiffness matrix of a mold. For the first With the The coupling stiffness matrix between the molds.

[0025] By analyzing the overall stiffness matrix Global eigenvalue decomposition can be represented as a superposition of several fundamental bias fields: (2) In the formula, For the synchronous rebound of the first The fundamental deviation field of order, For the first Generalized stiffness coefficient.

[0026] Each fundamental deviation field satisfies global orthogonality, that is: (3) The overall deviation field of synchronous rebound is composed of the basic deviation fields of each order: (4) Its total deviation field can be expressed as a linear combination of the basic deviation fields of each order: (5) In the formula, As the overall deviation factor, For the first The weighting coefficients of the fundamental deviation field.

[0027] The overall deviation factor is fitted to the actual deviation field using the least squares method. Solve for this: (6) In sequential springback, the springback behavior of each die is relatively independent, and the overall stiffness matrix is ​​in a block diagonal form with no off-diagonal coupling terms. (7) where, is the total number of dies, is the local stiffness matrix of the th die.

[0028] The local stiffness matrix is locally eigen-decomposed, whose eigenvalues and eigenvectors are block-independent. The stiffness matrix of the th die can be decomposed as: (8) where, is the th order fundamental deviation field of the th die, is the th order generalized stiffness coefficient of the th die.

[0029] The global deviation field of the sequential springback is the superposition of the fundamental deviation fields of each die in block structure: (9) where, is the fundamental deviation field of the th die: (10) The total deviation field is the linear combination of the fundamental deviation fields of each die: (11) where, is the global deviation factor, is the local deviation factor of the th die, is the weight coefficient of the th order fundamental deviation field of the th die.

[0030] Based on the block independence, the local deviation factor can be solved by least square fitting the actual deviation field of each die: (12) For the global description of the chain die forming springback, the stiffness coupling effect of path dependence in multi-step continuous plastic deformation needs to be considered, and its incremental tangent stiffness matrix is: (13) where, is the load increment of the th step, is the displacement increment of the th step.

[0031] No. The incremental stiffness matrix of the process is: (14) In the formula, For the total number of molds, For the first Step 1 The local stiffness matrix of a mold. For the first Step 1 With the The coupling stiffness matrix of the mold: right Perform eigenvalue decomposition: (15) In the formula, For the first Step 1 The fundamental deviation field of order, For the first Step 1 Generalized stiffness coefficient.

[0032] No. The overall deviation field of a process is composed of basic deviation fields of various orders: (16) No. The total deviation field of each process is a linear combination of the basic deviation fields of each order: (17) In the formula, For the first Overall deviation factor of each process For the first Step 1 The weighting coefficients of the fundamental deviation field.

[0033] No. The overall deviation factor of each process is fitted to the actual deviation field using the least squares method. Solve for this: (18) The total deviation field of springback during chain die forming can be obtained by summing the deviation fields of each process: (19) Equations (1) and (7) are two ideal decoupling models constructed to reduce computational complexity, corresponding to the two physical mechanisms of synchronous springback and sequential springback, respectively; while Equation (14) is a complete model that is closer to the actual physical process.

[0034] Step S2, the local cumulative springback amount generated by the measured part in the chain die forming process is calculated by using the piecewise equivalent beam theory, and a deviation representation of sequential springback is obtained.

[0035] In this step, first, the part is divided into different characteristic regions such as simply supported beam section, cantilever beam section, linear deformation section based on part features and longitudinal axis. Then the deflection curve equation of each characteristic region is established based on material mechanics. Specifically: The deformation of the outer region is decomposed into a right-end bending moment acting on a two-end simply supported beam and a linear beam with constant slope , and its deflection curve equation is: (20) (21) In the formula, is the elastic modulus, is the cross-sectional moment of inertia, is the length of the beam element.

[0036] The middle region is equivalent to a two-end simply supported beam with support settlement and asymmetric bending moment acting on the right side, and its deflection curve equation is derived using the superposition principle as: (22) The parameterized rewriting of formula (22) can be obtained as: (23) Step S3, the global coupling springback amount generated by the measured part in the forming process is calculated by using the point sequence neural network, and the deviation representation of synchronous springback is obtained.

[0037] In this step, first, the part surface is meshed to form a number of nodes, and a point sequence containing the coordinates of each node is generated. Then an LSTM neural network is constructed, which takes the spatial geometric features of the point sequence as input and outputs the global coupling springback amount of synchronous springback.

[0038] The LSTM neural network structure includes an LSTM layer composed of 128 units and a three-layer fully connected network, where the LSTM layer is used to extract spatial features, and the three-layer fully connected network is composed of 256 nodes, 64 nodes and 1 output node respectively, for predicting springback error.

[0039] Step S4, based on the mold sequence in the forming stage, the weights of the two springback modes are distributed by using the exponential decay weight function, and the sequential springback and synchronous springback are dynamically coupled based on the weights to obtain the final prediction result.

[0040] In the chain die forming process, the physical mechanism is also evolving as the dies are unloaded one by one, for example, the early unloaded dies are less, the unloading behavior of each die is relatively independent, the transmission and release of residual stress is mainly local accumulation, i.e. sequential springback, while the later unloaded dies increase, the released residual stress interacts and cooperates with each other, the global coupling, i.e. synchronous springback effect, becomes more and more significant. Therefore, the results of sequential springback and synchronous springback cannot be simply added fixedly, but the contribution proportion of the two springback modes to the final springback result needs to be intelligently and dynamically adjusted according to the specific forming stage, i.e. the die sequence number, therefore, the embodiment of the application adopts an exponential decay weight function to distribute the weights of the two springback modes: (24) By fitting the actual data by the least square method, the optimal and are obtained: (25) wherein, is the number of die sequences, is the actual springback cross section.

[0041] The predicted cross section based on the coupling of sequential and synchronous springback is: (26) wherein, is the sequential springback weight, is the sequential springback cross section, is the synchronous springback cross section, is the correction error term.

[0042] In summary, the embodiment of the application decomposes the springback generated by chain die forming into sequential springback and synchronous springback, adopts the segmented equivalent beam theory (SEBT) to calculate the local cumulative springback of sequential springback, predicts the global coupling springback of synchronous springback based on the point sequence neural network, and realizes the dynamic coupling of the two by an exponential decay weight function. Sequential springback decomposes the longitudinal axis deformation into responses of characteristic regions such as simply supported beam and cantilever beam by SEBT, and calculates the per-die springback with local cumulative effect. Synchronous springback generates point cloud sequences by grid division, and calculates the multi-die springback with global coupling characteristics during the cooperative release of stress field by a neural network based on point sequences.

[0043] Based on this, the embodiment of the application takes the chain die forming of a typical hat-shaped part as the research and verification object, establishes a finite element simulation model of the chain die forming of the hat-shaped part, adopts a dynamic explicit algorithm to simulate the mold pressing process in the dynamic stage, and adopts a static implicit algorithm to solve the elastic recovery in the springback stage. The deviation between the predicted profile and the simulation profile is compared, so as to verify the springback deviation prediction accuracy of the sequential synchronous springback coupling analysis method, reveal the coupling mechanism of the local springback and the overall springback of the chain die forming of the hat-shaped part, and provide a theoretical basis for the springback deviation prediction and control of the chain die forming part. Specifically: The embodiment of the application takes the chain die forming of a hat-shaped part of ultra-high strength steel, the material of which is MS1500, the thickness of which is 1 mm, and the cross-sectional height of which is 40 mm, as an example. The specific process is as follows: Step S11, obtain the chain die forming process parameters, and build a finite element simulation model.

[0044] Five groups of discrete die blocks (numbers #1~#5) are adopted, the upper die block is pressed down, the lower die block is fixed, and the pressing amount of #1~#5 die blocks at the same time is 0 mm, 20 mm, 40 mm, 20 mm and 0 mm respectively, and the die chamfer radius is 3 mm. The three-dimensional model of the hat-shaped part, the upper die and the lower die is constructed in Abaqus, as shown in Figure 2 . Among them, the sheet metal grid adopts an 8-node reduced integration element (C3D8R), and the die adopts an analytical rigid body model. The transition area between the side wall and the flange of the hat-shaped part is refined, the unit size is 1.5 mm×1.5 mm, and the unit size of other areas is 1.5 mm×1.0 mm. The dynamic explicit algorithm is adopted in the forming stage, and the integral step is 0.01 s; the static implicit algorithm is adopted in the springback stage, and the solution accuracy is set to 1e-5.

[0045] Step S21, calculate the sequential springback of the part to be tested based on the piecewise equivalent beam theory.

[0046] According to the deformation law of the longitudinal axis of the hat-shaped part, the part at the fourth process is divided into five feature areas, as shown in Figure 3 . Area #1 is equivalent to a cantilever beam, areas #2~#4 are equivalent to simply supported beams with non-symmetrical bending moment at the right end, and area #5 is equivalent to a simply supported beam plus a linear beam. The specific parameters in the simply supported beam model are identified by formula (23) as follows: Among them, area #2: A1=-0.118, B1=-7.146×10 -4 , C1=8.496×10 -5 ; Area #3: A2=-0.094, B2=1.849×10 -4 , C2=5.026×10 -5 ; Area #4: A3=-0.085, B3=1.620×10 -3C3 = 5.826 x 10 -6 ; Region #5: A4 = -0.049, B4 = 1.505 x 10 -4 , C4 = 5.534 x 10 -5 ; Settlement height: h1 = 1.196, h2 = 0.865, h3 = 0.651.

[0047] The longitudinal axis obtained by parameterized fitting is discretely sampled in MATLAB, and the spatial distribution of the simulation longitudinal axis scatter data is evaluated for consistency to quantify the model prediction accuracy. To improve the reconstruction accuracy of the part surface, an adaptive piecewise interpolation method based on multiple key point constraints is used in the embodiments of the application, and an additional key point is introduced to optimize the traditional cross-section interpolation model. After preprocessing and coordinate normalization of the point cloud data of the real surface, the surface point cloud predicted based on the piecewise equivalent beam theory is systematically compared and analyzed, as shown in Figure 4 .

[0048] Step S31, the synchronous springback of the part to be measured is calculated based on the LSTM neural network of the point sequence.

[0049] The surface of the hat-shaped part is scanned to generate a point cloud containing all nodes, and the coordinate difference of adjacent nodes is calculated using the eight-neighborhood method to form an input sequence, as shown in Figure 5 . The springback error of the finite element simulation, i.e., the deviation of the measured surface from the initial surface, is used as the label. The global average of the synchronous springback is 2.04 mm, mainly concentrated in the midspan region corresponding to the #1 die, as shown in Figure 6 .

[0050] Step S41, the sequential springback and synchronous springback of the part to be measured are dynamically coupled to obtain the prediction result.

[0051] The center section positions of each die are selected, as shown in Figure 7 . The half-sections at the center section positions of each die under the three springback modes are extracted, as shown in Figure 8 . An exponential decay weight function is used to assign weights to the two springback modes, and the measured deviation is fitted by the least squares method: The sequential weights of the 1# to 5# dies are as follows: Step S51, the prediction result is verified and analyzed.

[0052] As shown in Figure 9 , the spatial matching degree of the overall predicted surface and the measured surface is high.

[0053] The embodiment of the present application shows that the chain die forming springback deviation prediction method based on sequential synchronization springback coupling can efficiently and accurately predict the chain die forming surface of the hat-shaped part, thereby providing a quantitative basis for process parameter optimization.

[0054] In summary, the chain die forming springback deviation prediction method of the present application has the following significant advantages through deep analysis of the springback mechanism and algorithm innovation: (1) The prediction accuracy is significantly improved. In the prior art, the traditional sequential model ignores the stiffness coupling between dies, resulting in global deviation, and the synchronization model lacks time series stiffness evolution analysis, which cannot reflect the path dependence characteristics. Both of them have the problem of incomplete representation of physical mechanism. The present application realizes a breakthrough in accuracy through three innovations: based on the stiffness decomposition theory, the springback is decoupled into sequential springback and synchronous springback, corresponding to two physical mechanisms of local accumulation and global cooperation respectively; when calculating the sequential springback by using the segmented equivalent beam theory, the residual stress transfer law of each die unloading is accurately captured through characteristic region division and deflection equation; when processing the synchronous springback by using the LSTM neural network, the stiffness matrix eigenvalue is included in the input, thereby strengthening the spatial-stiffness coupling feature extraction, and the prediction accuracy is significantly improved.

[0055] (2) The calculation efficiency is improved. The existing finite element full-process simulation needs repeated modeling and iterative calculation, which is time-consuming and lengthy, and cannot meet the real-time optimization needs of industrialization. The present application discards the full-process high-fidelity simulation and adopts the strategy of combining segmented modeling and intelligent prediction. The segmented equivalent beam SEBT model is based on the principle of material mechanics, and the efficiency of calculating the sequential springback is much higher than that of nonlinear finite element analysis. At the same time, once the neural network model is trained, the synchronous springback prediction speed for new working conditions is extremely fast. The independent calculation mode of the two has good parallel potential, and when combined, it avoids the repeated iterative calculation of full-process simulation.

[0056] (3) Strong universality and easy to extend. The existing methods are mostly for specific cross sections or die layouts, and need to be re-modeled when changed, which has poor adaptability. The present application realizes a cross-scene compatible framework through modular design, which has the characteristics of modularity and parameterization. Since the decoupling idea of sequential springback and synchronous springback, the dynamic coupling algorithm and the neural network prediction architecture are universal, for chain die forming parts of different cross-sectional shapes, only the division method of the characteristic region and the corresponding segmented equivalent beam calculation parameters need to be adjusted according to the structural characteristics of the part. The input parameters of the neural network can also be trained according to the new cross-sectional shape, while the network structure and coupling algorithm framework remain unchanged. Therefore, this method is not only suitable for hat-shaped parts, but also can be easily extended and applied to the chain die forming springback prediction of other beam-shaped parts, only the above local adjustments are needed, without re-writing the core algorithm.

[0057] Based on the sequential synchronization rebound coupling based chain die forming rebound deviation prediction method provided in the above embodiments, correspondingly, the application also provides a sequential synchronization rebound coupling based chain die forming rebound deviation prediction device, which is used for decoupling the rebound generated by a part in a chain die forming stage into sequential rebound and synchronization rebound, and then performing rebound prediction. The specific implementation of the device is as follows: As shown in Figure 10 The sequential synchronization rebound coupling based chain die forming rebound deviation prediction device provided in the embodiments of the application includes: A sequential rebound calculation module 101, which is used for calculating the local cumulative rebound generated by a to-be-tested die based on the to-be-tested die to obtain a deviation representation of the sequential rebound.

[0058] A synchronization rebound calculation module 102, which is used for calculating the global coupling rebound generated by the to-be-tested die based on the to-be-tested die to obtain a deviation representation of the synchronization rebound.

[0059] A dynamic coupling prediction module 103, which is used for performing dynamic coupling on the sequential rebound deviation representation obtained by the sequential rebound calculation module 101 and the synchronization rebound deviation representation obtained by the synchronization rebound calculation module 102 by adopting an exponential decay weight function allocation method, so as to obtain a final rebound prediction result of the to-be-tested die.

[0060] Based on any of the above embodiments, another embodiment of the application further provides an electronic device, which can include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor can invoke the logic instructions in the memory to execute the above method.

[0061] In addition, the logic instructions in the memory described above can be implemented in the form of a software functional unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the application or parts of the application that make contributions to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the embodiments of the application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0062] On the other hand, the embodiments of the application also provide a storage medium having a plurality of instructions stored thereon, and the instructions are suitable for being loaded by a processor to execute the sequential synchronization rebound coupling based chain die forming rebound deviation prediction method provided in the above embodiments.

[0063] In one aspect, the technical solutions of the present application can be embodied in software product, which is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program code storage media.

[0064] The device embodiments described above are merely illustrative, and units described as separate components can or can not be physically separated, and components shown as units can or can not be physical units, i.e., can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0065] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be implemented by means of software and the necessary general hardware platform, and of course can also be implemented by hardware. Based on such understanding, the above technical solutions, which are essential or contribute to the prior art, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of the embodiments or some parts of the embodiments.

[0066] In summary, although the present application has been disclosed as above with preferred embodiments, the above preferred embodiments are not intended to limit the present application, and those skilled in the art can make various modifications and decorations without departing from the spirit and scope of the present application, therefore the protection scope of the present application is subject to the scope defined by the claims.

Claims

1. A method for predicting springback deviation in chain die forming based on sequential synchronization springback coupling, characterized in that, The method comprises the following steps: Step 1: decoupling the springback of a part in chain die forming into sequential springback and synchronous springback; Step 2: calculating the local cumulative springback of a part to be tested to obtain a deviation representation of the sequential springback and calculating the global coupled springback to obtain a deviation representation of the synchronous springback; Step 3: dynamically coupling the deviation representation of the sequential springback and the deviation representation of the synchronous springback of the part to be tested to obtain a final springback prediction result of the part to be tested.

2. The method of claim 1, wherein, In the step 2, the local cumulative springback is calculated by using a segmented equivalent beam theory, and the global coupled springback is calculated by using a point sequence neural network.

3. The method of claim 1, wherein, The process of obtaining the local cumulative springback of the part to be tested is as follows: The longitudinal axis deformation of the part to be tested is decomposed into a plurality of characteristic regions by using a segmented equivalent beam theory; The local springback of each characteristic region is calculated by using a deflection curve equation; The local springback of each characteristic region is added to obtain the local cumulative springback of the sequential springback.

4. The method of claim 1, wherein, The process of obtaining the global coupled springback of the part to be tested is as follows: The surface of the part to be tested is meshed to generate a point sequence containing node coordinates; A deviation prediction model of the synchronous springback is constructed; The spatial geometric features of the point sequence are taken as input, and the global coupled springback of the synchronous springback is output.

5. The method of claim 4, wherein, The deviation prediction model of the synchronous springback is constructed based on a point sequence and an LSTM neural network, and the LSTM neural network comprises a spatial feature extraction layer and a fully connected network layer.

6. The method of claim 1, wherein, In the dynamic coupling, an exponential decay weight function is used to assign the weights of the sequential springback and the synchronous springback based on the sequence of the die in the forming stage: And through the least square method fitting actual data, obtain the optimal and : wherein is the number of mold sequences, is the actual rebound section; The predicted cross section based on the coupling of the sequential springback and the synchronous springback is: wherein is the sequential back-elasticity weight, is the sequential back-elasticity section, is the simultaneous back-elasticity section, is the correction error term.

7. The method of claim 1, wherein, Further comprising: Step 4: establishing a finite element simulation model of chain die forming, and verifying the prediction accuracy of the coupling of the sequential springback and the synchronous springback by simulation data; The finite element simulation model comprises a forming stage and a springback stage, the forming stage adopts a dynamic explicit algorithm, and the springback stage adopts a static implicit algorithm.

8. A device for predicting springback of a part produced in a chain die forming stage based on sequential synchronization springback coupling, characterized in that, The device comprises: a sequential springback calculation module for calculating the local cumulative springback of a part to be tested to obtain a deviation representation of the sequential springback; a synchronous springback calculation module for calculating the global coupled springback of the part to be tested to obtain a deviation representation of the synchronous springback; a dynamic coupling prediction module for dynamically coupling the deviation representation of the sequential springback and the deviation representation of the synchronous springback of the part to be tested to obtain a final springback prediction result of the part to be tested. 9.An electronic device comprising a memory and a processor, the memory storing a computer program, wherein, The processor executes the computer program to implement the steps of the method in any one of claims 1 to 7.

10. A storage medium, characterized by The storage medium stores a plurality of instructions, and the instructions are adapted to be loaded by the processor to execute the steps of the method in any one of claims 1 to 7.

Images