A new energy automobile battery bottom guard plate simulation test method

Abstract

The application relates to the technical field of vehicle engineering and computer-aided engineering, and discloses a simulation test method for a new energy automobile battery bottom protection plate. The method constructs a variable elastic modulus constitutive model based on plastic strain induction, performs forming simulation, extracts stress-strain data of a forming final state and geometric coordinate data of a springback final state, obtains equivalent membrane stress by performing membrane bending decoupling analysis on Cauchy stress tensor, calculates virtual damping coefficients of grid nodes by combining stiffness degradation degree and multi-physical field mapping, uses the coefficients to perform reverse weighting correction and virtual strain energy smoothing processing on a geometric deviation vector, generates a mold compensation grid, and finally iteratively executes simulation by taking the compensation grid as a boundary condition. The application can accurately predict the influence of material stiffness degradation on springback, automatically optimizes mold surface precision under the premise of ensuring forming safety through a damping mechanism, and solves the contradiction between high-strength plate forming cracking and springback control.

Description

Technical Field

[0001] This invention relates to the fields of vehicle engineering and computer-aided engineering technology, specifically to a simulation testing method for the bottom protection plate of a new energy vehicle battery. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the battery underbody protection plate, as a core component ensuring battery pack safety, faces increasingly stringent requirements for structural strength and dimensional accuracy. To balance lightweight design and collision safety, underbody protection plates are typically manufactured using ultra-high-strength steel or high-grade aluminum alloys. However, these materials generally exhibit significant springback during forming, making it difficult for the final geometric dimensions of the stamped parts to meet design tolerances, severely impacting subsequent welding and assembly quality. Therefore, utilizing finite element simulation technology for accurate springback prediction and surface compensation during the mold design stage has become a crucial step in shortening the development cycle.

[0003] In existing sheet metal forming simulation techniques, it is typically assumed that the elastic modulus of the material remains constant during unloading. However, actual studies in metal physics show that with the accumulation of plastic deformation, the increase in microscopic damage within the material leads to a non-negligible nonlinear decay of the macroscopic elastic modulus, i.e., stiffness degradation. Existing simulation methods ignore this effect, often resulting in a significant deviation between the predicted springback value and the actual physical state, making it difficult for mold compensation schemes calculated based on this to succeed in a single trial run.

[0004] Furthermore, current mainstream mold surface compensation strategies mostly employ iterative methods based on geometric deviation reverse correction, directly applying reverse deformation to the full extent of the geometric error after springback. While this method theoretically aims to eliminate dimensional deviations, it faces severe process contradictions in engineering practice: to forcibly correct the large springback deviations of high-strength sheet metal, the algorithm often applies excessive reverse deformation in local areas, causing the material stress state in those areas to deteriorate sharply and exceed the forming limit, thus inducing cracking defects. Existing technologies struggle to automatically balance the two mutually constraining goals of dimensional accuracy and forming safety at the algorithmic level. Once the risk of cracking occurs, engineers typically have to manually modify the digital model based on experience, lacking an automated closed-loop control mechanism that can intelligently constrain compensation behavior based on stress state. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a simulation testing method for the bottom protection plate of new energy vehicle batteries. This method solves the problems of distorted springback prediction caused by ignoring material stiffness degradation in the molding simulation of new energy vehicle battery bottom protection plates, and the difficulty of traditional reverse compensation strategies in automatically avoiding the risk of molding cracking while eliminating dimensional deviations.

[0006] To achieve the above objectives, the present invention provides a simulation test method for the bottom protection plate of a new energy vehicle battery, which is implemented in a computer-aided engineering calculation system.

[0007] This simulation testing method first establishes a constitutive model of the material. Based on the plastic strain-induced mechanism, this model defines the evolution of the material's instantaneous elastic modulus as it nonlinearly decreases with increasing cumulative plastic strain. In the finite element simulation environment, this evolution law is set as a material property, replacing the traditional constant elastic modulus setting. Based on this variable elastic modulus constitutive model, the system performs forming simulation calculations on the battery bottom plate, simulating the macroscopic stiffness degradation phenomenon caused by the accumulation of microscopic damage during the forming process of high-strength steel or aluminum alloy materials.

[0008] After the simulation calculation is completed, the method extracts the cumulative plastic strain data of the elements in the final forming state, the Cauchy stress tensor data, and the nodal geometric coordinate data in the final rebound state. To address the technical problem that traditional rebound compensation is prone to local cracking, this method introduces a physics mapping mechanism. Specifically, the system performs membrane bending decoupling analysis on the extracted Cauchy stress tensor data to separate the equivalent membrane stress component that leads to the risk of material fracture, and calculates the stiffness degradation degree in conjunction with the cumulative plastic strain of the elements. Based on these two physical quantities, the equivalent membrane stress and the stiffness degradation degree, the system calculates the virtual damping coefficient of each mesh node through multiphysics mapping. This virtual damping coefficient is a dimensionless scalar used to characterize the risk of local compensation.

[0009] Based on this, the system calculates the geometric deviation vector between the formed workpiece and the design model, and uses the virtual damping coefficient obtained from the previous calculation to perform a reverse weighted correction on this geometric deviation vector. For areas identified as high-risk, the virtual damping coefficient suppresses the magnitude of the reverse compensation. To prevent discontinuities in the mold surface caused by differences in damping coefficients at various nodes, the method further introduces virtual strain energy smoothing to equalize the initial compensation amount after damping correction, thereby generating the final mold compensation mesh.

[0010] Finally, the system uses the generated mold compensation mesh as the new boundary condition and iteratively executes the molding simulation. This process continues until the simulation results simultaneously satisfy the geometric tolerance conditions and the molding safety conditions. At this point, the output mold profile is the final mold process plan.

[0011] In some implementations, constructing a constitutive model based on the variable elastic modulus induced by plastic strain involves characterizing the physical properties of the material. By extracting the unloading stiffness values ​​of the material under different pre-plastic deformation degrees, an exponential decay function relationship between the instantaneous elastic modulus and the cumulative plastic strain is established. This function relationship is integrated into the solver through a user material subroutine, enabling the simulation calculation to update the elastic stiffness matrix in real time according to the current cumulative plastic strain state of the element at each time integration step, thereby improving the accuracy of springback prediction.

[0012] In some implementations, the data extraction steps explicitly distinguish between the final forming state and the final springback state. Final forming state data extraction includes acquiring the in-plane normal stress components, shear stress components, and cumulative plastic strain scalars at each integration point along the thickness direction of the shell element at the moment the mold closes and the process ends. Cauchy stress tensor data characterizes the material's true load-bearing limit state before springback occurs, while the element's cumulative plastic strain data characterizes the local stiffness damage state of the material.

[0013] In some implementations, to accurately identify cracking risk, the membrane bending decoupling analysis step performs numerical integration on the Cauchy stress components distributed along the thickness direction, calculating the zero-order moment of each stress component, thereby separating the membrane force tensor acting on the neutral layer plane from the complex stress state. By substituting the membrane force tensor into the stress invariant formula to calculate the scalar equivalent membrane stress and comparing it with the material yield strength, a membrane stress distribution field reflecting the true cracking risk is generated. This process eliminates the interference of bending stress on cracking determination.

[0014] In some implementations, the calculation of the virtual damping coefficient constructs a mapping from material state to process constraints. Specifically, the stiffness degradation rate parameter is obtained by calculating the ratio of the final elastic modulus to the initial elastic modulus. The damping function includes a stiffness sensitivity term and a stress safety barrier term. When the stiffness degradation rate parameter increases, indicating severe material softening, or when the equivalent membrane stress approaches the material's yield limit, the virtual damping coefficient output by the function increases. Preferably, the stress safety barrier term uses a hyperbolic tangent nonlinear function to map the normalized equivalent membrane stress, so that when the stress approaches a critical value, the damping coefficient rises sharply and approaches a value of one, thereby freezing the geometric compensation operation in this region at the numerical calculation level and preventing cracking due to excessive stretching.

[0015] In some implementations, the generation process of the mold compensation mesh employs a strategy combining weighted correction and smoothing constraints. The system calculates the geometric deviation vector and its inverse vector, and uses the difference between the deviation vector and the virtual damping coefficient as a weighting coefficient to scale the magnitude of the inverse vector. This means that for nodes in high-risk areas, the geometric compensation is significantly reduced or even reduced to zero.

[0016] In some implementations, to ensure the machinability and smoothness of the mold surface, the method applies a virtual strain energy smoothing constraint to the initial compensation amount. The initial compensation amount is considered as a virtual displacement applied to the mesh nodes, and the resulting locally added virtual strain energy density is calculated. For nodes with energy densities exceeding a preset threshold, the system performs Laplace smoothing on them and their neighboring nodes, eliminating abrupt changes by amortizing the deformation compensation amount. The final node compensation vector is then superimposed onto the original mesh to reconstruct the mold surface.

[0017] In some implementations, iterative control employs a dual-criteria approach: first, a geometric tolerance condition, requiring the maximum normal distance to be less than the tolerance value; and second, a forming safety condition, requiring the equivalent membrane stress of all units to be less than the safety threshold and with no local exceedances. Logically, if forming is safe but geometric tolerances are not met, conventional compensation iterations continue. If forming safety conditions are not met, regardless of whether the geometric dimensions meet the standards, the virtual damping coefficient of the corresponding region is forcibly increased in the next iteration. By reverting a portion of the geometric compensation, forming safety is prioritized, thereby achieving an automatic balance between accuracy and safety in the process design.

[0018] This invention provides a simulation testing method for the bottom protection plate of a new energy vehicle battery. It has the following beneficial effects: This invention constructs a constitutive model of variable elastic modulus based on plastic strain-induced elasticity, defining the instantaneous elastic modulus of a material as a variable that decreases nonlinearly with increasing cumulative plastic strain. This overcomes the calculation bias caused by the assumption of a constant elastic modulus in traditional simulation methods. This method can accurately characterize the stiffness degradation characteristics of high-strength steel or aluminum alloy materials used in battery bottom plates after undergoing large deformation, significantly improving the physical reality and calculation accuracy of molding springback prediction, thereby reducing the number of trial and error attempts in mold development.

[0019] This invention utilizes the membrane bending decoupling analysis of Cauchy stress tensor and the virtual damping coefficient mapping mechanism to achieve intelligent constraint on mold compensation. By separating the equivalent membrane stress that leads to cracking from the complex stress state and mapping it together with the stiffness degradation degree as a virtual damping coefficient, this method can automatically identify and freeze the geometric compensation operation of high-risk areas at the algorithm level. This processing effectively solves the technical contradiction of simply pursuing dimensional accuracy in traditional reverse compensation methods, which leads to local breakage of the component, and ensures the safety of the molding process.

[0020] This invention introduces virtual strain energy smoothing constraints and dual convergence criteria to ensure the machinability of the final mold surface and the feasibility of the process scheme. It uses smoothing processing based on the Laplace operator to eliminate the abrupt changes in the compensation mesh caused by damping differences, generating a smooth and continuous mold surface. At the same time, based on the dual iterative logic of geometric tolerance and molding safety, a convergence path prioritizing safety is established, which can automatically search for the optimal process window that balances dimensional accuracy and molding quality, avoiding the virtual iteration from falling into an infinite loop. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the compensation vector generation algorithm with energy smoothing constraints of the present invention. Figure 3 This is a schematic diagram of the virtual iterative closed-loop control and convergence determination logic of the present invention. Detailed Implementation

[0022] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] See attached document Figure 1 This invention provides a simulation test method for the bottom protection plate of a new energy vehicle battery, which runs on a computer-aided engineering calculation system containing a processor and a memory.

[0024] This method mainly includes the following steps: First, a constitutive model of variable elastic modulus based on plastic strain-induced deformation is constructed.

[0025] The computer system receives uniaxial tensile test data and cyclic loading and unloading test data for 550 grade high-strength steel.

[0026] The processor extracts the unloading stiffness values ​​of the material under different pre-plastic deformation degrees based on the experimental data of cyclic loading and unloading.

[0027] The processor establishes a functional relationship between instantaneous elastic modulus and cumulative plastic strain through numerical fitting, and defines the evolution law of nonlinear decay of the material's elastic modulus with the increase of cumulative plastic strain.

[0028] The evolution law is written as a user material subroutine that can be called by the finite element solver, which is used to update the elastic stiffness matrix of each element in real time during the simulation calculation.

[0029] Next, initial modeling simulation and multiphysics data extraction are performed.

[0030] The computer system imports the digital model file of the battery bottom cover plate and creates a finite element mesh model that includes the concave die, convex die, and pressure ring.

[0031] In the finite element model, assign the material properties defined by the above user material subroutine, and set process boundary conditions including blank holder force, friction coefficient and stamping speed.

[0032] The processor performs explicit or implicit finite element calculations to simulate the entire process of sheet metal from its initial state to the end of forming.

[0033] After the simulation, the computer system extracts the nodal geometric coordinate data of the formed workpiece, the cumulative plastic strain data of each unit integration point, and the Cauchy stress tensor data.

[0034] Subsequently, a membrane bending decoupling analysis of the residual stress tensor was performed.

[0035] The processor reads the Cauchy stress components at each integration point in the thickness direction for each shell element.

[0036] The processor performs numerical integration on the Cauchy stress components distributed along the thickness direction, separating the membrane stress components acting on the neutral layer plane and the bending moment components acting around the neutral axis.

[0037] The computer system calculates the equivalent membrane stress value based on the membrane stress components, and compares the equivalent membrane stress value with the yield strength of the material to generate membrane stress distribution field data.

[0038] Then, the virtual damping coefficient based on the local stiffness degradation gradient is calculated.

[0039] The processor calculates the ratio of the final elastic modulus to the initial elastic modulus of the element to which each grid node belongs, and obtains the stiffness degradation rate parameter.

[0040] The processor combines stiffness degradation rate parameters with membrane stress distribution field data to calculate the virtual damping coefficient of each grid node.

[0041] The virtual damping coefficient is positively correlated with the stiffness degradation rate and the equivalent membrane stress. That is, in regions where stiffness decay is severe or membrane stress is close to the yield limit, the virtual damping coefficient has a larger value.

[0042] Next, a compensated mesh with energy smoothing constraints is generated.

[0043] The normal geometric deviation vector between the geometric coordinates of the formed workpiece nodes obtained from computer system simulation and the design target digital model.

[0044] The processor uses a virtual damping coefficient to perform a weighted calculation on the normal geometric deviation vector to obtain the initial compensation vector after damping correction. For nodes with a large virtual damping coefficient, the magnitude of their initial compensation vector is reduced relative to the magnitude of the geometric deviation vector.

[0045] The processor calculates the local additional elastic strain energy density introduced after the initial compensation vector is applied.

[0046] When the local additional elastic strain energy density exceeds a preset threshold, the processor performs Laplace smoothing on the node and its neighboring nodes, distributing the deformation of the node to the neighboring nodes to generate the final node compensation vector.

[0047] Finally, virtual iterative closed-loop control and convergence determination are performed.

[0048] The computer system superimposes the final node compensation vector onto the original mold mesh data to generate the corrected mold surface data.

[0049] The computer system uses the corrected mold surface data as new boundary conditions to re-execute the molding simulation step, data extraction step, and decoupling analysis step.

[0050] The processor determines whether the current simulation results simultaneously meet the geometric tolerance conditions and the molding safety conditions.

[0051] The geometric tolerance condition is that the maximum normal distance between the node of the formed workpiece and the design target digital model is less than the preset tolerance value.

[0052] The molding safety condition is that the equivalent membrane stress value of all units is less than the set safety threshold.

[0053] If both of the above conditions are met, the current mold surface data will be output as the final process design scheme; otherwise, the virtual damping coefficient calculation and compensation mesh generation steps will continue to be executed using the current data until the convergence condition is met.

[0054] This embodiment details the specific process of constructing a variable stiffness constitutive model applicable to high-strength steel of 550MPa grade and above. This process transforms physical experimental data into numerical algorithms executable in a computer simulation system.

[0055] First, experimental data on the stiffness degradation of the material were obtained. A uniaxial cyclic loading and unloading test was performed on standard plate specimens using an electronic universal testing machine. The machine's control program was set to a multi-stage strain control mode, meaning that after the specimen underwent a preset proportion of plastic deformation, the stress was completely unloaded, and then reloaded until a higher proportion of plastic deformation was achieved. This process was repeated until the specimen necked or fractured. The preset plastic deformation proportions covered the entire plastic domain from the micro-yield stage to before necking.

[0056] The processor receives stress-strain time-series data acquired by the extensometer and force sensor of the testing machine. For each unloading-reloading hysteresis loop, the processor extracts stress and strain data points along the unloading path, calculates the average slope of the path using a linear regression algorithm, and determines the slope value as the instantaneous elastic modulus corresponding to the current cumulative plastic strain state. By processing all graded loop data, a discrete dataset containing the cumulative equivalent plastic strain independent variable and the instantaneous elastic modulus dependent variable is generated.

[0057] Next, a mathematical model of stiffness evolution is established. The processor calls the nonlinear least squares fitting module and uses an exponential decay function to identify the parameters of the discrete dataset.

[0058] The exponential decay function used is as follows: ; in, Defined as the current cumulative equivalent plastic strain The corresponding elastic modulus value; Defined as the initial elastic modulus when the plastic strain is zero; Defined as the saturated residual elastic modulus after plastic deformation has fully developed; Defined as the modulus degradation rate constant, it characterizes how quickly stiffness decreases with increasing plastic deformation. The processor calculates the output parameters through fitting. , and The optimal solution is obtained, thus completing the mathematical definition of the material stiffness degradation characteristics.

[0059] Subsequently, the mathematical model is compiled into a user material subroutine for the finite element analysis system. This subroutine is configured to be called in each time integration step of the solver, and its execution logic is as follows: The subroutine first defines and initializes the state variable, which is used to record and transmit the history value of the cumulative equivalent plastic strain at each element integration point during the calculation process.

[0060] At the start of any computational increment step, the subroutine reads the cumulative equivalent plastic strain value from memory at the previous moment of that integration point.

[0061] The subroutine substitutes the accumulated equivalent plastic strain value into the aforementioned exponential decay function to calculate the instantaneous elastic modulus at the current moment of the integral point.

[0062] Based on the calculated instantaneous elastic modulus, the subroutine updates the elastic stiffness matrix (or flexibility matrix) in Hooke's Law. For the isotropic material assumption, the Young's modulus term in the matrix is ​​updated; for the anisotropic material assumption, the stiffness components in each direction are synchronously scaled according to the preset anisotropy coefficient ratio.

[0063] Finally, the subroutine performs a tensor product operation using the updated stiffness matrix and the strain increment tensor of the current increment step to calculate the trial stress increment. It then performs stress loop integration by combining the yield criterion (such as Hill48 or VonMises criterion) and the flow rule, and outputs the updated Cauchy stress tensor and the updated cumulative plastic strain state variables for use in the next increment step.

[0064] This embodiment details the specific technical implementation of using a finite element analysis system to perform full-process numerical simulation of the battery bottom protection plate and obtain key physical field data.

[0065] First, a discretized numerical model of the process system is constructed. The computer system imports the geometric model of the battery bottom cover plate and generates the geometric surfaces of the die, punch, and pressure ring based on this model. For the die components, rigid shell elements are used for meshing, ensuring that they do not undergo elastic or plastic deformation during the calculation. For the battery bottom cover plate material, four-node elasto-plastic shell elements are used for mesh discretization. To accurately capture the stress gradient distribution along the thickness direction of the plate material, at least five Gaussian numerical integration points are set in the thickness direction of the shell elements.

[0066] In the finite element preprocessing module, the user material subroutine with variable stiffness characteristics generated in Example 2 is associated with the sheet metal mesh. Simultaneously, contact friction conditions are set, employing either a contact model based on Coulomb's law of friction or an enhanced friction model where the friction coefficient dynamically varies with contact pressure and sliding velocity, to simulate the actual contact behavior between the sheet metal and the mold surface. The loading curve of the blank holder is set to apply a constant or time-varying blank holder force during the forming process.

[0067] Next, a multi-step sequential coupled solution calculation is performed. The processor first calls the explicit dynamics solution module to simulate gravity settlement, edge closing, and drawing processes in sequence. During the drawing calculation, the user material subroutine is activated, dynamically adjusting the stiffness matrix based on the real-time deformation history of each element. When the drawing process is completed, the die movement stops, and the sheet metal is in a closed-die holding pressure state, the system records the current time step as the final forming state.

[0068] Next, the processor calls the implicit statics solution module to perform springback calculation. The system removes the mold constraints and external loads, retaining only the residual stress field inside the sheet metal. It uses the Newton-Raphson iteration method to calculate the process by which the sheet metal reaches a new mechanical equilibrium position under the drive of internal forces. The state corresponding to this equilibrium position is recorded as the final springback state.

[0069] Finally, the directional extraction and mapping of multiphysics data are performed.

[0070] For the final state of springback, the processor reads the spatial coordinates of all nodes in the sheet metal mesh and performs normal projection calculations on them and the surfaces of the original design model. For each node, it calculates the shortest Euclidean distance and direction vector to the design surface, generating a geometric deviation vector field.

[0071] For the final forming state (i.e., the moment before springback occurs), the processor extracts the Cauchy stress tensor components at all thickness integration points within each shell element, specifically including in-plane normal stress and shear stress components. Simultaneously, it extracts the cumulative equivalent plastic strain scalar data for each integration point. These final forming stress and strain data are labeled and stored as input sources for subsequent calculations of the stiffness degradation field and for membrane-bending decoupling analysis. This step ensures that the data used to analyze crack risk and stiffness damage reflects the material's true load-bearing limit state before springback deformation.

[0072] This embodiment details a specific algorithm for mechanically decoupling the complex stress state inside a sheet metal, aiming to separate the tensile dominant factor that leads to material fracture from the bending dominant factor that leads to geometric springback.

[0073] For each shell element, the processor reads the Cauchy stress tensor data at all Gaussian integration points along the thickness direction. For general three-dimensional shell structures, the focus is on the in-plane stress components, i.e., the normal stresses along the principal axes of the element's local coordinate system. , and shear stress .

[0074] The processor uses a numerical integration algorithm to calculate the zero-order moment and first-order moment of each stress component.

[0075] The zeroth moment corresponds to the membrane force tensor. This characterizes the in-plane tensile or compressive force flow acting on the neutral surface of an element. The calculation process involves multiplying the stress values ​​at each integration point by the corresponding thickness integration weighting coefficient and then summing them. Mathematically, this is expressed as the stress flow along the thickness coordinate... Integrating the stress distribution function: ; in, The instantaneous thickness of the sheet metal unit. In-plane coordinate index, For coordinate variables along the thickness direction, the range of values ​​is from arrive In discrete computation, this integral is transformed into a weighted sum of the contributions at all integration points.

[0076] The first moment corresponds to the bending moment tensor This characterizes the internal force couple that causes bending deformation of the element. Its calculation process involves combining the stress value at each integration point with its distance from the neutral surface. Integrating the product along the thickness direction: ; After completing the tensor decomposition, the processor calculates the equivalent membrane stress in scalar form based on the membrane force tensor. This is used to assess the risk of cracking. The calculation is based on the stress invariant principle of the Von Mises yield criterion, substituting the components of the membrane force tensor into the equivalent formula: ; The computer system will calculate the equivalent membrane stress of each element. By comparing the stress distribution with the material's yield strength or forming limit stress, membrane stress distribution field data is generated. In this data field, each element is assigned a stress load coefficient, which quantitatively characterizes how close the region is to tensile failure under the current forming state, providing a mechanical state constraint basis for determining the virtual damping coefficient in subsequent steps.

[0077] This embodiment details the specific processing procedure for converting discrete plastic strain data obtained from molding simulation into a global stiffness degradation field to guide mold compensation.

[0078] The processor first accesses the final state database recorded in memory, traversing all shell units that make up the battery bottom cover. For any shell unit, the system reads the cumulative equivalent plastic strain value at all Gaussian integral points in the thickness direction. Because the battery bottom cover undergoes complex tensile deformation during stamping, the degree of plastic deformation varies significantly in different areas. For example, there may be a large accumulation of plastic strain at the root of the reinforcing rib or at the corner, while the plastic strain is smaller in the bottom planar area.

[0079] In order to construct a macroscopic physical field that can characterize the local resistance to deformation of the sheet metal, the processor calls the variable stiffness constitutive model constructed in Example 2, substitutes the cumulative equivalent plastic strain value of each integration point into the exponential decay function, and calculates the instantaneous elastic modulus of the integration point at the unloading time.

[0080] Subsequently, the processor performs element-level stiffness homogenization. Since the stiffness at different points along the thickness direction of the shell element may vary due to bending deformation (e.g., compression on the inside, tension on the outside), to obtain a scalar value that represents the overall in-plane stiffness characteristics of the element, the processor calculates a weighted average of the instantaneous elastic modulus at all integration points along the thickness direction. The weighting coefficients depend on the weight values ​​corresponding to each integration point in Simpson's integral rule or Gaussian integral rule. This weighted average is defined as the equivalent unloading modulus of the shell element.

[0081] Next, a smooth mapping of field data from the element center to the mesh nodes is performed. Since strain and modulus in finite element calculations are usually stored at the element center or integration point, and the mold compensation operation is based on the displacement adjustment of the mesh nodes, the processor uses shape function interpolation algorithms or inverse distance weighting algorithms to project the equivalent unloading modulus of all adjacent elements around a shared node onto that node, thus obtaining the node-level final modulus of that node. .

[0082] Finally, the processor performs dimensionless normalization on the modulus data of all nodes to generate a stiffness degradation field. The system calculates the stiffness retention rate of each node (i.e., the ratio of the node's final modulus to the material's initial modulus). The ratio of the two values ​​is used to construct a stiffness degradation factor. This factor intuitively quantifies the spatial heterogeneity of material properties: in In the region where the value approaches 0, it indicates that the material stiffness has not significantly decreased; while in the region where the value approaches 0, it indicates that the material stiffness has not significantly decreased. Regions with larger values ​​indicate a significant decrease in material stiffness, i.e., the material softens. This stiffness degradation field data is stored in memory as a scalar contour plot, serving as a direct input variable for subsequent calculations of the virtual damping coefficient.

[0083] This embodiment details how to construct a mathematical function that can simultaneously respond to material stiffness degradation and membrane stress critical states in order to generate a virtual damping coefficient for controlling the compensation step size. .

[0084] The processor targets each mesh node. Retrieve the normalized stiffness value at that node stored in memory. ) and equivalent membrane stress value ( The constructed damping function aims to output a dimensionless scalar between 0 and 1. The larger the value of this scalar, the stronger the suppression of geometric compensation operations at that location.

[0085] The specific damping function construction includes two core coupled terms: the stiffness sensitivity term and the stress safety barrier term.

[0086] The first term is the stiffness sensitivity term, used to respond to macroscopic softening caused by microscopic damage to the material. (Processor calculation) This value directly quantifies the proportion of stiffness loss. At a physical level, when the material stiffness in a certain region decreases significantly (i.e.,...) much smaller Its resistance to external forced displacement weakens, and even a tiny geometric overcompensation can induce local buckling or undulation. Therefore, the mechanism of this term is: as the proportion of stiffness loss increases, the damping value is linearly increased, forcibly reducing the geometric correction magnitude of this region in a single iteration.

[0087] The second term is the stress safety barrier term, used to prevent fracture induced by compensation operations. The processor introduces a nonlinear mapping function, typically a hyperbolic tangent function. The membrane stress state is normalized and mapped. The calculation formula is as follows.

[0088] ; in, The yield strength of the material is used as the benchmark threshold for stress normalization. The stiffness influence weighting coefficient, The stress influence weighting coefficients are read in through a preset configuration file and typically satisfy the following conditions: ; This is a truncation function to ensure the final output is... It is strictly limited to the closed interval [0,1].

[0089] In this functional logic, the hyperbolic tangent function It plays a crucial role in the soft threshold function. When the equivalent membrane stress value of the node... At a low level, The function value grows slowly and contributes little to the stress term, allowing the algorithm to perform large geometric compensations to eliminate springback; as the membrane stress gradually approaches the yield strength... hour, The slope of the function increases sharply, leading to The value rapidly climbs to near 1. This nonlinear response mechanism forms a numerical safety barrier, automatically freezing the geometric position of the node when the stress approaches the critical point, prohibiting any tensile compensation operations that could further worsen the membrane stress, thus mitigating the risk of physical cracking at the algorithmic level through mathematical constraints.

[0090] This embodiment deeply analyzes the physical orientation of the virtual damping field generated by the algorithm in the actual stamping process design, and clarifies how the field data serves as an intelligent process constraint map to establish a dynamic balance between springback compensation and forming safety.

[0091] The processor displays the virtual damping coefficient in the visualization module. The image is rendered as a colored cloud map overlaid on the 3D mesh of the battery bottom cover. This cloud map is not merely an intermediate result of mathematical calculations, but rather a visual reflection of the correctability or process sensitivity of the material in its current forming state.

[0092] High damping region ( Approaching 1.0): In the cloud map, these areas typically correspond to the reinforcing rib roots of the battery bottom plate, small rounded corner radius features (R corners), and deeply stretched vertical sidewall areas.

[0093] From a physical perspective, these regions undergo intense plastic flow during the forming process, resulting in extremely high dislocation density within the material and a significant decrease in macroscopic elastic modulus (severe stiffness degradation). Simultaneously, these regions typically bear membrane tensile stresses close to the material's limits.

[0094] Generating a high damping value at this location signifies that the algorithm has physically identified this region as being in a critical sub-failure state. The damping coefficient essentially acts as a geometric freeze command: it informs the compensation system that although there may be geometric springback deviations at this location, it is strictly forbidden to correct these deviations by forcibly moving the mold surface in the opposite direction. This is because, with the stiffness already significantly reduced and the stress extremely high, any further geometric displacement adjustment is highly likely to lead to excessive local thinning or even direct fracture. Therefore, the physical meaning of the high damping region is a constrained area that sacrifices the accuracy of local geometric compensation for process safety.

[0095] Low damping region ( Approaching 0): These areas typically correspond to the large, flat bottom of the underbody plate, the flange face, and shallow stretching areas with gentle curvature changes.

[0096] From a physical perspective, these regions primarily undergo elastic deformation or slight plastic deformation, with the material's elastic modulus remaining near its initial value and the membrane stress level far below the yield limit. However, due to the low stiffness of large-area flat plate structures, these regions are often the areas where springback deformation (such as warping and collapse) is most significant.

[0097] Generating a low damping value here means the algorithm determines that this area is a safe and adjustable zone. The system will allow the full application of the calculated reverse geometry compensation, making full use of the material's remaining plastic deformation capacity and offsetting springback through over-crowning of the mold surface.

[0098] Transition damping region ( Between 0 and 1: Located at the junction of the two extreme regions mentioned above. The damping coefficient in this region exhibits a gradient distribution. Its physical significance lies in maintaining the continuity and smoothness of the mold surface. It prevents surface defects such as steps or wavy lines caused by adopting drastically different compensation strategies for adjacent nodes (one completely stationary, the other moving significantly).

[0099] In summary, the construction of a virtual damping field essentially transforms traditional full-field equivalent compensation into an adaptive differentiated compensation based on the material's load-bearing state. Numerically, it mimics the decision-making logic of experienced process engineers: applying gentle pressure or leaving it alone in areas prone to cracking, while applying heavy pressure to areas with simple springback, thereby maximizing the dimensional accuracy of the component without introducing new cracking risks.

[0100] See attached document Figure 2 This embodiment details how to use the virtual damping coefficient calculated in the aforementioned steps to correct the geometric springback deviation, and how to solve the problem of mold surface discontinuity caused by uneven damping through an energy constraint mechanism, ultimately generating a mold compensation mesh that can be used for actual processing.

[0101] Phase 1: Damped Initial Vector Correction The processor first reads the geometric deviation field obtained from the simulation calculation. This field includes the normal position errors of all grid nodes relative to the design model. Subsequently, the system performs a damped weighted calculation node by node. For any node... The system calls its corresponding virtual damping coefficient. (Generated from Example 5), the primary compensation vector is calculated through linear scaling. ; In this step, the technical solution achieves hard constraints on local deformation. For nodes located in high-risk areas (high-damping zones), even if they have large springback deviations (e.g., (Very large), the amount of compensation that takes effect It will also be forcibly compressed to near zero. While this treatment ensures that the single point will not break due to overcompensation, direct application... (The text abruptly shifts to a different topic) ...because the damping coefficients of adjacent nodes may differ (e.g., from the crack risk zone to the safe zone),... This can cause drastic changes in the spatial distribution of the compensated vector field, resulting in tiny steps, spikes, or wavy patterns on the generated mold surface, which is unacceptable in stamping die processing.

[0102] Phase 2: Smoothing detection based on additional strain energy density To eliminate the aforementioned discontinuities, the processor introduces a smoothing detection mechanism based on additional virtual strain energy. The system treats the primary compensation vector field as a virtual displacement field applied to the mold mesh. For each node, the system calculates the local additional elastic strain energy density caused by this virtual displacement field. This calculation is not based on actual physical deformation, but rather on energy functional analysis of the mesh topology. The processor utilizes the node's current degradation modulus. As a stiffness weight, the virtual elastic energy generated between this node and its first-order neighboring nodes due to the difference in compensation is calculated: ; in, The summation symbol. It represents summation over all nodes. Adjacent nodes Perform cumulative calculations. Represents a node The compensation vector, The square of the Euclidean norm.

[0103] If a certain node If the energy smoothing threshold is exceeded, it indicates that the compensation amount at this point has a sudden change relative to the surrounding neighborhood, and equalization processing is required.

[0104] The third stage: Laplacian diffusion smoothing. For detected high-energy nodes, the processor executes an iterative smoothing algorithm based on the Laplacian operator. The physical essence of this algorithm is the distribution of deformation requirements. When a critical node is restricted from movement due to high damping, the smoothing algorithm drives its neighboring nodes to share part of the deformation compensation task, thus forming a smooth transition surface. The iterative update formula is as follows: ; in, The relaxation factor controls the speed of smooth convergence and typically takes a value between 0.1 and 0.5. For nodes The set of topologically adjacent nodes; Distance weights are typically taken as the reciprocal of the node spacing. Represents a node exist The compensation vector at time (the next time step), Represents a node exist The compensation vector at time step, Represents a node exist The compensation vector at time step.

[0105] Through several iterations (usually 3 to 5), the high-frequency abrupt changes in compensation are diffused to the surrounding area. The resulting compensation vector field not only preserves the protective effect of the damping mechanism on the dangerous area (i.e., the dangerous point itself does not undergo large displacement), but also achieves the C1 continuity (tangential smoothness) of the overall mold surface through the coordinated deformation of the neighborhood.

[0106] Phase 4: Mold Mesh Reconstruction Finally, the processor will smooth the final compensation vector. The reconstructed mesh is superimposed onto the node coordinates of the original mold mesh. The system performs a quality check on the reconstructed mesh to ensure that there is no mesh distortion caused by the compensation operation (e.g., a negative Jacobian determinant). The generated corrected mesh is exported in STL or IGES format and used as the input geometric boundary for the next round of virtual iterative simulation.

[0107] See attached document Figure 3 This embodiment describes in detail how to construct an automated closed-loop control system. This system connects the aforementioned variable stiffness simulation, stress decoupling analysis and damping compensation algorithm, and through multiple rounds of virtual trial and error, finds the optimal mold surface that balances dimensional accuracy and molding safety.

[0108] Mold Mesh Update and Simulation Reset: At the start of each iteration, the processor first receives the energy-smoothed compensation mesh data from the previous iteration. The system uses this mesh data to replace the geometric surfaces of the die and punch in the original finite element model, while keeping the blank holder geometry, initial sheet shape, friction coefficient, and blank holder force unchanged.

[0109] In the restarted simulation, the variable stiffness constitutive model based on plastic strain-induced deformation in Example 2 remains active. This means that when a small change in the mold profile (such as overcorrection) causes a change in the material flow path and local plastic strain accumulation of the sheet metal, the system recalculates the stiffness degradation history of each element. This dynamic update mechanism ensures that each round of springback prediction is based on the latest material stiffness state, rather than using old stiffness data, thus eliminating the iteration non-convergence problem caused by ignoring changes in the stiffness field in traditional methods.

[0110] Construction and determination of the dual convergence criterion: After the simulation, the processor performs an automated convergence determination. Unlike traditional methods that only consider a single criterion for dimensional deviations, this embodiment constructs a dual convergence criterion for geometric accuracy forming safety, requiring the simultaneous satisfaction of the following two inequality constraints: Geometric accuracy convergence criterion: The system calculates the maximum value (or a specific percentile value, such as the 95th percentile) of the normal distance deviation between the current formed workpiece and the design model at all nodes in the field. The judgment formula is as follows: ; in, This means finding the node with the greatest change (the most unstable node) among all nodes in the model. For the first Residual geometric deviation after round iteration Preset manufacturing tolerances (e.g.) This guideline ensures that the dimensions of the final product meet assembly requirements.

[0111] Forming safety convergence criterion: The system checks the membrane stress state of all elements to prevent material tearing due to overcompensation. The judgment formula is as follows: ; in, This means extracting the value with the highest stress ratio among all elements. For the first The unit equivalent membrane stress after round iteration, The forming limit stress or tensile strength of the material. A safety factor (e.g., 0.9 or 0.95) is used. This guideline ensures that the compensated mold surface will not induce new process defects.

[0112] Iteration strategy and termination mechanism: The processor executes the logical branch based on the decision result of the dual criteria: Scenario 1: Double Convergence (Pass). If the geometric deviation is within the tolerance range and the membrane stress in all regions is below the safety threshold, the system determines that the iteration is successful. At this time, the processor exports the current mold mesh data as a standard CAD surface in IGES or STEP format as the final CNC machining data.

[0113] Scenario 2: Geometric out-of-tolerance, but safe (Fail-Safe). If springback out-of-tolerance exists, but the membrane stress is at a low level (safe zone), the system will directly use the current residual deviation to calculate the new compensation amount and apply it in full in the next iteration to eliminate springback.

[0114] Scenario 3: Fail-Risk. If local membrane stress exceeds the safety threshold (i.e., there is a risk of cracking), the system will forcibly trigger a high-damping mechanism, even if the geometry may meet the requirements or still deviate. In the next round of calculation, the virtual damping coefficient of this region will be used. The algorithm will automatically push it up to close to 1.0, thereby forcing a partial reduction in deformation from that area in subsequent compensation calculations to prioritize molding safety.

[0115] Termination condition: To prevent infinite loops caused by physical inconsistencies (i.e., material properties not meeting the geometric design requirements), the system sets a maximum number of iterations. (For example, 10 times). If convergence is not achieved after reaching this number of times, the system will output the current optimal solution and mark the deadlock area that cannot simultaneously meet the size and safety requirements, prompting the engineer to modify the product design (such as changing the radius or adding process cuts) instead of continuing to adjust the mold.

Claims

1. A simulation test method for the bottom protection plate of a new energy vehicle battery, characterized in that, Includes the following steps: A constitutive model of variable elastic modulus based on plastic strain-induced elasticity is constructed, and the evolution law of instantaneous elastic modulus decaying nonlinearly with cumulative plastic strain is set, and the evolution law is set as the simulated material property. Based on the aforementioned variable elastic modulus constitutive model, a forming simulation was performed on the battery bottom cover plate, and the unit cumulative plastic strain data, Cauchy stress tensor data, and nodal geometric coordinate data of the final forming state and the rebound state were extracted respectively. The equivalent membrane stress is calculated by performing membrane bending decoupling analysis on the Cauchy stress tensor data, and the virtual damping coefficient of the mesh nodes is calculated by multiphysics mapping in combination with the degree of stiffness degradation. The mold compensation mesh is calculated by using the virtual damping coefficient to perform reverse weighted correction on the geometric deviation vector and virtual strain energy smoothing; the mold compensation mesh is used as the boundary condition to iteratively execute the molding simulation until the dual criteria of geometric tolerance and molding safety are met, and the mold process scheme is output.

2. The simulation test method for the bottom protection plate of a new energy vehicle battery according to claim 1, characterized in that, The steps for constructing the constitutive model of variable elastic modulus based on plastic strain-induced deformation specifically include: Extract the unloading stiffness values ​​of the material under different pre-plastic deformation degrees; Establish an exponential decay function relationship between the instantaneous elastic modulus and the cumulative plastic strain; The exponential decay function is written as a user material subroutine, and the forming simulation is set to call the subroutine in each time integration step, so as to update the elastic stiffness matrix in real time according to the current cumulative plastic strain of the element.

3. The simulation test method for the bottom protection plate of a new energy vehicle battery according to claim 1, characterized in that, The steps of extracting the cumulative plastic strain data and Cauchy stress tensor data of the unit at the final state of molding specifically include: At the moment when the molding process is completed and the mold is kept closed, extract the in-plane normal stress component, shear stress component and cumulative plastic strain scalar of each shell element at each integration point in the thickness direction. The Cauchy stress tensor data is used to characterize the material's true load limit state before springback occurs, and the unit cumulative plastic strain data is used to characterize the material's local stiffness damage state.

4. The simulation test method for the bottom protection plate of a new energy vehicle battery according to claim 3, characterized in that, The steps for calculating the equivalent membrane stress by performing membrane bending decoupling analysis on the Cauchy stress tensor data specifically include: Read the Cauchy stress components at each integration point in the thickness direction for each shell element; The zero-order moment of each stress component is calculated by numerical integration of the Cauchy stress components distributed along the thickness direction, thereby separating the membrane force tensor acting on the neutral layer plane. The equivalent membrane stress in scalar form is calculated by substituting the membrane force tensor into the stress invariant formula, and the equivalent membrane stress is compared with the yield strength of the material to generate a membrane stress distribution field.

5. The simulation test method for the bottom protection plate of a new energy vehicle battery according to claim 1, characterized in that, The step of calculating the virtual damping coefficient of the mesh nodes through multiphysics mapping, taking into account the degree of stiffness degradation, specifically includes: The stiffness degradation rate parameter is obtained by calculating the ratio of the final elastic modulus to the initial elastic modulus of the element to which the mesh node belongs. Construct a damping function that includes a stiffness sensitivity term and a stress safety barrier term; The damping function is used to map the stiffness degradation rate parameter and the equivalent membrane stress into a virtual damping coefficient with a value between zero and one. The value of the virtual damping coefficient increases when the stiffness degradation rate parameter increases or the equivalent membrane stress approaches the material yield limit.

6. The simulation test method for the bottom protection plate of a new energy vehicle battery according to claim 5, characterized in that, The stress safety barrier term uses a hyperbolic tangent nonlinear function to map the normalized equivalent membrane stress, so that when the equivalent membrane stress approaches the yield limit, the virtual damping coefficient shows a sharp upward trend and approaches the value of one, thereby freezing the geometric compensation operation of the region at the algorithm level.

7. The simulation test method for the bottom protection plate of a new energy vehicle battery according to claim 1, characterized in that, The steps for calculating the mold compensation mesh by using the virtual damping coefficient to perform reverse weighted correction of the geometric deviation vector and virtual strain energy smoothing include: The geometric deviation vector and its inverse vector are calculated by comparing the node geometric coordinate data with the design model. The difference between the virtual damping coefficient and the value of the difference is used as a weighting coefficient to scale the magnitude of the reverse vector, and the initial compensation amount is obtained by calculation. In particular, for high-risk area nodes with large virtual damping coefficients, the magnitude of the initial compensation is reduced relative to the magnitude of the geometric deviation vector.

8. The simulation test method for the bottom protection plate of a new energy vehicle battery according to claim 7, characterized in that, The specific steps of applying virtual strain energy smoothing constraints to the initial compensation amount include: The initial compensation amount is considered as a virtual displacement applied to the mesh node, thereby calculating the local additional virtual strain energy density introduced between the node and its neighboring nodes. Nodes whose local additional virtual strain energy density exceeds a preset threshold are detected, and Laplace smoothing is performed on the nodes and their neighboring nodes. The final node compensation vector is calculated by distributing the deformation compensation of the node to the neighboring nodes. The final node compensation vector is superimposed on the original mold mesh data to reconstruct the mold surface and thus form the mold compensation mesh.

9. The simulation test method for the bottom protection plate of a new energy vehicle battery according to claim 1, characterized in that, The criteria for meeting both geometric tolerances and molding safety include determining the following two conditions: Geometric dimensional tolerance condition: The maximum normal distance between the node of the formed workpiece and the design digital model is less than the preset tolerance value; Molding safety conditions: The equivalent membrane stress value of all units is less than the preset safety threshold, and there is no local stress concentration caused by geometric compensation that exceeds the material fracture limit.

10. A simulation test method for a new energy vehicle battery bottom protection plate according to claim 9, characterized in that, The following logic is executed during the determination process: If the forming safety condition is met but the geometric tolerance condition is not met, then the next round of compensation iteration is performed using the current residual geometric deviation; If the molding safety conditions are not met, then regardless of whether the geometric tolerance conditions are met, the virtual damping coefficient of the corresponding region will be forcibly increased in the next iteration, and molding safety will be guaranteed by reverting back a portion of the geometric compensation amount.

Images