A new energy vehicle battery pack multi-material multi-objective intelligent optimization design method

Abstract

This invention provides a multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs, relating to the field of new energy vehicle battery pack structural design and intelligent optimization technology. The method first establishes a parametric initial finite element model of the battery pack and extracts target responses such as total mass, first-order natural frequency, maximum equivalent stress under vertical bump conditions, and maximum deformation under Y-axis extrusion conditions under static, modal, and extrusion conditions. Then, a candidate material database is constructed and a constitutive model is established, and sensitivity analysis is performed on battery pack components to screen out key components. Next, range analysis is performed on the key components to obtain the optimal material combination. Finally, an intelligent algorithm is used for multi-objective optimization to obtain the optimal Pareto solution set. This invention can uniformly optimize lightweighting, modal vibration isolation, static strength, and extrusion safety, while also considering the connection of dissimilar materials and the feasibility of engineering manufacturing. It is applicable to the structural design of new energy vehicle battery pack housings and their key connecting components.

Description

Technical Field

[0001] This invention relates to the field of new energy vehicle battery pack structure design and intelligent optimization technology, and in particular to a multi-objective intelligent optimization design method that integrates multi-material matching, joint optimization of structural parameters, and manufacturing verification for battery pack housing and its key connecting components. Background Technology

[0002] There is a significant coupling relationship between the structural weight, stiffness, collision safety, modal characteristics, and manufacturing feasibility of new energy vehicle battery packs. Existing battery pack designs typically use a single material or empirically based thickening designs to meet strength requirements, resulting in significant mass redundancy. Another approach, while employing finite element analysis or single-objective optimization, often only makes local adjustments to the thickness, failing to uniformly model and jointly decide on candidate material systems, component functional differences, modal vibration damping, static conditions, and extrusion safety constraints. This makes it difficult for the optimization results to simultaneously achieve lightweighting, safety, and engineering feasibility.

[0003] Especially for key components such as the upper cover, lower body, connecting brackets, and lifting lugs, their stress mechanisms and process requirements differ: the upper cover and local connection areas are sensitive to external stiffness and first-order modes, the lower body is sensitive to overall load-bearing capacity and compressive strength, and the lifting lugs must also meet the local high-strength requirements of handling and assembly conditions. If a uniform material and uniform thickness strategy is adopted without distinguishing the functions of the components, problems such as excessive local stress, low modal frequencies, or waste of material properties can easily occur.

[0004] Therefore, there is an urgent need to propose a systematic approach that links "component-level material screening, size-level multi-objective optimization, manufacturing finishing and backtesting" to achieve a balanced optimization of the battery pack structure among multiple objectives such as mass, stiffness, strength, modality, and compression protection. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a multi-material, multi-objective intelligent optimization design method for battery packs of new energy vehicles, so as to solve the problems of extensive material selection, single optimization objective, difficulty in balancing lightweight and safety, and insufficient engineering applicability of optimization results in the existing battery pack structure design.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs includes the following steps: S1. Establish the initial parametric model of the battery pack: Taking the battery pack housing as the object, establish a three-dimensional parametric geometric model including the upper cover, lower housing, module bracket, module partition, connecting bracket, fixing strip, heat insulation board, and external lifting lugs. Based on load-bearing function, the components are divided into shell element optimized components, solid support components, and non-load-bearing equivalent mass components. BMS, cables, and local fastening accessories can be treated as equivalent mass or ignored. Cell modules are replaced by mass elements or mass-inertia equivalent modules. Subsequently, mesh generation is performed according to the component manufacturing method: shell elements are preferred for sheet metal parts, solid elements are preferred for die-cast or locally reinforced areas, and local densification is applied to hole edges and connection transition areas to obtain an initial finite element model suitable for static, modal, and extrusion analysis.

[0007] S2. Construct a multi-condition performance evaluation system: Apply typical operating conditions and safety conditions to the initial finite element model obtained in S1 to obtain the target response set of the battery pack structure. The typical operating conditions include at least vertical bump conditions, cornering braking conditions, emergency braking conditions, and reversing braking conditions; the safety conditions include at least X-direction compression conditions along the length of the battery pack and Y-direction compression conditions along the width of the battery pack. The extracted target responses include at least the total mass of the battery pack, the first-order natural frequency, the maximum equivalent stress under the vertical bump conditions, and the maximum deformation of the housing under the Y-direction compression conditions; when necessary, the maximum displacement, connection zone stress, constraint reaction force, and module safety clearance can also be recorded simultaneously.

[0008] S3. Establish a candidate material database and mechanical constitutive model: For key components of the battery pack, construct a candidate material database consisting of basic steel grades and high-performance alternative materials. Candidate materials should include at least three of the following: DC01 low-carbon steel, DP800 grade dual-phase steel, DP1000 grade dual-phase steel, and 6061-T6 aluminum alloy. Quasi-static and low-to-medium strain rate tensile tests will be conducted on the candidate materials to obtain engineering stress-strain curves, which will then be converted to obtain the actual stress-strain relationship. Based on the test results, a material constitutive model matching the extrusion analysis will be established to support subsequent multi-material simulation and selection.

[0009] S4. Perform component sensitivity analysis and discrete material pre-screening: Use the material properties or thickness parameters of each component of the battery pack as perturbation variables, calculate the direct sensitivity or normalized sensitivity of each variable to the target response set, and screen out the key components that have a significant impact on the total mass, first-order natural frequency, vertical bump stress and Y-direction extrusion deformation; then carry out multi-factor multi-level orthogonal or combined experiments for the key components, discretely map different material allocation schemes to the key components, obtain the correspondence between the material combination scheme and the target response, and select the component-level multi-material combination that meets the safety constraints and has the advantage of lightweighting.

[0010] S5. Establish manufacturing constraints and dissimilar material compatibility constraints: Based on the multi-material combination obtained in S4, further introduce manufacturing and assembly feasibility constraints. These constraints include: upper and lower limits for thickness, minimum forming thickness constraint, thickness rounding constraint, dissimilar material connection constraint, heat-affected zone safety factor constraint, local reinforcement constraint for connection area, and anti-corrosion isolation constraint. Among them, the dissimilar material connection constraint is used to limit the optional connection forms between steel and aluminum, and steel and high-strength steel to at least one of the following: mechanical connection, bolt connection, flow drill screw connection, riveting connection, or composite connection with insulating isolation layer, in order to avoid galvanic corrosion and connection failure.

[0011] S6. Construct a surrogate model for continuous size variables: After determining the material types of key components, the thickness of key components is taken as a continuous design variable. The Latin hypercube sampling method is used to generate sample points within the value range of each design variable. The finite element solver is called to perform batch simulations on each set of sample points to obtain the target response dataset corresponding to each sample point. Based on the dataset, a response surface surrogate model from design variables to target response is established, and the accuracy of the surrogate model is checked using the coefficient of determination and root mean square error. The surrogate model is only used for subsequent optimization search when it meets the preset accuracy threshold.

[0012] S7. Multi-objective joint optimization is performed using the PSO-GA hybrid algorithm: The optimization objectives are to minimize the total mass of the battery pack, maximize the first-order natural frequency, minimize the equivalent stress under vertical turbulence conditions, and minimize the maximum deformation under Y-direction extrusion. The manufacturing constraints, connection constraints, and safety threshold constraints in S5 are used as boundary conditions. A hybrid intelligent algorithm formed by coupling the particle swarm algorithm and the genetic algorithm is used to iteratively optimize the thickness of key components to obtain the Pareto front solution set that satisfies the constraints.

[0013] S8. Comprehensive decision-making on the Pareto front solution set: For the Pareto front solution set obtained in S7, game theory is introduced to improve the TOPSIS decision-making method. Subjective preference weights and objective data weights are integrated through game theory. The relative closeness of each candidate solution to the ideal solution and the negative ideal solution is calculated, thereby selecting the compromise optimization scheme with the best comprehensive performance.

[0014] S9. Perform manufacturing rounding and full-condition verification: Round the thickness and correct the specifications of the compromise optimization scheme obtained in S8 according to the engineering manufacturing requirements to form the final manufacturable design parameters; recalculate the rounded battery pack model using the same static, modal and extrusion conditions as in S2, and verify its total mass, first natural frequency, maximum equivalent stress under vertical bump conditions, maximum deformation under Y-direction extrusion and module safety clearance. Output the final design result when all indicators meet the preset targets, otherwise return to S6 or S7 for re-optimization.

[0015] Preferably, the key components selected in S4 are the upper cover, the lower body, the external lifting lugs, and the connecting bracket.

[0016] Preferably, the upper cover is made of 6061-T6 aluminum alloy, the lower body is made of DP800 grade duplex steel, the external lifting lugs are made of DP1000 grade duplex steel, the connecting bracket is made of 6061-T6 aluminum alloy, and the remaining non-critical components retain the basic steel type or existing materials.

[0017] Preferably, in S6 and S7, the thickness of the upper cover, the thickness of the lower body, the thickness of the external lifting lugs, and the thickness of the connecting bracket are denoted as X1, X2, X3, and X4, respectively, and their value ranges are [1, 4] mm, [1, 4] mm, [1, 5] mm, and [1, 4] mm, respectively.

[0018] Compared with the prior art, the present invention has at least the following beneficial effects: Firstly, this invention does not separate material selection and size optimization, but rather constructs an integrated process of multi-material screening at the component level and multi-objective joint optimization at the size level, which can significantly reduce the repetitive design costs caused by "selecting materials first and then trying and failing".

[0019] Secondly, this invention incorporates mass, modal stress, static stress, and extrusion deformation into the target response system simultaneously, which can more realistically reflect the multi-constraint balance relationship in battery pack engineering applications.

[0020] Third, this invention introduces manufacturing constraints, dissimilar material connection constraints, and rounding back verification steps, enabling the optimization results to be directly applied to engineering projects.

[0021] Fourth, by combining the proxy model with the PSO-GA hybrid algorithm, this invention can improve optimization efficiency and solution quality while keeping computational costs under control. Attached Figure Description

[0022] The accompanying drawings, which are incorporated herein and form part of the specification, illustrate embodiments of the invention and, together with the specification, further serve to explain the principles of the invention and enable those skilled in the art to practice and use the invention.

[0023] Figure 1 This is a schematic diagram of the overall process of the method of the present invention; Figure 2 This is a parameterized schematic diagram of key components of the battery pack of the present invention; Figure 3 This is a schematic diagram of the multi-material pre-screening and sensitivity analysis of the present invention; Figure 4 This is a schematic diagram illustrating the proxy model, multi-objective optimization, and decision-making process of the present invention; Figure 5 This is a schematic diagram comparing the main performance indicators of the present invention before and after optimization. Detailed Implementation

[0024] The present invention will be further described below with reference to the accompanying drawings and embodiments, but the scope of protection of the present invention is not limited to the following embodiments. Without departing from the concept of the present invention, those skilled in the art can make equivalent substitutions or conventional adjustments to the candidate material set, the number of target responses, the constraint threshold, and the optimizer parameters.

[0025] Example 1: Intelligent Optimization Process for Multi-material and Multi-objective Battery Packs Reference Appendix Figure 1 and attached Figure 2 First, a parametric initial model of the new energy vehicle battery pack structure is established. The initial model includes an upper cover, lower housing, module bracket, module partition, fixing strip, heat insulation plate, connecting bracket, and external lifting lugs. The model is slightly simplified: electrical accessories that do not bear the main load are ignored, and the cell modules are equivalent in terms of mass and inertia; the upper cover, lower housing, and connecting bracket are described using shell elements, while local pressure-bearing areas or die-cast support areas are described using solid elements, and the hole edges and connection transition areas are densified using washer elements. Preferably, the reference mesh size for the upper cover, lower housing, and heat insulation plate is 10mm, and the reference mesh size for the remaining components is 5mm.

[0026] Subsequently, a multi-condition performance evaluation system was established. The static condition includes at least vertical bumps, cornering braking, emergency braking, and reversing braking; the extrusion condition includes at least X-axis extrusion and Y-axis extrusion, where the extrusion head is preferably a semi-cylindrical rigid body with a radius of not less than 75 mm, and the extrusion speed is not greater than 2 mm / s. Based on the simulation results, the total mass M of the battery pack, the first-order natural frequency f1, the maximum equivalent stress σb under the vertical bump condition, and the maximum deformation δy under the Y-axis extrusion condition were extracted. Here, M represents the total mass of the battery pack structure; f1 represents the first-order natural frequency of the battery pack; σb represents the maximum equivalent stress under the vertical bump condition; and δy represents the maximum displacement or deformation of the battery pack casing along the compression direction under the Y-axis extrusion condition.

[0027] Example 2: Establishment of Material Constitutive Model Reference Appendix Figure 3 Quasi-static and low-to-medium strain rate tensile tests were conducted on candidate materials DC01, DP800 grade duplex steel, DP1000 grade duplex steel, and 6061-T6 aluminum alloy. The obtained engineering stress-strain curves were then converted to real stress-strain curves. To adapt to extrusion safety simulation, a simplified form of the Johnson-Cook constitutive model was preferred. Where A is the material's yield strength, B is the hardening modulus, n is the material's hardening exponent, and C is the material's strain rate coefficient. For equivalent plastic strain, For equivalent stress, For equivalent plastic strain rate, For reference strain rate, The melting temperature. The reference temperature is used for the experiment. This study does not consider the influence of temperature on the mechanical properties of the material; therefore, the coefficient of the temperature term is 0. Using the stress-strain curves of the material under quasi-static conditions, numerical fitting methods are employed to determine the values ​​of A, B, and n. Combined with the stress-strain curves under medium strain rates, the value of C is determined.

[0028] In this embodiment, the material constitutive parameters can be obtained by fitting the measured data of the original batch of the original plate material. If the material supply batch is changed, the parameters can be reconstructed according to the same test procedure. Therefore, this invention does not depend on a fixed set of batch parameters and has good implementability and transferability.

[0029] Example 3: Screening of Key Components and Pre-matching of Materials Reference Appendix Figure 3 Using each component of the battery pack as candidate variables, the sensitivity of each component's material properties or thickness to four types of responses—mass, frequency, stress, and deformation—is determined. A normalized sensitivity index is preferred. Among them, S ij x represents the normalized sensitivity of the i-th design variable to the j-th response; i Represents the baseline value of the i-th design variable; Δx i y represents the disturbance of the i-th design variable; j Represents the baseline value of the j-th response; Δy j This indicates the change in response before and after the disturbance. Based on the sensitivity results, key components that significantly affect multiple targets are retained. In this embodiment, the key components selected are the upper cover A, the lower housing B, the external lifting lug C, and the connecting bracket D.

[0030] Based on this, a three-level orthogonal experiment was conducted on the four key components. For each component, one of three candidate materials was selected: DP800 grade dual-phase steel, DP1000 grade dual-phase steel, and 6061-T6 aluminum alloy. These materials were then combined with the base material DC01 to form a hybrid material battery pack. Each experimental scheme was substituted into a finite element model to calculate its M, f1, σb, and δy. The optimal material combination that satisfies the strength and stiffness constraints was selected through range analysis or variance analysis. The preferred combination in this embodiment is: the upper cover is made of 6061-T6 aluminum alloy, the lower casing is made of DP800 grade dual-phase steel, the external lifting lugs are made of DP1000 grade dual-phase steel, the connecting bracket is made of 6061-T6 aluminum alloy, and the remaining components are made of DC01 or the original base material.

[0031] Example 4: Proxy Model Construction and Accuracy Determination After determining the material combination, the thicknesses of the upper cover, lower body, external lugs, and connecting brackets are set as continuous variables X1, X2, X3, and X4, respectively, with preferred ranges of [1, 4] mm, [1, 4] mm, [1, 5] mm, and [1, 4] mm. Latin hypercube sampling is used to generate sample points, with a preferred number of 80–150 sets, more preferably 100 sets. For each set of sample points, static, modal, and extrusion finite element solvers are called to establish a sample database, and the response surface method is used to construct an approximate mapping between design variables and the target response.

[0032] To evaluate the accuracy of the surrogate model, the coefficient of determination R² and the root mean square error RMS are introduced: Where n is the number of samples. For the predicted results, For actual results, The mean of the actual results is denoted as R². The closer R² is to 1 and the closer RSME is to 0, the higher the fitting accuracy of the surrogate model. Typically, R² is required to be no less than 0.9 and RSME to be no more than 0.2.

[0033] Preferably, when the R² of each response is not less than 0.9 and the RMSE is not higher than a preset threshold, the surrogate model is considered to meet the requirements for optimized use.

[0034] Example 5: Multi-objective optimization, integrated decision-making and backtesting Reference Appendix Figure 4 Using the thickness vector X=[X1, X2, X3, X4] of key components as optimization variables, a multi-objective optimization model is established: Where F(x) represents the set of multi-objective functions; M(x) represents the total mass of the battery pack determined by the design variable x; σb(x) represents the maximum equivalent stress under vertical bump conditions; δy(x) represents the maximum deformation under Y-axis compression; and f1(x) represents the first-order natural frequency. The frequency objective is negative in order to unify it into a minimization form.

[0035] The constraints include at least: upper and lower limits for thickness, maximum equivalent stress not exceeding the allowable value of the corresponding material, first-order natural frequency higher than the upper limit of the main excitation frequency band of the road surface, a safe gap between the box and the internal module after Y-axis extrusion, the requirement that the connection area of ​​dissimilar materials meets the connection form and corrosion protection requirements, and the manufacturability constraint of the rounded standard plate thickness. Preferably, in the PSO-GA hybrid algorithm, the population size is 100-200, the maximum number of iterations is 300-600, the inertia weight decreases linearly from 0.9 to 0.4, the crossover probability is 0.75-0.90, and the mutation probability is 0.005-0.03.

[0036] After obtaining the Pareto front solution set, game theory is used to improve TOPSIS for comprehensive decision-making. Its core idea is to integrate subjective and objective weighting through game theory to obtain equilibrium weights and calculate relative proximity, thereby determining the compromise solution closest to the ideal solution. Preferably, after determining the compromise solution, the thickness parameter is rounded, for example, from 1.871mm to 1.87mm or the corresponding standard plate thickness, and the static, modal, and extrusion analyses, consistent with the previous steps, are re-executed to verify that the rounded solution still meets the design requirements.

[0037] In a preferred embodiment, the optimized upper cover thickness is 1.87 mm, the lower casing thickness is 1.56 mm, the external lifting lug thickness is 3.49 mm, and the connecting bracket thickness is 2.71 mm. Retest results show that the total battery pack mass is reduced by approximately 4% compared to the initial design, the first-order natural frequency is significantly increased, the maximum equivalent stress under vertical turbulence conditions decreases to within the material's allowable range, and the maximum deformation of the casing under Y-axis compression conditions is less than the safety clearance between the shell and the module, thus achieving a balance between lightweight design and structural safety.

Claims

1. A multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs, characterized in that, Includes the following steps: Establish a parametric initial finite element model of the battery pack, including the upper cover, lower body, connecting bracket, external lifting lugs, and internal support components; Static, modal, and extrusion conditions are applied to the initial finite element model, and the total mass of the battery pack, the first natural frequency, the maximum equivalent stress under vertical turbulence condition, and the maximum deformation under Y-direction extrusion condition are extracted as target responses. A candidate material database consisting of basic steel grades and high-performance alternative materials was constructed, and at least one material constitutive model suitable for extrusion analysis was established based on the material test results. Sensitivity analysis was performed on each component of the battery pack to identify key components that significantly affect the target response. Multi-material orthogonal experiments or combined experiments were conducted on the key components to determine the material allocation scheme for the key components. After determining the material allocation scheme, the thickness of key components is used as a continuous design variable. A proxy model between the design variables and the target response is established by using Latin hypercube sampling and finite element batch calculation. Based on the surrogate model and combined with manufacturing constraints, connectivity constraints and safety threshold constraints, the PSO-GA hybrid algorithm is used to perform multi-objective optimization on the thickness of key components to obtain the Pareto front solution set. The TOPSIS method is improved using game theory to select the compromise optimal solution from the Pareto front solution set; The optimal compromise solution was rounded to the thickness and tested under all operating conditions to obtain the final optimized design result of the battery pack.

2. The multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs according to claim 1, characterized in that, The key components include the upper cover, the lower body, external lifting lugs, and the connecting bracket.

3. The multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs according to claim 1, characterized in that, The static operating conditions include vertical bumps, turning braking, emergency braking, and reverse braking; the dynamic operating conditions include the first six modal frequencies; the extrusion operating conditions include X-axis extrusion and Y-axis extrusion. In addition to the total mass of the battery pack, the first-order natural frequency, the maximum equivalent stress under vertical turbulence conditions, and the maximum deformation under Y-direction compression conditions, the target response also includes at least one of the following: maximum displacement, stress in the connection area, module safety clearance, and constraint reaction force.

4. The multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs according to claim 1, characterized in that, The candidate material database includes at least three of the following: DCO1 low-carbon steel, DP800 duplex steel, DP1000 duplex steel, and 6061-T6 aluminum alloy. The material constitutive model is the Johnson-Cook constitutive model or its equivalent constitutive model, established based on the results of quasi-static and medium-low strain rate tensile tests.

5. The multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs according to claim 1, characterized in that, The sensitivity analysis employs at least one of the following: direct sensitivity analysis, normalized sensitivity analysis, or local perturbation analysis. The multi-material orthogonal test or combination test is used to obtain the total mass, first natural frequency, maximum equivalent stress and maximum deformation corresponding to different material allocation schemes, and to determine the material combination of key components based on range analysis, mean analysis or variance analysis.

6. The multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs according to claim 1, characterized in that, The manufacturing constraints include at least the upper and lower limits of thickness, the minimum forming thickness constraint, the thickness rounding constraint, the dissimilar material connection constraint, and the corrosion protection and isolation constraint. When key components are made of a mixture of steel and aluminum alloy, the connection constraint between the dissimilar materials is limited to at least one of the following: bolt connection, riveting connection, flow drill screw connection, mechanical locking connection, or composite connection with an insulating layer.

7. The multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs according to claim 1, characterized in that, The proxy model adopts at least one of the response surface model, the Kriging model, and the radial basis function model. The surrogate model must meet the accuracy requirements of having a determination coefficient of not less than 0.90 and a root mean square error of not more than a preset threshold before it can be used for optimization search in the PSO-GA hybrid algorithm.

8. The multi-material, multi-objective intelligent optimization design method for new energy vehicle battery packs according to claim 1, characterized in that, After the PSO-GA hybrid algorithm outputs the Pareto front solution set, the game theory-improved TOPSIS method is used to calculate the relative proximity of each candidate solution and select the compromise optimal solution. The compromise solution, after thickness rounding, is retested under static, modal, and extrusion conditions. Only when the total mass, first-order natural frequency, maximum equivalent stress under vertical turbulence, maximum deformation under Y-axis extrusion, and module safety clearance all meet the preset requirements will it be output as the final design result.

Images