Multi-Objective Optimization Design for a Battery Pack of Electric Vehicle with Surrogate Models


In this investigation, a systematic surrogate-based optimization design framework for a battery pack is presented. An air-cooling battery pack equipped on electric vehicles is first designed. Finite element analysis (FEA) results of the baseline design show that global maximum stresses under X-axis and Y-axis transient acceleration shock condition are both above the tensile limit of material. Selecting the panel and beam thickness of battery pack as design variables, with global maximum stress constraints in shock cases, a multi-objective optimization problem is implemented using metamodel technique and multi-objective particle-swarm-optimization (MOPSO) algorithm to simultaneously minimize the total mass and maximize the restrained basic frequency. It is found that 2nd order polynomial response surface (PRS), 3rd order PRS and radial basis function (RBF) are the most accurate and appropriate metamodels for restrained basic frequency, global maximum stresses under x-axis and y-axis shock conditions respectively. Results demonstrate that all the optimal solutions in Pareto Frontier have heavier weight and lower frequency compared with baseline design due to the restriction of global maximum stress response. Finally, two optimal schemes, "Knee Point" and "lightest weight", satisfied both of the stress constraint conditions, show great consistency with FEA results and can be selected as alternative improved schemes.