Crash Safety Design for Lithium-ion Vehicle Battery Module with Machine Learning

Lithium-ion battery systems have been used as the main power source for electric vehicles due to their lightweight and high energy density. The impact safety of these battery systems has been a primary issue. In this work, the crashworthiness design of a typical vehicle battery module is implemented through numerical (finite element) simulations integrated with machine learning algorithms (decision trees). The module with multiple layered porous cells is modeled with a simplified, homogeneous material law, and subjects to the impact of a cylindrical indenter. The main protective component on the module - cover plate is designed as an energy absorbing sandwich structure with a core of cellular solids. Large scale simulations are conducted with various design variable values for the sandwich structure, and the results form a design (simulation) dataset. Based on the dataset, machine learning is applied to the sandwich cover plate design to: (1) correlate the design variables to the response; (2) investigate the complex inter-relationship between design variables; and (3) derive decision-making rules to achieve the designs with highest energy absorbing capability.