ANN-based optimization of TPMS diamond sandwich structures for lightweight battery enclosure in electric vehicles

Electric conversion vehicles offer a promising solution for transforming internal combustion engine (ICE) platforms into electric mobility, where battery safety and structural efficiency are key challenges. This study introduces a novel application of triply periodic minimal surface (TPMS) Diamond-type cores in lightweight sandwich structures for vehicle battery pack enclosures. To overcome computational limitations in simulation-based optimization, an artificial neural network (ANN) surrogate modeling framework was developed to predict structural deformation and mass responses from finite element (FE) data. The ANN models were integrated with multi-objective optimization using the non-dominated sorting genetic algorithm II, with optimal designs selected via TOPSIS methodology. Key design variables included core and face sheet thicknesses and TPMS unit cell length, targeting simultaneous minimization of deformation and mass. Results demonstrate that reducing upper plate thickness significantly decreases mass, while increasing unit cell length and minimizing TPMS wall thickness enhances energy absorption within displacement constraints. The optimized TPMS sandwich structure achieves 43.6% weight reduction compared to conventional steel enclosures while maintaining crashworthiness under impact speeds up to 95 km/h. This work demonstrates the transformative potential of combining TPMS geometries with machine learning-based optimization for developing lightweight, energy-absorbing structures in electric vehicle (EV) applications.