digilib@itb.ac.id +62 812 2508 8800

ABSTRAK Michael A. S. Biharta
PUBLIC Irwan Sofiyan

This research study involves designing and optimizing sandwich-based auxetic honeycomb structure to protect battery cells for the battery pack system of electric vehicles (EV) undergoing axial impact load using the machine learning method. This research explored the methods to help expedite the process of design and rapid prototyping that is becoming more important today. The pouch battery system was chosen due to its significant potential to be the next Li-Ion battery for EV with regard to its simplicity, higher energy density, and higher space efficiency compared to cylindrical or prismatic battery cells. However, just like the other Li-Ion battery, the pouch battery cells have a high fire risk during large battery deformation. The sandwich structure was designed to minimize the deformation of the battery pack undergoing axial impact loading. Both battery’s and sandwich structure’s performance were analyzed numerically by using the non-linear finite element method. The optimization was done using two machine learning algorithms: Artificial Neural Network (ANN) and Non-Dominated Sorting Genetic Algorithm Type II (NSGA-II). ANN predicted the specific energy absorption of the auxetic honeycomb and maximum battery stress during the deformation. The optimized design maximizes the auxetic honeycomb's specific energy absorption and limits the maximum battery stress during the deformation with the help of Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) methodology. The optimized design has a geometric shape of Double-U, a length of 6 ????????????????, a width of 4.2 ????????????????, cross section’s thickness of 0.6 ????????????????, and consists of 1 layer. The optimum design has SEA of 47,997.84 ???????? and can maintain the battery’s von Mises stress to a maximum of 43.16 ????????????????????????, well below the designated battery’s von Mises stress limit of 67.97 ????????????????????????.