Comparative Review of Multi-Objective Optimization Algorithms for Design and Safety Optimization in Electric Vehicles

Despite the widespread use of established optimization algorithms like Non-Dominated Sorting Genetic Algorithm-II (NSGA-II), Non-Dominated Sorting Genetic Algorithm-III (NSGA-III), and Multi-Objective Evolutionary Algorithm based on Decomposition (MOEA/D) in real-world engineering optimization problems, newer algorithms such as Two-Stage NSGA-II (TS-NSGA-II), Dynamic Constrained NSGA-III (DCNSGA-III), MOEA/D with Virtual Objective Vectors (MOEA/D-VOV), Large-Scale Evolutionary Multi-Objective Optimization Assisted by Directed Sampling (LMOEA-DS), and Super-Large-Scale Multi-Objective Evolutionary Algorithm (SLMEA) remain underexplored in the context of Battery Electric Vehicle (BEV) safety, particularly in optimizing complex, non-linear, and constrained multi-objective problems like crashworthiness and thermal management. This study addresses this gap by comparing these newer algorithms against traditional methods using a newly introduced benchmark problem focused on BEV battery protection (RWMOP-BEV). The design problem aimed to maximize energy absorption during impact, enhance crash force efficiency, and optimize temperature difference, all while adhering to design space and operational constraints. The comparative results, based on four performance indicators—hypervolume (HV), inverted generational distance (IGD), averaged Hausdorff distance $\left ({{ \Delta _{p} }}\right)$ , and spread—reveal that SLMEA emerged as the best algorithm, not only for RWMOP-BEV but also across other benchmark sets, including DTLZ problems and other real-world multi-objective optimization problems.