Optimizing Vehicle Joints through Adaptive Stacking Model and Discrete Marine Predator Optimization Algorithm

Abstract

The design of vehicle body joints is a critical aspect of the conceptual design process. Joint structures significantly affect the mechanical performance of vehicle bodies. However, due to the nonlinear relationship between joints and body performance, it is challenging to develop an explicit expression for optimization. Furthermore, traditional finite element analysis is impractical due to the vast number of possible joint configurations. Therefore, we propose a surrogate model-based optimization method to address this problem. First, we propose an intelligent adaptive stacking method (IASM) to establish the surrogate model. We evaluate the performance of IASM and other competitors on 34 benchmark functions and 3 open engineering projects, and IASM demonstrates the best predictive performance overall. Next, we construct joint modules with different configurations as candidate modules, which we connect to the vehicle body using beam units to build the simplified vehicle body (JMBB). JMBB significantly reduces the computational cost of finite element simulation, generating training samples for IASM. We then propose a discrete marine predator algorithm (DAMPA) to optimize the joints based on IASM. Compared to the genetic algorithm, DAMPA identifies joint modules with better mechanical performances. To validate the effectiveness of our method, we modify the base vehicle body using the optimized joints, resulting in a 7.4 kg reduction in body mass while enhancing four other mechanical metrics.