Multi-Objective Optimization of Truss Structures Using NSGA-II and SHAMODE Algorithms

Abstract

This study investigates the performance of Success-History Adaptive Multi-Objective Differential Evolution (SHAMODE) and Non-dominated Sorting Genetic Algorithm II (NSGA-II), methods in solving a large-scale, multi-objective truss optimization problem. The objective is to minimize the structural weight while maximizing displacement performance, subject to stress and displacement constraints. Four widely used performance metrics including Hypervolume, Generational Distance (GD), Inverted Generational Distance (IGD), and Spacing-to-Extent (STE) are employed to evaluate the quality and distribution of the Pareto fronts obtained. Results from independent runs show that SHAMODE consistently produces superior Pareto fronts, as evidenced by higher HV values and significantly lower GD and IGD scores compared to NSGA-II. Furthermore, SHAMODE achieves a more uniform distribution of solutions, indicated by its lower STE values. These findings demonstrate SHAMODE's effectiveness and robustness in handling complex structural optimization problems