Advancing truss structure optimization—A multi-objective weighted average algorithm with enhanced convergence and diversity

Abstract

• Presents MOWAA, a novel algorithm designed for enhanced convergence and diversity in truss structure optimization. • Demonstrates superior performance of MOWAA compared to NSGA-II, MOEA/D, MOLCA, MOEDO, and MORIME across various benchmark truss structures. • Balances exploration and exploitation through adaptive strategies, resulting in high-quality Pareto fronts with increased coverage of objective space. • Empirical results indicate MOWAA's capability to effectively handle complex, multi-objective truss problems, achieving better hypervolume, IGD, and spacing metrics. • Validated across eight diverse truss structures, MOWAA exhibits robustness in both simple and complex configurations, establishing it as a reliable tool for structural optimization tasks. The challenge of achieving equilibrium between exploration and exploitation stands as a critical barrier in multi-objective metaheuristic optimization when applied to complex engineering problems such as truss structure design. The Multi-Objective Weighted Average Algorithm (MOWAA) presents a new methodology which employs adaptive weighted average position control to optimize population movement for enhanced solution quality. The performance evaluation of MOWAA relies on benchmarking it against five state-of-the-art multi-objective optimization algorithms NSGA-II, MOEA/D, MOLCA, MOEDO and MORIME through eight truss structure optimization problems of increasing complexity. The evaluation of performance relies on three key metrics: Hypervolume (HV), Inverted Generational Distance (IGD) and Spacing (SP). MOWAA demonstrates superior performance compared to competing algorithms through its ability to generate Pareto fronts with higher HV values and lower IGD values and more uniform distribution. The enhanced performance of MOWAA demonstrates its superior capability to efficiently explore the objective space for finding optimal weight-minimization and compliance trade-offs. The robustness of MOWAA is proven through statistical validation with the Friedman rank test which establishes MOWAA as the leading approach with statistically significant advantages. MOWAA demonstrates runtime efficiency throughout truss optimization tasks of varying sizes which enables its practical application for real-world structural optimization problems. MOWAA emerges as a sophisticated and efficient optimization method which demonstrates strong capabilities for engineering applications and computational design.

Topics & Concepts