An elitist multi-objective particle swarm optimization algorithm for composite structures design

Abstract

Optimization is an important area of research in Engineering, usually due to the potentiality of saving costs and improving structural safety. Composite structures are typically complex, and the Finite Element Method is frequently required to evaluate such structures. From another perspective, Robust Design Optimization (RDO) is an approach that aims to consider the variability of the composite structures response due to uncertainty in design variables or material properties. Under these conditions, the problem of maximizing the robustness is added to the optimality problem related to minimizing the structure's weight. This work combines the advantages of Particle Swarm Optimization (PSO), such as simplicity and greater exploration and exploitation capabilities, with fitness assignment methodologies and elitist strategies commonly applied to Genetic Algorithms. The purpose is to achieve a more perceptible Pareto front and faster. The development is applied to the RDO bi-objective optimization problem in composite structures. Results for optimal design variables, critical displacements and stresses are discussed. The results show that elitist-based PSO approaches lead to a Pareto front with a larger number of optimal solutions, with more robust and lighter solutions when compared to other methodologies.