Projeto de sistemas de engenharia usando o algoritmo de partículas vibrantes

Abstract

Systems design is one of the most important pillars in engineering. This is due to number of applications that can be developed from this principle. Designing means reducing costs, saving time and resources to guarantee of quality, efficiency and safety of products. To assist decisionmaking, population-based optimization tools have been employed. Solving a design problem means treating a problem with more than one objective, most of which are conflicting, that is, an improvement in any one of these objectives doesn’t result, necessarily, in an improvement of the others. The optimal solution for these problems, unlike the optimization with an single objective, is the attainment of non-dominated solutions that form the Pareto Curve, also know as Pareto Optimal. In recent decades, numerous population-based algorithms have been proposed for the treatment of optimization problems. Among these, the Vibrating Particles System algorithm configures a promising strategy to solve mono-objective problems. This work aims to: i) evaluate the Vibrating Particles System algorithm as a tool to solve design problems in mono-objective context; and ii) extend the Vibrating Particles System algorithm to multi-objective context by incorporating three operators to original algorithm: Pareto dominance, crowd distance and a mutation strategy. To expedite the convergence process and avoid local optima, an external repository is considered. The performance of the proposed methodology is evaluated considering mathematical and design problems in engineering. The obtained results demonstrate that the Vibrating Particles System algorithm (in single and multi-objective contexts) configures an interesting strategy for the treatment of optimization and design problems when compared to other evolutionary strategies in terms of statistical parameters, performance metrics and graphical analysis.