Evolutionary Constrained Optimization: A Hybrid Approach

摘要: The holy grail of constrained optimization is the development an efficient, scale invariant, and generic constraint-handling procedure in single- multi-objective problems. Constrained a computationally difficult task, particularly if constraint functions are nonlinear nonconvex. As classical approach, penalty function approach popular methodology that degrades objective value by adding proportional to violation. However, has been criticized for its sensitivity associated parameters. Since inception, evolutionary algorithms (EAs) have modified various ways solve Of them, recent use bi-objective algorithm which minimization violation included as additional objective, received significant attention. In this chapter, we propose combination with manner complementary each other. provides appropriate estimate parameter, while solution unconstrained penalized method induces convergence property overall hybrid algorithm. We demonstrate working on number standard numerical test most cases, our proposed observed take one or more orders magnitude lesser evaluations find minimum accurately than some best-reported existing methodologies.