Robust PSO-Based Constrained Optimization by Perturbing the Particle's Memory
Angel Munoz, Arturo Hernández, Enrique Villa · 2007
In this chapter, we described a robust PSO for solving constrained optimization problems. We discussed the premature convergence problem, which still is an issue in evolutionary computation. A brief trip was made through several proposals to attain a balance between exploration and exploitation. Also, we briefly review recent works that contribute with interesting ideas for handling-constraints in PSO. This work presents an algorithm called PESO to handle constrained optimization problems. Based on the empirical and theoretical results of several works, we explain and validate every component applied in PESO. We empirically show the performance of PESO in a wellknow benchmark. PESO has shown high performance in constrained optimization problems of linear or nonlinear nature. Three important contributions of PESO are worth to mention: A new neighbourhood structure for PSO, the incorporation of perturbation operators without modifying the essence of the PSO, and a special handling technique for equality constraints.