Çok doruklu optimizasyon için particle swarm sistemleri

Many real world problems in science, engineering and economy, involve inoptimization. Multimodal optimization problems represent a subset of such problems,where the goal is to obtain multiple different solutions with equal (or preferable) quality ina single search space. Meta-heuristics, especially, evolutionary computation techniques arethe most commonly used approaches for solving difficult optimization problems. However,due to the multimodality, they might require some enhancements. Particle SwarmOptimizer (PSO), as a swarm intelligence technique, has proven its success in solvingoptimization problems. Additionally, several modified versions of the PSO algorithm makeit a good choice for attacking multimodal problems as well. In this thesis, the performanceof existing multimodal PSO techniques are analyzed over a set of well-known benchmarkfunctions. Moreover, a new PSO algorithm based on craziness and hill-climbing isproposed for solving multimodal optimization problems.