Particle Swarm Optimizer with Catfish Effect as Scout Strategy for Global Optimization Problem

• Main Authors: Sheng-Wei Tsai, 蔡昇偉
• Other Authors: Cheng-Hong Yang
• Format: Others
• Language: en_US
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/20089122740417662289

碩士 === 國立高雄應用科技大學 === 電子工程系 === 97 === Particle swarm optimization (PSO), a population-based stochastic optimization technique, was developed by Eberhart and Kennedy in 1995 via simulating the social behavior of organisms. The efficiency of PSO has been demonstrated by solving optimization problems in various areas, e.g. function optimization, fuzzy system control, parameter optimization, artificial neural network training, travel sales problems, pattern recognition, and optimizing power flow. With its superb performance in nonlinear function optimization, PSO has drawn the attention of many researchers. However, PSO exhibits poor local search capabilities and often leads to premature convergence, especially in complex multi-peak search problems. In order to overcome the premature convergence of PSO, this thesis proposes catfish particle swarm optimization (CatfishPSO), in which the catfish effect is applied as a scout strategy to improve the performance of the PSO algorithm. The proposed method was applied to two types of optimization problems, namely a numerical optimization problem (twenty-two benchmark functions with 2, 4 and 30 different dimensions) and a combinatorial optimization problem (ten data sets taken from the University of California, Irvine repository), respectively. The results obtained from the experiments and statistical analyses thereof indicate that the catfish strategy is capable of enhancing the performance of the PSO to a significant level.