| Summary: | 碩士 === 國立高雄第一科技大學 === 系統資訊與控制研究所 === 96 === In this study, we proposed two kinds of improved particle swarm optimization (PSO) Algorithms, named improved particle swarm optimization (IPSO) and sliding levels particle swarm optimization (SLPSO), respectively. The algorithms are applied to solve the benchmark single-multi function of problems, which make experiments on the characteristics of variant PSO and the effect of the differences between the proposed IPSO and SLPSO algorithms. We also utilize Taguchi method, which has the excellently experienced ability of inference and the analysis of variance to achieve the performance of fast convergence and searching the optimal solutions in the large searching solution space.
In order to obtain the convergence and stability, however, we employ the sliding levels of orthogonal array to reduce the standard derivation caused by the interaction of PSO coefficients. Moreover, the proposed algorithm is generalized and is solved for multi- parameter.
We employ five kinds of often-used benchmark functions to illustrate that the IPSO and SLPSO have the better abilities of searching optimal solutions and searching speed than the typical PSO.
|
|---|
