Convergence Analysis of PSO with Nonlinear Time-Varying Evolution using Uniform Design

• Main Authors: Chin-Hsien Yang, 楊京憲
• Other Authors: Jyh-Horng Chou
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/16749166577117305768

碩士 === 國立高雄第一科技大學 === 電機工程研究所碩士班 === 103 === In this paper, the convergence analysis of Particle swarm optimization (PSO) is studied. For optimization problems, how to get a set of solutions with good convergence and diversity is a difficult and challenging work. Uniform design can be used to construct a set of experimental points which are scattered uniformly in a continuous design space. Uniform design has been widely used in the optimization processes for many engineering applications. PSO with nonlinear time-varying evolution (PSO-NTVE) is used to adjust inertia weight and acceleration coefficients. The approach depends on the choice of tuning factors α, β and γ. The uniform design can provide an efficient way to obtain the optimal parameters of optimization problems. The advantage of the uniform design is that the required experiment times is considerably less than other known experimental-design techniques when the number of factor levels is large. In this paper, by using the uniform design to optimize the PSO-NTVE method, the major objective is to achieve fast speed of convergence and better solution accuracy with minimal computational burden. To demonstrate the performance of the proposed uniform design to optimize the PSO-NTVE method, five well-known benchmarks are used for illustration. First, each factor of α, β and γ are separate into 11 levels, then the uniform design is used to gather the experimental data. Finally, regression analysis is used to study the influence of formulation parameters. The results show that the average values and the standard deviations of the PSO-NTVE method with uniform design is better than the PSO-NTVE with Taguchi Method.