Particle Swarm Optimization Algorithms Based On the Behavior Analyses

• Main Authors: Z. L. Huang, 黃子倫
• Other Authors: C. W. Tao
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/54876858152420247148

碩士 === 國立宜蘭大學 === 電機工程學系碩士班 === 97 === In this thesis, a new behavior analyzed multimodal particle swarm optimization (BAMPSO) algorithm is proposed for not only unimodal problems but multi-modal problems. The main idea is to find the local minima by analyzing the variation of the fitness value when the particles are moving. Since almost all the local minima are found, the global minimum can be obviously obtained. That is, the BAMPSO can avoid converging to local solution and efficiently find the global solution. Moreover, the behavior analyzed adaptive particle swarm optimization (BAAPSO) algorithm based on the same idea to on-line search the global minimum is also provided. BAAPSO algorithm can on-line to adjust parameters and improve the accuracy on searching for multi-objection problem. Experiment results and comparisons with other PSO algorithms are included to indicate the effectiveness of the proposed BAMPSO and BAAPSO algorithms.