在本論文中，我們提出了一種改良的粒子演算法(PSO)，名為自調整非線性慣性權重粒子群演算法(SNPSO)。SNPSO是一種針對慣性權重改良的方法，利用非線性和自我調整的特性來改善粒子最佳化方法易落入區域最佳解的缺點。非線性具有較好的能力避免粒子落入區域最佳解，而自我調整性則能增加粒子的靈活性，使粒子具有較大的能力往全域最佳解作搜尋。本文亦提出一種針對SNPSO參數最佳化的搜索策略，使得我們在選取參數時更具有策略性。最後，我們使用16個目標函數對SNPSO演算法進行模擬與測試，並且與幾個已提出的PSO演算作比較。經由模擬結果顯示，本文所提出的自調整非線性慣性權重粒子群演算法在目標函數中的表現，整體來說均有較優越的表現，同時也顯示本文所提出的方法能有效的改善PSO演算法的搜索效能並改善PSO演算法易落入區域最佳解的缺點。In this thesis we have presented an improved algorithm for Particle Swarm Optimization (PSO) named Self-adjusted Nonlinear inertia weight PSO algorithm (SNPSO). SNPSO algorithm is an improved method of the inertia weight, utilize nonlinear and self-modulation characteristics to improve PSO algorithm that is easy to trap into the local optimal solution, The thesis also presents a method of searching parameters in the SNPSO. Finally, The performance of SNPSO is fairly demonstrated by applying sixteen benchmark problems and comparing it with several popular PSO algorithm. The analysis of result shows that our proposed methods is effective and gain better performance than other popular PSO algorithms.Furthermore, our method can efficiently improve the performance of standard PSO and more ability to prevent the particle fall into some local optimal solutions.
