粒子群最佳化演算法是一個隨機、群體行為的演化式搜尋技巧,為Kennedy與Eberhart於1995年提出,是一個從鳥群和魚群得到靈感的演算法。由於粒子群具有容易實現的優點(如需設定的參數少)而受到歡迎,但在粒子群演算法中常常難以在全局探索和局部搜尋間取得平衡。為增進粒子群演算法效能及維持粒子多樣性,本研究提出二個改良式演算法,第一個演算法為 Velocity-Adjustable PSO,根據粒子與全域最佳解間的距離動態改變粒子的速度,第二個演算法為Elitist-Communicative PSO,藉由菁英粒子間的交換產生更優秀的粒子帶領所有的粒子朝最佳解的位置飛近。第三個演算法為Distance-Based PSO,根據粒子與全域最佳解間的距離動態改變粒子的演化方法。
Particle Swarm Optimization (PSO) is a stochastic, population-based evolutionary search technique proposed by Kennedy and Eberhart in 1995, which is inspired by flocks of birds and shoals of fish. It is popular due to its simplicity in its implementation, as a few parameters are needed to be tuned. PSO has difficulties in controlling the balance between exploration and exploitation. In order to improve the performance of PSO and maintain the diversity of particles, we proposed three improved algorithm, the first algorithm called VAPSO (Velocity-Adjustable PSO) adjusts the velocity of the particle according to its distance from itself to the gbest. The second algorithm called ECPSO (Elitist-Communicative PSO) generates the better solutions to take the lead by communication between elitist and other elitists. The last algorithm called DBPSO (Distance-Based PSO) adjusts the evolve method of the particle according to its distance from itself to the gbest.