CAJ | 학술논문

为了克服传统粒子群算法（PSO）的早熟和局部最优问题,提出了一种新的自适应惯性权重的混沌粒子群算法（ACP-SO算法）.该算法采用分段Logistic混沌映射的方法产生初始种群,并根据种群的进化状态来动态调整惯性权重.在详细阐述算法的种群初始化过程和动态调整惯性权重的过程之后,对经典的测试函数分别采用几种改进的PSO算法和ACPSO算法对其进行了测试,与其他几种方法相比,ACPSO算法的全局搜索能力有了显著的提高,并且能有效地避免早熟收敛问题,同时也说明ACPSO算法应用的可行性和有效性.
위료극복전통입자군산법（PSO）적조숙화국부최우문제,제출료일충신적자괄응관성권중적혼돈입자군산법（ACP-SO산법）.해산법채용분단Logistic혼돈영사적방법산생초시충군,병근거충군적진화상태래동태조정관성권중.재상세천술산법적충군초시화과정화동태조정관성권중적과정지후,대경전적측시함수분별채용궤충개진적PSO산법화ACPSO산법대기진행료측시,여기타궤충방법상비,ACPSO산법적전국수색능력유료현저적제고,병차능유효지피면조숙수렴문제,동시야설명ACPSO산법응용적가행성화유효성.
To overcome the problem of premature convergence and local optimal in conventional particle swarm optimization（PSO）,a new adaptive inertia weight chaos particle swarm optimization（ACPSO） is presented.The algorithm generates initial population with segmented logistic map,and varies inertia weight dynamically based on the evolutionary state of the population.After the detailed illustrations of how to generate initial population and how to adjust the inertia weight,this paper tests some classical functions with some improved PSO algorithms and ACPSO algorithm.Compared with other algorithms,the ACPSO algorithm not only has a great advantage of convergence property,but also avoids the premature convergence problem effectively,and at the same,it shows the feasibility and validity of the ACPSO algorithm.