CAJ | 학술논문

为解决粒子群算法搜索精度不高,特别是对高维函数优化性能不佳问题,提出一种信息充分交流的扩散粒子群算法(DPSO-FCI).该算法在粒子更新方式上引入周围极值,在演化过程中粒子以一种递增方式进行扩散操作,使得种群信息得到更加充分的利用.同时,通过非线性调整惯性权重、扩散操作引导极值变化来增强群体对信息的利用能力.采用4个基准测试函数对DPSO-FCI算法进行测试,并与几种不同类型的改进粒子群优化算法进行对比.实验结果验证了DP-SO-FCI算法的有效性.
위해결입자군산법수색정도불고,특별시대고유함수우화성능불가문제,제출일충신식충분교류적확산입자군산법(DPSO-FCI).해산법재입자경신방식상인입주위겁치,재연화과정중입자이일충체증방식진행확산조작,사득충군신식득도경가충분적이용.동시,통과비선성조정관성권중、확산조작인도겁치변화래증강군체대신식적이용능력.채용4개기준측시함수대DPSO-FCI산법진행측시,병여궤충불동류형적개진입자군우화산법진행대비.실험결과험증료DP-SO-FCI산법적유효성.
To improve the poor search quality of the standard particle swarm optimization(PSO)especially for optimizing high-dimensional function, a diffusion particle swarm optimization with fully communicated information (DPSO-FCI) was prop-osed. Circumference extreme value was introduced into particle update mode and an increasing dissipative mode into particle dif-fusion operation in evolvement was inducted so that information can be fully utilized in a community. Meanwhile, a nonlinear function was used to adjust the inertia weight and diffusion oper-ation for leading extremum changing so as to enhance the ability of the group to utilize the information. The DPSO-FCI was tested on four benchmark functions, and then compared with other particle swarm optimization algorithms. Finally, the re-suits demonstrate the effectiveness of the proposed algorithm.