摘要
人工鱼群算法(AFSA)存在收敛精度低、易陷入局部最优、后期收敛速度慢等问题,因此难以得到精确的全局最优解。经过对人工鱼群算法、模拟退火(SA)算法和差分进化(DE)方法的研究,提出将SA算法和DE思想引入AFSA算法的后期搜索中,从而得到基于差分进化与模拟退火的人工鱼群算法(DESA-AFSA)。该算法首先通过鱼群算法搜索全局最优解;然后,在公告板最优记录的基础上,采用SA算法对全局极值满意解域进行局部优化,进而跳出局部极值。当SA算法的问题规模较大时,会降低寻优的最优值精度。因此,在SA算法中,通过DE操作增大个体差异性,凸显优秀个体,使优化值更接近最优解。仿真结果表明,与基于模拟退火的人工鱼群算法(SA-AFSA)和AFSA相比,DESA-AFSA在收敛速度、寻优精度和跳出局部极值的能力方面都有所改善,证明了DESA-AFSA的有效性。
Artificial fish swarm algorithm（AFSA has many disadvantages, such as low convergence accuracy, easy to fall intolocal optimum and slow convergence rate. Therefore, it is difficult to find the precise global optimal solution by using AFSA. TheAFSA, simulated annealing（SA） algorithm and differential evolution（ DE） method are studied, and it is proposed that introducingthe Sa algorithm and DE method into the post-searching of AFSA, a new artificial fish swarm algorithm（ DESA- AFSA） based onDE and SA is obtained. The algorithm searches the global optimal solution by fish swarm algorithm firstly, and on the basis ofoptimal records on bulletin board, the SA algorithm is used to optimize locally the global extreme satisfaction solution region, andthen jumps out of the local extremum. When the SA algorithm is facing large scale, the precision of the optimal solution will bereduced.Therefore, in the SA algorithm, the individual difference is enhanced by DE operation, and the outstanding individual ishighlighted, so that the optimization is closer to the optimal solution. The simulation results show that comparing AFSA based onSA（SA-AFSA） and AFSA, the DESA- AFSA improves the capability of convergence speed, optimization accuracy and jumping outof the local extremum： the effectiveness of desa-afsa algorithm is verified.
出处
《自动化仪表》
CAS
2018年第2期72-76,85,共6页
Process Automation Instrumentation
基金
特殊环境机器人技术四川省重点实验室基金资助项目(13ZXTK07)
关键词
人工鱼群算法
差分进化
模拟退火
全局最优解
局部极值
鲁棒性
自适应
Artificial fish swarm algorithm （ AFSA ）
Differential evolution （ DE ）
Simulated annealing （ SA ）
Gsolution
Local extremum
Robustness
Adaptation
