一种自适应混合变异的引力搜索算法  

Adaptive Hybrid Mutation Gravitational Search Algorithm

作　　者：逯清玉[1] 张晓明[2] 

机构地区：[1]青海建筑职业技术学院基础及素质教育部,西宁810012 [2]长春建筑学院基础教学部,长春130607

出　　处：《重庆师范大学学报（自然科学版）》2017年第3期85-90,共6页Journal of Chongqing Normal University:Natural Science

摘　　要：【目的】针对引力搜索算法在求解优化问题时易陷入局部极值问题,提出了一种自适应混合变异的引力搜索算法。【方法】采用动态调整粒子速度和位置的更新公式,提高算法搜索精度。引入变异算子,对最优粒子进行高斯变异,对非最优粒子进行自适应t分布变异。【结果】提高算法在求解函数优化问题时的全局探索能力和局部开发能力。【结论】用9个标准测试函数的仿真实验,与标准GSA及改进算法进行比较,结果表明所提出算法具有较强的收敛精度和鲁棒性。[Purposes]In view of the shortcomings of Gravitational Search algorithm, such as easy to fall into local extremum in function optimization, an adaptive hybrid mutation search algorithm was proposed. [Methods]The dynamic updating formula of particle velocity and position is adopted to improve the accuracy of the algorithm. Mutation operator is introduced to optimize Gauss variation of the optimal particle, and the adaptive T distribution variation of the non-optimal particles is improved. [Findings]The global exploration ability and local development ability of the algorithm are improved in function optimization. [Conclusions]The simulation results on 9 standard test functions show that the proposed algorithm has better convergence accuracy and robustness compared with the standard GSA and its improved algorithms.

关 键 词：局部极值 引力搜索算法 T分布 函数优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]