惯性质量衰减的引力搜索算法  

Gravitational search algorithm with inertial mass attenuation

作　　者：钱伟懿[1] 张丽佳[1] QIAN Weiyi ZHANG Lijia(College of Mathematics and Physics, Bohai University, Jinzhou 121000, China)

机构地区：[1]渤海大学数理学院,辽宁锦州121013

出　　处：《沈阳师范大学学报（自然科学版）》2017年第2期138-144,共7页Journal of Shenyang Normal University:Natural Science Edition

基　　金：国家自然科学基金资助项目(11371071);辽宁省教育厅科学研究一般项目(L2013426)

摘　　要：针对引力搜索算法易陷入局部最优的缺点,提出一种惯性质量衰减的引力搜索算法。该算法认为粒子的惯性质量有一定衰减,把粒子惯性质量的衰减率看成一个模糊变量,用其隶属度函数定义惯性质量的衰减率,把衰减率与粒子惯性质量乘积作为粒子的惯性质量,从而提高算法的开发能力。另外,为了提高算法的探索能力,给出一个新的变异算子。最后,把所提出算法应用到经典测试函数中,并与引力搜索算法及其他改进的引力搜索算法比较,数值结果表明所给出的算法能够提高求解精度和收敛速度。In order to overcome the shortcomings that gravitational search algorithm（GSA）traps into local optima easily,agravitational search algorithm with inertial mass attenuation is proposed.In this algorithm,the particle’s inertial mass is considered to have certain attenuation,the attenuation rate of particle’s inertial mass is regarded as a fuzzy variable,the attenuation rate is defined based on membership functions,the particle’s inertial mass is redefined by the product of attenuation rate and particle’s inertial mass.Such exploitation ability of the proposed algorithm is improved.In addition,a new mutation operator is proposed to improve the exploration ability.Finally,the proposed algorithm is tested on several benchmark test functions and compared with GSA and the other improved GSA.The numerical results indicate that the proposed algorithm can improve the convergence speed and precision.

关 键 词：引力搜索算法 隶属度函数 变异算子 全局优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]