摘要
蝙蝠算法(BA)收敛精度不高、收敛速度缓慢、易陷入局部极值,借鉴布谷鸟算法和正弦余弦算法的优化策略,提出了一种基于Sobol序列和间歇Levy跳跃的改进蝙蝠算法(LZBA).利用Sobol序列初始化蝙蝠的位置,重新定义自适应惯性权重和寻优因子,平衡局部和全局搜索能力,采用概率性替换策略,增加种群多样性,通过Levy跳跃和混沌扰动全局最优解,试图跳出局部最优.选择几个不同维度的标准测试函数进行对比寻优实验,实验分析表明:LZBA无论在普适性、稳定性还是收敛精度方面均优于BA算法和CS算法.
Bat algorithm(BA) has low convergence precision,slow convergence speed and easy to fall into local extremum.By referring to the optimization strategies of cuckoo algorithm and sine and cosine algorithm,an improved bat algorithm(LZBA) based on Sobol sequence and intermittent Levy jumping is proposed.Sobol sequence was used to initialize bat position,redefine adaptive inertia weight and optimization factor,balance local and global search ability,adopt probabilistic substitution strategy,increase population diversity,and try to jump out of local optimal solution by Levy jumping and chaos.Several standard test functions of different dimensions were selected for comparative optimization experiment.Experimental analysis showed that LZBA was superior to BA algorithm and CS algorithm in terms of universality,stability and convergence accuracy.
作者
李志军
LI Zhi-jun(School of Information Engineering,Guangxi University of Foreign Languages,Nanning 530222,China)
出处
《数学的实践与认识》
2021年第8期313-320,共8页
Mathematics in Practice and Theory
基金
广西高校中青年教师科研基础能力提升项目(2019KY0902)。
