基于时变Morlet小波变异的蝙蝠算法及函数优化

Bat Algorithm Based on Time-varying Morlet Wavelet Mutation for Function Optimization

摘要:

为了提高蝙蝠算法的全局收敛能力和收敛速度,提出了一种时变Morlet小波变异的蝙蝠算法(TVMWMBA)；构建Morlet小波分布函数来描述小波变异因子分布的概率密度,然后利用小波变异因子的波动性和周期性,在每次迭代中对蝙蝠种群的个体进行小波变异,避免陷入局部最优,再通过时变系数动态收缩小波变异因子的变异空间,加快算法的收敛,提高收敛精度；函数优化仿真实验结果表明,改进算法有效的保持了种群的多样性,显著提高了优化稳定性,具有全局收敛能力强、寻优精度高和收敛速度快等特点。

Abstract:

To improve Bat Algorithm's global converging capability and speed, this paper presents a novel time-varying-Morlet-Wavelet-Mutation-based Bat Algorithm (TVMWMBA). It first builds a Morlet wavelet distribution function to describe the probability density of the wavelet mutation factor distribution. Afterwards, it takes advantage of the waving property and periodicity of Morlet wavelet to carry out wavelet mutation on each member of the bat population in each iteration to avoid localized optimization. Finally, it uses time-varying coefficient to dynamically contract the mutation space of the wavelet mutation factor to improve the algorithm's converging speed and converging precision. Simulated experiment for function optimization shows that the improved algorithm has effectively maintained the diversity of the population and significantly raised the optimization stability. It has also achieved strong global converging capability, excellent optimum-searching precision, and fast converging speed.