基于自适应小生境的改进入侵性杂草优化算法

Improved Invasive Weed Optimization Based on Adaptive Niche Algorithm

为提高入侵性杂草优化算法(IWO)的种群多样性,使算法在处理高维多峰问题时具有更好的全局收敛性。结合小生境思想提出一种小生境杂草优化算法(NIWO)。该算法根据种群内个体间的欧式距离对杂草种群进行分类,并采用自适应小生境数策略确定分类个数,对种群进行繁殖竞争等其他操作,从而增强种群的多样性,提高算法的全局寻优能力,保证算法的收敛精度。利用4个标准测试函数测试算法的寻优能力,仿真结果表明,无论对于低维还是高维多峰函数,NIWO算法的收敛精度和稳定性都优于标准IWO算法。

This paper aims to improve population diversity of the standard invasive weed optimization（IWO） to ensure better global convergence of the algorithm in dealing with the high dimension multimodal problems.By combining the Niche algorithm,the IWO algorithm is improved,named Niche invasive weed optimization（NIWO）.This algorithm is enlightened by the idea of birds of a feather flock together.Individuals in the weed populations are first adaptively classified according to the Euclidean distance,and other operations are then completed.As a result,diversity of population is enhanced to improve the algorithm＇s capability of global optimization and convergence precision.The searching capability of the algorithm is verified based on four standard test functions.Experimental results show that,regardless of the low dimensional or high dimension multimodal function,The NIWO algorithm＇s search accuracy and stability are significantly better than the standard IWO algorithm.