摘要
麻雀搜索算法是一种基于麻雀捕食与反捕食行为的新型群体智能优化算法。本文针对此算法后期容易陷入局部最优的问题,提出了基于混沌扰动和精英反向学习策略的改进麻雀搜索算法。该算法通过加入自适应权重策略平衡算法的全局探索和局部挖掘能力,融入改进的Tent混沌初始化种群以提升初始解的质量,在发现者位置更新时引入精英反向学习策略,从而增加发现者的多样性。此外,在算法进入停滞状态时使用混沌扰动以产生新解,使算法拥有跳出局部最优的能力,从而提高了算法的全局搜索能力。11个基准测试函数的仿真结果表明,与其他算法相比ISSA算法在迭代速度、寻优精度和稳定性上更具优势。最后,将ISSA算法应用于Hammerstein系统的辨识问题,验证了该算法的有效性和可靠性。
Sparrow search algorithm(SSA)is a novel swarm intelligence optimization algorithm based on foraging and anti-predation behaviors of sparrows.In this paper,an improved sparrow search algorithm(ISSA)integrating chaotic distur-bance and elite opposition-based learning strategies are proposed to solve the problem that SSA is easy to fall into local opti-mum in the late iteration.Based on the SSA,adaptive weights are added to balance its global exploration and local mining ca-pabilities.The population is initialized by improved Tent chaos to improve the quality of the initial population,and elite oppo-sition-based learning method is introduced while the location of the discoverer is updated to increase the diversity of discover-ers and improve its global optimization ability.Furthermore,the chaotic perturbation is used to generate new solutions when ISSA is stagnated,which enables the algorithm to jump out of the local optimum and improves its global optimization ability.The test results of eleven test functions show the ISSA outperforms the other algorithms in the convergent speed,optimization accuracy and solution stability.Finally,ISSA is applied to the Hammerstein system identification problem,which verifies its effectiveness and reliability.
作者
王德凯
江善和
邢翔宇
WANG Dekai;JIANG Shanhe;XING Xiangyu(Department of Electronic Engineering and Intelligent Manufacturing,Anqing Normal University,Anqing 246133,China)
出处
《安庆师范大学学报(自然科学版)》
2023年第3期74-82,共9页
Journal of Anqing Normal University(Natural Science Edition)
基金
国家自然科学基金项目(51607004)
安徽省自然科学基金项目(2008085MF197,1708085ME132)
安庆师范大学研究生学术创新项目(2021yjsXSCX104)。
关键词
群体智能
麻雀搜索算法
精英反向学习
自适应权重
混沌扰动
swarm intelligence
sparrow search algorithm
elite opposition-based learning
adaptive weights
chaos per-turbation