摘要
针对花朵授粉算法(Flower pollination algorithm,FPA)易陷入局部极值、后期收敛速度慢的不足,提出一种基于引力搜索机制的花朵授粉算法.该算法在基本花朵授粉算法的全局寻优部分,采用花朵个体间的万有引力和算法本身的莱维飞行共同实现个体位置的更新,使花朵受莱维飞行和个体间引力的双重影响,个体在通过优化信息的共享向质量大(最优位置)的个体靠近,且个体间的万有引力牵制莱维飞行的随机游走.同时又利用莱维飞行的跳跃及不均匀性步长避免个体陷入局部极值,从而提高算法的寻优能力.通过对高维单峰函数、高维多峰函数、低维函数及多峰复杂函数的优化实验结果表明,改进算法的寻优性能显著优于基本的花朵授粉算法,其收敛速度、收敛精度、鲁棒性均较对比算法有较大提升.最后,利用改进算法对弹簧张力设计问题、压力管设计问题2个工程实例进行测试,获得了较好的结果.仿真实验结果佐证了改进算法的有效性和可行性.
A flower pollination algorithm （FPA） based on gravity search mechanism is presented to overcome the problems of being easily trapped into local extremum and low speed of convergence. In the global optimization of FPA, the algorithm uses the gravity between individuals and the Lévy flight of algorithm itself to update individual locations. The flower is influenced by both gravity and Lévy flight, and individual is close to the individual which is in a higher quality （optimal position） through optimization of the information, and the random walk of Lévy flight is contained by the gravity between individuals. At the same time, the improved algorithm uses the jump and irregular step length of Lévy to limit the individual into local extremum so as to improve the algorithm＇s optimization ability. High dimensional unimodal function, high dimensional multimodal function, low dimensional function and multi peak complex function optimization results show that the improved algorithm has better global searching ability than the basic flower pollution algorithm, and faster convergence and more precise convergence than those of comparison algorithm. Finally, the improved algorithm is applied to two engineering examples, the tension spring design problem and the pressure vessel design problem, and obtains desired outcomes. Simulation results have proven the effectiveness and feasibility of the improved algorithm.
出处
《自动化学报》
EI
CSCD
北大核心
2017年第4期576-594,共19页
Acta Automatica Sinica
基金
国家自然科学基金(61562032)
河池学院计算机应用技术重点学科(2016-91)资助~~
关键词
花朵授粉算法
寻优性能
万有引力
适应度值
Flower pollination algorithm （FPA）, optimization ability, gravity, fitness
