基于引力搜索机制的花朵授粉算法

发布时间：2018-05-28 00:06

本文选题：花朵授粉算法 + 寻优性能　；参考：《自动化学报》2017年04期

【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.
[Abstract]:A flower pollination algorithm based on gravity search mechanism is proposed to solve the problem that flower pollination algorithm (Flower pollination algorithm) is prone to fall into local extremum and slow convergence rate. In the global optimization part of the basic flower pollination algorithm, the universal gravitation between flower individuals and the Levi flight of the algorithm itself are used to update the individual position together, and the flowers are affected by the double influence of Levy flight and individual gravity. Individuals are moving closer to those with high mass (optimal position) through the sharing of optimized information, and the gravitational pull between individuals restrains the random walk of Levy's flight. At the same time, the jump and non-uniformity step size of Levi flight is used to avoid individual falling into local extremum, so as to improve the optimization ability of the algorithm. The experimental results of high-dimensional single-peak function, high-dimensional multi-peak function, low-dimensional function and multi-peak complex function show that the performance of the improved algorithm is significantly better than that of the basic flower pollination algorithm, and the convergence rate and accuracy of the improved algorithm are better than that of the basic flower pollination algorithm. The robustness is much better than the contrast algorithm. Finally, two engineering examples of spring tension design and pressure pipe design are tested by using the improved algorithm, and good results are obtained. Simulation results demonstrate the effectiveness and feasibility of the improved algorithm.
【作者单位】：江西财经大学信息管理学院;河池学院计算机与信息工程学院;
【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~~
【分类号】：TP18

【相似文献】