一种基于Lévy飞行的细菌觅食优化算法

Bacterial foraging optimization algorithm based on Lévy flight

分析了细菌觅食优化(BFO)算法的基本原理,为了改善算法局部搜索能力突出而全局搜寻能力欠佳、算法结构复杂等缺点,在BFO算法的复制操作中引入Lévy飞行机制,并将算法的迁徙操作移入趋向操作内部,简化了算法结构,变原来的三层嵌套循环结构为两层循环,提出一种新的基于Lévy飞行的BFO算法——LBFO算法。该算法的复制操作中,保留当前种群中50%的优良细菌个体,然后对余下的个体全部用Lévy飞行进行位置更新,保证了算法全局收敛性的同时又加强了算法的随机搜索能力,有助于保持种群多样性和减少早熟收敛的现象发生。最后将LBFO算法对选取的六个基准测试函数和0-1背包问题进行实验仿真,实验结果表明新提出的LBFO算法不仅收敛速度快,而且优化精度高,运行速度也得到很大程度的提升。

This paper analyzed the principle of the bacterial foraging optimization （BFO） algorithm. It improved the standard BFO algorithm to mainly overcame its poor global search capability and complex structure. And it introduced Levy flight into re- production operation and moved the operation of elimination and dispersal to chemotaxis operation to simplify the algorithm structure, and translated the triple loops into dual loops. This paper proposed a new BFO （LBFO） algorithm based on Levy flight mechanism. In the reproduction operation, the algorithm reserved half of the elite bacterial, and updated the other by Levy flight. It guaranteed not only the global convergence but also the random search ability of LBFO algorithm, thus help to keep the diversity of the bacterial population and reduce the situation of premature convergence. Finally, the experimental results on 6 Benchmark function optimization problems and 0-1 knapsack problems show that the improved LBFO algorithm has the higher searching performance and convergence rate than the standard BFO algorithm.