双种群协同下带混沌闪烁机制的萤火虫算法研究

A Firefly Algorithm with Chaotic Flicker Mechanism under Double Population Collaboration

针对萤火虫算法处理高非线性、多极值的复杂工程优化问题所存在的快速收敛与早熟、全局探索和局部探索之间的矛盾,提出了一种在双种群策略下具备混沌闪烁机制的萤火虫改进算法。首先,通过引入混沌闪烁因子ξ调制萤火虫运动状态,模拟萤火虫发光习性,能够在保持种群内个体自主动力性的前提下大幅提升算法的收敛速度;同时,使用双种群策略进行全局种群和局部种群的划分,保持种群间信息交互,有效平衡了算法全局探索和局部探索的能力,降低了陷入局部最优的风险。采用经典单模、多模测试函数集对算法进行测试,结果表明,在相同种群规模和迭代次数下,算法能够提高收敛速度,避免了局部最优,从而达到更好的寻优效果。部分测试函数收敛精度相比于其他算法,可得到5、6个数量级以上的提升,而且算法也能够在相对最少的函数评价次数内满足精度要求。

The firefly algorithms could not solve the contradiction between convergence rate and early maturing and that between global exploration and local exploration when handling complex engineering optimization problems with high nonlinearity and multi-extreme values.On this basis,a kind of dual population firefly algorithm is proposed based on firefly“glowing”and“extinguishing”flicker mechanism.In the algorithm,the moving state of firefly is modulated with chaotic flicker factorξto simulate firefly biological habits,which can greatly enhance the convergence rate of the algorithm under the premise of keeping the individual autonomous power in the population.At the same time,the population is divided into global population and local population through the dual population strategy,which can maintain the interaction between groups of information and balance the global exploration ability and local exploration ability of the algorithm,and hence reducing the risk of falling into the local optimal risk.The classic single-mode and multi-mode test functions are used to verify the algorithm.The results showed that the algorithm can achieve better optimization effect while ensuring the convergence rate and avoiding local optimum.The convergence accuracy can be improved more than4-5orders of magnitude.Meanwhile,the algorithm can meet the requirement of accuracy in the minimum number of function evaluations.