采用多精英指导的烟花算法

Fireworks algorithm with multi-elite guidance

分析了造成烟花算法中个体之间信息共享不足的原因,提出一种基于多精英指导的改进烟花算法.在每轮迭代中,较差子群中的个体接受随机选择的多个精英的指导,越差的个体被精英指导的概率越大.改进算法增强了个体之间的信息共享.在标准测试函数上的实验结果表明,改进算法在收敛精度和速度方面优于基本烟花算法.

The formation reason of information sharing insufficiency among individuals in fireworks algorithm was analyzed and a multi-elite guidance-based improved fireworks algorithm was presented.During every round of iteration in the algorithm,the individuals in more inferior sub-swarm would accept the guidance of randomly selected multiple elites and the worse the individual was,the higher the probability of guidance from the elites would be.The proposed algorithm would enhance the information sharing among the individuals and the result of experiment on benchmark testing functions showed that the improved algorithm would be superior to the elementary fireworks algorithm in convergence accuracy and speed.