类电磁机制算法研究与改进

Study on and improvement of electro-magnetism-like mechanism algorithm

针对原算法中初始种群不均匀、局部搜索随机性强及移动公式效率低等问题,提出了一种改进的类电磁机制算法。采用随机化均匀设计方法构造初始种群,将混沌搜索与试探性移动方法相结合进行局部搜索,运用改进后的公式计算粒子之间的合力,将遗传变异思想融入移动公式来更新粒子。实验结果表明,改进后的算法具有更好的收敛效果和更高的求解精度。

An improved electro-magnetism-like mechanism algorithm is proposed to solve the problems of the original EM algorithm, such as non-uniform initial population, strong randomness in the phase of local search and low efficiency in the movement according to the total force. The new algorithm generates the i^itial population with the randomized uniform design method, improves the local search algorithm with the chaotic search method and exploratory mobile method,and calculates the total force between particles with the modified equation. Besides, the idea of heredity and variation is used to update the loca- tions of those particles. Experimental results show that the improved algorithm has a better convergence effect and higher solution accuracy.