Abstract: A new Electromagnetism-like Mechanism (EM) algorithm was proposed in this paper in order to prevent data overflow and reduce the computation load. The formulas of the particle charge and the total force vector were improved. The lower bound of objective function and the formulas of the particle charge filtration were introduced. The standard test functions were tested and the new algorithm was compared with EM algorithm, which proved that the new algorithm converges faster. Furthermore, the numerical results show that the approach is efficient and valid.

Key words: attraction-repulsion, global optimization, Electromagnetism-like Mechanism (EM)

摘要： 针对类电磁机制算法中数据溢出、计算量过大的问题,改进了电量计算公式和合力计算公式,引入了函数值最小下界,增加了粒子过滤公式,从而得到一种新类电磁机制算法。从测试标准测试函数与经典类电磁算法的比较可看出,新算法收敛速度快,并从数值上验证了该算法的可行性和有效性。

关键词: 吸引-排斥, 全局优化, 类电磁机制