应用微粒群算法确定有机化合物分子式

Determining molecular formulas of organic compounds by Particle Swarm Optimization algorithm

针对基本微粒群优化算法(PSO)存在容易陷入局部最优和收敛速度慢的缺点,在整数空间使用带收缩因子的微粒群优化算法基础上,提出了一种带变异概率的微粒群优化算法(IPSO),用于提高微粒群的多样性,避免算法陷入局部最优解。实验证明,改进后的微粒群优化算法在防止早熟和加快收敛方面优于基本PSO算法和基本PSO算法加一半微粒随机初始化算法(PSO_HPO算法)。IPSO算法应用到确定有机化合物分子式时,取得了很好的效果。

To avoid the premature problem and the slow convergence of particle swarm optimization algorithm（PSO）,an improved particle swarm optimization algorithm（IPSO） is presented to used for determining molecular formulas of organic compounds.On the basic of integer programming and the PSO with contraction factor,the IPSO with mutation probability is proposed to get a good population diversity and to avoid PSO getting into local best result.The algorithm applied to determine molecular formulas of organic compounds is much better than those of PSO and PSO_HPO.