一种基于分子动理论的改进粒子群优化算法

Improved Particle Swarm Optimization Algorithm Based on Theory of Molecular Motion

提出了一种新颖的基于分子动理论的粒子群优化算法(MMT-PSO)。类比于物理学中质心的概念本文定义了群质心,MMT-PSO把种群中的每个粒子类比成分子,根据粒子与种群目前的质心之间的距离远近,粒子与质心间的分子作用力控制粒子的飞行方向以决定其是朝着群质心的方向飞行还是远离它,从而有效地协调了种群的多样性,使算法能够有效地平衡全局和局部搜索。通过解决典型的多峰、高维函数优化问题来证实算法的有效性,实验结果表明MMT-PSO比标准PSO具有更高的性能。

A novel particle swarm optimization based on theory of molecular motion （MMT-PSO） was proposed, and the population was regarded as molecule system. The molecular force between the molecules was proposed as an attractive or repulsive force determined by the distance of the molecules; the molecular force was introduced in the MMT-PSO to decide the particles to move towards the swarm centroid which was defined by analogy to the centroid in physics, or to keep away from it for maintenance of high diversity, hence the MMT-PSO could effectively balance the global and local search. Experimental results on multi-modal, high-dimensional numerical optimization problems show that MMT-PSO outperforms standard PSO.