粒子以概率收敛的粒子群算法的分析与实现

粒子群算法的收敛过程是通过粒子向收敛目标点的移动实现的。粒子向目标点的移动既可以按一定的轨道实现,也可以按给定的概率密度随机移动实现。通过分析随机移动时概率密度函数所应遵循的条件,给出了两大类共四种符合要求的概率密度函数,使用随机模拟的方法,将其中三种转化成为粒子的移动方程,从而给出了不同于传统粒子群算法的三种算法。经过在相同条件下对三个标准测试函数的优化运算,除算法2外,算法1与算法3表现均显著优于标准粒子群算法。

The Analysis and Implementation of Particle Swarm Algorithm Convergence in Probability

The Through the movement of the particles to the target point, the convergence of particle swarm algorithm is achieved, the movement of the particles can along a given track or according to the probability density. By analyzing the conditions which the probability density function should satisfy, Four kinds of probability density function is given, by using the stochastic simulation method, three kinds of the particle movement equation is deduced according to the probability density function. Then get three kinds of new particle swarm algorithm. Under the same conditions, comparing the results of three standard test functions, in addition to the algorithm 2, algorithm 1 and 3 were significantly better than the standard particle swarm algorithm.