一种基于竞争策略的粒子群优化算法

Particle Swarm Optimization Based on Competition Strategy

为提高粒子群优化(Particle Swarm Optimization,PSO)算法的收敛精精度与速度,提出了一种基于竞争策略的粒子群优化算法。算法通过对两粒子相似度的判定,来决定是否对粒子进行变换操作,能够提高粒子的多样性,避免局部最优,提高了收敛精度,并且当两个粒子被判定为同一个粒子时,根据适者生存的思想,适应度较优的粒子保留下来,适应度较差的粒子则需进行高斯变异变换,在保证粒子多样性的基础上减少了运算量,提高了收敛速度。并且通过多峰函数(Achley函数、Schaffer函数、Grienwank函数)验证,结果表明,改进后的粒子群优化算法在收敛精度与收敛速度方面都优于基本的粒子群优化算法。

In order to improve the precision and rate of convergence of PSO, Particle Swarm Optimization Based on Competition Strategy is proposed. Whether the transformation operation is preceded is determined by the similarity of two particles. So it can improve the diversity. And it has the capability of avoiding premature convergence and im- proves the convergence precision, According to the principle of survival of the fittest, the particle with better fitness performance survives and GA operation is only applied to the particle with bad performance, so convergence velocity is improved. And it is tested by Ackley function, Shcaffer function and Grievank function. The result shows that the improved PSO is better than the basic one in the performance of convergence and precision