鲶鱼二进制粒子群优化算法研究

Study on Catfish Binary Particle Swarm Optimization

在粒子群优化算法的研究中,粒子群算法是根据鸟群觅食过程中的迁徙和群集模型提出的,用来解决优化问题的一种随机方法。由于传统二进制粒子群算法具有早熟收敛的缺点,为提高算法的速度和稳定性,提出将"鲶鱼效应"引入到算法中,有效增加优化过程中种群的多样性,进而使粒子能够跳出局部最优点,进入解空间的其它区域继续进行搜索,有效避免早熟收敛的发生,直到最后找到全局最优解。数值仿真表明,通过恰当地选取算法的相关参数,鲶鱼二进制粒子群优化算法在收敛率、寻优速度及稳定性方面均有所提高。

The traditional binary PSO （BPSO） algorithm has the shortcoming of premature convergence. The idea of ＂catfish effect＂ was introduced into the BPSO algorithm, and the diversity of the population was effectively in- creased. This may make the particles skip from the local optimization and search in other area of solution space until finding the global optimization. The improved BPSO can avoid the problem of premature convergence effectively. The experimental simulations show that if corresponding parameters of the catfish BPSO algorithm are selected correctly, its performances including convergence rate, research speed and stability can be improved.