量子势阱粒子群优化算法的改进研究

Research on the improvement of quantum potential well-based particle swarm optimization algorithm

为提高量子势阱粒子群优化算法的优化能力,通过分析目前量子势阱粒子群优化算法的设计过程,提出了改进的量子势阱粒子群优化算法.首先,分别基于Delta势阱、谐振子和方势阱提出了改进的量子势阱粒子群优化算法,并提出了基于统计量均值的控制参数设计方法.然后,在势阱中心的设计方面,为强调全局最优粒子的指导作用,提出了基于自身最优粒子加权平均和动态随机变量的两种设计策略.实验结果表明,三种势阱粒子群优化算法性能比较接近,都优于原算法,且Delta势阱模型略优于其他两种.

To enhance the optimization ability of quantum potential well-based particle swarm optimization algorithm,the improved quantum potential well-based particle swarm optimization algorithms are proposed by analyzing the design process of current quantum potential well-based particle swarm optimization algorithms.Firstly,three improved quantum particle swarm optimization algorithms are proposed based on delta potential well,harmonic oscillator and square potential well,respectively,and then a statistic mean-based control parameter design method is presented for the proposed models.Secondly,to highlight the guiding role of the global optimal particle in designing potential well centers,two strategies are presented based on a weighted average of all self-optimal particles and dynamic random variables.The experimental results show that the performances of three improved algorithms are relatively close, the model based delta potential well are slightly better than the other two kinds of model,and the performances of three improved algorithms are superior to that of the original algorithm.