混沌粒子群算法对GM(1,1)模型参数的优化

Chaos particle swarm optimization algorithm for optimizing the parameter of GM(1,1)

鉴于基本粒子群算法存在初始化过程的随机性以及容易陷入局部最优解的不足,对基本粒子群算法进行改进.利用混沌运动的遍历性,产生大量初始群体,从中择优出初始群体,并在粒子群优化算法执行的过程中,对当前粒子个体产生混沌扰动,以使解跳出局部极值区间.用混沌粒子群算法对综合GM(1,1)参数优化模型的参数进行优化,认为利用优化所得参数值进行预测能取得更好的结果.

There are some problems of the randomness of initialization process and easy to fall into a local optimal solution when using Particle Swarm Optimization（PSO） algorithm.In this paper,PSO algorithm is improved.The basic idea is： at the beginning,a great deal of initial population are produced by ergodicity of chaos,and then the initial particle swarm is preferred out：during the running time of PSO algorithm,to adjust the current particale by chaos so that the solution jumps out local convergence.In the end,Chaos PSO algorithm is proposed to optimize the integrated GM（1,1） paramster optimization model and we get good result in the example.