基于时间序列的多岛遗传算法在水泥窑中的应用

摘要

研究时间序列在水泥回转窑中的建模问题,水泥回转窑熟料的煅烧是一个涉及物理、化学反应的复杂多变量、多扰动非线性过程.针对最小二乘在多参数高阶时间序列模型中难以寻找到较好的系数组,从而不能建立较精确的时间序列模型的情况,采用传统遗传算法也很难寻找出最优参数组,当利用多岛遗传算法对其进行参数寻优时,只要适当地设置算法里的一些关键参数即能得出满意的组解.实验结果表明:经过多岛遗传算法寻优得出的参数建立的时间序列模型,其拟合度满足模型曲线的精度要求.%Time series modeling in cement rotary kiln was carried out.The clinker calcination in cement rotary kiln is a complex process which boasting of physical and chemical reaction and multi-perturbed nonlinearity.Considering the fact that the least squares has difficulty in finding better coefficient group in multi-parameter' s higher order time series model and fails to establish a more accurate time series model,saying nothing of employing traditional genetic algorithm to find out the optimal parameter set.When using the multi-island genetic algorithm for parameter optimization,the satisfactory group solution can be obtained as long as the key parameters in the appropriate setting algorithm are set properly.Experimental results show that the time series model established by the parameters derived from the multi-island genetic algorithm can satisfy model accuracy requirements in the fitting degree.