灰色预测模型参数估计的优化方法

OPTIMAL METHOD OF ESTIMATION TO PARAMETERS IN THE GREY FORECASTING MODEL

基于拟合误差绝对值和最小及折扣系数建立了GM(1,1)模型参数估计的折扣最小一乘法,给出了求解折扣最小一乘法的小生境遗传算法,它克服了用线性规划方法求解时,折扣系数靠经验确定而易导致偏差的问题。

On the basis of least absolute value and discount coefficient,the discount least absolute value is offered on the parameter estimation in the grey model GM(1,1);in the meanwhile,it is solved by niche genetic algorithm which overcomes a deviation of discount coefficient determined only by one's experience when it is solved by linear programming.The Analysis of some examples indicates that the new method improves the precision of fitting and forecasting of the model.