GM(1,1)灰色预测模型的改进与应用

Application of improved GM(1,1)grey prediction model

针对传统GM(1,1)模型预测精度低的问题,采用变权构造背景值,并将模拟拟合的残差构建多项式作为修正值引入模型中,建立改进的GM(1,1)模型。利用该模型对2个实例进行了模拟和预测。结果表明,针对非指数函数特征的数据序列,通过该文改进的GM(1,1)模型预测,在保持模型精度相对稳定的情况下,拟合平均相对误差、残差均方差和预测平均相对误差值都更小。

Aiming at the problem of low prediction accuracy of traditional GM(1,1)model,variable weight is used to construct background value,and the polynomial constructed by simulation fitting residual error is introduced into the model as the modified value to establish an improved GM(1,1)model.The model is used to simulate and predict two examples.The results show that,for the data series with non exponential function characteristics,the fitting average relative error,residual mean square error and prediction average relative error are smaller under the condition of keeping the model accuracy relatively stable.