GM组合模型用于城市生活用水量预测

Prediction of urban domestic water demand by GM integrated model

以天津市为例,采用GM组合模型预测城市生活用水量,力求提高预测的精度。首先,通过对往年城市用水特点的分析,运用多元逐步回归的方法和等维灰数递补动态模型对天津市城市生活用水量进行预测,预测的平均误差分别为7.59%和11.55%;然后,采用上述两种模型的GM组合模型对天津市城市生活用水量进行预测,预测的平均误差降低为5.06%。实践证明,GM组合模型适用于城市生活用水量的预测,精度令人满意。

Taking Tianjin for example,urban domestic water demand was predicted by GM integrated model so as to improve prediction accuracy.Firstly,according to analysis of the characteristics of urban water usage in recent years,multiple stepwise regression method and the same dimensional gray recurrent dynamic model were applied to the prediction of urban domestic water demand in Tianjin City,and the average errors of the methods were 7.59% and 11.55% respectively.Then GM integrated model of the above two models was used and the error came down to 5.06%.It was proved that GM model was suitable for prediction of urban domestic water demand and the precision was satisfying.