灰色系统与多元逐步回归耦合模型在年用水量预测中的应用

Prediction of Annual Water-consumed Quantity Based on Coupling Model of Grey-system and Multiple Stepwise Regression

探讨了将多元回9-3分析模型和灰色系统GM（1，1）耦合应用于北京市总用水量预测的方法和技术。对北京市1980-2001年的序列资料，利用灰色系统的关联度分析和逐步回归的思想，通过比较和选择，建立了多元线性逐步回归方程；将用GM（1，1）模型拟合反推得到的预测值带入方程求出总用水量的未来值。结果表明：人均日生活用水量和工业用水量是影响北京市总用水量变化的主要因素。灰色系统与多元线性回归的耦合模型用于对城市总用水量的预测是行之有效的。

The coupling model of Grey-system and multiple stepwise regression was discussed and applied to the prediction of annual water-consumed quantity of Beijing. Firstly, using the method of correlation analysis and stepwise regression of grey system, the data sequences of Beijing from 1980 to 2001 were compared and selected to set up multiple stepwise regression formula. Then, the prediction value gained through GM（1,1） model was brought to the formula to get the future value of the total water quantity. The result showed that the water-consumed quantity per day, per person and industrial water-consumed quantity were the main influence factors for the change of total water quantity. It is feasible to apply the coupling model of grey-system and multiple stepwise regression for the prediction of total water quantity of the city.