基于理想边界和多元逐步回归模型的城市生活需水量预测

Prediction of Urban Domestic Water Demand Based on Ideal Boundary and Multiple Stepwise Regression Model

精准的生活需水量预测可为水资源管理部门调控供需提供数据支撑。由于测量过程中受诸多不确定因素的影响,观测值的测量结果不以单一的数值存在,利用区间数值可以很好描述这一不确定性的范围。现有预测区间的算法需要繁琐的参数搜索过程,本文算法可简化该过程。首先,利用一种初始区间设定办法得到区间宽度,作为本文初始绝对宽度构造区间上下界。其次,对上界集和下界集分别建立多元逐步回归模型预测城市生活需水量的区间上下界,以预测区间覆盖率、预测区间宽度、预测区间对称性和均方根误差作精度评估指标与已有模型结果对比,截取较优的绝对宽度,建立模型。与此同时,以邯郸市生活需水量为例,与已有模型对比,预测精度优于已有模型。

The prediction of accurate domestic water can provide more reasonable reference for water resources management departments to regulate supply and demand.The measurement results of the observed values do not exist as a single value due to the influence of many uncertainties during the measurement process,so the range of this uncertainty can be perfectly described by using interval values.The existing prediction interval algorithm needs a tedious parameter search process which can be simplified by the algorithm in this paper.Firstly,an initial interval setting method is adopted to obtain the interval width,which is used as the initial absolute width to construct the upper and lower bounds of the interval.Secondly,a multiple stepwise regression model shall be established to predict the upper and lower bounds of urban water demand respectively,and it is used to predict interval coverage,interval width,interval symmetry and root mean square error,and then use it as accuracy evaluation index,and compare with the existed model results,the absolute width is intercepted and the model shall be established.Finally,taking the urban domestic water demand of Handan as an example,the prediction accuracy of this model is better than that of existing models.