城市日用水量预测的非参数模型研究

The nonparametric multi regression model of urban daily water demand

本文针对传统的线性回归模型误差较大的特点,利用核估计与局部线性估计方法,以气温、节假日为自变量,以用水量为因变量建立了城市日用水量的多元非参数回归模型.经西安市实例验证表明,相对于线性回归模型而言,多元非参数回归模型预测精度较高,预测效果好.

According to the effecting factor and the character of urban daily water and because the line model dosen＇t reflect the nonlinear problem really, we set up the nonparametric multi regressive model of daily water demand. The regression function m（·） is estimated by the kernel estimation and local linear estimation. By the experiment of Xi＇an city, it is proved that the nonparametric multi regression model can meet the practical requirement of water supply dispatch system.