城市年用水量的分类预测探析

在对某市年用水量随机时间序列原始数据进行预处理的基础上,发现该市年生活用水量的一次累加时间序列数据具有明显的线性趋势;工业年用水量一次累加用水时间序列服从一元多项式。将该市年总用水量分为工业、生活、其他三类进行分类预测,分别用一次累加乘幂指数预测、一次累加一元多项式预测对该市生活和工业年用水量进行了预测,复相关系数达0.99。为便于分析比较,对该市年总用水量预测上采用对年用水量预测效果较好的灰色预测模型进行了预测。结果表明:分类用水量预测比总体用水量预测具有精度高,结果稳定的特点,可用于该市的年用水量预测。

Research on the classification prediction yearly water demand in urban

Based on pre-processing the time-series data of annual water consumptions in one city,we has found out that this city's one-accumulate of the time-series data of living water quantity possessed the obvious linear trend,the time-series data of industry water quantity submitted one kind of variety polynomial.And this paper divided the total annual water consumption into three types: industry,living and the other water consumption and carryed on the classification forecast,using one-accumulate exponent smoothing forecast and one-accumulate variety polynomial forecast model respectively,its compound related coefficient amount to 0.99.For the sake of comparing and analysising,gray forecasting model is applied in this city annual water consumption forecast.The result indicated that the classification forecast have the higher accuracy and steady than total forecast,it can be used for this city's annual water consumption forecast.