基于时间序列模型ARMA的水厂逐日需水量过程预测方法  

Water Plant Daily Water Demand Forecasting Method Based on Time Series Model ARMA

作　　者：孙平[1] 王丽萍[1] 陈凯 蒋志强[1] 张璞[1] 

机构地区：[1]华北电力大学可再生能源学院,北京102206 [2]深圳市水务规划设计院,深圳518036

出　　处：《中国农村水利水电》2013年第11期139-142,共4页China Rural Water and Hydropower

基　　金：国家自然科学基金资助项目(51279062);中央高校基本科研业务费专项资金资助(13XS24;13QN22;13XS22;13XS23)

摘　　要：水厂逐日需水量预测作为城市供水优化调度的基础,满足水厂需水量是供水系统调度的基本目标,因此准确预测需水量过程是进行供水系统优化调度的前提条件。通过分析深圳市主要水厂日需水量的变化规律,提取影响水厂逐日需水量关键因素,以时间序列模型ARMA(1,1)为基础,嵌入自适应机制进行滚动预测,提出水厂逐日需水量过程预测方法。实例应用表明ARMA模型能够适用于深圳市水厂逐日需水量预测,预测平均相对误差小于10%,预测值与实测值分布规律基本一致,预测精度较高。Water plant daily water demand forecasting is the basis for the optimization of city water supply dispatching, to meet the water plant requirement is the basic goal of dispatching of water supply system, therefore the accurate prediction process of water de- mand is a prerequisite for the optimal scheduling of water supply. Through an analysis of the changes of water daily water demand of Shenzhen waterworks, the extraction about the key influencing factors for water plant daily water demand forecasting, with the time series model ARMA （1,1） as the foundation, at the same time embedded adaptive prediction mechanism, the prediction method of water daily demand process is put forward. The examples show that the ARMA model can be applied to Shenzhen City＇s daily water demand forecast, the average relative error of forecast is less than 10%, predictive value distribution is basically consistent with the measured values, prediction accuracy is higher.

关 键 词：时间序列模型 日需水量 预测 

分 类 号：TU991.3[建筑科学—市政工程]