STUDY ON ARTIFICIAL NEURAL NETWORK FORECASTING METHOD OF WATER CONSUMPTION PER HOUR

作     者：刘洪波 张宏伟 田林 王新芳 LIU Hong-bo;ZHANG Hong-wei;TIAN Lin;WANG Xin-fang

作者机构：天津大学建筑工程学院天津300072 天津港保税区天津300456 兰州电力学校兰州730070 

出 版 物：《Transactions of Tianjin University》 (天津大学学报（英文版）)

年 卷 期：2001年第7卷第4期

页      面：233-237页

学科分类：08[工学] 0815[工学-水利工程] 

基　　金：Supported by Foundation for University Key Teacher by Ministryof Education 

主　　题：artificial neural network consumption per hour forecast BP algorithm Matlab 

摘      要：An artificial neural network (ANN) short term forecasting model of consumption per hour was built based on seasonality,trend and randomness of a city period of time water consumption *** hidden layer nodes,same inputs and forecasting data were selected to train and forecast and then the relative errors were compared so as to confirm the NN structure.A model was set up and used to forecast concretely by *** is tested by examples and compared with the result of time series trigonometric function analytical *** result indicates that the prediction errors of NN are small and the velocity of forecasting is *** can completely meet the actual needs of the control and run of the water supply system.