城市用水量预测的SIMULINK仿真技术研究  

Research on SIMULINK Simulation Technology for Urban Water Consumption Forecasting

作　　者：雷鸣[1] 张宏伟[1,2] 闫静静[1] 

机构地区：[1]天津大学环境科学与工程学院,天津300072 [2]天津工业大学环境与化学工程学院,天津300160

出　　处：《中国给水排水》2010年第15期54-57,共4页China Water & Wastewater

基　　金：天津市科技支撑重点项目(09ZCGYSF02200)

摘　　要：为保证城市供水优化运行的安全性和可靠性,提出了基于时间序列和神经网络理论的城市用水量预测的SIMULINK仿真模型。基于时间序列预测法的SIMULINK仿真模型依据回归算法确定模型参数,得到预测结果和误差,可通过调整SIMULINK模块参数提高仿真精度;在基于神经网络的SIMULINK仿真模型中,根据BP神经网络原理分别建立输入层、隐含层和输出层模型,得到预测结果和误差,可通过增加训练样本数提高仿真精度。仿真结果表明:基于时间序列和神经网络的水量预测SIMULINK仿真模型,不仅预测精度达到要求,而且还具有模块直观、参数易调和结果可视化等优点。In order to ensure the safety and the reliability of optimal operation of water supply system, the SIMULINK simulation modes based on time series and neural network for urban water consump- tion forecasting are put forward. In the simulation model based on time series, the parameters are determined based on regression algorithm, and the results and errors are obtained. Simulation precision can be improved by adjusting SIMULINK module parameters. In the simulation model based on neural network, input layer, hidden layer and output layer models are built based on BP neural network theory, and the results and errors are obtained. Simulation precision can be improved by increasing training sample quantity. Simulation results show that the SIMULINK simulation models based on time series and neural network for urban water consumption forecasting can meet the precision requirements and have the advantages of intuitive modules, easy adjustment of parameters and visual results.

关 键 词：水量预测 SIMULINK仿真模型 时间序列 神经网络 

分 类 号：TU991.31[建筑科学—市政工程]