RBF神经网络在阿拉尔垦区需水预测中的应用

利用RBF神经网络,建立了阿拉尔垦区需水量预测模型。选取农业用水灌溉定额、工业用水重复利用率、城镇生活人均日需水量、农村生活人均日需水量作为模型输入,农业、工业、城镇生活、农村生活需水量作为输出。将2001—2007年用水量数据作为训练样本,用2008—2009年用水量数据对模型进行检验。在农业、工业、城镇生活、农村生活4类需水量中,2009年工业需水量预测的相对误差最大,为-16.24%,总需水量的最大误差仅为1.80%,取得了较满意的结果,表明RBF神经网络模型用于该区需水量预测是可行的。

Application of RBF Neural Network of Water Demand Forecasting in Alar Reclamation Area

By using RBF neural network,this paper established the Alar Reclamation Area water prediction model.It selected agricultural water irrigation quota,industrial water reuse ratio,daily water demand per capita in urban and rural as model input and water demand in agriculture,industry,urban life and rural as output.It took 2001-2007 water consumption data as the training sample,and 2008-2009 data as test data.In agriculture,industry,urban life,and rural life output,the relative error of industrial water demand forecast in 2009 was the largest,which was-16.24%,the total amount of water needed to the maximum error was only 1.80%,showing the model achieved satisfactory results,which means that RBF neural network model for water demand prediction in the area was feasible.