摘要
为解决城市用水量预测中单一方法预测精度不高的问题,建立了灰色径向基(RBF)神经网络组合模型。对比实验结果表明,灰色GM(1,1)模型、RBF神经网络模型和灰色RBF神经网络组合模型的平均相对误差分别为2.122 2%,1.256 2%和0.682 1%。与灰色GM(1,1)模型和RBF神经网络相比,灰色RBF神经网络组合模型充分发挥了灰色系统的贫乏数据建模和RBF神经网络的高度非线性映射能力的双重优势,具有较高的预测精度,更适合用于城市用水量预测。
In order to overcome the low forecasting precision by a single method, a new combination model based on grey-RBF neural network was developed for forecasting of urban water consumption. The result of comparison tests showed that the average relative errors of grey GM （ 1,1 ） model, RBF neural network model and grey-RBF neural network combination model were 2. 122%, 1. 256% and 0. 682% respectively. Compared with grey GM （1,1） model and RBF neural network model, the grey-RBF combination model could bring into full play the double-edged advantages of grey system constructing forecasting model with poor information and a highly nonlinear mapping uniquely of RBF neural network. It was effective with the advantage of high prediction precision, and suitable for the prediction of urban water consumption.
出处
《供水技术》
2011年第4期34-37,共4页
Water Technology
基金
国家自然科学基金资助项目(61074069)
关键词
灰色预测
RBF神经网络
组合模型
用水量预测
grey prediction
RBF neural network
combination model
prediction of water consumption
