广义回归神经网络预测天津城市用水量

Generalized Regression Neutral Network Model Applied in Predicting Tianjin Urban Annual Water Consumption

通过对1991—1994年、1996～1999年、2001～2004年、2006—2007年天津城市年用水量相关的历史数据进行分析，利用广义回归神经网络进行训练和拟合，用1995年、2000年、2005年的实际数据进行模型检验，结果证明了广义回归神经网络用于城市年用水量预测的有效性和可行性。

By analyzing generalized regression neutral network model structure and Tianjin urban annual water consumption from 1991 to 2007, the annual water consumption in Tianjin urban area was predicted by training and simulating the network. The model was tested by the actual data from the year of 1995 and 2000 and 2005. The results showed that it was feasible to predict urban water consumption by generalized regression neural network model.