灰色神经网络组合模型(GNN)在涝灾预测中的应用

Grey Neural Network Model Application on the Forecast of Flood Disaster

洪涝是对人类及社会危害较大的一种自然灾害,洪涝严重影响农业生产和生态平衡,涝灾预测已成为防灾减灾的重要内容。应用BP神经网络模型对灰色模型预测精度较低的问题进行了改进,该方法是将灰色模型的预测值作为神经网络的输入,以实际值作为输出而构成灰色神经网络组合模型(GNN)。以辽阳地区50年的年降水量作为历史数据,建立GNN涝灾预测模型。预测结果表明:该方法与传统的灰色预测方法相比提高了预测精度,这种新的信息处理和预测方法是有效可行的。

Flood disaster forecast is the important content of disaster prevention and reduction.The basic principles and methods of conventional Gray Model are presented,then the Back Propagation neural network model was used to improve Gray Model.Gray neural network model was built by use grey prediction value as the input layer of neural network,and with the real value as output layer.This paper was based on the rainfall data of 50 years in Liaoyang,then established the GNN model for flood.The forecast result showed that the model had had high forecasting precision compared with the traditional forecast method,therefore flood forecasting based on GNN was feasible.