LVQ学习矢量量化神经网络模型在旱涝预测中的应用

The Application of LVQ Network Model to the Forecast of Droughts and Floods

旱涝灾害的预测对预防灾害和保护环境有着重要的作用。运用神经网络中LVQ学习矢量量化神经网络建立了旱涝预测模型,模型输入层神经元数目为3,输出层神经元数目为1,隐含层神经元数目采用实验法,最终确定为7。预测结果表明:该方法与传统的预测方法及BP神经网络相比,具有更强的容错性和鲁棒性,预测精度较高且计算简单等优点。

Forecasting droughts and floods plays an important role in preventing disasters and protecting the environment. This paper uses LVQ network to establish a model of drought and flood prediction with three importing and exporting units. The determination of concealed level neural single number uses the cut-and-try method. Seven units are used. The forecast results show that the model is more fault-tolerant and robust and the forecasting is more precise compared with the traditional forecasting method and BP networks.