摘要
降雨径流问题的相关关系研究通常采用统计学和水文模型方法，本文尝试采用具有强大非线性特点和容错能力的人工神经网络（ANN）技术，以福建福清龙江小流域为研究区域，利用长时间序列的降雨和水位资料作为研究数据，建立反向传播神经网络（BPNN），对降雨和水位的相关关系进行模拟，结果与实测数据较为吻合，达到了良好的效果。BPNN可推广应用于其他自然流域或气温、降雨预测等其他复杂的非线性问题中。
Statistical method or hydrological model commonly be applied for researching of correlation between precipitation and runoff. While this study applied artificial neural networks （ANN） which is known for its strong nonlinear and fault tolerance feature, to construct the back propagation neural network （BPNN）, and used precipitation and water level data of Longjiang river as input data, to simulate the correlation of the two factors. The results indicate that ANN can predict the water level well and the correlation coefficient is good enough to reflect changing of water level. BPNN can be applied in a natural river watershed or for other complicated nonlinear problems, such as air temperature and precipitation prediction.
