根据江苏辐射沙洲东大港水道4^#站位连续2个潮周期的流速、含沙量及水深的时序测量资料,取时序测量(计算)数据中的前21组数据为学习样本,后5组数据为检验样本,建立了该潮流水道的4种输沙BP神经网络动力模型。验证表明,运用BP神经网络模型可以建立精度较高的水体含沙量非线性动力关系,并可利用建立模型进行相应问题的预测计算。
Based on the field data of flow and suspended sediment at No.4 station in Dongdagang tidal channel, the Back Propagation (BP) model of artificial neural network is applied to predict sediment concentration and its transport in No.4 station. Four non-linear relationships between sediment, sediment transport and its affecting factors are discovered. The simulated and predicted results reveal that the BP model not only possesses high accuracy of fitness but also attains precise prediction as well.