流域年均含沙量的人工神经网络模型

本文引入人工神经网络BP网络模型对流域产沙进行了定量研究。针对小流域的土壤、地质、地貌在一定的时间范围内具有相当稳定的特性，选取采伐面积、采伐量、降雨量和年均径流量这四个代表植被、气候的主要因子对流域年均含沙量进行了建模预测。建模结果表明采伐面积、采伐量对流域产沙具有较强的延迟效应，得出的BP网络模型不仅拟合精度高，而且预测效果好。这为泥沙方面的定量研究提供了一条新的途径。

Artificial neural networks model of annual average sediment concentration in a waterhed

The Back Propagation (BP) model of artificial neural networks is applied to predict the annual average sediment concentration in a watershed. The area of tree felling, quantity of felled trees, quantity of rainfall and annual average runoff reflecting the vegetation and climate of the watershed are selected as the main factors to establish the annual average sediment concentration model for prediction. The result shows that the delay effects of tree felling area and quantity of felled tree on annual average sediment concentration are obvious. The BP model not only possesses high accuracy of fitness but also attains precise prediction as well.