摘要
神经网络模拟采用三层前馈网络模型 ,输入层有 3个结点 ,分别代表 3 0min降雨强度、降雨历时和降雨量 ,表示冲刷外因 .输出层只有一个结点 ,代表坡面冲刷量 ;隐层结点数采用 7个 ,即土压实度、坡度坡长、径流流速、径流流量、土颗粒粘聚力、植被指数和人为干扰等 ,为影响冲刷量的主要因素 .为了验证模型 ,在连徐高速公路路堤边坡用SR型人工降雨设施进行冲刷试验 ,得到冲刷量实测资料 ,将通过模型计算的冲刷量值与之比较 ,显示了模型具有较好的模拟预测效果 .
The model's first layer consists of 3 nodes, respectively representing the 30 minutes raining intensity, duration, and volume as the external factors. The third layer applies one output node, denoting the slope erosion. There are 7 nodes in the hidden layer. They are compaction, slope length and gradient, flow velocity, flow volume, viscidity, vegetation index, and jamming. In order to verify the mode, the SR artificial rainfall simulator was utilized to conduct a series of experiments at Pizhou segment of Lian Xu freeway. The comparison analysis between the data collected in field and simulated by the model showed an ideal forecast effect.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2002年第6期960-963,共4页
Journal of Southeast University：Natural Science Edition
