摘要
黄河干流大型水库的修建改变了进入下游河道的水沙过程,对河道演变具有显著影响。基于滞后响应理论,运用常规水沙因子拟合、逐步回归分析及BP神经网络等多种方法,分别建立了黄河下游典型断面平滩流量与水沙过程的多因素响应关系,对比分析了各方法针对平滩流量的计算效果。结果表明：常规水沙因子拟合可以5年滑动平均来反映前期水沙过程对平滩流量的影响;考虑平滩流量的多影响因子,逐步回归分析最终引入5年滑动全年流量和当年最大洪峰流量两个主要因子,平滩流量受水量的影响更为突出。就总体计算精度而言,BP神经网络〉逐步回归〉常规水沙因子拟合,但BP神经网络对于平滩流量极值点的预测效果较差,需要更大范围数据的训练。
Reservoir operation schemes could influence flow-sediment conditions in the downstream of river, and cause large impact on the response of channel. Based on delayed response theory, the nonlinear regres- sion, stepwise regression and BP neural network have been used to study the response relationship between bank-full discharge and the process of flow-sediment in the Lower Yellow River. Results show that the bank- full discharge is fitted with the flow and sediment coefficient based on 5 years of moving average values. Con- sidering multiple factors, the stepwise regression introduces the 5 years of moving average flow and peak flow as two main factors, in which the bank-full discharge is mainly related to the flow. In terms of overall predic- tion accuracy, the BP neutral network model 〉 stepwise regression 〉 common fitted formula, but the BP model is weak in prediction of extreme flow, and more data are needed to train the model.
作者
陈琳
胡春宏
陈绪坚
CHEN Lin;HU Chun-hong;CHEN Xu-jian(State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin,China Institute of Water Resources and Hydropower Research,Beijing 100048,China)
出处
《泥沙研究》
CSCD
北大核心
2018年第4期1-7,共7页
Journal of Sediment Research
基金
国家重点研发计划课题(2016YFC0402408)
关键词
黄河下游
平滩流量
水沙过程
滞后响应
逐步回归
BP神经网络
Lower Yellow River
bank-full discharge
process of flow-sediment
delayed response
stepwise re-gression
BP neutral network
