摘要
影响输沙量演变规律的前期因素比较多,且难于确定和提取,这就造成对输沙量的拟合和预测精度较低。在对输沙量时间序列混沌特性分析的基础上,利用嵌入相空间来确定前期影响因子,建立了基于混沌相空间技术的BP神经网络模型。模型既能考虑到影响输沙量时间序列的动力因子,又能解决网络的输入单元数确定的困难和利用神经网络超强的非线性映射功能,通过对龙川江流域控制站月输沙量的拟合与预测表明其结果合理,预测精度较高。
There are lots of factors which influence the evolvement law of sediment discharge, and it is very difficult to be known and gained, it results in low precision of simulation and forecast. Based on analysis of chaos characteristic of sediment discharge time series, BP neural networks model based on chaos phase space is proposed to forecast the sediment discharge through embedding dimension. Considering the influence of dynamical factor of sediment discharge as well as the difficulty of calculating number of input cell, the model is provided with strong nonlinear mapping capacity and applied to simulate and forecast monthly sediment discharge of Longchuan river watershed, the outcome is reasonable with higher precision.
出处
《水力发电学报》
EI
CSCD
北大核心
2007年第1期24-27,共4页
Journal of Hydroelectric Engineering
基金
国家重点基础研究发展计划项目(2003CB415202)
四川省学术带头人培养基金项目(2200118)
关键词
河流泥沙工程学
混沌
输沙量
神经网络
相空间
river sediment engineering
chaos
sediment discharge
neural networks
phase space
