摘要
本文运用复杂非线性回归方法,采取量纲归一化方式,选择采伐面积、采伐量、降雨量、年均含沙量、粒径等五种因子,建立了某流域年平均含沙量的预测模型。该复杂非线性回归方法的拟合合格率高达100%;在预测样本中,误差较现存的BP神经网络、BP、PP回归有了较大的改善,相对误差最低仅为0.0128%,有着非常好的精度,该方法可以为以后定量开展河流年均含沙量预测研究提供参考。
In this paper,the prediction model of annual sediment concentration is established by using complex nonlinear regression.The prediction model of average sediment concentration in a basin is established by using six factors such as felling area,felling amount,rainfall,average annual sediment concentration and grain size.The qualified rate of the complex nonlinear regression is up to 100%.In the prediction sample,the error is greatly improved compared with the existing BP neural network,BP and PP regression,and the lowest relative error is only 0.0128%,which has a very good accuracy.This method can provide a reference for the quantitative prediction of river annual sediment concentration in the future.
作者
于世龙
杨奉广
Yu Shilong;Yang Fengguang(College of Water Resource&Hydropower,Chengdu510100,China)
出处
《吉林水利》
2022年第8期6-14,共9页
Jilin Water Resources
基金
国家自然科学基金(51979180)。
关键词
流域
年均含沙量
非线性回归
预测
river basin
average annual sediment concentration
non-linear regression
predict
