基于RBF的组合模型在水文流量预报中的应用研究
Research on the Application of RBF Based on Combined Models in Hydrological Flow Forecasting
韩雪强1
作者信息
- 1. 河北省承德水文勘测研究中心,河北 承德 067000
- 折叠
摘要
针对目前水文流量预报准确度不佳问题,研究提出基于径向基函数(Radial Basis Function,RBF)神经网络与长短期记忆(Long Short-Term Memory,LSTM)神经网络的组合模型,用于提高预报的准确性,并基于最小二乘法,提出流量预报实时校正方法.结果表明,相较于RBF与LSTM单模型,RBF-LSTM预报30 天后流量的误差分别降低60.87%和65.27%;相较于校正前,校正后RBF-LSTM预报误差减小48.15%,验证了研究所提方法的有效性.研究结果可为提升水文流量预报准确度以及防汛抗洪工作提供参考.
Abstract
In view of the poor accuracy of hydrological flow forecast at present,a combined model based on radial basis function(RBF)neural network and long short term memory(LSTM)neural network is proposed to improve the accuracy of forecast.And a real-time correction method for traffic prediction was proposed based on the least squares method.In the experimental results,compared to RBF and LSTM single models,the error of RBF-LSTM in predicting traffic after 30 days decreased by 60.87%and 65.27%,respectively;compared to before correction,the RBF-LSTM prediction error after correction decreased by 48.15%,verifying the effectiveness of the proposed method in the study.The research provides reference for improving the accuracy of hydrological flow forecasting and flood prevention and control work.
关键词
水文流量/RBF/LSTM/最小二乘法Key words
hydrological flow/RBF/LSTM/least square method引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据