| 基于LH-BP松散耦合模型的日径流预测 |
| |
| 引用本文： | 方玉杰,万金保,罗定贵,陈春丽,邓长涛.基于LH-BP松散耦合模型的日径流预测[J].水电能源科学,2014,32(8):30-33. |
| |
| 作者姓名： | 方玉杰 万金保 罗定贵 陈春丽 邓长涛 |
| |
| 作者单位： | 南昌大学 鄱阳湖环境与资源利用教育部重点实验室， 江西 南昌 330029;南昌大学 鄱阳湖环境与资源利用教育部重点实验室， 江西 南昌 330029;广州大学 环境科学与工程学院， 广东 广州 510006;南昌大学 鄱阳湖环境与资源利用教育部重点实验室， 江西 南昌 330029;珠江水利科学研究院， 广东 广州 510611 |
| |
| 基金项目： | 国家“十一五”水污染控制与治理科技重大专项（2008ZX07526 008 03）；国家自然科学基金项目（41261097） |
| |
| 摘 要： | 鉴于基流过程对降雨不敏感,致使传统的BP神经网络日径流预测性能受到制约的问题,结合LyneHollick(LH)数字滤波算法和BP算法的优点,建立了基于LH分割基流与BP神经网络日径流预测的松散耦合模型(LH-BP)。先采用LH数字滤波算法分割出基流,再利用BP神经网络预测锦江流域四个水文站的直接径流和基流。结果表明,LH-BP耦合模型较传统的BP模型性能更优,弥补了传统的BP模型对日径流模拟与预测的不足。
|
| 关 键 词： | 基流分割 日径流预测 数字滤波算法 BP神经网络 耦合模型 |
Daily Runoff Forecasting Based on Loose Hybrid LH BP Model |
| |
| FANG Yujie,WAN Jinbao,LUO Dinggui,CHEN Chunli and DENG Changtao.Daily Runoff Forecasting Based on Loose Hybrid LH BP Model[J].International Journal Hydroelectric Energy,2014,32(8):30-33. |
| |
| Authors: | FANG Yujie WAN Jinbao LUO Dinggui CHEN Chunli and DENG Changtao |
| |
| Abstract: | The performance of traditional BP neural network is limited in daily runoff forecasing because the base flow is not sensitive to rainfall process. Therefore, a loose hybrid model called LH BP which combined the advantages of the Lyne Hollick (LH) digital filter algorithm and BP algorithm was built for daily runoff forecasting. Firstly, base flow was separated by LH digital filter algorithm. And then BP neural network was built for daily runoff forecasting of four hydrological stations in Jinjiang river basin. The results show that the hybrid LH BP model has a better performance than the traditional BP model, which makes up for the traditional BP model in daily runoff simulation and forecasting. |
| |
| Keywords: | base flow separation daily runoff forecasting digital filter algorithm BP neural network hybrid model |
| 本文献已被 CNKI 等数据库收录！ |
