计算机工程

• 软件技术与数据库 • 上一篇    下一篇

基于控制因子ADL模型的短期水位预测方法

董文永,盛康   

  1. (武汉大学计算机学院,武汉 430072)
  • 收稿日期:2015-03-16 出版日期:2016-03-15 发布日期:2016-03-15
  • 作者简介:董文永(1972-),男,教授,主研方向为演化计算、机器学习;盛康,硕士。
  • 基金项目:

    国家自然科学基金资助面上项目(61170305)。

Short-term Water Level Prediction Method Based on Control Factors ADL Model

DONG Wenyong,SHENG Kang   

  1. (Computer School,Wuhan University,Wuhan 430072,China)
  • Received:2015-03-16 Online:2016-03-15 Published:2016-03-15

摘要:

为有效提高水位预测精度,利用自回归分布滞后模型,结合站点水深调控计划、水位站流量因素等控制因子及其他相关站点水位信息,提出一种通过分析站点水位时间序列进行预测的方法。针对水位时间序列的特点,从模型选择、模型建模、模型实现开展研究。将该模型与其他常用时间序列预测模型应用于沙市水位站提前一天的水位预测实验及预测时间的扩展性实验,并对实验效果进行分析,结果表明,该模型能较好地拟合水位的变化趋势,提高模型预测的精确度。

关键词: 自回归分布滞后模型, 时间序列, 短期预测, 相关性分析, 神经网络

Abstract:

In order to effectively improve the prediction accuracy,this paper proposes a prediction method by analyzing the time series of relevant water level station,which is based on Autoregressive Distributed Lag(ADL) model,combining with the control factors of water flow and the other relevant site location.According to the characteristics of the time series of water level,an investigation is conducted from the model determining,modeling and implementing.The model is applied in the one-day ahead prediction of water level in Shashi water level station with several other commonly used time series forecasting models,and in the tests on the expansibility of prediction length.Experimental results show that the proposed model can well fit the trend,and achieve high prediction accuracy.

Key words: Autoregressive Distributed Lag(ADL) model, time series, short-term prediction, correlation analysis, neural network

中图分类号: