摘要
水文时间序列具有确定成分和随机成分,运用有效的数学方法提取序列样本中的各个成分,通过数学建模的思想,模拟和预测时间序列是近年来水文预报的重要发展方向。与以往过程驱动模型不同,近年来水文预报的模型主要是数据驱动模型,这类模型分为传统方法和新方法两大类。前者主要有成因分析和水文统计方法等,后者包括近年来新兴的模糊分析、人工神经网络、灰色系统分析、支持向量机等。对目前运用于时间序列预测的几个代表方法的模型特点、适用范围和不足进行了简要评述,展望了水文时间序列预测模型未来发展,期望能为水文预报工作提供帮助。
The hydrological time series is featured with certain components and random components.It is an important development direction of hydrological forecast to extract the components in the sequence samples by effective mathematical methods, and simulate & forecast the time series through the idea of mathematical modeling in recent years.Different from the previous process-driven model, the models of hydrological forecast in recent years are mainly data-driven models, on which the methods based are divided intotraditional methods and new methods.The former mainly includes cause analysis and hydrological statistics, while the latter mainly includes fuzzy analysis, artificial neural network, gray system analysis and support vector machine.This paper briefly reviews the model characteristics, scope of application and deficiencies of several representative methods currently applied to time series prediction, and forecasts the development of hydrological time series prediction models in the future. It is expected to provide assistance for hydrological forecast.
作者
程扬
王伟
王晓青
CHENG Yang;WANG Wei;WANG Xiaoqing(Hohai College,Chongqing Jiaotong University,Chongqing 400074,China;Southwest Water Transport Engineering Institute,Chongqing Jiaotong University,Chongqing 400042,China)
出处
《人民珠江》
2019年第7期18-23,共6页
Pearl River
基金
重庆市基础与前沿研究计划项目(cstc2016jcyjA0544)
关键词
水文时间序列
预测
数据驱动模型
hydrological time series
forecast
data-driven model