摘要
为提高径流预报精度,采用滑动窗口二次自回归模型进行径流中长期预报,提出自适应优化、平稳滑动窗口和均方根误差最小的综合选取模型参数的方法.实例应用结果表明:该方法不仅提高了模型的预报精度,而且保证了模型的稳定性;与人工神经网络模型相比,滑动窗口二次自回归模型的1步预报具有更高的精度,可用于中长期径流预报.
In order to improve the precision of prediction of runoff, the moving window quadratic auto-regressive model was adopted for the prediction of medium and long-term runoff. A comprehensive method for choosing the model parameters is put forward based on self-adapting optimization, smoothly moving windows and the minimum root mean square error. The results show that the present method not only improves the prediction precision, but also ensures the model stability. Compared with the artificial neural network model, the present model has higher prediction precision for its first step and ean be applied in the prediction of medium and long-term runoff.
出处
《河海大学学报（自然科学版）》
CAS
CSCD
北大核心
2009年第3期267-270,共4页
Journal of Hohai University(Natural Sciences)
基金
国家自然科学基金(4083063970471083)
关键词
中长期径流预报
二次自回归
滑动窗口
自适应
medium and long-term runoff prediction
quadratic auto-regression
moving window
self-adapting
