摘要
提出了一种基于支持向量回归的动态区域迁移建模方法,该方法使模型随着区域的动态迁移进行在线更新,并采用动态调整ε的方法对同一区域内的数据给予不同程度的考虑,使得当前时刻加入的新数据要比以前时刻的旧数据在建模中所占的比重大。针对LF炉钢水温度预报问题,将此方法应用于预报建模。研究结果表明,该建模方法十分有效。
The dynamic region migration modeling method based on support vector regression was proposed. In this method, the model of the system is refreshed on-line with the migrating region dynamically. An adaptive tube is applied to deal with the changes in the data of the same region that the weights of the new coming data are bigger than that of the older data. The method was employed to construct model, which is applied to Ladle Furnace＇s temperature prediction. The effective result indicates that the proposed method is a powerful method.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2009年第5期1224-1226,共3页
Journal of System Simulation
基金
北京市教育委员会重点学科共建项目(XK100080537)
关键词
支持向量回归
动态区域
建模
LF炉
support vector regression
dynamic region
modeling
Ladle Furnace
