摘要
本文将稳健估计方法引入时间序列建模,提出了基于稳健估计的自回归建模方法。采用某实测边坡两个监测点连续30期数据对该方法进行了验证计算与分析,结果表明当监测序列没有异常值时,稳健与常规自回归模型的预报精度相当;而当监测序列含有少量异常值时,稳健比常规自回归模型的预报精度有较明显的提高。
Traditional steady time-series models are briefly introduced in this paper,and the robust AR model is developed based on Huber estimation.Then modeling process and forecast analysis are discussed in detail.Finally,a case is studied from the 30-period continuous monitoring point data of a slope.The computation analysis shows that the robust AR model is more effective than traditional AR model in the forecast analysis when little gross error is buried in the raw monitoring data.
出处
《测绘科学》
CSCD
北大核心
2007年第2期73-74,80,共3页
Science of Surveying and Mapping
基金
国家自然科学基金项目(50579013)
现代水利科技创新项目(XDS2004-07)
关键词
边坡监测
时序分析
稳健自回归模型
slope monitoring
time-series analysis
robust AR model
