基于时间序列预测修正“新息”的测量数据的野值点去除  

Real Time Outliers Detection Based on Kalman Filter and Time Serial

作　　者：史兆良 王鹏[2] 

机构地区：[1]海军七零二厂 [2]安徽合肥电子工程学院,部队96161

出　　处：《电子技术（上海）》2008年第7期68-70,共3页Electronic Technology

摘　　要：在实际测量中,由于野值的影响,Kalman滤波新息的特性遭到破坏,滤波不再准确甚至发散。针对此一现象,提出了基于"新息"序列和时间序列预测联合修正的方法对测量数据进行处理的新算法。该算法运用"新息"序列进行野值点判别,利用时间序列观测的方法对野值点处的"新息"进行修复。仿真证明,该算法可使状态估计与野值点判别同时进行,并能很好的抑制滤波的发散。The innovation of the Kalman filter is often damaged because of the outliers in the practice and a new algorithm based on innovation and time serial predicting is proved in order to solve this question.The algorithm uses Kalman filter to detect the outliers and predicts the innovation in the outliers.And the simulation results show the algorithm can estimate the statement and detect the outliers at the same time and get correct result.

关 键 词：时间序列 KALMAN滤波 新息 测量数据 野值 

分 类 号：TN713[电子电信—电路与系统]