摘要
针对带未知参数的有色公共干扰噪声的自回归滑动平均(ARMA)模型,专门提出一种多段辨识算法:首先用递推辅助变量法获得自回归参数的局部估值,并采用平均局部的方法得到融合估值,再用带死区的GeversWouters算法和求解线性方程组得到滑动平均参数和噪声方差的局部和平均融合估值,并证明估值结果以概率1收敛于真实值。通过Matlab仿真显示单传感器得到的辨识结果均收敛于真实值,并且综合多个传感器的信息可以得到更为精确的结果。
For the unknown AutoRegressive Moving Average （ARMA） model with the colored common disturbance noise and the white measurement noise, a multi-stage information fusion method was specially presented. First, the local estimates of the Auto-Regressive （AR） paraments were obtained by Recursive Instrumental Variable （RIV） method, and the fusion estimates were obtained by taking the average of the local estimates. And then, the local and fusion estimates of Moving Average （MA） paraments and noise variance were obtained by using Gevers-Wouters algorithm to solve linear equations. The convergence of the fusion estimates was proved. By the simulation of Matlab, good convergence with the real value has been achieved by every single sensor, and the results of multiple sensors can be more accurate.
出处
《计算机应用》
CSCD
北大核心
2013年第A02期296-298,共3页
journal of Computer Applications
基金
安徽省自然科学基金资助项目(1208085MF111)
安徽省教育厅自然科学研究项目(KJ2011B123)
阜阳师范学院科研启动项目(2012FSKJ98)
安徽省博士后基金资助项目
关键词
多传感器信息融合
系统辨识
有色公共干扰噪声
自回归滑动平均模型
多段信息辨识算法
muhisensory information fusion
system identification
colored common disturbance noise
Auto-Regressive and Moving Average （ARMA） model
multi-stage information fusion method
