多传感器系统噪声统计辨识的一种相关方法  

Correlation Approach of Noise Statistics Identification for Multisensor Systems

作　　者：高媛[1] 王伟玲[1] 王强[1] 邓自立[1] 

机构地区：[1]黑龙江大学自动化系,哈尔滨150080

出　　处：《科学技术与工程》2009年第1期11-15,共5页Science Technology and Engineering

基　　金：国家自然科学基金(60874063);黑龙江省教育厅科学技术项目(11521124);黑龙江省电子工程重点实验室项目(DZZD2006-16)资助

摘　　要：对于带未知噪声统计和相关噪声的多传感器线性离散定常随机系统,通过左素分解将观测过程表为两个滑动平均(MA)过程之和,利用解相关函数矩阵方程组方法得到系统的噪声方差、相关阵及互协方差的在线估计器。基于观测过程的采样相关函数的遍历性证明了噪声统计估值器是强一致的。一个两传感器带相关噪声系统的仿真例子说明了方法的有效性。For the multisensor linear discrete time-invariant stochastic systems with unknown noise statistics and correlated noises, the new measurement processes can be described by the sum of two moving average （MA） processes by a left-coprime decomposition. The on-line estimators of the noise variances,correlated matrices, and cross-covariances are obtained by solving the matrix equations for correlation functions, and based on the ergodicity of sampled correlation functions of the new measurement processes, the strong consistent estimation of the noise statistics is proved. A simulation example for a two-sensor system with correlated noises shows effectiveness of the proposed approach.

关 键 词：噪声统计信息估值器 相关方法 收敛性 强一致性 

分 类 号：O211.64[理学—概率论与数理统计]