基于两阶段TKF的测量死区下离散时间系统的故障估计  

Fault estimation for discrete-time systems with output dead-zone using two-stage Tobit kalman filter

作　　者：黄洁[1] 何潇[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《山东大学学报（工学版）》2017年第5期136-142,156,共8页Journal of Shandong University（Engineering Science）

基　　金：国家自然科学基金资助项目(61473163;61522309;61490701)

摘　　要：针对出现测量死区的离散系统,提出一种基于两阶段TKF的故障估计方法。引入2个Bernoulli随机向量描述输出死区,并设计了增广状态Tobit卡尔曼滤波器(augmented state Tobit Kalman filter,ASTKF)。通过两步U-V变换方法对ASTKF的协方差矩阵解耦,从而获得两阶段Tobit卡尔曼滤波器(two-stage Tobit Kalman filter,TSTKF),并且利用TSTKF解决了系统故障估计问题。对所提出方法进行仿真,并与标准卡尔曼滤波器、间歇观测下的卡尔曼滤波器进行比较,说明了该方法的可行性和准确性。The problem of estimating the fault for discrete-time systems with output dead-zone was addressed via two- stage Kalman filtering approach. Two Bernoulli random vectors were introduced to model the dead-zone effect. A twostage Tobit Kalman filter （TSTKF） was derived to solve the filtering problem. The covariance matrices of the augmented state Tobit Kalman filter （ASTKF） was decoupled by using a two-stage U-V transformation technique to obtain the TSTKF. A numerical example was provided to illustrate the feasibility and accuracy of the proposed filter in the end which was compared with both standard Kalman filter and Kalman filter with intermittent observations.

关 键 词：故障估计 数据删失 测量死区 TOBIT回归模型 递推估计 两阶段Tobit卡尔曼滤波器 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]