基于有限测量区间TKF和χ^2检验的传感器间歇故障检测  

Detecting Intermittent Sensor Fault Based on TKF with Limited Measurement Interval and χ^2 Test

作　　者：王子健[1] 时丽平 WANG Zi-jian;SHI Li-ping(Department of Modern Education Technology Centre,Henan Quality Polytechnic,Pingdingshan 467000,China;Department of Information Engineering,Henan Quality Polytechnic,Pingdingshan 467000,China)

机构地区：[1]河南质量工程职业学院现代教育技术中心,河南平顶山467000 [2]河南质量工程职业学院信息工程系,河南平顶山467000

出　　处：《控制工程》2020年第7期1180-1185,共6页Control Engineering of China

摘　　要：在系统输出测量范围有限情况下,为了解决传感器间歇故障检测问题,提出一种基于有限测量区间TKF和χ^2检验的故障检测方法。引入3个伯努利随机变量描述系统输出的有限测量区间,推导有限测量区间条件下TKF的互协方差、输出协方差和状态协方差的计算公式,结合残差χ^2检验,检测有限测量区间TKF的残差输出。弹道滚动角速度模型的传感器间歇故障仿真结果表明,相比于标准Kalman滤波,有限测量区间TKF能够有效地估计正常情况下的状态值和检测传感器间歇故障,而标准Kalman滤波的状态估计和故障检测失效。To solve the problem of intermittent sensor fault detection under limited measurement range of system output,the fault detecting algorithm based on TKF with limited measurement interval andχ^2 test is proposed.Firstly,three Bernoulli random variables are introduced to describe the limited measurement interval of system output.Then the cross covariance,output covariance and state covariance of TKF are deduced under limited measurement interval.Finally,the residual output of TKF with limited measurement interval is detected by combining residualχ^2 test.Simulation case of sensor intermittent fault from trajectory rolling angular velocity model indicates that TKF with limited measurement interval can effectively estimate state under normal condition,and detect sensor intermittent fault.While standard Kalman filter fail to estimate state and detect fault.

关 键 词：KALMAN滤波 测量丢失 传感器故障 残差检验 伯努利随机变量 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TP273[自动化与计算机技术—控制科学与工程]