Detection of Soft Sensor Fault Using EKF Algorithm for Two Tank Interacting System

Abstract

This paper deals with the identification of fault acting in the highly non-linear two-tank interacting system. This two-tank interacting system is reported with a sensor fault that is acting on the system which makes the system to be highly unstable. For detecting the fault, the estimation algorithm is used. Here the Extended Kalman Filter algorithm is used for identifying the sensor fault acting in the system. In the EKF algorithm residual is generated that indicates the presence of the fault. The residual that is generated gives the difference between the actual and the predicted measurement. This residual generation shows the magnitude of fault that is present in the system and also the nature of the fault that is acting in the system. This EKF based detection of fault indicates the occurrence of fault at the instant of time.