Robust Fault Detection Using Subspace Aided Data Driven Design

The paper deals with the design of robust fault detection system using subspace aided data driven techniques. Because of unavailability of system matrices for the complex processes, a new algorithm has been proposed to identify a robust parity vector directly from the process data. The identified parity vector is used to construct a residual generator having robustness against the process and sensor noises and increased sensitivity to actuator and sensor faults. The performance of the proposed fault detection scheme has been analyzed by simulation studies on a coupled liquid three tank system (CLTS).