PI observer gain optimization for fault detection with disturbance attenuation

In this paper, the problem of fault detection under disturbances for multi-input and multi output linear systems is investigated. The considered fault detection approach relies on the recently developed PI observer. A new algorithm is proposed to calculate the gain of the PI observer with the purpose of optimizing the fault detection performance. In the proposed approach, part of the gain matrix is determined by using the system state space model matrices to reduce the disturbance's adverse impacts on the fault detection observer. The rest of the gain matrix is determined by the pole assignment and eigen-structure optimization method. The proposed method is verified by a simulation of a DC Motor.