Fault diagnosis for a UAV using unknown input observer

In this paper a fault tolerant control strategy for a quadrotor is proposed in the presence of actuator faults. A linear mathematical model which is derived from nonlinear model by calculating the Jacobian matrix of the system is used to represent the quadrotor dynamics in MATLAB/Simulink environment. Observer-based state estimation approach which is widely used in fault tolerant area is chosen to detect and isolate faults in quadrotor system. Generalised observer scheme (GOS)-based unknown input observer (UIO) whose principle to make the state estimation error de-coupled from the unknown inputs (disturbances) is utilised to detect and isolate the actuator faults. In fault-free condition, linear quadratic tracking (LQT) is preferred to stabilise the quadrotor to obtain faster system response. When it comes to faulty case, LQT cannot handle compensation of steady-state error owing to power loss in actuator. Therefore linear quadratic regulator (LQR) with integral action is selected by fault diagnosis unit to compensate steady-state error due to actuator fault. Simulation results are presented to demonstrate the performance of the proposed fault tolerant control strategy.