Robust Sensor Fault Estimation for LPV systems: Application to Quadrotor UAV

This work is dedicated to handling the problem of quadrotor sensors fault diagnosis where the vehicle model is expressed in an LPV framework and accompanied by a robust self-scheduled LPV controller which minimizes the quadratic performance H-infinity norm. Afterward, a new observer scheme including a virtual residual signal beside an auxiliary output is introduced to achieve quadrotor sensor fault estimation and isolation. Through the synthesis of such an observer gains assignment, some structural conditions are discussed to guarantee exact or at least asymptotic convergence of the residual to the fault followed by an enhanced H-/H-infinity approach when the decoupling conditions are not satisfied. Finally, the efficiency of the developed approach in fault estimation is demonstrated by applying it in simulation on a quadrotor UAV subjected to sensor faults.