The paper focuses on the estimation of the sensor faults and the state variables, using an unknown inputs observer (UIO), applied to an induction machine. A model based on LPV (linear parameter varying) systems of the machine is used where the rotation speed is considered as a variable parameter. Then based on Lyapunov theory, a feasible algorithm is explored to ensure the stability of the proposed approach. Furthermore, the observer efficiency is investigated in presence of the current sensor faults. It’s done by using the calculation of the observer gains based on the LMI (Linear Matrix Inequalities). The contribution of this study lies on the development of an extended unknown inputs observer (UIO) to estimate the sensors faults. In addition, an augmented system is constructed, using a first filter, to transform the sensor faults to actuator faults and the noise to disturbance. The performance of this method is compared either in terms of state observation errors or in terms of fault estimation. The results obtained by simulation illustrate the effectiveness of the proposed approach. Keywords: Induction motor; unknown inputs observer; linear