Expiremental validation of a fault tolerant control of induction motor using a voting algorithm

In this paper, we describe a method for designing an active fault tolerant control system for speed sensor fault of an induction motor. The aim of this work is to develop an algorithm receiving multiple outputs from different version (sensor or observers) to choose between the most probably adequate for such situation (nominal or faulty). This algorithm is called maximum-likelihood voting (MLV) algorithm. Two adaptive observers are used (Luenberger and sliding mode observers). The proposed approach is verified by experiment results.