PROCEEDINGS OF C I E I 2016 7 ONLINE OBSERVER-BASED INTER-TURN SHORT-CIRCUIT FAULT DETECTION AND IDENTIFICATION IN INDUCTION MACHINE AMAL GUEZMIL 1 , REMUS PUSCA 2* , HANEN BERRIRI 1 , ANIS SAKLY 1 , RAPHAEL ROMARY 2 , MOHAMED FAOUZI MIMOUNI 1 1 Monastir University, Res. Unit on Study of the Industrial System & Renewable Energies (ESIER), Ibn Eljazzar City, 5019, Monastir, Tunisia. 2 Artois University, EA4025, Laboratory of Electrical Systems and Environment (LSEE), Bethune, F-62400, France. Abstract: This paper presents an observer-based method for induction machine stator inter- turn short-circuit fault detection and identification. The mathematical model of induction machine with inter-turn short-circuit fault is developed. A high order sliding mode observer is used to estimate rotor flux and three-phase stator current under healthy and faulty conditions. The proposed strategy is based on calculating the difference between the measured three-phase stator current and estimated ones. That produces sets of residuals sensitive to fault. Theorical and experiment results are included to show the proposed strategy ability for detection and locating phase under fault. Keywords: Induction machine, Inter-turn short-circuit fault, High order sliding mode observer, Fault detection and identification. 1. INTRODUCTION Induction Machines (IM) are one of the most important parts for many industrial process that are widely used as electromechanical energy conversion devices due to their low cost, ruggedness and low maintenance [1-2]. Despite these superiorities, IM are prone to different internal faults. Stator winding faults is a common as well critical fault in IM, the main cause of the fault is insulation failures [3]. The Inter-Turn Short-Circuit (ITSC) fault is an initial stage of most of the stator winding failures. When an ITSC fault occurs, extremely high currents flowing in the short-circuit coil cause the propagation of destruction and unexpected shutdown of the IM. Hence, it’ s important to develop a reliable technique to detect the ITSC fault at early stage to prevent sever faults and damage. A large number of researchers have focused their efforts on the online early detection of the ITSC fault [4-8]. One of the most promising ones is model-based technique. The difference between measurement outputs and reconstructed outputs by the observer produces a set of residuals which should be sensitive to faults. In this context, observer-based methods have been proposed in literature for resolving ITSC Fault Detection and Identification (FDI) problem [9-11]. In [9], a model-based strategy for stator-ITSC detection on IM is presented. The proposed strategy is based on the generation of a vector of specific residual using a state observer. In [10], Duvvuri propose a technique based on the whiteness of innovation sequence developed by the standard extended Kalman filter. Also, a nonlinear Generalized Likelihood Ratio method is applied to identify the faulty phase along with its severity. An adaptive observer-based fault detection and diagnosis scheme is developed in [11], which can not only provide a rapid detection when stator ITSC fault occurs but also give an accurate diagnosis of the fault position and level. Thus, sliding mode techniques have been widely studied for state estimation problem * Corresponding author, email puscaremus@hotmail.com