DEVELOPMENT OF A SPEED SENSORLESS INDUCTION MOTOR DRIVES USING AN ADAPTIVE NEURO-FUZZY FLUX OBSERVER S. CHEKROUN *† , M. ZERIKAT ** , A. MECHERNENE * , N. BENHARIR ** * Dept. of Electrical Engineering, University of Science and Technology of Oran, Algeria ** Dept. of electrical engineering, Polytechnic School of Oran, Algeria Abstract In this paper, we propose an adaptive neuro-fuzzy inference system for high performance induction motor drive. The simultaneous observation of rotor speed and stator resistance in induction drive is obtained through a neuro-fuzzy observer trained with a backpropagation algorithm. The dynamic performance and robustness of the proposed neuro-fuzzy adaptive observer are evaluated under a variety of operation conditions. The suggested approach is designed and simulated in the laboratory and its effectiveness in tracking application is verified. The results have shown excellent tracking performance of the proposed speed sensorless control system and have convincingly demonstrated the usefulness of the hybrid neuro-fuzzy flux observer in high performance drives with uncertainties. Keywords: Induction Motor, Sensorless Control, Adaptive flux observer, ANFIS, Estimation 1. Introduction The sensorless control of induction motor drives based on the properties of the observability constitutes a vast subject, and the technology has further advanced in recent years. The control of the asynchronous machine is complex because the dynamics of the machine are non linear, multivariable, and highly coupled. Furthermore, there are various uncertainties and disturbances in the system. The induction motor is controlled through field orientation technique. The field oriented control method of sensorless vector control has been generally applied to drive the induction motor vector-controlled induction motor drives have been widely † Corresponding Author: schekroun@hotmail.fr University of Science and Technology of Oran, Algeria used in high-performance applications. Conventional vector control methods [1] require motor speed as a feedback signal. To obtain the speed information, transducers such as shaft-mounted tachogenerators, resolvers, or digital shaft position encoders are used, which degrade the system's reliability, especially in hostile environment. However, the speed accuracy is generally sensitive to model parameter mismatch if the machine is loaded, especially in the field-weakening region and in the low- speed range. The parameter contributing to this variation is [2]: · Rotor resistance variation with temperature, · Stator resistance variation with temperature, · Stator inductance variation due to saturation of the stator teeth. Conventional speed-sensorless flux estimators, such as the speed-adaptive full- order ux observer [3], are based on the standard dynamic motor model. Performance comparable to that of drives equipped with the speed sensor can be achieved in a wide speed and load range. However the application of the Neuro-Fuzzy observer have been successfully used for a few numbers of non linear and complex processes, ANFIS are robust and their performances are insensible to parameter variations contrary to conventional observer [4]. This work deals with sensorless control of induction motor drives and in particular with the stator resistance and rotor speed estimation by means of Adaptive flux observer and Adaptive Neuro-Fuzzy Inference System [5], based on the fundamental dynamic model of the induction machines. This paper is organized as follows: the adaptive flux