Mohammad Jawad ghorbani - student Sharif University of Technology Tehran, Iran
mahdi Akhbari - Assistant Professor, Faculty of Engineering, Shahed University Tehran, Iran
hossein Mokhtari - Associate Professor, Faculty of EE Engineering, Sharif University of Tech Tehran, Iran

This paper presents a high performance direct torque control (DTC) theme for the induction motor (IM). To solve those problems associated with conventional DTC, such as flux and torque ripple， variable switching frequency, inaccuracy in motor model and other parts of system. The Active Learning Method (ALM) is implemented on the DTC. In the Active Learning Method for information modeling, a method known as Ink Drop Spread (IDS) is used. The simulation results of DTC system based on ALM and the comparison of motor performance under the proposed control system with respect to those obtained under conventional DTC confirms its effectiveness and accuracy.

active learning method, direct torque control, fuzzy modeling, induction motor, reinforcement learning