ARTIFICIAL NEURAL NETWORK BASED DIRECT TORQUE CONTROL OF DOUBLY FED INDUCTION GENERATOR

Abstract

Direct Torque Control (DTC), without inner current control loops, is capable of simplifying the structure of control system and improving its dynamic performance when applied to the Doubly Fed Induction Generator (DFIG) based wind power generation system. However, the most significant drawback of the Conventional Direct Torque Control (C-DTC) strategy is the variable switching frequency which mainly depends on the sampling frequency, the lookup table structure, hysteresis bands, and the converter switching status. This paper proposes an improved DTC strategy by using an approach intelligent artificial technique such as Artificial Neural Networks (ANN), applied in switching select voltage vector; in this way, the ripple in current and torque can be reduced. The Levenberg- Marquardt back-propagation algorithm has been used to train the neural network and the simple structure network facilitates a short training and processing times. Finally, simulation results show that the proposed ANN-DTC strategy effectively reduces the torque and flux ripples at low switching frequency, even under variable speed operation conditions.