Neural Network Based Field Oriented Control for Doubly-Fed Induction Generator

This paper discusses the use of the concept of classical and predictive direct power control for shunt active power filter. These strategies are used to improve the active power filter performance by compensation of the reactive power and the elimination of the harmonic currents drawn by non-linear loads. A theoretical analysis followed by a simulation using Matlab/Simulink software for the studied techniques have been established. Moreover, two tests bench have been carried out using the dSPACE card 1104 for the classic and predictive DPC control in order to evaluate the studied methods in real time. Obtained results are presented and compared in the present paper to confirm the superiority of predictive technique. To overcome the pollution problems caused by the consumption of fossil fuels, renewable energies are the alternatives recommended to ensure green energy. In the same context, the tested predictive filter can be easily supplied by a renewable energy source that will give its impact to enhance the power quality.

Direct torque control of asynchronous motors has good dynamic performance of torque and flux. However, the use of hysteresis controllers leads to a variable switching frequency. This lack of control over frequency is the origin of torque and flux ripples. To overcome these problems, this article proposes modified fuzzy direct torque control for an induction motor. In addition, a three-phase pulse width modulation rectifier controlled by the fuzzy direct power control is used. This rectifier draws a sinusoidal current from the grid and operates with a power factor close to unity. The pulse width modulation rectifier feeds the inverter of the direct torque control, ensuring a speed control loop. The proposed direct torque control–direct power control method is investigated based on theoretical analysis, computer simulation, and experimental validation that provide acceptable results.