Compensation Current of Active Power Filter Generated by Artificial Neural Network Approach

The semiconductor switches are presented in many applications and they can be considered the main source of harmonic distortion presented in the electrical power system. The use of filters - active or passive - has played an important role in order to minimize the harmonic effects injected in the power system. The proposal of this work is to present an alternative approach to estimate the harmonic content of a single-phase system with non-linear loads. It uses artificial neural networks to determine the compensation current. The system is composed of AC single-phase controllers and parallel active power filter. Simulation results are presented to validate the proposed approach