Neural network and fuzzy logic controllers for three-phase three-level shunt active power filter

This paper describes the application of neural network and fuzzy logic based artificial intelligence techniques for the controller design for three-phase three-level shunt active power filter. The fuzzy logic based controllers are used for the regulation of dc voltage and current harmonic cancellation. The neural network is used for obtaining harmonics from non-linear load current. The instantaneous power theory is applied for reference current calculation and sinusoidal pulse width modulation is used for gate signals of active filter power circuit. The complete design procedure with mathematics for each component is described in detail. The results demonstrate the efficiency of fuzzy logic controller. The total harmonic distortion in current is being used as assessment criteria. The harmonic distortions obtained with PI controller and artificial intelligence based controllers are compared and are well within the IEEE519 std. limits. The simulation is done in MATLAB/SIMULINK for diode bridge rectifier with RL/RC as nonlinear loads, using power system and fuzzy/neural toolbox. The simulation results confirm that the proposed controller scheme significantly improves the performance by minimizing the harmonic content in non-linear load current. The proposed method is verified through experimentation by developing prototype with dSPACE1103.