Adaptive harmonic elimination in a five level z-source inverter using artificial neural networks

In this paper an adaptive harmonic elimination method is proposed for 5-level Z-source inverter (ZSI). The main objective of harmonic elimination method is to eliminate low order harmonics and satisfies fundamental component by solving nonlinear equations. For any DC voltages variation, this nonlinear equation must be solved to obtain switching angles and shoot through duty cycle of inverter. Genetic algorithm is used in this method to obtain switching angles and shoot through duty cycle of ZSI offline for any input DC voltages. Then an artificial neural network is trained according dataset obtained offline in GA to determine switching angles and shoot through duty cycle adaptively according variation of input voltages. In this method fundamental component of AC voltage is kept constant in variation of input voltage while low order harmonics is eliminated. The proposed method is verified using simulation in MATLAB/Simulink.