Implementation of Algebraic Method Based Selective Harmonic Elimination of Multilevel Inverter Using Artificial Neural Networks

Selective harmonics elimination (SHE) is one of the prominent technique that is used to produce required fundamental voltage of multilevel inverters with elimination of selective harmonic orders. The SHE equations are nonlinear equations and are also termed transcendental equations owing to the fact that these equations contain trigonometric terms. Attempt has been made in previous works to solve the SHE equation using broadly three techniques numeric, heuristic and algebraic. But solving SHE equations with minimum computational time and also maintaining minimum total harmonics distortion (THD) in output voltage still remains a challenging task. This is required for real time implementation of closed loop control of multilevel inverter. In this paper an artificial neural network switching control is designed for seven level multilevel inverter. The artificial neural network is trained with solutions of SHE obtained from algebraic method. The training set is collected from algebraic method because it gives best solutions of switching angles for minimum THD. The proposed neural network based switching control system is implemented on Simulink platform. The proposed neural network based switching control can easily be extended for real time closed loop control of applications like grid connected PV cells and variable frequency drives.