Selective harmonic elimination in multi-level inverters by using neural networks

Abstract

In this study, it is aimed to eliminate the harmonics selected by the selective harmonic elimination (SHE) method in a 7-level cascade multilevel inverter using artificial neural networks (ANNs). A control algorithm has been developed in which the 3rd and 5th harmonics or 5th and 7th harmonics can be eliminated according to the selection while adjusting the output voltage amplitude of the inverter. The required switching angles for SHE are calculated in real time using ANN. These angles were first obtained offline training of ANN using Newton-Raphson method. ANN was trained in MATLAB® environment according to the obtained data. The resulting ANN algorithm and practical implementation using the STM32F429 ARM microcontroller® and inverter switching was provided. Experimental results of the system with RL load were tested.