Implementation of ANN-based Selective Harmonic Elimination PWM using Hybrid Genetic Algorithm-based optimization

Abstract

A PWM technique with Selective Harmonic Elimination (SHE) is used to control fundamental harmonic and eliminate harmonics of chosen lower-order in voltage source inverters (VSI). Therefore, this PWM technique requires the determination of the optimum switching angles by solving the nonlinear equation set. The determined angles are recorded on a look-up table to generate PWM signals in real-time systems. The paper proposes two Artificial Neural Networks (ANN) based solution for determining angles and generating PWM signals. ANN generates optimum switching angels for all modulation index between 0 and 1.20 because of it has learning capability differently from the look-up table. Primarily, the optimum 11-switching angles for three-phase two-level inverter are determined by using offline Hybrid Genetic Algorithm (HGA). The first ANN was trained by the data obtained from HGA to calculate the switching angles without using a look-up table. Second ANN was trained by using these switching angles to generate PWM signals. The ANN-based SHEPWM was designed to obtain inverter output voltage which has a bipolar waveform with quarter-wave symmetry. The algorithm of ANN-based SHEPWM is performed by using TMS320F28335 Digital Signal Processor (DSP). The experimental results related to dc-link voltage, inverter output voltage and load current are measured for different Ma by using scope and power quality analyzer. The waveform of inverter output voltage are also analyzed with FFT for an induction motor load. The low-order harmonics are successfully eliminated by proposed ANN based SHEPWM. (c) 2016 Elsevier Ltd. All rights reserved.