JOURNAL OF LIAONING TECHNICAL UNIVERSITY

GRID-CONNECTED PHOTOVOLTAIC SYSTEM WITH ARTIFICIAL NEURAL NETWORK CONTROL USING THREE-PHASE MULTILEVEL INVERTER

Wale Akinyele, Olusegun Ajayi, Rafiu Adebisi, Oluwafemi Olowoyo

Abstract

This study explores the performance comparison of a grid-tied photovoltaic system utilizing three-phase Multilevel Inverter (MLI) topology Voltage Source Inverter (VSI) governed by an Artificial Neural Network (ANN). Multilevel inverters have undergone significant advancements, offering numerous benefits compared to classical topologies, generating alternating AC output voltage with multiple levels, resulting in reduced Total Harmonic Distortion (THD). The Multilevel Inverter (MLI) configuration utilizes diodes, switches and power sources, which can be modelled to optimize control signals with the presented controller. The grid-tied photovoltaic system efficiency relies heavily on efficient DC–AC conversion to optimize power output from PV generators amidst environmental fluctuations; a novel neural network-based Maximum Power Point Tracking (MPPT) technique has been launched. The fuzzy logic controller-based MPPT generates training datasets for the neural network MPPT. The presented procedure targets improved power quality and network responsiveness under shifting environmental and grid parameters. The presented methodology is realized on MATLAB/Simulink and benchmarked against existing techniques. Performance metrics illustrate improved performance, with reduced switching losses and lower Total Harmonic Distortion (THD).

Keywords: Photovoltaic (PV) system; voltage source inverter; artificial neural network; total harmonic distortion; fuzzy logic controller; MATLAB/Simulink; DC to DC converter; MPPT.