Optimization of solar energy using recurrent neural network controller with dc-dc boost, Cuk, and single-ended primary inductor converter (SEPIC) Converters

Abstract

The pressing issue of the greenhouse effect demands strategies to reduce carbon dioxide (CO2) emissions, a detrimental gas with widespread adverse effects. The sun, as the ultimate renewable energy source, generates energy without CO2 emissions. Harnessing solar power necessitates a photovoltaic (PV) system equipped with a Maximum Power Point Tracker (MPPT) to optimize energy output. The MPPT adapts to changing environmental conditions and communicates through a Pulse Width Modulator (PWM) to an Insulated Gate Bipolar Transistor (IGBT), which alters its duty cycle to align system resistance with the load. Traditional Perturbation and Observation (P&O) algorithms struggled with environmental variations, but advanced AI-based Recurrent Neural Network (RNN) controllers enhance efficiency. This research compares RNN controllers using three data sets of 104, 201, and 1001 entries with three DC-DC converters: Boost, Cuk, and Single-Ended Primary Inductor Converter (SEPIC).