Design and Implementation of Artificial Neural Network Controlled DC-DC Converter for Solar Cell-Based Hydrogen Production System

In this study, hydrogen production, an alternative energy source, was achieved using a solar energy system. An artificial neural network (ANN) assisted DC-DC buck converter and electrolysis system were used for the design and control of this system. The ANN assisted DC-DC buck converter was designed to achieve the desired output voltage. The electrolysis system was designed and implemented to produce hydrogen gas. A 545 W model was used as the photovoltaic (PV) panel. The maximum panel voltage was 41.80 volts, and the maximum power current was 13.93 amps. The PV panel control system uses the Espressif ESP32 S3 processor, trained with an artificial neural network algorithm, to instantly generate the desired PWM signal and drive the DC-DC buck converter system. The theoretical DC-DC buck converter output values for a 156 W system are 6.25 V and 25 A. These values were chosen to ensure the electrolysis system operates efficiently at high current and low voltage. Finally, the electrolysis system was constructed from stainless steel plates and a potassium hydroxide (KOH) solution to maintain low internal resistance, aiming to produce hydrogen gas. The circuit was first tested in MATLAB/Simulink, LTspice, and the Arduino IDE. The system design and implementation were then tested in a physical environment. Consequently, a DC-DC buck converter integrated with artificial neural networks to respond to environmental changes using renewable energy sources, including solar energy, and a proton exchange membrane (PEM) electrolysis system for hydrogen production were designed and the results were obtained.