Neural network – genetic algorithm optimization of a hybrid renewable energy system (HRES) for a primary school in a rural area

The main attention of the present paper is producing the energy demand of a primary school located in a remote area of the eastern province of Iran, Zabol. As remote districts have less access to the grid electricity, the energy demand of the primary school has been generated through renewable resources. Therefore, a hybrid renewable energy system (HRES) comprising of PV panels, wind turbines, as the generator of energy, hydrogen energy storage system, as energy storage, and batteries, as backup energy storage, is proposed. Afterward, an artificial neural network (ANN) has been trained based on simulated HRES to predict required grid energy and loss of power supply probability (LPSP). Then, trained ANN has been optimized with the genetic algorithm to find the lowest Life cycle cost (LCC), highest LPSP, and lowest grid power. The results indicated that system configuration which is comprised of 848 PV panels, 68 wind turbines, 30 batteries, a 152.215 kWh electrolyzer can have the most optimum LCC, LPSP and grid power. The mentioned system has LPSP of 77.29 %.