Enhancing energy efficiency and profitability in microgrids through a genetic algorithm approach, analyzing the use of storage systems

Abstract

Due to intermittent, renewable energy systems struggle to meet demands efficiently and reliably. This research is rooted in photovoltaic systems, incorporating demand response optimization via genetic algorithms, generation forecasting using an artificial neural network, and integrating a storage system, looking for the optimal configuration to increase efficiency and system profitability. The investigation analyzes sunny and clouded seasons. Four scenarios are considered and compared to the baseline case. The baseline is the first scenario, which involves photovoltaics and a grid. In the second scenario, optimization focuses on photovoltaic generation using neural networks and incorporates demand response. The third scenario enhances the first one by introducing batteries. The fourth scenario refines the second one by including batteries. The simulations are developed using MATLAB, and an economic analysis utilizing HOMER is conducted. The findings reveal that the proposed artificial neural network exhibits superior root mean square errors at 443.62 W for photovoltaic forecasting, with an R-value of 0.9751. Applying the genetic algorithm results in a 15 % increase in self-consumption and a 52 % reduction in imported energy costs. The results showcase the model’s ability to minimize grid dependency and enhance efficiency. Economically, scenario two is the most favorable, achieving a payback of 3.6 years.