Revisão bibliográfica do uso de inteligência artificial em sistemas de energias renováveis

Abstract

The widespread integration of renewable energy sources into the grid raises significant technical challenges due to their intermittent and stochastic nature. This work investigates the application of artificial intelligence as a strategic tool to manage and optimize these systems through a literature review in the Scopus database, spanning the period from 2015 to 2024. The qualitative analysis of 40 selected publications allowed for the identification of established trends in the literature, highlighting a prevalence of applications directed towards solar and wind energy. It was found that deep learning methodologies have established themselves as the primary choice for generation forecasting and predictive maintenance; more precisely, the examined studies highlight the primacy of Long Short-Term Memory neural networks in managing the non-linearity of climate data, outperforming traditional statistical approaches. Within the context of predictive maintenance, Convolutional Neural Networks stand out as the ideal model for automatic defect identification. Concomitantly, the literature indicates that meta-heuristic optimization algorithms are fundamental for maximizing efficiency in photovoltaic and wind systems. However, this integration does not occur without facing challenges: the analysis signals technical limitations regarding data quality and availability, in addition to addressing the energy paradox inherent in training large-scale models and knowledge fragmentation. Future projections indicate the indispensability of more integrated strategies. It is concluded that artificial intelligence constitutes a transformative vector capable of accelerating the energy transition on a global scale; however, cooperation among researchers, the industrial sector, and policymakers is necessary to overcome the identified challenges.