فهرست نشریات علمی دسترسی آزاد ایران

Modern energy systems are facing growing complexities, including the integration of renewable resources, the expansion of decentralized networks, and dynamic changes in consumption patterns. While traditional physics-based models may perform well under steady conditions, they exhibit clear limitations when confronted with uncertainties and real-time variations. In this context, data-driven and machine learning approaches have emerged as innovative tools that, by leveraging historical and real-time data, enable the analysis of nonlinear relationships and the identification of hidden patterns. The purpose of this study is to examine and compare two categories of artificial intelligence models in energy systems: supervised learning and unsupervised learning. The findings indicate that each category has its own strengths and limitations: supervised learning models are effective in load forecasting and energy generation prediction, whereas unsupervised learning models are valuable for pattern discovery and anomaly detection. The novelty of this paper lies in presenting an integrated analytical framework for comparing the applications of these models in energy systems, addressing both practical applications and theoretical challenges. Despite significant progress, a key research gap remains: the need for scalable and transparent models that can ensure both predictive accuracy and interpretability. The results show that while each approach individually addresses part of the requirements of energy systems, combining them in semi-supervised methods or hybrid frameworks can be an effective step toward improving efficiency, resilience, and sustainability. This advancement not only contributes scientifically but also leads, in practice, to optimized resource management, cost reduction, and enhanced grid security.