Forecasting the potential and output of building‐integrated photovoltaic (BIPV) and traditional photovoltaic (PV) systems, including rooftop, ground‐mounted and industrial‐shed installations, has become increasingly important, as these technologies hold substantial potential for meeting a significant share of energy demand. Artificial intelligence (AI) approaches, including machine learning (ML) and deep learning (DL) models, are widely recognised as powerful tools for forecasting solar resource potential and system performance. These models play an essential role in accelerating the integration of renewable energy within urban energy planning frameworks. In this context, forecasting for BIPV–PV systems can be broadly classified into three domains: potential, power and energy (PPE). Given the rapid advances in the field of DL over the past few years, numerous studies have made targeted efforts to improve the forecasting accuracy for both BIPV–PV systems by enhancing input data quality and applying advanced, complex and hybrid models. Most of these efforts have mainly narrowed their focus to one of the three forecasting domains rather than adopting a more integrated approach. This systematic literature review (SLR) aims to provide a comprehensive review of PPE forecasting approaches to enable more robust assessment and deeper insights into the feasibility and viability of BIPV–PV systems. The review further highlights key methodological challenges, outlines limitations and offers practical guidance for researchers, policymakers and developers, while identifying emerging trends and future opportunities in AI‐based forecasting for BIPV–PV applications.
Sadeq et al. (Thu,) studied this question.