A comprehensive review on the artificial intelligence for the development of thermal concentrating photovoltaic systems

Abstract

Under the effects of increased climate urgency and fossil fuel depletion, solar energy can become a fundamental renewable energy source. Although photovoltaics (PVs) have limited efficiency, concentrating photovoltaic-thermal (CPV/T) systems generate electricity and heat simultaneously, achieving 60 to 80 % total efficiency, considerably higher than those obtained from PVs. However, the performance of CPV/T highly depends on operational parameters such as irradiance fluctuations, cell temperature, and tracking inaccuracies, limiting their use in real world applications. Artificial intelligence (AI) techniques including deep neural networks (DNNs), reinforcement learning (RL), and hybrid algorithms solve these problems by enabling adaptive thermal regulation, fault detection, real-time optimization and precision solar tracking. This paper has reviewed recent progresses in AI, with special focus on CPV/T systems, analyzing approaches for dynamic cooling control, predictive maintenance, irradiance forecasting and system design. Experimental validations revealed that AI-driven control increased thermal stability by > 35, %decreased mirror misalignment by <= 85 %, and obtained R-2 > 0.99 for energy prediction. However, serious limitations such as climate-specific model transferability, computational constraints in off-grid settings, interdisciplinary gaps between AI and solar engineering, and data scarcity still exist. This paper reveals new opportunities to accelerate this high-impact synergy toward global renewable energy goals.