Experimental study of an optimally tilted solar PVT system combined with a thermal energy storage supported by performance modelling through probabilistic Gaussian process regression

Today, green energy generation has become the path to sustainability. Photovoltaic (PV) panels are key contributors to solar energy harvesting. The current study focuses on enhancing the efficiency of solar PV systems for power generation by integrating front-side cooling with thermal energy storage (TES), which produces hot water as a by-product for low-temperature applications. An experimental PVT-TES setup is developed at an analytically optimised tilt to boost solar radiation. The performance of the PVT–TES system is compared to that of a uncooled PV panel. The PVT–TES system is modelled using a Gaussian Process Regression (GPR) method and evaluated against the SVR (Support Vector Regression) and ANN (Artificial Neural Network) models to predict the PVT unit’s performance under varying unknown weather and to assess their impact on overall efficiency, aiming to develop a more efficient system. The proposed PVT–TES system provided a daily average power output of 30–33 W and an electrical efficiency of 16–17%. The PVT system shows a 5.35% power gain, 5.44% higher electrical efficiency and a 5.09% drop in surface temperature relative to the uncooled PV panel. An 8% TES temperature rise over ambient shows the PVT-TES systems, heat recovery potential. The GPR model effectively modelled the relationship between PVT–TES performance measures (power output, TES temperature) and input variables (solar irradiation, ambient temperature, panel surface temperature) with minimal prediction errors of 1.4W and 0.7036°C, while the SVR and ANN models achieved power prediction RMSEs of 2.7249W and 1.6807W, and TES temperature prediction RMSEs of 3.2197 and 0.9970°C, respectively.