This study explores the performance augmentation of a solar adsorption desalination system (SADS) powered by a hybrid solar thermal system with evacuated tubes and photovoltaic thermal (PV/T) collectors, both numerically and experimentally. The system concurrently generates both electrical power via the PV/T system and desalinated water through the SADS. The cooling of photovoltaic cells is achieved through the utilization of chilled water produced during the desalination process of the SADS, which aims to enhance the electrical efficiency of photovoltaic cells and optimize the overall utilization of the adsorption distillation cycle. The SADS is examined under different operational scenarios and thoroughly assessed in terms of specific cooling power, specific daily freshwater product, and the coefficient of performance (COP). More so, a hybrid machine learning framework integrating an adaptive network-based fuzzy inference system (ANFIS) fine-tuned through the utilization of a manta ray foraging optimization algorithm (MRFOA) is also developed to predict the performance parameters of the SADS. The experiments indicated that the modified PV/T system improved the PV efficiency by 18 % at peak time, where the yielded electrical power and electrical efficiency reached 112 W/m2 and 11.13 %, respectively. The specific freshwater product (SWP), specific cooling power (SCP), and COP of the SADS are estimated as 53.3 L/ton-cycle (6.30 m3/ton-day), 153 W/kg, and 0.25, respectively. Furthermore, Furthermore, the simulated findings showed that the deterministic coefficient and RMSE of the predictive SWP are 0.989 and 1.314 for ANFIS-MRFOA and 0.965 and 2.323 for typical ANFIS, respectively. Hence, the ANFIS-MRFOA exhibited the highest prediction accuracy and can be deemed a potent optimization tool for forecasting the energetic performance of adsorption distillation systems.