Machine learning prediction and 4E analysis of PV/T coupled with glass drying chamber system

Abstract

In order to maximize the utilization efficiency of solar energy while meeting the drying requirements, the PV/T coupled with glass drying chamber system was firstly proposed and employed to dry lemon slices in this study. The 4E methods and ten drying kinetics models were employed to assess drying performance and drying kinetics of lemon slices, meanwhile, the color variation was analyzed. Furthermore, three machine learning algorithms were selected to forecast the drying chamber temperature, PV electric efficiency, PV/T thermal efficiency, PV/T overall efficiency and moisture ratio. The results shown that Two-term model exhibited the best description under the system and open sun drying. The SEC was 3.55 kW h/kg, and the exergy efficiency of PV/T and drying chamber were 0.9 % and 60.04 %, respectively. The system achieved 59.23 tons in CO₂ mitigation and earned 1184.69 $ in carbon credit over 30yr lifecycle. The economic payback period was 3.59 years. Furthermore, the optimized GRU model was comprehensively evaluated as the superior model, and generalization capability had also been validated with higher values of R². Finally, the values of color variation were 5.83 and 4.94 under the system and open sun drying, respectively.