Machine learning, mathematical modeling and 4E (energy, exergy, environmental, and economic) analysis of an indirect solar dryer for drying sweet potato

Abstract

A developed indirect solar dryer is built and operated to dry sweet potato cubes. Since, numerous instruments have gathered experimental data to comprehensively evaluate the system's energy, exergy, environmental, and economical aspects. Additionally, four machine learning algorithms, namely Decision Trees (DT), Gradient Boosting Regression (GBR), Multiple Linear Regression (MLR), and Random Forest (RF), are evolved to forecast the solar collector's energy (RSAH,η) and exergy efficiency (RSAH,ηEX) as well as the drying chamber's mean drying temperature and exergy efficiency (DC,ηEx). In addition, ten drying kinetics mathematical models were employed to fit with sweet potato moisture ratio variation over the experiment. Also, Color and bioactive compounds were analyzed. Results show that, RSAH,η and RSAH,ηEX was 72.9 %, and 5.6 %, respectively. Storage unit thermal (ηTh.,SUPCM)and exergy efficiency (ηEx,SUPCM) were 43.4 %, and 18.4 %, respectively, the discharging lasted around 5.5 h. Theoretical drying chamber thermal efficiency (DC,ηth) was from 21.9 to 97.2 %. And av. DC,ηEx was 46.1 %. RF algorithm achieved the best results for RSAH,η, RSAH,ηEX , DC,Tmean, and DC,ηEx forecasting, because of its superior adaptability and generalization. The overall dryer efficiency was 15 % with a specific energy consumption of 4.509 kWh/kg moisture. The Page model showed the best fitting with sweet potato moisture ratio data. In addition, CO2 mitigation reached 20.2 with earned carbon credit is 56771 RMB. The economic payback period is 29.24 months, the annual total revenue is 8464 RMB and 0.7 RMB as a Return on investment.