A comparative study on life cycle assessment and economic analysis of photovoltaic-based air heating systems based on machine learning prediction

To analyze the environmental impact of photovoltaic-based air heating systems, the photovoltaic/thermal air heater system (PV/T-AHS), flat plate air heater system (FP-AHS) and electric air heater system (E-AHS) were respectively constructed and tested for building heating. A direct coupling framework that integrates machine learning predictive outputs with life cycle assessment models was presented, the system runtime performance data predicted by machine learning was directly used for the comparative evaluation of the environmental impacts of three solar air heating systems throughout their life cycles. Furthermore, sensitivity analysis was also performed for system lifetime and region. Moreover, economic analysis was used to assess the feasibility of system. The results shown that the convolutional neural network (CNN) model was considered as the optimal model with R² of 0.99. The average daily heat and electricity prediction of three systems were obtained and applied to LCA analysis based on the optimal CNN model. The LCA results shown that PV/T-AHS demonstrated superior environmental impact compared to FP-AHS and E-AHS, with total environmental impact values of 7.826, 15.9314 and 14.6408 mPt/year, respectively. The annual environmental impact of all three systems consistently decreased with increasing lifetime. Additionally, the payback period of PV/T-AHS was 1.56 years. The results of this study will provide referenceable environmental impacts for practical application of solar air heaters.