Electrical performance and power prediction of a roll-bond photovoltaic thermal array under dewing and frosting conditions

Abstract

In this paper, a roll-bond photovoltaic thermal heat pump (PVT-HP) system was constructed, and its photovoltaic (PV) module was cooled via heat pump circulation (HPC). Experiments and power prediction models of the PVT array were realized. First, comparative tests of the PVT array with and without HPC were conducted to investigate the electrical performance under dewing and frosting conditions. Then, the time vector-based genetic algorithm-back propagation (GA-BP-t) neural network model and the weight effect ratio (WER) were proposed to analyse the effects of the input variables. The results reveal that the average photoelectric conversion efficiency (PCE) of the PVT array with HPC under clear and cloudy skies respectively reached 14.2 % and 10.1 %, thereby exhibiting respective improvements of 14.8 % and 3.1 % as compared to the PVT array without HPC. Moreover, the influence of dewing and frosting was found to decrease the PCE; the PCE decreased by 11.7 % when the relative humidity increased by 85.7 % from 27.9 % to 51.8 %. The average mean absolute percentage error of the GA-BP-t model was 6.08 %, which was a decrease of 76.12 % relative to the engineering model. Furthermore, the inverse WER results of the PV temperature and relative humidity were found under clear and cloudy skies.