인공 신경망을 이용한 GTS 기반 PVT 집열기의 성능 예측 및 평가 = Predicting and testing the performance of the PVT collector based on GTS as a back sheet using an artificial neural network and experimental data

다국어 초록 (Multilingual Abstract)

Recently, Deep learning methods have been widely used in PVT collector systems based on artificial neural networks (ANNs) due to their high precision in evaluating the performances without using physical modeling. A photovoltaic thermal (PVT) collector is a hybrid system that merges a photovoltaic (PV) module and a thermal collector in one unit to generate electricity and heat simultaneously. In addition to that, it increases efficiency and reduces required costs and space. In this study, three different models of PVT collectors were fabricated based on Tedlar Polyester Tedlar (TPT), Graphite Thermal Sheet (GST), and TPT/GST as a back sheet and tested. An Ethylene Glycol Coolant (EGC) was also utilized with a water tank and a heat exchanger. The ANN model is used to predict PVT collectors’ electrical and thermal efficiencies based on measured weather datasets. Besides, both electrical and thermal models have been tested experimentally. Accurate modeling methods have been applied to generate the output results for training and validation. The actual climatic data and PVT collectors’ outputs of one year with a 5-minute step have been used to calculate the thermal and electrical efficiencies. The solar irradiance, wind speed, humidity, PVT collectors’ outputs, and temperature are the most critical variables to consider as input in the ANN-based model. This study’s expected results are increasing the PVT’s total efficiency by using high thermal conductivity material (GTS) instead of low thermal conductivity material (TPT). The electrical efficiency of the PVT module will increase due to the cooling effect. We also expect congruous results between the ANN analysis and experimental test.