Thermography and machine learning combination for comprehensive analysis of transient response of a photovoltaic module to water cooling

Water flow cooling is a cost-effective and environmentally friendly way to enhance the performance of a photovoltaic (PV) system. However, most of the studies have investigated a PV unit with water flow cooling in the steady-state condition. Therefore, as the novelty, this study aims at modeling and investigation of the system performance in the transient condition. For a 50W polycrystalline PV module, the annual experimental data is utilized to find its transient response to water flow cooling, which covers temperature and power improvement (ΔT and ΔP). An artificial neural network (ANN) is developed using the experimental data for prediction of ΔT and ΔP. ANN is validated and then, utilized to provide a comprehensive discussion about the impact of effective parameters on the system response. According to the results, for all the investigated cases, there is a threshold value for water flow rate. For instance, when wind velocity and ambient temperature are 1 ms􀀀 1 and 25 ◦C, the threshold values for the solar radiation values of 800, 1000, and 1200 Wm-2 are 0.009, 0.013, and 0.015 kgs􀀀 1, respectively. Moreover, irradiance and ambient temperature have great impacts on dimensionless ΔT and ΔP, while wind velocity is the least effective one.