PV power prediction in Qatar based on machine learning approach

PV output power is highly sensitive to many environmental parameters, hence, power available from plants based on this technology will be affected, especially in harsh environments such that of Gulf countries. In order to conduct the PV performance evaluation and analysis in arid regions in terms of predicting the output power yield, proper acquisition, recording and investigation of relevant environmental parameters are considered to guarantee accuracy in the predictive models. In this paper, the authors analyze and predict the effects of these relevant environment parameters (e.g. ambient temperature, PV surface temperature, irradiance, relative humidity, dust settlement and wind speed) on the performance of PV cells in terms of output power. Different predictive models based on Machine Learning approach are trained and tested to estimate the actual PV output power in reference with an adequate time frame. Results show that the developed models could predict the PV output power accurately. ? 2018 IEEE.Scopu