Enhanced performance Gaussian process regression for probabilistic short-term solar output forecast

With increasing concerns of climate change, renewable resources such as photovoltaic (PV) have gained popularity a means energy generation. The smooth integration in power system operations is enabled by accurate forecasting mechanisms that address their inherent intermittency and variability. This paper proposes probabilistic framework to predict short-term PV output taking into account the uncertainty weather. To this end, we make use datasets comprise meteorological data irradiance, temperature, zenith, azimuth. First, categorise four groups based on solar time using k-means clustering. Next, correlation study performed choose weather features which affect greater extent. Finally, determine function relates aforementioned selected with Gaussian Process Regression Matern 5/2 kernel function. We validate our method five generation plants different locations compare results existing methodologies. More specifically, order test proposed model, two methods are used: (i) 5-fold cross-validation; (ii) holding out 30 random days data. confirm model accuracy, apply independent times each clusters. average error follows normal distribution, 95% confidence level, it takes values between -1.6% 1.4%.