Improvement of Tracking Algorithms using Machine Learning

This paper explores innovative solutions to enhance the efficiency of photovoltaic (PV) tracking systems using deep learning based tracking algorithms. Conventional solar trackers perform well on sunny days but struggle with optimal orientation during cloudy weather and the dynamic path of the sun. We propose dynamic solutions that adapt to real-time weather conditions using on-site measured and satellite data. By leveraging deep learning algorithms and historical time series data, our approach forecasts the optimal angle for the upcoming hour, maximizing the energy output of single-axis solar trackers.