Image processing methods for predicting the time of cloud shadow arrivals to photovoltaic systems

This document presents a combination of image processing tools used for detecting clouds from sky images and for predicting the arrival time of cloud shadows to an area of interest, for example an area of solar panels. Solar panels are used to produce energy from the sun in areas without access to electricity from central power plants and are increasingly being used by people who want to have the flexibility of being disconnected from the main power grid. Unfortunately, the volatility of cloud motion makes photovoltaic (PV) power unreliable for those who want a steady production of electricity. Knowledge about expected drops in power generation due to cloud shadows in PV systems can improve the performance of microgids by granting users the flexibility to turn on alternative forms of power generation, such as fuel cells, or power storage devices such as batteries, to keep the same loads working uninterruptedly. The solution presented was designed to address the need to improve the predictability of solar power generation for possible PV users.