Short term solar irradiance prediction model using ground based sky imaging

Abstract

With the growing energy demands and depleting fossil fuel deposits world is moving towards renewable energy. Solar energy has become a focus on this transition, especially in tropical countries where solar power harvesting is economical. The major disadvantage of solar power is its intermittent behavior due to cloud movement. Several researches are ongoing to predict the solar power in different temporal and spatial resolutions as a solution to this drawback. This research discusses on a short term solar prediction model using total sky images captured with a specifically developed hardware. A MATLAB software model is developed using image processing techniques to track and predict cloud movement throughout the visible span. The cloud movement behavior is mapped with the output current of a fixed solar panel to develop the prediction algorithm. The developed software model is capable of predicting the solar panel output on real time total sky images with a 10 second temporal resolution. Additionally, the model parameters can be fine-tuned to predict the solar irradiance for a different location.