Short-term forecasting, All Sky Imager, Image Processing, Cloud Segmentation
Summary / Abstract:
Accurate prediction of photovoltaic energy remains a challenge, as PV production is dependent on fluctuating weather conditions, such as solar irradiance which relies on cloudy conditions. In this context, our work is based on a specific “All Sky Imager” (ASI) , integrating “fish-eye” concave lenses with a 180° field of view. This paper presents a novel approach of cloud segmentation using ASI to improve - PV production forecasting. First, we improve the identification of various components in the images (clouds, sun, noise, etc.). Then, we study the very short-term impact of clouds through a sky images segmentation and tracking processing. Finally, we correlate the skylevel segmented image to fluctuations in the actual on-site solar irradiance measurements. The results show that the segmentation is efficient in clear and overcast sky conditions. However, high precision irradiance accuracy in partially cloudy sky conditions is difficult to obtain due to the chaotic impact of the circumsolar region and different cloud opacities.