Research on a cloud image forecasting approach for solar power forecasting

Accurate cloud image forecasting is very necessary for sky image based solar power forecasting. To provide a means that can track the cloud deformation process and then forecast the cloud shape and position in a future sky image, a cloud image forecasting approach is proposed in this paper. Firstly, the cloud pixels in sky images are identified using Otsu's method. Secondly, a mathematical description method of cloud location, shape, and deformation is suggested, so that the key information of cloud and its deformation process can be extracted from original images and then digitized. Thirdly, genetic algorithm (GA) is applied to optimize the cloud deformation process according to digitized historical cloud information. In the end, the cloud shape and position in a future image is forecasted using linear extrapolation. The feasibility and effectiveness of the proposed method are validated by simulation.