Cloud Position Forecasting Based on ConvLSTM Network

A technique for cloud position prediction for intra-hour Irradiance prediction is presented. This cloud motion prediction technique is mainly used as an ultra-short-term photovoltaic power prediction in the field of electrical engineering. However, due to the impact of complex cloud motion processes, it is difficult to obtain accurate results using traditional algorithms. Many scholars have applied deep learning to predict and process images and achieved good results. This paper proposes a machine learning algorithm-ConvLSTM to predict cloud position. The cloud position and corresponding irradiance fluctuations were predicted ten minutes ahead.