Solar Energy Forecasting Using Deep Learning Techniques

Abstract

The extent of renewable power generation usage has increased in recent times, and it will become inevitable soon to generate clean and green power. Microgrid units use renewable energy for power generation. The inclusion of Distributed energy resources (DER) will lead to intermittent power generation because of the continuously changing weather and seasonal conditions. The electricity consumption of the end-user also changes according to the time and the seasons. The source forecasting and the load forecasting becomes very important to schedule the energy storage device operations. In this paper, we use Solar energy as the source,solar irradiance changes with respect to place and time. In this article, Solar forecasting is performed for one month. If in case there are occurrences of an event like days of autonomy, effective load management can be performed provided a prior possession of knowledge about the source availability. We have different Artificial intelligence techniques like Fuzzy Logic, Machine learning (ML), and Deep learning (DL) mechanisms for source forecasting. When the microgrid is set up at a particular place, the irradiance, data and other parameters are considered for building the dataset. This article consists of a brief introduction about the microgrid and different source forecasting techniques. Machine learning algorithms and Deep learning algorithms are discussed and the efficiency of the various algorithms are compared using the Root Mean Square Error (RMSE) values.