Energy Predictions for System on a Chip Solutions.

The active participation of end-users in the smart grid brings opportunities to those who want to participate. The participation and remuneration of end-users are issues that have been studied over the years in scientific publications. However, they heavily depend on energy predictions for participation requests and participation assessment for remuneration. These energy predictions are crucial to remunerate end-users according to their participation. The study of energy prediction models is needed and must take into account data privacy and security issues. One possible solution is to include prediction models in energy management systems. This paper proposed five energy prediction models that are able to be executed in a system on a chip, such as a single-board computer. The models, based on a mathematical approach and on a learning approach, were tested in a physical deployment using a Raspberry Pi 3 Model B. The results show the ability of having energy prediction models being executed in energy management systems deployed in a system on a chip, resulting in satisfactory prediction values.