再生可能エネルギーの大量導入に伴う課題解決のための機械学習の活用

抄録

Introduction of variable renewable energy systems (VRE), such as solar and wind power, into power systems is growing rapidly around the world. While this growth provides a step towards more sustainable societies, it also brings challenges. One important challenge regards the VREs output weather-related variability. If deployed in large scale, such variability can cause issues in the balancing of power demand and supply. To deal with this challenge on the power system side, one possibility is to increase the system's flexibility, which means the capability to deal with potential mismatches between VRE and demand. For example, conventional power generators could be operated in such a way to compensate partially for the VREs output variability. On the VRE side, forecasting, curtailment, and battery-coupled operation are often considered. Regardless of the measures employed, they typically require tackling complex problems using massive databases, and modeling natural phenomena, such as the weather. Such problems are particularly fit for machine learning (ML) techniques, and they have been the front-runner in research and applications related with the integration of VREs to power systems. In this report, we introduce examples of the latest ML techniques, and the most recent trends regarding their applications in VREs.