Intermittency caused by renewable energy sources such as PVs as well as highly stochastic power profiles of electric vehicle charging stations are gradually causing electrical grid challenges. A solution to minimise the impact of these new sources is to couple them with storage and control frameworks.
However, storage is costly and hardly profitable if it serves a single service for a single entity.
Smart planning methods based on current market regulation can allow battery storage systems to serve multiple services either sequentially or simultaneously by deciding in advance the optimal operation mode.
The goal of this project is to assess the performance of different forecasting methods as well as different planning algorithm techniques. It would be a pleasure to host a student onboard to chip in the process of enhancing this planning and control framework. The following tasks are suggestion topics and may be refined or reorientated based on the student’s expertise and desires.
Tasks of student
- Build month-ahead load forecasting methods (SARIMAX,
convolutional neural networks, …) and compare performances.
- Formulate, implement, and compare planning algorithms to combine
storage services within the current market framework.
- Programming experience in Python.
- Knowledge on statistics, machine learning and convex optimisation.
- Electrical grid knowledge is not required but appreciated.
Max Chevron – [email protected]
Enea Figini – [email protected]