Reinforcement Learning for Micro-Grid Control and Optimization

This project aims at optimizing the use of energy in micro-grids. Reinforcement learning, a branch of artificial intelligence, will be used to reduce the consumption of fossil fuel by better deciding when to charge or discharge batteries. Using reinforcement learning for this automated decision making problem could be interesting as it should allow to handle factors of difficulty that other types of methods struggle to deal with. For example, it should be able to cope with more complex micro-grids, or to better handle uncertainty in the systems predicting the future demand. A key focus of the project is to demonstrate the ability of reinforcement learning to be deployed on real world industrial cases where reliability and respect of industrial standards are critical.

Faculty Supervisor:

Sarath Chandar Anbil Parthipan;Irina Rish

Student:

Partner:

Institut de Recherche Hydro-Québec;Hydro-Quebec

Discipline:

Computer science

Sector:

Energy and Utilities; Green/Alternative Energy; Artificial Intelligence

University:

Université de Montréal

Program: