Congestion Prediction in Energy Networks Using Machine Learning

Energy networks are often very complex which results in highly unpredictable congestion patterns in the physical constraints of the network. Most efforts to model congestion in energy networks are made on toy problems. Here, the objective is to model and predict congestion in a real physical energy network using automated machine learning systems, in particular deep neural networks. Physical properties of the network will also be properly modelled. The proposed method is expected to predict congestion with higher accuracy than the existing methods used by the partner organization. The intern will learn how to apply his knowledge in deep neural networks to a real-world complex problem.

Faculty Supervisor:

Sue Becker

Student:

Rory Finnegan

Partner:

Invenia Technical Computing

Discipline:

Psychology

Sector:

Information and communications technologies

University:

McMaster University