Supervisor:
Professor Birgitte Bak-Jensen

Co-Supervisor:
Assistant Professor Jayakrishnan Pillai

Assessment Committee:
Professor Huai Wang, AAU Energy (Chair)
Professor Sami Repo, Tampere University
Associate professor Peiyuan Chen, Chalmers University of Technology

Abstract:

The global energy crisis and environmental problems necessitate the development of renewable energy sources. However, the stochastic and fluctuating behaviors of the renewable generation bring risks to the operation and planning of active distribution networks, which become more sophisticated and face more uncertainties than ever before. Traditional models have difficulty in fully meeting the analysis and control requirements of active distribution networks.

As one of the artificial intelligence technologies, deep learning has shown outstanding performance in various fields, since it can automatically learn latent features from high dimensional data without requiring simplifications and assumptions of the system’s physical models. The strong learning and representing capabilities of deep learning bring new opportunities to address gaps in existing models for analysis and control requirements of active distribution networks.

In this context, the Ph.D. project focuses on using deep learning to transform the massive collected data into knowledge, and provide deeper insights into the past, better understandings of the future, and practical suggestions on possible decisions for the economy and secure operation of active distribution networks. To achieve this objective, the project aims to model and optimize the power loads and renewable energy sources in active distribution networks from multiple perspectives at the data level (e.g., missing data imputation), the algorithm level (e.g., point prediction and uncertainty prediction), and the decision-making level (e.g., risk-based day-ahead optimal scheduling).