Encognito – Energy Analytics for Power Grids and Distributed generation: This is a five-year project funded by Equinor under their Academia Programme agreement with UiT The Arctic University of Norway. The project focuses on analyses of the power grid, with a specific interest in integration of renewable energy sources. The long-term goal is to expand the research activities of the UiT Machine Learning Group in this field to make us a nationally leading academic provider of machine learning algorithms in support of the energy transition.
The project is managed by Stian Normann Anfinsen and funds one researcher (Huamin Ren) and an adjunct associate professor (Christopher Coello, data analyst at Elvia Nett). In addition to research and technical development, the ambition is to increase the domain knowledge of the Machine Learning Group by interaction and networking with industry, research institutes and academia, and to further expand our activities in energy analytics by fostering master’s projects and seeking funding for new research proposals.
Smart Senja: This project is a large-scale demonstration of a smart energy system for the future in fishing villages Husøy and Senjahopen on the island of Senja. This is an interdisciplinary effort involving technologists, social scientists, economists, industry and local inhabitants. The aim is to improve reliability of local power distribution networks and make grid investments redundant.
One part of this ENOVA-funded project is to develop prognostic tools for load forecasting, power flow computation, stability analysis and optimised intervention using flexibility assets, prosumer trading and demand response. In parallel, the UiT Machine Learning Group is undertaking research on probabilistic algorithms for the same tasks, which will allow risk-based management of the grid. We employ modern techniques from machine learning with a joint emphasis of accuracy, speed and explainability, with a particular focus on deep neural networks.
Prediction of wind power production with physics-informed machine learning: This is a Ph.D. project that uses data from wind parks in the counties of Nordland, Troms and Finnmark to develop improved methods for short to medium-range forecasts of produced wind power. The ambition is to design hybrid algorithms that combine physical models of local meteorology in complex terrain with statistical approaches. Ph.D. student Hao Chen is co-supervised by Stian Normann Anfinsen from the Machine Learning Group.
Vision-based Power Line Inspection: The UiT Machine Learning Group has supervised an industrial Ph.D. candidate from eSmart Systems, a Norwegian provider of software solutions for inspections of powerlines, grid maintenance planning and energy flexibility optimization. This project has developed the innovative deep learning-based object recognition technology which underlies the power line inspection services offered by eSmart Systems. Ph.D. student Nhan Van Nguyen defended his thesis in December 2019 with Robert Jenssen as main supervisor.