Self-supervised learning for predictive maintenance

The aim of this project is to develop self-supervised and continual learning methods that will improve the accessibility of data-driven predictive maintenance in power networks. By incorporating continual learning, the methods will be able to adapt to evolving conditions in power networks and accurately handle the aging effects of electrical components, leading to more informed and accurate maintenance decisions. The efficacy of the proposed methods will be evaluated through case studies using data collected from power stations. This project will enable wider adoption of predictive maintenance practices in power networks, leading to improved reliability and reduced maintenance costs.

Project manager at MDU: Ning Xiong