Research project Our project aims to develop an AI-based decision support system to predict boiler leaks in combined heat and power plants. By enhancing previously validated prediction models, the system will improve prediction accuracy and reduce costly unplanned downtime.
A joint project between the Department of Computing Science and Umeå Energi aiming to improve operational reliability in combined heat and power plants. An important part of the project is to ensure compliance with the EU's AI Regulation, which means that transparency, accountability, and data integrity must be taken into account throughout the project's life cycle. Our project is funded by Vinnova as part of the research programme "Advanced Digitalisation – Needs-driven Innovation 2025", and has a total budget of SEK 4,020,000.
This project aims to develop and implement an ethical AI-based decision support system, (DSS) to predict boiler leaks in combined heat and power plants, compliant with the . It builds on previously validated prediction models and enhances them with advanced AI algorithms to improve prediction accuracy, thereby reducing the negative impacts of unplanned downtime.
By integrating advanced AI algorithms, the project aims to significantly improve prediction accuracy, thereby creating the conditions to reduce the negative impacts of unplanned outages. The goal is to develop a robust and reliable decision support system that not only enhances operational safety in combined heat and power plants, but also meets the regulatory requirements imposed on AI systems in critical societal sectors—thus ensuring an ethically sound and trustworthy solution.
The project will employ agile development to enable iterative design and testing of the AI-based DSS. Teams from and Umeå University will coordinate with stakeholders at production sites. The research group "Formal methods for trustworthy hybrid intelligence" at Umeå University will apply agile design increments based on co-creation methods and the introduction of both EU's Trustworthy AI guidelines and AI regulatory frameworks. Our project is funded by as part of the programme "Advanced Digitalisation – Needs-driven Innovation 2025", and has a total budget of SEK 4,020,000.
From , key participants include Måns Kjellander (Project Leader), Magnus Torrkulla, Ville Lintamo, Johnny Karlsson, and Dan Bohman.
