Established in 2020, but around for years
The section was established on 1st of January 2020 out of a working group of the former unit Embodied Systems for Robotics and Learning at the Mærsk Mc-Kinney Møller Institute. Located at the Mærsk building at the Odense campus of SDU, the section hosts the creative space for currently 11 researchers.
Our backgrounds span from computer science and electrical engineering to epidemiology and physics, giving us a broad perspective to research new, innovative solutions in a variety of fields, including clinical applications in a variety of diseases and life cycle management of wind turbines.
Due to our long presence at SDU, we have established over the years strong collaborations with clinical researchers and companies all over Denmark. In this environment, we are conducting research in clinical and energy harvesting projects.
Our focus lies on the application of artificial intelligence and data science to propel and advance practice and research in these fields. Together with our collaborators, we are pushing the boundaries of what we can get out of existing data, or devise strategies and devices to provide further insight into our research topics.
AI in health
As AI is beginning to make its mark on healthcare, a few questions keep raising:
- How can we best advance AI in healthcare and healthy living to the benefit of people, patients, and care professionals?
- To what extent we can rely on AI algorithms when it comes to matters of safety-critical applications (life and death) or personal health and well-being?
- How do we ensure that AI does not inadvertently discriminate against specific minorities, race and gender, or other groups, thereby perpetuating inequalities in access and quality of care?
Our research is centered around answering these questions keeping in mind that we should move from a reactive sick care to a proactive health care. This requires AI in its outmost definition, featuring adaptive and embedded solutions while being precise, transparent, interpretable, adaptive, versatile and robust.
A useful demarcation line that makes the distinction between our research and others, crisp and easy to apply can be formulated as follows. In addition to the design of predictive algorithms for early detection and prediction of high‑impact diseases and their associated complications, our research benefits from the use of causal inference to interpret the outcome of the algorithms. Thus, we enable the data‑crunching power of AI to go hand in hand with domain knowledge from human experts and established clinical sciences, equipping them with a strong decision‑support tool while keeping human oversight over the generated recommendations.
Danish Electronic Health Records is the gold mine to make better clinical predictions and explanations, which we extensively use in areas such as endocrinology with focus on diabetic complications, urology (prostate cancer), hepatology (liver disease) , gynecology (preeclampsia), gastroenterology (colorectal cancer, Crohn’s disease), neurology (epilepsy & seizure) and cardiology with focus on Atrial fibrillation.
AI and Big data analytics in renewable energy
These activities center around monitoring the behavior of wind turbines under normal and abnormal conditions and operational states, to provide performance benchmarks for the manufacturers and their customers, and a predictive framework to facilitate remaining lifetime estimations and optimal operation and maintenance plans.
Details about a fault’s progression, including the remaining‑useful‑lifetime (RUL), are key features in monitoring industrial operation and maintenance (O&M) planning. In order to avoid increase in O&M costs through subjective human involvement and over‑conservative control strategies, access to reliable models to estimate the RUL for wind turbines is of great importance. A substantial amount of effort has been put into studying main bearing failures and to find a strategy for an optimal stop.
In one study, we showed that on average, wind turbines are stopped 13 days prior to their failure, accumulating 786 days of potentially lost operations across a wind farm of 67 wind turbines. In another study, we showed that by utilizing the volume of data and analyzing it using artificial intelligence, to present a case study on main bearing failures for 108 turbines, predictions of the remaining useful lifetime of more than 90 days can be expected on average, outperforming the closest state‑of‑the‑art estimate by almost a factor of two on average.
We have, through our research on wind turbine condition monitoring, been crucial to establishing and strengthening SDU’s collaboration with Siemens Gamesa Renewable Energy, which has resulted inter alia in a valuable collaboration between the two. Siemens Gamesa maintains the industry’s largest collection of historical machine condition data, gathered from over 28,000 turbines worldwide.
Inside each turbine, more than 600 sensors transmit over 200 gigabytes of data each day. In close collaboration with the Diagnostic Center at Brande,we could therefore access the industry’s largest to date historical dataset; using the technology of Big Data Analytics, machine learning and the most novel techniques in AI, we were able to convert that sensor data to useful knowledge.
This research has resulted in many high impact publications describing novel solutions that enable our industrial partner to predict significant failures, remaining lifetime in terms of time and production, ahead of time.
Our research pushes the limits of prediction and prevention of unscheduled downtime and thereby extends the useful life of each wind turbine, which strengthens the competitive advantage of Siemens Gamesa Renewable Energy in all three areas of their wind power business: Onshore, Offshore and Services.
The Diagnostic Center at Brande, in collaboration with our team is currently migrating the suggested solutions and algorithms to installed wind turbines. The societal impact is huge: optimization of the performance of wind turbines means that the society will have access to more efficient, cheaper, renewable and green energy. This will drive the energy transition towards a sustainable world.