The adoption of renewable energy sources is becoming an increasing need across the globe. Such energy systems have several properties that make them particularly relevant and attractive for the implementation of data-driven operation and maintenance (O&M) algorithms such as condition-based and predictive maintenance.

In my talk I will describe our work done in collaboration with the company Fluence Energy to develop scalable algorithms for both wind farms and solar plants. I will elaborate on the use of AI for developing and scaling the algorithms to multi-component heterogeneous devices that are operated under diverse environmental conditions. Special attention will be given to the combination of data-science with engineering domain knowledge to bridge the gap between advanced state-of-the-art concepts in research and their deployment in real-world operational systems. In particular, the use of transfer learning, uncertainty quantification and physics informed deep learning will be demonstrated through concrete examples of anomaly detection and fault diagnostics in operational renewable energy systems.

Speaker: Dr. Lilach Goren Huber, Zurich University of Applied Sciences, Switzerland