The paper describes a new approach to democratize access to results from expensive high-performance computer simulations of atmospheric boundary layer flow interacting with wind turbines, in large wind farms. Users interact with the data using a virtual sensor array methodology and essentially stream the data on demand to their analysis or visualization programs rather than having to download files and worrying about data formats, etc.

This research aims to improve wind turbine maintenance by detecting faults early using only data from normal operation. The method analyzes vibration signals in both time and frequency domains and uses a variational autoencoder, a type of deep learning model, to learn normal behavior. It then detects anomalies by measuring how much new data deviate from this learned model. Tests on real turbine data show early and accurate detection of faults such as pitch issues and icing.