Operation & Maintenance (O&M) costs may account for 30% of the total cost of energy for offshore wind power. Alarmingly, only after a few years of installation, wind turbines (WT) - offshore ones in particular - may require emergency repairs. These systems further feature a short lifespan hindering a fruitful investment in green energy. Motivated by this need, we have designed and implemented a monitoring and diagnostics web-based platform in the context of operation and maintenance of wind turbines, applicable across both onshore and offshore systems. We call this the WINDMIL RTDT framework. Using the algorithms hosted on this platform, owners and operators of wind turbines can quantify the risk of future failure of components and trace back the root-cause of failure.
This is business-critical information for energy companies and wind farm owners and operators. The platform consists of a software-hardware solution, implementing various structural health monitoring algorithms, machine learning models and frameworks for detecting faults, errors, damage patterns, anomalies and abnormal operation. Once trained on the linked data, our platform seamlessly facilitates the process of deploying (serving) these algorithms for monitoring on live-streamed data. The platform can serve as a decision-support tool for wind turbine manufacturers, wind farm Operators, service companies and insurers. WINDMIL RTDT establishes a technology proof of concept for an easy to deploy and easy to use web-based user-facing application for condition monitoring of wind turbines.
The algorithms included in the WINDMIL RTDT application correspond to four Analysis Categories, namely Deep Learning, System Identification, Modal Analysis and Probabilistic Machine Learning, with some of the algorithms already fully implemented, and some of the modules scheduled to be functional in a second version of the tool.
The detailed description, and further information on the overall structure and the different services enabled by the WINDMIL RTDT platform, can be found in this detailed final report:
https://www.research-collection.ethz.ch/handle/20.500.11850/437457(opens in new window)A demo video showcasing the workflow of the software can be viewed here:
https://www.youtube.com/playlist?list=PLqXfgM2e4ShR7Z4xjFjArYvpSiRoENoaX(opens in new window)A production version of the software web UI portal is hosted on Azure Cloud, and can accessed here: 51.103.129.34
