Upcoming Clean Energy Engineering
A Multi-agent Explainable Detection and Inference for Collaborative Wind Turbine Systems
Summary
Original abstract (not yet simplified)Wind energy is critical to achieving the European Union's carbon neutrality goal by 2050 and has experienced rapid development. Currently, there is a great challenge to large-scale wind farms due to low reliability and high operation and maintenance costs. Datadriven methods can offer a powerful and sustainable way to monitor wind turbine health. However, existing approaches primarily reliant on centralized...
View original technical description
Wind energy is critical to achieving the European Union's carbon neutrality goal by 2050 and has experienced rapid development. Currently, there is a great challenge to large-scale wind farms due to low reliability and high operation and maintenance costs. Datadriven methods can offer a powerful and sustainable way to monitor wind turbine health. However, existing approaches primarily reliant on centralized learning, are often inadequate in providing the necessary transparency for actionable insights, fall short in consistently delivering accurate results across diverse operating conditions, and are unable to fully accommodate data privacy concerns of independent wind farms. To address these challenges, this project aims to develop a Multi-agent Explainable Detection and Inference framework for Collaborative wind turbine systems (MEDIC). Specifically, MEDIC includes three progressive tasks: 1) explainable turbine-level diagnostics powered by knowledge-guided graph modeling; 2) robust cross-turbine generalization through causal disentangled representation learning; 3) privacy-preserving collaborative learning facilitated by multi-agent federated learning. MEDIC will empower multiple agents (wind turbines) to collaboratively contribute to and benefit from a globally optimized anomaly detection model while maintaining control over their sensitive data. MEDIC specifies the resources needed for this project, including the quality and capacity of the host, mentors, data, and experimental facilities. This project will enhance the applicant's scientific skills and innovation capability, expand research horizons, and establish research collaborations. A two-way knowledge transfer approach in energy big data, distributed modeling and energy system analysis is proposed to ensure benefits between theapplicant and the host. MEDIC will make a significant contribution to the state-of-the-art in wind energy system monitoring and predictive maintenance and the EU climate goals.
Related Research
Grants with similar aims, by meaning.
Scalable Digital Twins with Advanced Machine Learning for Offshore Wind Turbine Operation and Maintenance
Multidisciplinary optimisation of wind farms through experiments and data-driven modelling
Offshore Wind Turbine Blade Monitoring Using Computer Vision and AI
Research on Virtual Sensor Technology for Advancing Wind Farm Digitalisation
Wind turbine sensor placement optimisation for digital twin development (OSP)
Original classification
HORIZONPlain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research. Is something wrong? Let us know