#1 Wind turbine predictive maintenance
The testbed is aimed at creating a digital twin of a wind farm by aggregating simulation and machine learning models of single turbines for predictive maintenance. Data are used to detect the health status of the turbine itself, to predict failures and to plan wind farm maintenance operations for reducing unexpected breakdowns and downtime.
The main activities foreseen for this testbed are:
1. to design and install redundant IoT modules able to log data at high sample rates, with best in class security for transferring data to the backend where data is stored;
2. to fuse data coming from individual subsystems in a wind turbine simulation model;
3. to train machine learning models on the cloud and migrate them dynamically to edge nodes for localized pre-trained control at the wind farm (smartification of wind farm edges at the plant);
4. to build an accurate machine learning -based performance model for wind turbines;
5. to early detect failures of monitored subsystems;
6. to determine and execute predictive maintenance plans;
7. to optimize wind turbine control systems.
Objectives
Smartifying the wind turbine, and especially Yaw and Pitch systems, by integrating sensors, edge nodes, and the associated software components
Providing efficient data transmission solutions
Optimizing the wind turbine control system
Developing a digital twin to simulate the best orientation of blades to increase energy production and reduce mechanical stresses on the wind turbine
Extending residual useful life of critical components in the wind turbine
Producing failure models for each turbine and for the whole wind-farm
Developing predictive maintenance algorithms for each component of the turbine and for the whole farm, based on the information gathered/elaborated and the scenarios simulated at the cloud side
Involved Partners
