RT-DT platform
WINDMIL RT-DT: An autonomous Real-Time Decision Tree framework for monitoring and diagnostics on wind turbines.
As part of the WINDMIL project, we developed a proof of concept, which combines various modules under one single “commercialisable” platform as follows:
- Innovative hardware-software (HW-SW) package, able to acquire and fuse in real-time both condition monitoring (CM) and specialised structural health monitoring (SHM) data from WT components.
- Innovative computational core: object-oriented implementation of the Decision Tree algorithm, Bayesian network based computation of predictive probabilities, automated updating of the decision tree, automated discovery of unclassified states, and an automated root-cause analysis algorithm in an ensemble of decision trees. Using this computational core, we can quantify the risk of future failure of a given component and trace back the root-cause of failure.
- Software implementation of an autonomous object-oriented DT learning algorithm, interfacing with both real-time telemetric output from the HW-measuring unit on a WT and the cloud based analytics data.
- Cloud data storage and analytics of long-term historical data; interface with the real-time DT SW unit
- Online user interface to visualise the decision tree and the corresponding output (decision support tool)
- Network and communication setup in addition to APIs to tap into the various stages of the data (a.) raw measured, (b.) post-process/filtered measured and (c.) computed output.