Specific Small Digital-Twin for Turbofan Engines

Ten years ago, we develop a numeric state vector to allow us following the performance of a small fleet of turbofan engines. As engine manufacturer with “flight by the hour” new contracts going, we needed to improve our technology; hence, building analytic digital-twin solutions. This improvement allows to monitor the engines as well as the behavior of the airlines. It becomes possible to identify use classes to help design systems better adapted to our clients. The main difficulty to upgrade this technology was to manage huge amount of data. At the time, the only observations we had were small 4 Kb snapshots broadcasted by satellite link. And even with such small amount of observations, the mathematics were limited to an evolving buffer of one year and a half of flights for a ten aircrafts small company. Today, we download full flight temporal measurements reaching more than 2 Gb per flight and we want to manage our fleet of more than 35,000 engines. Moreover, the life duration of the engine parts increases and the first shop visit is now more than three years away. The solution was to use new cluster technology and adapt algorithms. A main element is to be able to reduce the observation dimension hence allowing mathematic computation, but specifically for each class of behavior. The compression algorithm is now running an auto-encoder neural network on GPU and the cluster analysis still uses a self-organizing map but deployed in parallel on a cluster of computers. We propose to present this new improvement and conclude with our view about operational digital-twins.