Categorization of Aircraft Missions for Exploitation by a Digital Twin

The use of new recurrent neural models with layers of attentions has proven to be very effective in monitoring the internal state of an aircraft's engines. Our research work has shown the effectiveness of these methods for predicting corrosion or just measuring a deterioration in performance. However, until now, only the data broadcast by the engine has been readily available, but it seems logical that the description of the mission and the way the pilot handles the aircraft seem equally important. We have therefore developed a mathematical method to describe each mission, in this way it becomes possible to import new data helping to monitor engine wear. In the meantime, these new measurements also give us a new methodology to explore the use of our systems. For example, we are now able to categorize flights and it will become possible to adapt our design and our maintenance offer to the real needs of airlines.

Keywords: Aircraft engines, PHM, Neural network.

Download PDF