[5B1] Modelling of the vibratory response of an aircraft engine using AI
J Gomez and J Griffaton
Safran Aircraft Engines, France
This study presents an artificial intelligence (AI)-driven approach for predicting aircraft engine vibratory responses across flight phases. Traditional mechanical models often fall short in capturing the range of influencing factors, leading to inadequate predictions. Tuning mechanical parameters is often necessary to reproduce real values and the model obtained cannot predict the vibrational behaviour of an entire flight. By means of artificial intelligence techniques, our supervised model synthesises complex interactions between several physical variables, including shaft speeds, altitude, temperatures and pressures, among other influencing variables, to achieve comprehensive predictions. For this work, we trained the model with one year of vibrational data from an aircraft engine and used three different AI model techniques that yield very good results. The model built offers enhanced predictive capabilities and can serve as a virtual sensor or as a reference for normal behaviour to perform condition monitoring.
