Assessing Aircraft Engine Wear through Simulation Techniques
Abdellah Madane
1
, J
´
er
ˆ
ome Lacaille
2
, Florent Forest
3
, Hanane Azzag
4
, and Mustapha Lebbah
5
1,2
DataLab, Safran Aircraft Engines, 77550 Moissy-Cramayel , France
abdellah.madane@safrangroup.com
jerome.lacaille@safrangroup.com
3
IMOS Lab, EPFL, Lausanne, Switzerland
florent.forest@epfl.ch
4
LIPN, UMR CNRS 7030, Sorbonne Paris Nord University, 93430 Villetaneuse, France
azzag@univ-paris13.fr
5
David Lab, UVSQ, Paris-Saclay University, 78035 Versailles, France
mustapha.lebbah@uvsq.fr
ABSTRACT
In the field of aeronautical engineering, understanding and
simulating aircraft engine performance is critical, especially
for improving operational safety, efficiency, and sustainabil-
ity. At Safran Aircraft Engines, we were able to demonstrate
the effectiveness of using time series collected from the en-
gines after each flight to build a digital twin that provides a
dynamic virtual model able to mirror the real engine’s state by
using a transformer-based conditional generative adversarial
network. This virtual representation allows for advanced sim-
ulations under diverse operational scenarios like flight con-
ditions and controls, aiding in understanding the impact of
different factors on engine health. It is, therefore, possible
for us to provide virtual flights performed by our engines in
their actual state of wear. This research paper presents a ma-
chine learning model that effectively simulates and monitors
the state of aircraft engines in real-time, enabling us to track
the evolution of the engines’ health over their life cycle. The
model’s adaptability to incorporate new data ensures its ap-
plicability throughout the engine’s lifespan, marking a step
forward in proactive aeronautic maintenance and potentially
enhancing engine longevity through timely diagnostics and
interventions.
1. INTRODUCTION
In the realm of aircraft engine systems, strategically placed
sensors within the engines play an integral role by capturing
Abdellah Madane et al. This is an open-access article distributed under the
terms of the Creative Commons Attribution 3.0 United States License, which
permits unrestricted use, distribution, and reproduction in any medium, pro-
vided the original author and source are credited.
essential operational data throughout flight cycles. This data
is vital for implementing Prognostics and Health Manage-
ment (PHM) systems (Lacaille & Langhendries, 2022; For-
est et al., 2020). Notably, as demonstrated in recent studies
(Langhendries & Lacaille, 2022), such frameworks can ben-
efit from machine learning models like recurrent neural net-
works, which utilize temporal data to predict engine degrada-
tion patterns accurately.
The evolution of data acquisition technologies has facili-
tated the continuous collection of Engine Operational Data
(CEOD) during flights. This comprehensive dataset includes
a variety of sensor outputs and computational analyses per-
formed by onboard systems, with data processing occurring
post-flight. Leveraging this continuous data flow enhances
the development of algorithms that surpass the traditional
models based on snapshot data. This continuous monitor-
ing is particularly beneficial for improving anomaly detec-
tion techniques, as detailed in (Coussirou, Vanaret, Lacaille,
& DataLab, 2022). By harnessing state-of-the-art computa-
tional techniques, such as machine learning and big data ana-
lytics, engineers and researchers are now able to process and
interpret vast amounts of operational data in real-time. This
capability not only enhances the accuracy of predictive main-
tenance models but also facilitates a more proactive approach
to engine management. The granularity of CEOD allows for
a detailed understanding of engine performance under vari-
ous conditions, thus aiding in the optimization of engine effi-
ciency and reducing unscheduled maintenance.
The following research focuses on two main objectives. The
first is the development of a data-driven simulation frame-
work for aircraft engines that utilizes CEOD to replicate the
1
