PEER REVIEW
20
th
Australian International Aerospace Congress, 27
th
February – 1
st
March 2023, Melbourne
20th Australian International Aerospace Congress
ISBN number: 978-1-925627-66-4
Please select category below:
Normal Paper
Student Paper
Young Engineer Paper
Bearing-Fault Signature Generation for Equipment Health
Monitoring using a seven Degree-of-Freedom Bearing-
Vibration-Model under non-steady Conditions
Philipp Schildt
1,2,3
, Pier Marzocca
2
, Carsten Braun
3
, Wim Verhagen
2
1
Rolls-Royce Electrical, Günther-Scharowsky-Str. 1, 91508 Erlangen, Germany, philipp.schildt@rolls-royce-electrical.com
2
Royal Melbourne Institute of Technology, PO Box 71 Bundoora VIC 3083, Australia
3
FH Aachen University of Applied Sciences, Hohenstaufenallee 6, 52064 Aachen, Germany
Abstract
The advent of electric propulsion in aviation comes with the opportunity to further increase
propulsion system safety and reduce maintenance efforts as systems are becoming mechanically
less complex. In modern electrical machines, the remaining components subject to mechanical
wear are basically the rotor-bearings. Optimized designs use the same bearing for the rotor of
the electrical machine and the propeller. One key technology to reduce maintenance cost and
increase safety is to detect upcoming faults in a very early stage to enable condition-based
maintenance and calculate remaining useful life (RUL). Whilst conventional and artificial
neural network (ANN)-based monitoring techniques are well proven against available bearing-
fault-datasets in steady-conditions, bearing fault-detection and -classification becomes more
complex in a non-steady environment. In this work, a 7- degree of freedom (DoF) bearing-fault
model is used to synthesize high-resolution vibration data under different non-steady
environments: Based on well validated 5-DoF simulation approaches, a bearing fault vibration
model has been developed, allowing the synthesis of time-series signals for bearing-faults in a
simulated (flight simulator) or recorded (flight experiments) environment. The model,
implemented in MATLAB/Simulink, accounts for normal- and lateral-acceleration, pitch- and
yaw-rates, propeller rotational-speed, and thrust-force to the bearing. By using simulated- and
recorded-flight data, it can be shown that manoeuvring- and atmospheric-excitation has a
significant impact on the bearing-fault vibration-signatures. The synthesized data can then be
utilized to train ANN-based classifiers or regression-models, as well as to optimize detection
algorithms based on conventional feature extraction for new propulsion-systems, before
operational-data is available. Training and optimization of ANN-based classifiers is subject to
ongoing work and not part of this paper.
Keywords: Equipment-Health-Monitoring, Electric-Propulsion, Advanced-Air-Mobility,
Bearing-Fault-Diagnosis, Flight-Conditions, Remaining-Useful-Life, Distributed Propulsion.
Introduction
As the need for climate-neutral air transport evolves, electric propulsors as key enabler for
battery-electric- or hybrid- flights become more and more important. Early Demonstrators, as
the e-Genius [1] (maiden Flight in 2011) showed the general feasibility of battery-electric flight
for longer distance travel with small aircraft. However mass constraints, especially with respect
to the energy storage density of current battery technology, do not allow for longer commercial
