Publication: Automated vibration-based fault detection in octocopter arm
Loading...
Date
2023-11-01
Authors
Mohamad Hazwan Bin Mohd Ghazali
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
A multirotor is a type of unmanned aerial vehicle (UAV) that is prone to crashing,
posing a risk to individuals or structures in the vicinity. The presence of cracks in the
multirotor will deteriorate its performance and, at worst, cause it to crash. Therefore,
it is essential to regularly monitor the multirotor components prior to operating the
multirotor to minimize the risk of multirotor crashing. However, cracks are difficult
to be detected by visual inspection as it is normally unnoticeable. Existing studies or
approaches regarding fault detection in multirotor are limited to faulty propellers and
rotors, and most faults cannot be detected in real-time. The aim of this research is
to develop a real-time, user-friendly vibration-based fault detection approach in multirotor
that focuses on the cracks in multirotor arms and motor failure cases. The
vibration data of normal and faulty conditions are recorded by ADXL335 accelerometers,
and the time-domain outputs are analyzed by statistical means. Additionally,
infrared (IR) and WCS1800 current sensors are also installed to monitor the motor
rotational speed and current, which is useful in motor failure cases. A combination
of sensor fusion and fuzzy logic algorithms is adopted to provide the decision-making
regarding the multirotor condition and its corresponding countermeasures. The final
output is then transmitted to the user via long-range (LoRa) wireless communication
protocol. Experimental results demonstrated a clear distinction in the vibration levels,
where the cracked multirotor arm exhibits a greater maximum vibration amplitude and
root mean square (RMS) values compared to a normal arm in all multirotor test modes.
For instance, in the bench test with minimum power, the maximum amplitude differences are in the range of 60% to 100% and 140% to 250% for the minor and major
cracks, respectively. In terms of vibration signals, the faulty motor displayed a maximum
amplitude ranging from -0.43 g to -0.57 g, whereas the healthy motor registered
a maximum amplitude in the range of -1.2 g to -1.31 g during flight tests. In the short
term, a minor crack in the multirotor arm (≤0.09 g amplitudes recorded in the bench
tests (minimum power)) will have a negligible impact on the multirotor performance,
but if the crack is significant (≥0.1 g amplitudes recorded in the bench tests (minimum
power)), it can cause the multirotor to crash. Findings also indicate that the multirotor
can still fly even when two of its motors are malfunctioning, as long as these faulty
motors are located directly opposite each other. The real-time fault detection approach
developed in this research provides an effective and user-friendly solution to the condition
monitoring of multirotor, covering both mechanical and electrical fault cases. It
is also applicable in multiple test modes without the need to upload a new code