=Paper=
{{Paper
|id=Vol-3360/p03
|storemode=property
|title=Applications of Machine Learning Techniques for Fault
Diagnosis of UAVs
|pdfUrl=https://ceur-ws.org/Vol-3360/p03.pdf
|volume=Vol-3360
|authors=Luttfi A. Al-Haddad,Alaa Jaber
|dblpUrl=https://dblp.org/rec/conf/system/Al-HaddadJ22
}}
==Applications of Machine Learning Techniques for Fault
Diagnosis of UAVs==
https://ceur-ws.org/Vol-3360/p03.pdf
Applications of Machine Learning Techniques for Fault
Diagnosis of UAVs
Luttfi A. Al-Haddad1 , Alaa Jaber1
1
University of Technology, Baghdad, Iraq
Abstract
Due to the heavy usage of Unmanned Aerial Vehicles (UAVs) and the co-evolution of modern technologies, a crucial intro-
duction to fault diagnosis has taken place in recent studies in the avoidance of ravaging consequences. Machine Learning
techniques are one of the other major fault-diagnosing approaches in the field of Artificial Intelligence. This review article
delivers an elaborated overview of the latest studies concerning UAVs fault diagnosis utilizing Machine Learning and Deep
Learning techniques. A summarized comparison of the different methods is distinguishably elaborated where the conclusion
highlights that research on fault diagnosis systems is progressing and yet to end. Consideration should be given to a growing
number of research and methodologies.
Keywords
Unmanned Aerial Vehicles, Fault Diagnosis, Machine Learning, Artificial Intelligence, Artificial Neural Network
1. Introduction of variant applications have escalated rapidly in recent
years [6, 7, 8, 9, 10, 11, 12], as many application areas have
Usage of Unmanned Aerial Vehicles (UAVs) has exhibited taken an interest in their beneficent conclusions. Heli-
an expeditious escalation recently. UAVs are employed copter UAVs fault diagnosis [13] is one, and sensor fault
across many civil applications [1]. They provided a signif- diagnosis [14] is two. Fault diagnosis can be achieved
icant role in Infrastructural, Agricultural, Transporting, by signal processing or machine learning approaches, or
Security, telecommunications, and many other applica- based on both. Noting that, a recent study showed that
tions. In the past decade and in contrarily, UAVs have implementing machine learning onto signal processing is
been used in aerial surveillance for military purposes. sufficient [15]. Figure 1 describes the different methods
State, local and federal governments, including govern- of fault detection and isolation (FDI). Minimizing ma-
ment officials among many thrived countries, employed chine learning into hardware and analytical redundancy.
UAVs for aerial surveillance [2]. UAVs were also imple- This review paper will elaborate on machine learning
mented in monitoring Power transmission lines [3]. Now methods in fault diagnosis. Machine learning is one of
that UAVs are employed in both civil and military appli- the major data-driven approaches in fault diagnosis and
cations, studies regarding effectiveness and endurance has been used in many variant aspects regarding UAVs.
have risen in the past years [4]. Their ascendancy gives Figure 2 categorizes the machine learning methods into
them the privilege of replacing humans in jobs that can three approaches: supervised, unsupervised, and rein-
be repetitive, hard, or even dangerous [5]. While relying forcement learning. Artificial Intelligence (AI) is evolving
on UAVs for performance is increasing, faults started oc- in both the short and long-term processes [16, 17]. Ma-
curring despite the modern technologies and advanced chine learning (ML) methods have predicted the battery
manufacturing. A UAV system is partially composed of life of UAVs more efficiently than general methods of
other subsystems, which are consistently vulnerable to physics failure [18], especially in non-stationary vibra-
faults. In order to avoid defects, a prediction of faults in tions [19]. Usage of UAVs in communication has also
a manner of fault diagnosing methods has taken place in led to various problems, problems that were solved by
many recent studies on different fields and applications. adopting machine learning methods [20, 21]. Real-life
Fault diagnosis means diagnosing the event of defi- scenarios of security monitoring wildfires using machine
ciencies within the utilitarian units of the process, which learning methods have demonstrated the effectiveness of
leads to undesired or intolerable behavior of the com- fault detection in many aspects [22]. Another application
plete framework. Studies and reviews on fault diagnosis is the detection of the disastrous citrus greening, where
drones proved to be more efficient regarding inspections
SYSYEM 2022: 8th Scholar’s Yearly Symposium of Technology, Engi- due to their wide coverage. Machine learning methods
neering and Mathematics, Brunek, July 23, 2022 for citrus greening diagnosis were discussed, compared,
" Luttfi.A.AlHaddad@uotechnology.edu.iq (L. A. Al-Haddad);
and elaborated on, demonstrating their high accuracy in
Alaa.A.Jaber@uotechnology.edu.iq (A. Jaber)
0000-0001-5709-195X (A. Jaber) fault diagnosis [23, 24? ].
© 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License
Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings
http://ceur-ws.org
ISSN 1613-0073
CEUR Workshop Proceedings (CEUR-WS.org)
19
Luttfi A. Al-Haddad et al. CEUR Workshop Proceedings 19–25
2.1. Artificial Neural Network ANN
Artificial Neural Networks (ANNs) are known to be the
most commonly used method of machine learning ap-
proaches as they have evolved due to their flexibility and
ease of coding[48, 49]. In [29], a prototype of a fault diag-
nosis pattern to identify and recognize damaged blades
of a multirotor UAV was used. The ANN was introduced
to identify some particular features of emitted acoustic
emissions and signals. In the proposed technique, an
accurate fault classifier prospered. The recordings of the
noise emissions from a UAV were utilized to construct a
classification model to identify the unbalances of blades
in a UAV blade [36]. The authors have developed a model
Figure 1: Fault Detection and Isolation (FDI) Methods Clas- based on an artificial neural network to detect the unbal-
sification [5] ances of a quadcopter blade. The indoor test experiments
have shown a promising fault detection method in UAV
blades. Hence, below are the most popularly used NN
techniques in this regard.
2.1.1. Convolutional Neural Network CNN
CNN is of wide-range use [50]. In [30], the authors have
introduced a price-conscious fault detection method in a
large fixed-wing UAV. Six different classifiers were used
where the convolutional neural network-based classi-
fier reflected good accurate results despite the longest
time of these results. The experimental results have
demonstrated an effective model to reduce expenses on
computing equipment that ensures the same overall ef-
ficiency of the fault diagnosis system. The work in [31]
suggests a method to localize the acoustic emissions in
Figure 2: Machine Learning Overview[28] plate-like structures. One sensor and a convolutional
neural network algorithm were used where intentional
small damages were made to the system. This work can
2. Commonly Applied Machine be similar to a fixed-wing UAV structure. Audio noise
was recorded during the flight of a UAV with a damaged
Learning Methods For UAVs propeller, where the detection model was trained based
Fault Diagnosis on the convolutional neural network in [35]. Augmen-
tation of transfer learning with deep learning has made
The progression in machine learning techniques, sen- the CNN more functional based on experimental data
sors, and IT innovations have opened the entryways validation. The authors of [43] have taken actual test
for UAV applications in numerous divisions. The main flight data of a fixed-wing UAV and implemented them
divisions, be that as it may, are wireless networks, mil- in a compound fault diagnosis and labeling method. Five
itary, agribusiness, mining, and many others [25]. In a classifiers were used, including a fully convolutional neu-
short time, implementing machine learning techniques ral network (FCNN) and a modified CNN. The diagnosing
to detect faults in UAVs has taken the attention of numer- performance is improved according to the experimental
ous previous research studies that that involve. Where results and comparison of the five methods.
it is essential to consider the authenticity and original-
ity of the acquired dataset utilizing signal processing or 2.1.2. Long and Short-Term Memory Neural
other approaches [26, 27]. While different methods were Network LSTM NN
used, this review paper has considered the most com-
mon modern techniques and approaches. Overview of In [38], an airborne acceleration sensor is used to detect
exiting research studies on fault diagnosis of UAVs using faults of blades in a quadcopter using a long and short-
machine learning techniques are listed in Table 1. term memory neural network-based model. The accuracy
of this algorithm is proved to be sufficient compared to
20
Luttfi A. Al-Haddad et al. CEUR Workshop Proceedings 19–25
Table 1
Summary on existing studies on fault diagnosis of UAVs using machine learning
Machine Learning Method UAV Type Part of Fault Detection
Artificial Neural Network [29] Quadcopter Damaged Blades
Decision Tree and Convolutional Neural Network[30] Fixed-Wing Maintenance purposes
Convolutional Neural Network [31] - Health monitoring
Support Vector Machine and K Nearest Neighbor [32] Fixed-Wing Damaged Wing
Motor base loosening and Dam-
Self-Organizing Map [33] Quadcopter
aged blades
Damaged Blades Loosening of
K Nearest Neighbor [34] Quadcopter Motor Screw and Loosening of
Arm Screw
Convolutional Neural Network [35] Quadcopter Damaged Blades
Artificial Neural Network [36] Quadcopter Unbalanced Blades
Amplitude in normal achieved
K Nearest Neighbor[37] Fixed-Wing
flights
Long and Short-Term Memory Neural Network [38] Quadcopter Damaged Blades
Deep Residual Shrinkage Neural Network [39] Quadcopter Damaged Blades
Radial Basis Function Neural Network [40] Quadcopter Actuators
Long and Short-Term Memory Neural Network [41] Fixed-Wing Wing
Support Vector Machine[42] Quadcopter Gyro and Accelerometer
Convolutional Neural Network [43] Fixed-Wing Wing
Decision Tree, Support Vector Machines and K Nearest
Quadcopter Motor, Bearing and Blades
Neighbor [44]
Back Propagation Neural Network[45] Quadcopter Sensors
Radial Basis Function Neural Network [46] Fixed-Wing Sensors
Fuzzy Neural Network [47] Fixed-Wing Actuators
other neural networks, while the vibrations signals in the 2.1.4. Other Neural Networks
airframe were recorded experimentally and translated
In [39], the authors have developed and upgraded the
into codes using the fault diagnosis method. A fixed-
used neural networks. Damaged blades in a quadcopter
wing UAV fault diagnosis system based on five models,
diagnosis based on a deep residual shrinkage network
one of which was a long and short-term memory neural
and an extra convolution layer have both emerged to
network [41]. Convenient predictions were provided
produce an upgraded neural network algorithm named
based on numerical simulations.
1D-WIDRSN. The experimental statistical analysis has
shown the effectiveness and accuracy of the hybrid algo-
2.1.3. Radial Basis Function Neural Network RBF rithm compared to normal neural networks. have used a
NN back propagation neural network (BPNN) as a machine
In [40], the authors have introduced a fault-tolerant con- learning method to diagnose faults in a sensor of a quad-
trol approach to detect actuator faults in a quadcopter. A copter. Then, it was optimized by a genetic algorithm
normal adaptive sliding mode control is combined with to fasten the convergence. The results are shown ex-
a radial basis function neural network, introducing a perimentally, supporting that enhanced BPNN is more
modified adaptive sliding mode control approach. An efficient in fault diagnosis. A strategy of scattered fault-
experimental, numerical comparison between the two is tolerant cooperative control to acquire a synchronized
elaborated, showing the significant role of a radial basis track control of UAVs was introduced in [47] by using
function network. The authors of [46] have implemented fuzzy neural networks. An experimental approach where
machine learning neural networks into the fault detec- the following UAV tracks the behavior of the leading UAV
tion methods. A radial basis function neural network was is conducted regardless of the actuator faults. The simu-
used to minimize time due to the algorithm’s flexibility lation results are then discussed to prove the adequacy
when dealing with nonlinear environments. Sensor faults of the proposed strategy.
in fixed-wing UAVs are proved to be easily detected using
the proposed system experimentally and statistically.
21
Luttfi A. Al-Haddad et al. CEUR Workshop Proceedings 19–25
Table 2
Fault Diagnosis ML Methods comparative results
Machine Learning Method UAV Type Part of Fault Detection
ML Method Fault Detection Part UAV type
CNN 17.4% Blades 32% Quadcopter 57.89%
K-NN 17.4% Wing 16% Fixed-Wing 36.84%
SVM 13% Motor 12%
ANN 8.7% Actuators 8%
LSTM NN 8.7% Motor Bearing 8%
RBF NN 8.7% Sensors 8%
DT 8.7% Others 16%
SOM 4.3%
Other NN 13%
2.2. K Nearest Neighbor K-NN assembling method to exhibit a model for diagnosing
health status in a quadcopter UAV. Vibration features of
The work in [32] describes preliminary damage diag-
three flight conditions (normal, motor base mount loose,
nosing and classification system for a fixed-wing UAV.
unbalanced broken blades) were recorded and trained in
The system includes a description of data analysis from a
a system that has assembled variant vibration patterns
piezoelectric sensor system with independent component
of fault situations. The experimental results have proved
analysis and machine learning methods. One of which
that the method can also predict the occurrence of the
was the subspace K-nearest neighbor with the best results
fault, not only diagnose it.
and accuracy. In [34], variant faults in a quadcopter UAV
were examined in a fault diagnosis system. Damaged
parts were blades, armature eccentric, and motor base 2.4. Support Vector Machine SVM
loosening. Pulse and vibration signals were recorded and SVM is used in many different aspects [51]. In [42], the
analyzed using a machine learning method employing authors simulated an aircraft model and utilized it to
K-NN. Experimental results demonstrate the high effi- generate data and test some designed algorithms. The
ciency of the used method. Authors of [37] suggest an simulated measurements were collected from random
innovative system for fault diagnosis of fixed-wing UAVs flight data. A supervised fault diagnosing method based
(FW-UAVs), where the procedure dynamics, operation on SVM was utilized to identify the faulty and nominal
conditions, changing data density, and procedure dis- flight states in loss of effectiveness in control surfaces
turbance are evaluated. A modified algorithm utilizing of a drone UAV. Results encourage the use of SVM in
Shared Nearest Neighbor based Distance (SNND) and fault diagnosis due to accurate and effective acquired
a K-Nearest Neighbor algorithm hiring SNND (SNND- accuracy. Furthermore, as discussed in subsection 2.2.
KNN) was proposed to realize offline operation condition the authors of [32] have also adopted the use of support
classification and online identification. The results have vector machine in fault diagnosis. An average result
confirmed the suitability of algorithms for fault diagnosis accuracy was obtained using SVM. In addition, authors
of FW-UAVs. Generally, the malfunctions of blades, bear- of [44] and aand as discussed in subsection 2.2, have
ings, and eccentrics are well-known in motors of UAVs. adopted the SVM where the best results were acquired
The recorded sound data of the motors were analyzed in based on it
a fault diagnosis system of the mentioned malfunctions
in [44]. Important feature extraction employing signal
processing and different machine learning techniques, 2.5. Decision Tree
including K-NN, were used in the system network where As discussed in subsections 2.1.2 and 2.2, the authors in
all algorithms proved high result efficiency. High accu- [30] and [44] have adopted the decision tree method in
racy in the proposed approach demonstrated that the machine learning fault diagnosis where Gradient-based
study would put up to the reflections in the pertinent decision tree have showed better accuracy over normal
field. decision tree machine learning methods.
2.3. Self-Organizing Map SOM
The authors of [33] have embraced the self-organizing
map machine learning method, which is an unsupervised
22
Luttfi A. Al-Haddad et al. CEUR Workshop Proceedings 19–25
3. Conclusion advanced scince Engineering Information Technol-
ogy 10 (2020).
According to the statistical analysis of table 2 and based [7] R. Brociek, G. Magistris, F. Cardia, F. Coppa,
on this review article, we have concluded the following: S. Russo, Contagion prevention of covid-19 by
means of touch detection for retail stores, in: CEUR
• Neural Network Methods are the most used tech-
Workshop Proceedings, volume 3092, CEUR-WS,
niques concerning fault diagnosis of different part
2021, pp. 89–94.
of UAVs with a total percentage of 56.5
[8] J. S. Mohammed, J. A. Abdulhady, Rolling bearing
• Blades are more vulnerable to damage conditions fault detection based on vibration signal analysis
as their percentage is over 30% in the parts were and cumulative sum control chart, FME Transac-
recent studies conducted fault diagnosis on. tions 49 (2021) 684–695.
• Type of UAV percentages proves that drones are [9] C. Napoli, G. Pappalardo, E. Tramontana, Improv-
on a heavy usage term and hence more suscepti- ing files availability for bittorrent using a diffusion
ble to damage . model, IEEE Computer Society, 2014, pp. 191–196.
Adding up, the developing request for secure flights of doi:10.1109/WETICE.2014.65.
unmanned aerial vehicles requires modern and worldly- [10] A. A. Jaber, R. Bicker, The optimum selection of
wise fault diagnosis methods not only for faults in blades wavelet transform parameters for the purpose of
and wings but also in other UAV subsystems. In this fault detection in an industrial robot, in: 2014 IEEE
respect, a promising approach that appears to have cap- International Conference on Control System, Com-
tured the attention of researchers in recent years is the puting and Engineering (ICCSCE 2014), IEEE, 2014,
hybrid fault diagnosis methods that delicately address pp. 304–309.
the undesired behavior of an unmanned aerial vehicle [11] S. M. Jawad, A. A. Jaber, Bearings health monitor-
based on combined machine learning techniques or/with ing based on frequency-domain vibration signals
signal processing for important feature extraction. analysis, Engineering and Technology Journal 41
(2022) 86–95.
[12] N. Brandizzi, V. Bianco, G. Castro, S. Russo, A. Wa-
References jda, Automatic rgb inference based on facial emo-
tion recognition, in: CEUR Workshop Proceedings,
[1] H. Shakhatreh, A. Sawalmeh, A. Al-Fuqaha, Z. Dou, volume 3092, CEUR-WS, 2021, pp. 66–74.
E. Almaita, I. Khalil, N. Othman, A. Khreishah, [13] X. Qi, J. Qi, D. Theilliol, Y. Zhang, J. Han, D. Song,
M. Guizani, Unmanned aerial vehicles: A survey C. Hua, A review on fault diagnosis and fault toler-
on civil applications and key research challenges. ant control methods for single-rotor aerial vehicles,
arxiv 2018, arXiv preprint arXiv:1805.00881 (????). Journal of Intelligent & Robotic Systems 73 (2014)
[2] T. F. Abiodun, Usage of drones or unmanned aerial 535–555.
vehicles (uavs) for effective aerial surveillance, map- [14] Y. H. Gao, D. Zhao, Y. B. Li, Uav sensor fault diag-
ping system and intelligence gathering in combat- nosis technology: A survey, in: Applied Mechanics
ing insecurity in nigeria, African Journal of Social and Materials, volume 220, Trans Tech Publ, 2012,
Sciences and Humanities Research 3 (2020) 29–44. pp. 1833–1837.
[3] S. Y. Wong, C. W. C. Choe, H. H. Goh, Y. W. Low, [15] A. Bondyra, P. Gasior, S. Gardecki, A. J. Kasin-
D. Y. S. Cheah, C. Pang, Power transmission line ski, Development of the sensory network for the
fault detection and diagnosis based on artificial in- vibration-based fault detection and isolation in the
telligence approach and its development in uav: A multirotor uav propulsion system., in: ICINCO (2),
review, Arabian Journal for Science and Engineer- 2018, pp. 112–119.
ing 46 (2021) 9305–9331. [16] R. Giuliano, The next generation network in 2030:
[4] G. Capizzi, C. Napoli, S. Russo, M. Woźniak, Lessen- Applications, services, and enabling technologies,
ing stress and anxiety-related behaviors by means in: 2021 8th International Conference on Electrical
of ai-driven drones for aromatherapy, in: CEUR Engineering, Computer Science and Informatics
Workshop Proceedings, volume 2594, CEUR-WS, (EECSI), IEEE, 2021, pp. 294–298.
2020, pp. 7–12. [17] F. Bonanno, G. Capizzi, G. L. Sciuto, C. Napoli,
[5] G. K. Fourlas, G. C. Karras, A survey on fault di- Wavelet recurrent neural network with semi-
agnosis and fault-tolerant control methods for un- parametric input data preprocessing for micro-wind
manned aerial vehicles, Machines 9 (2021) 197. power forecasting in integrated generation systems,
[6] T. A. Dhomad, A. Jaber, Bearing fault diagnosis in: 5th International Conference on Clean Electri-
using motor current signature analysis and the ar- cal Power: Renewable Energy Resources Impact,
tificial neural network, International Journal on ICCEP 2015, 2015, p. 602 – 609.
23
Luttfi A. Al-Haddad et al. CEUR Workshop Proceedings 19–25
[18] S. S. Mansouri, P. Karvelis, G. Georgoulas, G. Niko- 3955.
lakopoulos, Remaining useful battery life predic- [30] K. Zheng, G. Jia, L. Yang, C. Liu, A cost-sensitive
tion for uavs based on machine learning, IFAC- diagnosis method based on the operation and main-
PapersOnLine 50 (2017) 4727–4732. tenance data of uav, Applied Sciences 11 (2021)
[19] A. A. Jaber, R. Bicker, A simulation of non- 11116.
stationary signal analysis using wavelet transform [31] A. Ebrahimkhanlou, S. Salamone, Single-sensor
based on labview and matlab, in: 2014 European acoustic emission source localization in plate-like
Modelling Symposium, IEEE, 2014, pp. 138–144. structures using deep learning, Aerospace 5 (2018)
[20] P. S. Bithas, E. T. Michailidis, N. Nomikos, 50.
D. Vouyioukas, A. G. Kanatas, A survey on machine- [32] M. Anaya, H. Cerón, J. Vitola, D. Tibaduiza, F. Pozo,
learning techniques for uav-based communications, Damage classification based on machine learning
Sensors 19 (2019) 5170. applications for an un-manned aerial vehicle, Struc-
[21] N. Dat, V. Ponzi, S. Russo, F. Vincelli, Supporting tural Health Monitoring 2017 (2017).
impaired people with a following robotic assistant [33] D.-L. Cheng, W.-H. Lai, Application of self-
by means of end-to-end visual target navigation organizing map on flight data analysis for quad-
and reinforcement learning approaches, in: CEUR copter health diagnosis system., International
Workshop Proceedings, volume 3118, CEUR-WS, Archives of the Photogrammetry, Remote Sensing
2021, pp. 51–63. & Spatial Information Sciences (2019).
[22] D. Alexandrov, E. Pertseva, I. Berman, I. Pantiukhin, [34] W.-H. Lai, S.-T. Tsai, D.-L. Cheng, Y.-R. Liang, Appli-
A. Kapitonov, Analysis of machine learning meth- cation of wavelet scattering and machine learning
ods for wildfire security monitoring with an un- on structural health diagnosis for quadcopter, Ap-
manned aerial vehicles, in: 2019 24th conference of plied Sciences 11 (2021) 10297.
open innovations association (FRUCT), IEEE, 2019, [35] W. Liu, Z. Chen, M. Zheng, An audio-based fault di-
pp. 3–9. agnosis method for quadrotors using convolutional
[23] Y. Lan, Z. Huang, X. Deng, Z. Zhu, H. Huang, neural network and transfer learning, in: 2020
Z. Zheng, B. Lian, G. Zeng, Z. Tong, Comparison American Control Conference (ACC), IEEE, 2020,
of machine learning methods for citrus greening pp. 1367–1372.
detection on uav multispectral images, Computers [36] G. Iannace, G. Ciaburro, A. Trematerra, Fault diag-
and electronics in agriculture 171 (2020) 105234. nosis for uav blades using artificial neural network,
[24] F. Bonanno, G. Capizzi, S. Coco, C. Napoli, A. Lau- Robotics 8 (2019) 59.
dani, G. Sciuto, Optimal thicknesses determination [37] S. Liang, S. Zhang, Y. Huang, X. Zheng, J. Cheng,
in a multilayer structure to improve the spp effi- S. Wu, Data-driven fault diagnosis of fw-uavs with
ciency for photovoltaic devices by an hybrid fem - consideration of multiple operation conditions, ISA
cascade neural network based approach, IEEE Com- transactions (2021).
puter Society, 2014, pp. 355–362. doi:10.1109/ [38] X. Zhang, Z. Zhao, Z. Wang, X. Wang, Fault de-
SPEEDAM.2014.6872103. tection and identification method for quadcopter
[25] A. I. Khan, Y. Al-Mulla, Unmanned aerial vehicle based on airframe vibration signals, Sensors 21
in the machine learning environment, Procedia (2021) 581.
computer science 160 (2019) 46–53. [39] P. Yang, H. Geng, C. Wen, P. Liu, An intelligent
[26] F. Fallucchi, M. Gerardi, M. Petito, E. W. De Luca, quadrotor fault diagnosis method based on novel
Blockchain framework in digital government for deep residual shrinkage network, Drones 5 (2021)
the certification of authenticity, timestamping and 133.
data property, in: Proceedings of the 54th Hawaii [40] N. P. Nguyen, S. K. Hong, Fault-tolerant control
International Conference on System Sciences, 2021, of quadcopter uavs using robust adaptive sliding
p. 2307. mode approach, Energies 12 (2018) 95.
[27] A. Simonetta, A. Trenta, M. C. Paoletti, A. Vetrò, [41] A. Cui, Y. Zhang, P. Zhang, W. Dong, C. Wang, In-
Metrics for identifying bias in datasets, SYSTEM telligent health management of fixed-wing uavs: A
(2021). deep-learning-based approach, in: 2020 16th In-
[28] M.-A. Lahmeri, M. A. Kishk, M.-S. Alouini, Artificial ternational Conference on Control, Automation,
intelligence for uav-enabled wireless networks: A Robotics and Vision (ICARCV), IEEE, 2020, pp.
survey, IEEE Open Journal of the Communications 1055–1060.
Society 2 (2021) 1015–1040. [42] E. Baskaya, M. Bronz, D. Delahaye, Fault detection
[29] A. Bondyra, M. Kołodziejczak, R. Kulikowski, & diagnosis for small uavs via machine learning,
W. Giernacki, An acoustic fault detection and isola- in: 2017 IEEE/AIAA 36th Digital Avionics Systems
tion system for multirotor uav, Energies 15 (2022) Conference (DASC), IEEE, 2017, pp. 1–6.
24
Luttfi A. Al-Haddad et al. CEUR Workshop Proceedings 19–25
[43] K. Zheng, G. Jia, L. Yang, J. Wang, A compound
fault labeling and diagnosis method based on flight
data and bit record of uav, Applied Sciences 11
(2021) 5410.
[44] A. Altinors, F. Yol, O. Yaman, A sound based method
for fault detection with statistical feature extraction
in uav motors, Applied Acoustics 183 (2021) 108325.
[45] Y. Chen, C. Zhang, Q. Zhang, X. Hu, Uav fault de-
tection based on ga-bp neural network, in: 2017
32nd Youth Academic Annual Conference of Chi-
nese Association of Automation (YAC), IEEE, 2017,
pp. 806–811.
[46] I. Samy, I. Postlethwaite, D.-W. Gu, I. Fan, Detection
of multiple sensor faults using neural networks-
demonstrated on an unmanned air vehicle (uav)
model (2010).
[47] Z. Yu, Y. Zhang, Z. Liu, Y. Qu, C.-Y. Su, Distributed
adaptive fractional-order fault-tolerant cooperative
control of networked unmanned aerial vehicles via
fuzzy neural networks, IET Control Theory & Ap-
plications 13 (2019) 2917–2929.
[48] G. Capizzi, G. Lo Sciuto, C. Napoli, E. Tramontana,
An advanced neural network based solution to en-
force dispatch continuity in smart grids, Applied
Soft Computing Journal 62 (2018) 768 – 775.
[49] G. Capizzi, G. Lo Sciuto, C. Napoli, R. Shikler,
M. Wozniak, Optimizing the organic solar cell man-
ufacturing process by means of afm measurements
and neural networks, Energies 11 (2018).
[50] G. C. Cardarilli, L. Di Nunzio, R. Fazzolari, D. Gi-
ardino, A. Nannarelli, M. Re, S. Spanò, A pseudo-
softmax function for hardware-based high speed
image classification, Scientific reports 11 (2021)
1–10.
[51] R. R. Sarra, A. M. Dinar, M. A. Mohammed, K. H.
Abdulkareem, Enhanced heart disease prediction
based on machine learning and 𝜒2 statistical opti-
mal feature selection model, Designs 6 (2022) 87.
25