Accès libre

Comparative Analysis of Deep Learning and Decision Tree Approaches for Predicting Aircraft Engine Remaining Useful Life

Hassina Madjour

Hassina Madjour

Laboratory of Automation and Manufacturing, Department of Industrial Engineering, University of Batna 2Batna, Algeria
Profil Orcid
Recherchez cet auteur sur
Sciendo|Google Scholar
Madjour, Hassina
, 
Hanane Zermane

Hanane Zermane

Laboratory of Automation and Manufacturing, Department of Industrial Engineering, University of Batna 2Batna, Algeria
Profil Orcid
Recherchez cet auteur sur
Sciendo|Google Scholar
Zermane, Hanane
, 
Djemaa Rahmouni

Djemaa Rahmouni

Laboratory of Automation and Manufacturing, Department of Industrial Engineering, University of Batna 2Batna, Algeria
Profil Orcid
Recherchez cet auteur sur
Sciendo|Google Scholar
Rahmouni, Djemaa
 et 
Mohammed Djamel Mouss

Mohammed Djamel Mouss

Laboratory of Automation and Manufacturing, Department of Industrial Engineering, University of Batna 2Batna, Algeria
Profil Orcid
Recherchez cet auteur sur
Sciendo|Google Scholar
Mouss, Mohammed Djamel
  
19 nov. 2024
Comparative Analysis of Deep Learning and Decision Tree Approaches for Predicting Aircraft Engine Remaining Useful Life's Cover Image
À propos de cet article
Article précédent
Article suivant
Citez
Télécharger la couverture
Catégorie d'article: Research Article
Publié en ligne: 19 nov. 2024
Pages: 183 - 200
DOI: https://doi.org/10.2478/fas-2023-0012
Mots clés
© 2023 Hassina Madjour et al., published by Sciendo
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Fig. 1.

Simple CNN-1D architecture with two convolutional layers (Frederick et al., 2007).
Simple CNN-1D architecture with two convolutional layers (Frederick et al., 2007).

Fig. 2.

Example of a decision tree.
Example of a decision tree.

Fig. 3.

Simplified engine diagram simulated in C-MAPSS (Heimes, 2008).
Simplified engine diagram simulated in C-MAPSS (Heimes, 2008).

Fig. 4.

The distribution of the dataset’s features.
The distribution of the dataset’s features.

Fig. 5.

Bar chart of influential features.
Bar chart of influential features.

Fig. 6.

Correlation heatmap for selected dataset features.
Correlation heatmap for selected dataset features.

Fig. 7.

Diagram of model degradation of all engines.
Diagram of model degradation of all engines.

Fig. 8.

Decision Tree structure.
Decision Tree structure.

Fig. 9.

Diagram of true and predicted RUL using CNN-1D.
Diagram of true and predicted RUL using CNN-1D.

Fig. 10.

Diagram of true and predicted RUL using DT.
Diagram of true and predicted RUL using DT.

The meanings of C-MAPSS data sources_

Description Symbol Units
Total temperature at the fan inlet T2 °R
Total temperature at the LPC outlet T24 °R
Total temperature at the HPC outlet T30 °R
Total temperature at the LPT outlet T50 °R
Pressure at the fan inlet P2 Psia
Total pressure in bypass-duct P15 Psia
Total pressure at the HPC outlet P30 Psia
Physical fan speed Nf rpm
Physical core speed Nc rpm
Engine pressure ratio (P50/P2) Epr
Static pressure at the HPC outlet Ps30 Psia
Ratio of fuel flow to Ps30 Phi PPS/psi
Corrected fan speed NRf rpm
Corrected core speed NRc Rpm
Bypass ratio BPR
Burner fuel-air ratio farB
Bleed enthalpy bleed
Demanded fan speed Nf_dmd Rpm
Demanded corrected fan speed PCNR_dmd Rpm
Coolant bleed (HPT) W31HPT 1bm/s
Coolant bleed (LPT) W32LPT 1bm/s
The total temperature at the HPT outlet Parameters for calculating Health Index
Fan stall margin SmFan
LPC stall margin SmLPC
HPC stall margin SmHPC

Description of FD001 datasets_

Dataset CMAPSS (FD001)
Training engine 100
Testing engine 100
Working condition 1
Fault modes 2

The evaluation metrics_

Model MSE RMSE R2
Train set CNN-1D 459.5114 21.4362 0.7461
Test set 735.5647 27.1213 0.5707
Train set DT 567.8965 23.8305 0.6761
Test set 837.1339 28.9332 0.5588

CNN-1D structure_

Layer (type) Output Shape Param #
conv1d (Conv1D) (None, 23, 32) 96
flatten (Flatten) (None, 736) θ
dense (Dense) (None, 64) 47168
dense_1 (Dense) (None, 1) 65
Langue:
Anglais
Périodicité:
1 fois par an
Sujets de la revue:
Ingénierie, Présentations et aperçus, Ingénierie, autres
RSS Feed de la revue