The Fault Detection of Aero-engine Sensor Based on Deep Belief Networks

DOI

10.2991/icmcm-16.2016.18How to use a DOI?

Keywords

Aeroengine sensor; Fault Detection; Deep learning; Deep belief network; Flight parameter

Abstract

Due to its high complexity and accuracy, the faults of aeroengine sensor can not be described with precise mathematical model and it is hard to detect the faults using traditional redundant method. Focus on this problem, a big data-deep learning based model is established to detect the faults of sensor of the engine air exhaust temperature. The classified model using deep belief network(DBN) is built firstly and is trained using a great number of data collected by flight parameter recorder. The model is able to classify the fault through feature learning layer by layer. The results of simulation experiment show that the accuracy with artificial feature extraction is 98% while 96.6% without it. The accuracy of this model is higher than the model using BP neural net and support vector machine(SVM) in both conditions which shows the superiority of the DBN algorithm in sensors fault diagnosis.

Copyright

© 2016, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Chuang Guo
AU  - Xiao-fei Zheng
AU  - Bin Yao
PY  - 2016/12
DA  - 2016/12
TI  - The Fault Detection of Aero-engine Sensor Based on Deep Belief Networks
BT  - Proceedings of the 2016 7th International Conference on Mechatronics, Control and Materials (ICMCM 2016)
PB  - Atlantis Press
SP  - 85
EP  - 92
SN  - 2352-5401
UR  - https://doi.org/10.2991/icmcm-16.2016.18
DO  - 10.2991/icmcm-16.2016.18
ID  - Guo2016/12
ER  -