Incipient Fault Diagnosis Method of Vehicle Power Supply Based on Feature Optimization and Deep Learning

被引：0

作者：

Li W. ^{[1
,2
,3
]}

Han Y. ^{[1
,2
]}

Sun X. ^{[4
]}

机构：

[1] College of Electrical and Information Engineering, Lanzhou University of Technology, Gansu, Lanzhou

[2] Key Laboratory of Gansu Advanced Control for Industrial Processes, Gansu, Lanzhou

[3] National Demonstration Center for Experimental Electrical and Control Engineering Education, Lanzhou University of Technology, Gansu, Lanzhou

[4] Lanzhou Power Supply Vehicle Research Institute Co., Ltd., Gansu, Lanzhou

来源：

Binggong Xuebao/Acta Armamentarii | 2022年 / 43卷 / 11期

关键词：

incipient fault diagnosis; recursive feature elimination; stacked auto-encoder; vehicle power supply;

D O I：

10.12382/bgxb.2021.0577

中图分类号：

学科分类号：

摘要：

The vehicle power supply is the main power source for the training and emergency management of military equipment. The accurate diagnosis of incipient faults can effectively prevent the occurrence of serious faults. However, monitoring data is often redundant, making it difficult to effectively extract the symptoms of incipient faults. Aiming at this issue, a new intelligent incipient fault diagnosis method is proposed based on a combination of recursive feature elimination (RFE) and stacked auto-encoders (SAE). The collected feature variables are sorted based on their importance by using RFE to eliminate redundancy and extract the optimal subset of feature features. Then, the feature subset is used as the input of the SAE deep network, and the incipient fault category is used as the output, resulting in the effective diagnosis of incipient faults in vehicle power supply. The simulation results show that compared with SAE and shallow neural networks, the diagnosis accuracy of the proposed method is significantly improved, reaching 95.4%. © 2022 China Ordnance Society. All rights reserved.

引用

页码：2935 / 2944

页数：9

共 30 条

[1] MA X J, YUAN D, XIANG Y, Et al., Research on inte-grated power system and its key techniques of ground combat platform, Acta Armamentarii, 38, 2, pp. 396-406, (2017)
[2] CAI H R, ZHUANG Y P, YANG S W., Low-noise mobile power plant design and selection [J], Telecom Power Technologies, 25, 3, pp. 73-78, (2008)
[3] GAO Q, LIAO Z L, YUAN D, Et al., Effects of key parameters of integrated power system on vehicle stability [J], Acta Armamentarii, 41, 9, pp. 1727-1735, (2020)
[4] LI J, ZHOU D H, SI X S, Et al., Review of incipient fault diagnosis methods, Control Theory & Applications, 29, 12, pp. 1517-1529, (2012)
[5] WEN C L, LU F Y, BAO Z J, Et al., A review of data driven-based incipient fault diagnosis [J], Acta Automatica Sinica, 42, 9, pp. 1285-1299, (2016)
[6] PENG K X, MA L, ZHANG K., Review of quality-related fault detection and diagnosis techniques for complex in-dustrial processes, Acat Automatica Sinica, 43, 3, pp. 349-365, (2017)
[7] BHALLA D, BANSAL R K, GUPTA H O., Function analysis based rule extraction from artificial neural networks for transformer incipient fault diagnosis [J], International Journal of Electrical Power & Energy Systems, 43, 1, pp. 1196-1203, (2012)
[8] ZHANG Z Y, WANG Y, WAMNG K S., Fault diagnosis and prognosis using wavelet packet decomposition Fourier transform and artificial neural network[J], Journal of Intelligent Manufacturing, 24, 6, pp. 1213-1227, (2013)
[9] HINTON G E, SALAKHUTDINOV R R., Reducing the dimensionality of data with neural networks [J], Science, 313, 5786, pp. 504-507, (2006)
[10] TAMILSELVAN P, WANG P F., Failure diagnosis using deep belief learning based health state classification[J], Re-liability Engineering & System Safety, 115, pp. 124-135, (2013)

← 1 2 3 →