|
CSME 2024/12
Volume 45 No.6
:
543-552
|
|
Fault Detection for Train Drive Shafts Based on Principal Component Analysis of Vibration Signals
Ping-Chun Tsai a and Ying-Chen Chuang b
aDepartment of Mechanical Design Engineering, National Formosa University, No. 64, Wunhua Rd., Huwei Township, Yunlin County 632, Taiwan (R.O.C.).
bTaoyuan Metro Corporation, No. 251, Sec. 4, Linghang N. Rd., Dayuan Dist., Taoyuan City 337, Taiwan (R.O.C.).
|
Abstract:
To develop a fault detecting model for monitoring the health of a drive shaft during operation of
trains on Taipei’s Wenhu train line, this study developed a train drive shaft fault detection method by
using principal component analysis (PCA). First,
numerous features were extracted from drive shaft
vibration signals collected during train operation.
Redundant features were removed through feature
ranking, and PCA was used to reduce the features’
dimensionality. The model was ultimately established
using PCA and Hotelling’s T2 statistics. The proposed
method is constructed using only vibration data from
a healthy drive shaft. The feasibility of the proposed
methodology was validated experimentally. The
detection rate and false positive rate were 100% and
4.7%, respectively. The main contributions of this
study are as follows: Interpretable drive shaft wear
signals were identified. The diagnosis model is constructed entirely from measurements of healthy drive
shafts. The model enables real-time drive shaft monitoring.
|
|
Keywords: principal component analysis, MRT train, drive shaft, fault detection.
|
|
|
*Corresponding author; e-mail:
|
©
2024
CSME , ISSN 0257-9731
|