Operating status of roller bearings directly affects the safety of the urban rail vehicle, so accurate identification of the state has considerably practical significance. The fault detection and isolation method of the roller bearing operational was performed by the comprehensive utilization of fast independent component analysis (Fast ICA), and kernel fisher discriminant. Fast ICA method was used to denoise and extract signal with fault frequency. The fault characteristic signal separation used Fast ICA with the normal signal applied as a signal source. The Pearson product-moment correlation coefficient had been used to determine which component was the extracted one we needed. Then eight indicators were extracted as roller bearings’ state features after Fast ICA. Based on the features under different states, the states of roller bearings on the urban rail vehicles were identified by the K-Fisher Discriminant. The experiment results indicated that the accuracies of the multiple states identification was more than 96 %, and verified the superiority of the proposed method.
[1]
Aapo Hyvärinen,et al.
A Fast Fixed-Point Algorithm for Independent Component Analysis
,
1997,
Neural Computation.
[2]
Terrence J. Sejnowski,et al.
Independent Component Analysis Using an Extended Infomax Algorithm for Mixed Subgaussian and Supergaussian Sources
,
1999,
Neural Computation.
[3]
M.R. Raghuveer,et al.
Bispectrum estimation: A digital signal processing framework
,
1987,
Proceedings of the IEEE.
[4]
Aapo Hyvärinen,et al.
Fast and robust fixed-point algorithms for independent component analysis
,
1999,
IEEE Trans. Neural Networks.
[5]
Andrew K. Chan,et al.
Neural pattern identification of railroad wheel-bearing faults from audible acoustic signals: comparison of FFT, CWT, and DWT features
,
1997,
Defense, Security, and Sensing.