Fault detection in linear electromagnetic actuators using time and time-frequency-domain features based on current and voltage measurements

In this paper an approach towards databased fault diagnosis of linear electromagnetic actuators is presented. Time and time-frequency-domain methods were applied to extract fault related features from current and voltage measurements. The resulting features were transformed to enhance class separability using either Principal Component Analysis (PCA) or Optimal Transformation. Feature selection and dimensionality reduction was performed employing a modified Fisher-ratio. Fault detection was carried out using a Support-Vector-Machine classifier trained with randomly selected data subsets. Results showed, that not only the used feature sets (time-domain/time-frequency-domain) are crucial for fault detection and classification, but also feature pre-processing. PCA transformed time-domain features allow fault detection and classification without misclassification, relying on current and voltage measurements making two sensors necessary to generate the data. Optimal transformed time-frequency-domain features allow a misclassification free result as well, but as they are calculated from current measurements only, a dedicated voltage sensor is not necessary. Using those features is a promising alternative even for detecting purely supply voltage related faults.

[1]  Keith Worden,et al.  TIME–FREQUENCY ANALYSIS IN GEARBOX FAULT DETECTION USING THE WIGNER–VILLE DISTRIBUTION AND PATTERN RECOGNITION , 1997 .

[2]  Ingrid Daubechies,et al.  Ten Lectures on Wavelets , 1992 .

[3]  Tshilidzi Marwala,et al.  Faults Detection Using Gaussian Mixture Models, Mel-Frequency Cepstral Coefficients and Kurtosis , 2006, 2006 IEEE International Conference on Systems, Man and Cybernetics.

[4]  Samuel H. Huang,et al.  System health monitoring and prognostics — a review of current paradigms and practices , 2006 .

[5]  Goutam Saha,et al.  Feature selection using singular value decomposition and QR factorization with column pivoting for text-independent speaker identification , 2010, Speech Commun..

[6]  I. Jolliffe.  Principal Component Analysis , 2005 .

[7]  Mohamed Ben-Daya,et al.  Handbook of maintenance management and engineering , 2009 .

[8]  Ian T. Jolliffe,et al.  Principal Component Analysis , 2002, International Encyclopedia of Statistical Science.

[9]  Daming Lin,et al.  A review on machinery diagnostics and prognostics implementing condition-based maintenance , 2006 .

[10]  E. Grall-Maes,et al.  Features extraction for signal classification based on Wigner-Ville distribution and mutual information criterion , 1998, Proceedings of the IEEE-SP International Symposium on Time-Frequency and Time-Scale Analysis (Cat. No.98TH8380).

[11]  F. Mörchen.  Time series feature extraction for data mining using DWT and DFT , 2003 .

[12]  Sagar V. Kamarthi,et al.  Feature Extraction From Wavelet Coefficients for Pattern Recognition Tasks , 1999, IEEE Trans. Pattern Anal. Mach. Intell..

[13]  Isabelle Guyon,et al.  An Introduction to Variable and Feature Selection , 2003, J. Mach. Learn. Res..

[14]  Lin Ma,et al.  Prognostic modelling options for remaining useful life estimation by industry , 2011 .

[15]  Fuhui Long,et al.  Feature selection based on mutual information criteria of max-dependency, max-relevance, and min-redundancy , 2003, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[16]  Bo-Suk Yang,et al.  Application of nonlinear feature extraction and support vector machines for fault diagnosis of induction motors , 2007, Expert Syst. Appl..