Selection of Input Parameters for Multivariate Classifiersin Proactive Machine Health Monitoring by Clustering Envelope Spectrum Harmonics

Abstract:

In condition monitoring (CM) signal analysis the inherent problem of key characteristics being masked by noise can be addressed by analysis of the signal envelope. Envelope analysis of vibration signals is effective in extracting useful information for diagnosing different faults. However, the number of envelope features is generally too large to be effectively incorporated in system models. In this paper a novel method of extracting the pertinent information from such signals based on multivariate statistical techniques is developed whichsubstantialy reduces the number of input parameters required for data classification models. This was achieved by clustering possible model variables into a number of homogeneous groups to assertain levels of interdependency. Representatives from each of the groups were selected for their power to discriminate between the categorical classes. The techniques established were applied to a reciprocating compressorrig wherein the target was identifying machine states with respect to operational health through comparison of signal outputs for healthy and faulty systems. The technique allowed near perfect fault classification.In addition methods for identifying seperable classes are investigated through profiling techniques, illustrated using Andrew’s Fourier curves.

• Periodical: Applied Mechanics and Materials (Volume 798)
• Pages: 308-313
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.798.308
• Online since: October 2015
• Authors: Ann Smith*, Feng Shou Gu, Andrew Ball
• Keywords: Classifiers, Cluster Analysis, Genetic Algorithms, Reciprocating Compressor, Relevance Vector Machines

References

[1] S King, P R Bannister, D A Clifton, and L Tarassenko'Probabilistic approach to the condition monitoring of aerospace engines'IMechE Vol. 223 Part G: J. Aerospace Engineering(2009).

DOI: 10.1243/09544100jaero414

[2] M. A Zaidan, A. R. Mills R.F. Harrison'Bayesian framework for aerospace gas turbine engine prognostics' IEEE Aerospace Conference Proceedings(2013).

DOI: 10.1109/aero.2013.6496856

[3] O.W. Laslett, A.R. Mills, M.A. Zaidan, R.F. Harrison Fusing an ensemble of diverse prognostic life predictions, Aerospace Conference IEEE (2014), pp.1-10.

DOI: 10.1109/aero.2014.6836267

[4] F. Gu, A.D. Ball'Use of the smoothed pseudo-Wigner-Ville distribution in the interpretation of monitored vibration data Maintenance' vol. 10 (1995), pp.16-23.

[5] B. S. Yang, W. W. Hwang, M. H. Ko, S. J. Lee'Cavitation detection of butterfly valve using support vector machines. ' Journal of Sound Vibration, vol. 287 (2005), pp.25-43.

DOI: 10.1016/j.jsv.2004.10.033

[6] B. S. Yang, W. W. Hwang, D. J. Kim, A. Chit Tan'Condition classification of small reciprocating compressor for refrigerators using artificial neural networks and support vector machines. ' , Mechanical Systems and Signal Processing, vol. 19(2005).

DOI: 10.1016/j.ymssp.2004.06.002

[7] M. Ahmed, A. Smith, F. Gu, A.D. Ball'Fault Diagnosis of Reciprocating Compressors Using Revelance Vector Machines with A Genetic Algorithm Based on Vibration Data' Proceedings of the 20th International Conference on Automation & Computing (2014).

DOI: 10.1109/iconac.2014.6935480

[8] M.E. Tipping Sparse Baysian learning and the relevance vector machine., The journal of Machine Learning Research, vol. 1(2001), pp.211-244.

[9] M.E. Tipping, A. Faul, Analysis of sparse Baysian learning, Advances in neural information processing systems, vol. 14 (2002), pp.383-389.

[10] Manley, Bryan F. J. Multivariate Statistical Methods: A Primer, 3rd Edition. Chatfield and Colins ISBN 9781584884149. (1986).

[11] B. S. Everitt, G. Dunn'Applied Multivariate Data Analysis', 2nd Edition. Arnold, London, ISBN 0 340 74122 8(2001).

[12] Rao, B. Handbook of Condition Monitoring. Elsevier Science Ltd, Oxford, UK, ISBN 1 85617 2341 (1996).

[13] Isermann, R., Fault-Diagnosis Applications. Model-based condition monitoring; Actuators, drives, machinery, plants, sensors, and fault-tolerant systems. Springer. ISBN 978-3-642-12766-3 Springer Berlin Heidelberg New York. (2001).

DOI: 10.1007/978-3-642-12767-0

[14] A. Smith, F. Gu, A.D. BallSparse Statistical Modelling in Prognostic Health Monitoring: Discriminant Analysis and Naive Bayes Classifiers Established Through Input Parameter Selection By Variable Clustering. ICAC'15 University of Strathclyde, Glasgow, UK, 11-12 September 2015(2015 Unpublished).