# Probabilistic Principal Components and Mixtures, How This Works

Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 9339)

## Abstract

Classical Principal Components Analysis (PCA) is widely recognized as a method for dimensionality reduction and data visualization. This is a purely algebraic method, it considers just some optimization problem which fits exactly to the gathered data vectors with their particularities. No statistical significance tests are possible. An alternative is to use probabilistic principal component analysis (PPCA), which is formulated on a probabilistic ground. Obviously, to do it one has to know the probability distribution of the analyzed data. Usually the Multi-Variate Gaussian (MVG) distribution is assumed. But what, if the analyzed data are decidedly not MVG? We have met such problem when elaborating multivariate gearbox data derived from a heavy duty machine. We show here how we have dealt with the problem.

In our analysis, we assumed that the considered data are a mixture of two groups being MVG, specifically: each of the sub-group follows a probabilistic principal component (PPC) distribution with a MVG error function. Then, by applying Bayesian inference, we were able to calculate for each data vector x its a posteriori probability of belonging to data generated by the assumed model. After estimation of the parameters of the assumed model we got means - based on a sound statistical basis - for constructing confidence boundaries of the data and finding outliers.

### Keywords

Probabilistic principal components Multi-variate normal distribution Mixture models Un-mixing multivariate data Condition monitoring Gearbox diagnostics Healthy state Probabilities a posteriori Outliers

## Preview

### References

1. 1.
Bartelmus, W., Zimroz, R.: A new feature for monitoring the condition of gearboxes in nonstationary operating systems. Mechanical Systems and Signal Processing 23(5), 1528–1534 (2009)
2. 2.
Bartkowiak, A., Zimroz, R.: Outliers analysis and one class classification approach for planetary gearbox diagnosis. Journal of Physics: Conference Series 305 (1), art. no. 012031 (2011)Google Scholar
3. 3.
Bartkowiak, A., Zimroz, R.: Data dimension reduction and visualization with application to multidimensional gearbox diagnostics data: Comparison of several methods. Diffusion and Defect Data Pt.B: Solid State Phenomena 180, 177–184 (2012)Google Scholar
4. 4.
Bishop, C.M.: Neural Networks for Pattern Recognition. Oxford University Press (1995)Google Scholar
5. 5.
Chen, J., Zhang, C., Zhang, X., et al.: Customized lifting multiwavelet packet information entropy for equipment condition identification. Smart Mater. Struct. 22, 095022 (14pp). IOPPublishing (2013)Google Scholar
6. 6.
Chen, J., Zhang, C., Zhang, X., et al.: Planetary gearbox condition monitoring of ship-based satellite communication antennas using ensemble multivawelet analysis methods. Mech. Syst. Signal Process. 54, 277–292 (2014)Google Scholar
7. 7.
Cocconcelli, M., Zimroz, R., Rubini, R., Bartelmus, W.: Kurtosis over energy distribution approach for STFT enhancement in ball bearing diagnostics. In: Condition Monitoring of Machinery in Non-Stationary Operations 2012, Part I, pp. 51–59 (2012)Google Scholar
8. 8.
Heyns, T., Heyns, P.S., de Villiers, J.P.: Combining synchronous averaging with a Gaussian mixture model novelty detection scheme for vibration-based condition monitoring of a gearbox. Mech. Syst. Signal Process. 32, 200–215 (2012)
9. 9.
Heyns, T., Heyns, P.S., Zimroz, R.: Combining discrepancy analysis with sensorless signal resampling for condition monitoring of rotating machines under fluctuating operations. Int. J. of Condition Monitoring 2(2), 52–58 (2012)
10. 10.
Jardine, A.K.S., Lin, D., Banjevic, D.: A review on machinery diagnostics and prognostics implementing condition-based maintenance. Mech. Syst. Signal Process. 20, 1483–1510 (2006)
11. 11.
Jedlinski, L., Jonak, J.: Early fault detection in gearboxes based on support vector machines and multilayer perceptron with a continuous wavelet transport. Applied Soft Computing Journal (2015) (in print)Google Scholar
12. 12.
Khan, S.S., Madden, M.G.: One-class classification: taxonomy of study. The knowledge Engineering Review. Cambridge Univ. Press (2014)Google Scholar
13. 13.
Lei, Y., Lin, J., Zuo, M.J., He, Z.: Condition monitoring and fault detection of planetary gearboxes: A review. Measurement 48, 292–306 (2014)
14. 14.
Montechiesi, L., Cocconcelli, M., Rubini, R.: Artificial immune system via Euclidean Distance Minimization for anomaly detection in bearings. Mech. Syst. Signal Processing (2015) (in print). http://dx.doi.org/101016/j.ymssp.2015.04.017
15. 15.
Nabney, I.T.: NETLAB, Algorithms for Pattern Recognition. Springer, Heidelberg (2002)
16. 16.
Pimentel, M.A.F., Clifton, D.A., Clifton, L., Tarassenko, L.: A review of novelty detection. Signal Processing 99, 215–249 (2014)
17. 17.
Tipping, M.E., Bishop, C.M.: Probabilistic principal component analysis. J. Roy. Statist. Soc. B 61, 611–622 (1999)
18. 18.
Zheng, J., Zhang, H., Cattani, C., et al.: Dimensionality reduction by supervised neighbor embedding using Lapacian search. Computational and Mathematical Methods in Medicine (Hindawi) 2014, 594379, 14pp (2014). http://dx.doi.org/10.1155/2014/594379
19. 19.
Zimroz, R., Bartkowiak, A.: Investigation on spectral structure of gearbox vibration signals by principal component analysis for condition monitoring purposes. Journal of Physics: Conference Series 305 (1), art. no. 012075 (2011)Google Scholar
20. 20.
Zimroz, R., Bartkowiak, A.: Multidimensional data analysis for condition monitoring: features selection and data classification. CM2012–MFPT2012. BINDT, June 11–14, London. Electronic Proceedings, art no. 402, pp. 1–12 (2012)Google Scholar
21. 21.
Zimroz, R., Bartkowiak, A.: Two simple multivariate procedures for monitoring planetary gearboxes in non-stationary operating conditions. Mech. Syst. Signal Process. 38(1), 237–247 (2013)