Applied Machine Learning: Reconstruction of Spectral Data for the Classification of Oil-Quality Levels

Abstract

In modern complex systems and machines - e.g., automobiles or construction vehicles - different versions of a "Condition Based Service" (CBS) are deployed for maintenance and supervision. According to the current state of the art, CBS is focusing on monitoring of static factors and rules. In the area of agricultural machines, these are for example operating hours, kilometers driven or the number of engine starts. The decision to substitute hydraulic oil is determined on the basis of the factors listed. A data-driven procedure is proposed instead to leverage the decision-making process. Thus, this paper presents a method to support continuous oil monitoring with the emphasis on artificial intelligence using real-world spectral oil-data. The reconstruction of the spectral data is essential, as a complete spectral analysis for the ultraviolet and visible range is not available. Instead, a possibility of reconstruction by sparse supporting wavelengths through neural networks is proposed and benchmarked by standard interpolation methods. Furthermore, a classification via a feed-forward neural network with the conjunction of Dynamic Time Warping (DTW) algorithm for the production of labeled data was developed. Conclusively, the extent to which changes in hyper-parameters (number of hidden layers, number of neurons, weight initialization) affect the accuracy of the classification results have been investigated.

Keywords

• Machine learning,Neural networks,Spectral analysis

References

1. K. Pöpping, “Das Betriebs- und Alterungsverhalten biologisch schnell abbaubarer Hydrauliköle" Dissertation, April, 2012. G. E. Newell, “Oil analysis cost‐effective machine condition monitoring technique,” Ind. Lubr. Tribol., vol. 51, no. 3, pp. 119–124, Jun. 1999. A. K. S. Jardine, D. Lin, and D. Banjevic, “A review on machinery diagnostics and prognostics implementing condition-based maintenance,” Mech. Syst. Signal Process., vol. 20, no. 7, pp. 1483–1510, Oct. 2006. A. D. Stuart, S. M. Trotman, K. J. Doolan, and P. M. Fredericks, “Spectroscopic measurement of used lubricating oil quality,” Appl. Spectrosc., vol. 43, no. 1, pp. 55–60, January 1989. S. Paul, W. Legner, A. Hackner, V. Baumbach, and G. Müller, “Multi-parameter monitoring System für hydraulische Flüssigkeiten in Offshore-Windkraftgetrieben,” Tech. Mess., vol. 78, no. 5, pp. 260–267, 2011. A. Agoston, C. Oetsch, J. Zhuravleva, and B. Jakoby, “An IR-absorption sensor system for the determination of engine oil deterioration,” in Proceedings of IEEE Sensors, 2004., pp. 463–466. “CRISP-DM: Ein Standard-Prozess-Modell für Data Mining – Statistik Dresden.” [Online]. Available: https://statistik-dresden.de/archives/1128. [Accessed: 06-Apr-2019]. R. Wirth and J. Hipp, “CRISP-DM: Towards a Standard Process Model for Data Mining.” W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery, “Chapter 14. Statistical Description of Data. 14.9 Savitzky-Golay Smoothing Filters,” pp. 766–772, 2007. A. L. Samuel,“Some studies in machine learning using the game of Checkers,” IBM J. Res. Dev., pp. 71--105, 1959. T. Mitchell, Machine Learning. .McGraw-Hill Education Ltd., 1997 U. Fayyad, G. Piatetsky-Shapiro, and P. Smyth, “From Data Mining to Knowledge Discovery in Databases,” AI Mag., vol. 17, no. 3, pp. 37–37, March 1996. S. S. Haykin and S. S. Haykin, Neural networks and learning machines. Prentice Hall/Pearson, 2009. P. Ramachandran, B. Zoph, and Q. V. Le, “Searching for Activation Functions,” October 2017.