An Adaptive Anomaly Detector used in Turbofan Test Cells

##plugins.themes.bootstrap3.article.main##

##plugins.themes.bootstrap3.article.sidebar##

Jérôme Lacaille Valério Gerez Rafik Zouari

Abstract

Airplane engines use sophisticated technologies to improve their efficiency, reduce their weight, reduce fuel consumption, limit NOx generation and reduce the generated noise. On another hand, airlines want to decrease their maintenance costs. These changes may have an effect on engine reliability and there is a greater need to understand and control the behavior of the engine. This is the goal of PHM algorithms. However, if such algorithms are "easy" to build, V&V stay a challenge. To increase their readiness level, Snecma, as engine manufacturer, tests all engines on bench cells during development phases and before reception. Now Snecma chooses also to use PHM algorithms on bench tests. It helps the maturation of the code itself but it is also a way to monitor the bench cells.

The present document describes an implementation on a partial bench test cell of a generic abnormality detector. The first section gives an outlook at the implementation of some algorithms on a real test cell. The second section is the description of the main algorithm: an online abnormality detector able to automatically update when new recurrent usual observations appear. Finally the last section sketches some results obtained during the execution of the algorithm.

How to Cite

Lacaille, J. ., Gerez, V. ., & Zouari, R. . (2010). An Adaptive Anomaly Detector used in Turbofan Test Cells. Annual Conference of the PHM Society, 2(1). https://doi.org/10.36001/phmconf.2010.v2i1.1865
Abstract 38 | PDF Downloads 20

##plugins.themes.bootstrap3.article.details##

Keywords

diagnostics, test cell, V&V, maturation, Turbofan

References
(Aaron, 2004) C. Aaron. Algorithme em et classifica- tion non supervisée. In ACSEG, 2004.
(Akaike, 1978) H. Akaike. A bayesian analysis of the minimum aic procedure. Ann. Inst. Statist. Math., 30, part A:9–14, April 1978.
(Ausloos and all., 2009) A. Ausloos and all. Estimation of monitoring indicators using regression methods; application to turbofan start sequence. In ES- REL, Prague, Poland, 2009.
(Azencott, 2003) R. Azencott. A method for monitoring a system based on performance indicators, 2003. U.S. Patent 6594618B1.
(Basseville et al., 2009) Michèle Basseville, Laurent Mevel, and Rafik Zouari. Variations on cusum tests for flutter monitoring. In Proceedings of the 2nd International Workshop in Sequential Methodologies (IWSM), Troyes, FR, June 2009.
(Bilmes, 1998) J. A. Bilmes. A gentle tutorial of the em algorithm and its application to parameter estimation for gaussian mixture and hidden markov models. International Computer Science Institute, April 1998.
(Blanchard et al., 2009) S. Blanchard, J. Lacaille, and M. Cottrell. Health monitoring des moteurs d’avions. In Les entretiens de Toulouse, France, 2009.
(Cômes et al., 2010a) E. Cômes, M. Cottrell, M. Ver- leysen, and J. Lacaille. Aircraft engine health monitoring using self-organizing maps. In ICDM, Berlin, Germany, 2010.
(Cômes et al., 2010b) E. Cômes, M. Cottrell, M. Ver- leysen, and J. Lacaille. Self organizing star (sos) for health monitoring. In ESANN, Bruges, 2010.
(Cottrell and all., 2009) M. Cottrell and all. Fault prediction in aircraft engines using self-organizing maps. In WSOM, Miami, FL, 2009.
(De Troyer et al., 2008) Tim De Troyer, Rafik Zouari, Patrick Guillaume, and Laurent Mevel. A new frequency-domain flutter speed prediction algorithm. In Proceedings of ISMA2008 - Noise and Vibration Engineering Conference, Leuven, BE, September 2008.
(Dempster et al., 1977) A. P. Dempster, N. M. Laird, and D. B. Rubin. Maximum likelihood from incomplete data via the em algorithm. Journal of the Royal Statistical society, 39(1):1–38, 1977.
(Flandrois and Lacaille, 2009) X. Flandrois and J. La- caille. Expertise transfer and automatic failure classification for the engine start capability system. In AIAA Infotech, Seattle, WA, 2009.
(Hartmann, 1998) W. M. Hartmann. Signals, Sound, and Sensation. Springer-Verlag, New York, 1998.
(Haykin, 1994) S. Haykin. Neural Networks, a Comprehensive Fondation. MacMillan, IEEE Press, 1994.
(Hazan et al., 2010) A. Hazan, M. Verleysen, M. Cot- trell, and J. Lacaille. Trajectory clustering for vibration detection in aircraft engines. In ICDM, Berlin, Germany, 2010.
(Klein, 2009) R. Klein. Model based approach for identification of gears and bearings failure modes. In PHM Conference, San Diego, CA, 2009.
(Lacaille and Dubus, 2005) J. Lacaille and H. Dubus. Defectivity analysis by a swarm of intelligent distributed agents. In AEC/APC, Palm Spring, CA, 2005.
(Lacaille and Nya Djiki, 2009) J. Lacaille and R. Nya Djiki. Model based actuator control loop fault detection. In Euroturbo Conference, Graz, Austria, 2009.
(Lacaille and Zagrebnov, 2006a) J. Lacaille and M. Zagrebnov. Building a robust model for process control using advanced mathematical techniques. In AEC/APC tutorial, Aix en Provence, France, 2006.
(Lacaille and Zagrebnov, 2006b) J. Lacaille and M. Zagrebnov. A statistical approach of abnormality detection and its applications. In AEC/APC, Denver, CO, 2006.
(LacailleandZagrebnov,2007) J.LacailleandM.Zagrebnov. An unsupervised diagnosis for process tool fault detection: the flexible golden pattern. IEEE Transactions on Semiconductor Manufacturing, 20(4):355–363, 2007.
(Lacaille, 2004) J. Lacaille. Industrialisation d’algorithmes mathématique. Habilitation à driger des recherches, Université Paris 1, Sorbonne, 2004.
(Lacaille, 2005) J. Lacaille. Mathematical solution to identify the causes of yield deterioration - a defectivity data based solution with an emergent computing technology. In ISMI, Austin, TX, 2005.
(Lacaille, 2006) J. Lacaille. Advanced fault detection. In AEC/APC tutorial, Denver, CO, 2006.
(Lacaille, 2007) J. Lacaille. How to automatically build meaningful indicators from raw data. In AEC/APC, Palm Spring, CA, 2007.
(Lacaille, 2008) J. Lacaille. Global predictive monitoring system for a manufacturing facility, 2008. U.S. Patent 20080082197A1.
(Lacaille, 2009a) J. Lacaille. An automatic sensor fault detection and correction algorithm. In AIAA ATIO, Hilton Beach, SC, 2009.
(Lacaille, 2009b) J. Lacaille. A maturation environment to develop and manage health monitoring al- algorithms. In PHM, San Diego, CA, 2009.
(Lacaille, 2009c) J. Lacaille. Standardized failure signature for a turbofan engine. In IEEE Aerospace Conference, Big Sky, MT, 2009.
(Lacaille, 2010) J. Lacaille. Validation of health- monitoring algorithms for civil aircraft engines. In IEEE Aerospace Conference, Big Sky, MT, 2010.
(Mardia et al., 1979) K.V. Mardia, J.T. Kent, and J.M. Bibby. Multivariate Analysis. Academic Press, 1979.
(Vapnik,1995) V.N.Vapnik.TheNatureofStatistical Learning. Springer Verlag, NY, 1995.
(Zouari et al., 2008a) Rafik Zouari, Tim De Troyer, Laurent Mevel, Michèle Basseville, and Patrick Guillaume. Flutter monitoring using a mixed model-based and data-based approach. In Proceedings of ISMA2008 - Noise and Vibration Engineering Conference, Leuven, BE, September 2008.
(Zouari et al., 2008b) Rafik Zouari, Laurent Mevel, and Michèle Basseville. An adaptive statistical approach to flutter detection. In Proceedings of the 17th IFAC World Congress, pages 12024–12029, Seoul, KR, July 2008.
(Zouari et al., 2008c) Rafik Zouari, Laurent Mevel, Andrzej Klepka, and Michèle Basseville. Adaptive flutter monitoring using wavelet filtering and recursive subspace-based detection. In Proceedings of the 4th European Workshop on Structural Health Monitoring, pages 1137–1144, Cracow, PL, July 2008.
Section
Technical Papers

Most read articles by the same author(s)