Turbo-Shaft Engine Model Compensation Based on LSSVR

Abstract:

Considering the larger modeling errors between the turbo-shaft engine and on-board model, the model correction method based on least squares support vector regression is proposed. Firstly, the modeling principle of on-board turbo-shaft engine model is introduced, and then the structure of model combined with a compensation module is designed. The algorithm of LSSVR is used to build up the model compensation module, which is trained off-line and corrected on-line. Simulation studies on turbo-shaft engine have shown that the LS-SVM method can effectively reduce the model errors, and comparison with the interpolation correction, neural network one, the method proposed has better precision.

• Periodical: Applied Mechanics and Materials (Volumes 246-247)
• Pages: 738-743
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.246-247.738
• Online since: December 2012
• Authors: Feng Lu, Yi Qiu Lv, Wei Lin
• Keywords: Interpolation, Least Square Support Vector Regression, Model Compensation, On-Board Model, Turbo-Shaft Engine

References

[1] Garg S. Propulsion Controls and Diagnostics Research at NASA Glenn Research Center [R]. NASA/TM-215028, (2007).

DOI: 10.2514/6.2007-5713

[2] Ballin M G. A High Fidelity Real-Time Simulation of a Small Turbo-shaft Engine [R]. NASA/TM- 100991, (1988).

[3] Orme J S, Conners T R. Supersonic flight test results of a performance seeking control algorithm on a NASA-15 spacecraft [R]. AIAA-94-3210, (1994).

DOI: 10.2514/6.1994-3210

[4] Volponi A. Enhanced Self Tuning On-Board Real-Time Model (eSTORM) for Aircraft Engine Performance Health Tracking [R]. NASA/CR-215272, (2008).

DOI: 10.1109/aero.2003.1234150

[5] Vapnik V. The Nature of Statistical Learning Theory [M]. New York: Springer, 1995: 16- 23.

[6] Vapnik V, Levin E, Le C Y. Measuring the VC-dimension of learning machines [J]. Neural Computation, 1994, 6(5): 851- 876.

DOI: 10.1162/neco.1994.6.5.851