Aeroengine Gas Path Performance Parameter Prediction Based on GM(1,1) Rolling-Markov Chain Approach

Abstract

As one of the most important gas path performance parameters, the exhaust gas temperature (EGT) can provide more effective information about the health state of aeroengine. However, the changing process of aeroengine EGT is often affected by many uncertain factors and the sample data are relatively less, which make it difficult to predict its trend accurately by the traditional regression analysis method. Aiming at this problem, the GM(1,1) rolling-Markov chain model is proposed and used for aeroengine EGT prediction in this paper. Based on the equal dimensional new information theory, GM(1,1) rolling model is utilized to predict the changing trend of aeroengine EGT firstly. Then the Markov chain theory is used to solve the influence of random fluctuation on prediction accuracy, which can achieve an effective estimate of the non-linear parameter. As an example, the historical monitoring data of EGT from one aeroengine of Air China is used to verify the prediction performance of GM(1,1) rolling-Markov chain model. The analysis results show that this model has higher prediction accuracy and can effectively reflect the random fluctuation characteristics of EGT, which provides a new method for aeroengine gas path performance parameter prediction.