Applied Research on Supervised Learning in the Judgment of Airfoil Transition

This paper combines the latest machine learning techniques to develop an effective and reliable supervised learning model for transition judgment. Firstly, the variable-interval time average (VITA) method is used to transform the fluctuating pressure signal into a sequence of states in the Markov state space. Then we describe it using Markov chain model, and obtain its feature vectors. Then the hidden Markov model is used to pre-classify the feature vectors labeled using the traditional RMS criteria. And finally a classification model based on probability density distribution is established. The research shows that the model developed in this paper is effective and reliable and possesses a generalization ability. Compared with the traditional RMS criterion, a reasonable 'transition zone' can be obtained using the developed classification model without comparing the signals at multiple locations.