e^N-NEURAL NETWORK MODEL FOR PREDICTING TRANSITION OF 3-D SUPERSONIC SWEPT WING

In order to improve the computational efficiency of 3-D supersonic boundary layer transition prediction, a neural network model for 3-D compressible boundary layer transition prediction using neural network models instead of linear stability analysis is developed. By the research on the linear stability analysis method and flowfield characteristics of supersonic swept wing, neural network model parameters of supersonic swept wing transition prediction are proposed. Using a series of supersonic swept blunt plate models as the sample set, the e^N-neural network model is established. The sensitivity of each input parameter is analyzed by taking the standard model of three-dimensional supersonic large swept back straight wing as the test set, and the calculation results and efficiency of e^N-neural network model and traditional stability analysis method are compared to verify the accuracy and efficiency of this method.