Mach Number Prediction of Multi-mode Wind Tunnel Flow Field Based on Model Migration

In the process of wind tunnel experiment, rapidity and accuracy of Mach number prediction directly affect the quality of wind tunnel process. In practical applications, the new operating conditions have few running times and only a small amount of data, so the established model is usually underfitted and deviates from the actual model. In this paper a RBFNN prediction model is established based on model migration using a small amount of new operating conditions data for the multi-mode wind tunnel flow field, where frozen, fine-tuning, adding gradient features and other migration methods are adopted. And experiments are carried out to analyze the transfer effect of the proposed model.