DATA FUSION UNSTEADY AERODYNAMIC MODELING BASED ON EXPERIMENTAL DATA

Dynamic stall prediction at high angles of attack is faced with the dual challenges of insufficient accuracy of calculation data and lack of experimental data. In order to make full use of the characteristics of different data sources to establish a dynamic stall aerodynamic time-domain prediction model, this paper proposed a data fusion modelling method, which combines a Computational Fluid Dynamics solver with a neural network model. By fusing experimental data and Computational Fluid Dynamics simulation data, combined with an integrated neural network model, an unsteady aerodynamic data fusion modelling framework for airfoil dynamic stall is established. Based on the NACA0012 airfoil dynamic stall test data, and the Computational Fluid Dynamics numerical simulation results, the proposed data fusion framework performs high precision in the prediction of wind tunnel test data, including lift and moment coefficients at different pitch angles, balanced angles of attack and reduced frequencies. Results show that the proposed data fusion framework not only has higher prediction accuracy, but also has strong abilities in both generalization and convergence.