NEURAL-NETWORK PREDICTION OF 3-DIMENSIONAL UNSTEADY SEPARATED FLOWFIELDS

Unsteady surface pressures mere measured on a wing pitching beyond sta tic stall. Surface pressure measurements confirmed that the pitching w ing generated a rapidly evolving, three-dimensional unsteady surface p ressure field. Using these data, both linear and nonlinear neural netw orks were developed. A novel quasilinear activation function enabled e xtraction of a linear equation system from the weight matrices of the linear network. This equation set was used to predict unsteady surface pressures and unsteady aerodynamic loads. Neural network predictions mere compared directly to measured surface pressures and aerodynamic l oads. The neural network accurately predicted both temporal and spatia l variations for the unsteady separated flowfield as well as for the a erodynamic loads. Consistent results were obtained using either the li near or nonlinear neural network. In addition, fluid mechanics modeled by the linear equation set were consistent with established vorticity dynamics principles.