UNSTEADY FLUID-MECHANICS APPLICATIONS OF NEURAL NETWORKS

The capability to harness or alleviate unsteady aerodynamic forces and moments could dramatically enhance aircraft control during severe man euvers as well as significantly extend the life span of both helicopte r and wind turbine blade/rotor assemblies, Using recursive neural netw orks, time-dependent models that predict unsteady boundary-layer devel opment, separation, dynamic stall, and dynamic reattachment have been developed, Further, these models of the flow-wing interactions can be used as the foundation upon which to develop adaptive control systems. The present work describes these capabilities for three-dimensional u nsteady surface pressures and two-dimensional unsteady shear-stress me asurements obtained for harmonic and constant-rate pitch motions. In t he near future, it is predicted that such techniques will provide a vi able approach for the development of six degree-of-freedom motion simu lators for severe vehicle maneuvers as well as a foundation for the ac tive control of unsteady fluid mechanics in a variety of systems.