Comparisons of Two Reduced-Order Models for Linearized Unsteady Aerodynamic Identification

摘要: This paper compares the performance of two unsteady aerodynamic reduced-order models (ROMs), namely linear Volterra series and autoregressive with exogenous input (ARX) model, on modeling dynamically behaviors. The difference between these methods is that latter model has an term while former only input-related term. first system a plunging cylinder in low-Reynolds number flow, where flow stable (Re < 47). Although training data can be fitted well both methods, method requires higher order than ARX for same accuracy. Comparison frequency response indicates approximates more closely, at high Reynolds over-fitted by series. second over pitching NACA0012 airfoil, including subsonic transonic states. convergence respect to delay orders, different Mach numbers mean angles attack, studied detail. As or angle attack increases, required will increase. But still this small terms level All results indicate outperforms most cases, especially when close unstable state.