Direct global aerodynamic modeling using output-error optimization with an embedded interpolation structure
Abstract
High-accuracy aerodynamic models covering the full flight envelope are essential for high-fidelity flight simulation and simulator certification. This paper focuses on global aerodynamic modeling based on flight data. Aerodynamic parameters are identified at discrete Mach-altitude conditions using the time-domain output-error method, and the resulting estimates are directly interpolated to construct a global model. Although the directly interpolated global model remains stable and captures the main response trends, its errors at the training conditions are larger than those of the corresponding single-point OEM models, because it recalculates aerodynamic parameters from the instantaneous Mach number and altitude instead of retaining the constant parameter set identified for each maneuver. To address this issue, a global aerodynamic-parameter identification method is proposed by embedding the final interpolation structure into multi-condition time-domain output-error optimization. Tensor-product B-splines parameterize the aerodynamic-parameter surfaces over Mach number and altitude, and their control points are jointly optimized using multi-condition data. Linear, quadratic, and cubic B-spline models are compared with direct linear interpolation. Results show that the optimized B-spline models reduce response errors, and the quadratic model provides the best balance between fitting accuracy at training conditions and prediction accuracy at independent Proof-of-Match (POM) conditions.