Directly-Physical Subspace Identification for Structure-Preserving Aeroservoelastic Model Updating

Kurzfassung

This paper presents a directly-physical subspace identification approach for structurepreserving aeroservoelastic model updating. The proposed method combines a physics-based continuous-time aeroservoelastic model with data-driven Markov-parameter information obtained from subspace identification. Instead of replacing the nominal model by a black-box identified realization, the method updates physically meaningful parameters of the nominal continuous-time model. For each candidate parameter vector, the structured continuous-time model is assembled, discretized, and compared with Markov parameters extracted from an experimentally identified discrete-time model. An optional time-domain fitting term may be added to improve transient response correlation. The method is applied to a flexible wind-tunnel aeroservoelastic model with control-surface inputs and acceleration outputs. The updated model preserves the physical partition into elastic modal states, actuator states, aerodynamic lag states, control-surface lag states, and output-delay states.