Bridging The Gap: Data Driven Method For Linearized Model Fidelity Enhancement

Abstract

Accurate flight dynamics models are necessary for cost and development time reduction in modern rotorcraft design. Yet, discrepancies frequently exist between high-fidelity, physics-based models and data-driven models identified from flight. Bridging this gap through physics-based model refinements often requires longer time frames than the tight development schedules allow. This paper presents a practical application of a data-driven Gain-Delay Correction Method (GDCM) to rapidly enhance the fidelity of a physics-based linearized rotorcraft model. The method optimizes a set of input and output gains and delays to address frequency responses discrepancies between linearized multi-body models and identified models derived from flight. This technique is applied the dataset of an AW family helicopter, covering a wide range of flight conditions. The results demonstrate that the method effectively identifies systematic gain discrepancies, for which a correction scheduled with airspeed is synthesized to improve the physics-based model accuracy. Also, a key limitation of the proposed approach is highlighted: the inability of a simple delay structure to compensate for phase underestimation in the physics-based model. The proposed workflow provides a pragmatic solution for rapid model fidelity improvement and offers a way to provide valuable cues to guide long term physics-based model refinements.