Interpretable Aeroelastic Models for Control at Insect Scale - presented by Dr. Michelle Hickner

Interpretable Aeroelastic Models for Control at Insect Scale

Dr. Michelle Hickner

Dr. Michelle Hickner
Slide at 16:29
Summary
Linear, low-rank, state-space aeroelastic
modelling method
Accurately predicts lift & deformation
Model is built from data
To be published in AIAA Journal
arxiv.org/abs/2111.11299
Thank you to our funders
Air Force Office of Scientific Research
(AFOSR FA9550-19-1-0386)
National Science Foundation Al Institute in Dynamic Systems (Grant No. 2112085)
UNIVERSITY of WASHINGTON
1
References
  • 1.
    M. K. Hickner et al. (2022) Data-Driven Unsteady Aeroelastic Modeling for Control. AIAA Journal
Share slide
Summary (AI generated)

Interpretable models help us avoid undesirable situations like the yellow and red curves. The test maneuver involves pitching up and holding pitch down at different speeds to measure the quasi-sty contribution. If the speed is too fast, the coefficient of lift cannot track the reference (light gray curve). The yellow curve falls short, while the orange curve tracks it well but with a constraint on curvature to prevent excessive stress. Increasing the angle of attack requires a higher coefficient of lift, which relies on angular acceleration for short maneuvers and pitching up for longer durations.

Interpretable coefficients show that deformation tracks the angle of attack, emphasizing the importance of control objectives that prioritize angular acceleration over maintaining high angles of attack. This approach simplifies control planning by quickly identifying potential issues based on coefficient values. The method discussed involves creating a linear low-rank model for an aeroelastic system with a flexible wing, accurately predicting lift and deformation by leveraging data. For more details, refer to our paper. Thank you for your attention.