This is carried out by Dr. Ajit Kumar, Department of Mechanical and Aerospace Engineering.
Aerodynamic parameter estimation is a crucial process in aircraft system identification that involves identifying stability and control derivatives from observed input-output data. Different methods, including analytical, wind tunnel, and flight test-based approaches, are used to estimate these parameters. However, the analytical method is prone to error, while the wind tunnel approach needs correction for boundary layer interference, Reynolds number, and scaling effects.
The presented paper compares regularized regression-based methods for aerodynamic parameter estimation using flight test data. The methods investigated are LASSO, Ridge, and Elastic net regression. LASSO performs variable selection and regularization simultaneously, while Ridge regression adds a penalty term to least squares regression to prevent instability. Elastic net regression combines the advantages of both methods. The study shows that LASSO, Ridge, and Elastic net are effective methods for aerodynamic parameter estimation. These methods provide more accurate and stable parameter estimates than the conventional LS and FEM methods. The paper highlights the usefulness of regularized regression-based methods for aerodynamic parameter estimation using flight test data, which is essential for aircraft design and performance evaluation.
In conclusion, aerodynamic parameter estimation is a critical process in understanding aircraft performance and designing control systems. The presented paper demonstrates that regularized regression-based methods, such as LASSO, Ridge, and Elastic net, are viable options for accurate and reliable parameter estimation using flight test data.
