Nate Michek

Nate Michek's work concentrates on accurately modeling unstable free-flight rigid body motion, consisting of 3D translational motion and large angular rates about all axes and orientations outside of typical flight envelopes, which is a significant challenge. Due to the complexity of these flights to accurately model the underlying dynamics experimental flight testing is needed. By using experimental data, an unconstrained flight can be studied and used to develop data-driven models with an emphasis on machine learning techniques specifically, Physics-Informed Neural Networks (PINNs). PINNs introduce the equations of motion for a free-flight body into the loss function of the neural network training enforcing the solution to follow the known physics. The neural network is developed such that it provides a Non-Determinant model of the aerodynamic coefficients for a given body within the constraints of the collected data and the known physics. To this point, multiple versions of PINNs have been investigated on the use of simulated data. Future work with this project includes further investigation into the developed methods on experimental data, uncertainty quantification, data augmentation, and comparisons with classical aircraft system identification techniques.