11/21/2023
By Danielle Fretwell

The Francis College of Engineering, Department of Mechanical & Industrial Engineering, invites you to attend a Doctoral Dissertation defense by Joel Walker on "Taking Off with Machine Learning for Flight Modeling."

Candidate Name: Joel Walker
Degree: Doctoral
Defense Date: Tuesday, Nov. 28, 2023
Time: 6:30-8:30 p.m.
Location: Dandeneau, Rm. 105 (DAN-105)

Committee:

  • Advisor David Claudio, IE Program Director, Mechanical & Industrial Engineering, UML
  • Carter Keough, Assistant Professor, Mechanical & Industrial Engineering, UML
  • Maru Cabrera, Assistant Professor, Computer & Information Sciences, UML

Brief Abstract:
Flight test for aircraft certification is a fundamental method to ensure safe aircraft and air travel worldwide. Flight test data is collected through a limited number of discrete subspace points in the flight envelope to verify and validate preselected linear model coefficients chosen before the test program. This model is presented as evidence that the platform meets the certification authority requirements throughout the expected flight envelope and to develop high-fidelity full-motion simulators, the primary training method for business aviation and airline industries. System identification utilizing local linear approximations is still the dominant flight test approach, exacerbating the development of a global model when trying to capture typical non-linear flight dynamics. This makes it appealing to leverage the robust and highly evolving machine learning techniques successfully used to find physical and physics-based insights. This work summarizes the historical developments of flight modeling. Coupling machine learning methodologies with airworthiness model validation to reduce the uncertainties of current approaches while providing dynamic and real-time model development with little programmatic overhead is proposed. The limited number of studies using machine learning for flight dynamics modeling are discussed, showing their strong potential and specific scope limitations. The constrained overlap of these two topics with comments and discussion toward potential areas of exploration are detailed.

This work surveys Machine Learning, specifically neural network approaches, as applied to dynamic flight test modeling. Neural Network structural and architectural elements and associated processing parameters are investigated. Noise is introduced through industry-standard simulation methods to provide insight into real-world applicability and determine the robustness of the various approaches to noise effects. Additionally, Machine Learning probabilistic representations in the form of Gaussian Process Regression and Neural network-based approaches that address incremental and sequential learning are explored. By leveraging flight dynamic simulation data, these are evaluated to show their promise as potential flight dynamics modeling approaches. This is also appealing as machine learning allows for active capture of flight dynamics independently of the historical preselected model validation and reduces risks to the flight test crew.