INTEGRATION OF PHYSICS-INFORMED APPROACHES WITH MACHINE LEARNING TECHNIQUES FOR AERODYNAMICS

The thesis utilizes a fusion of machine learning techniques and physics laws to tackle aerodynamic challenges. Specifically, it employs machine learning methodologies to enhance various aspects of aerodynamics, including the optimization of airfoil designs, prediction of 2D flow fields, and refinement of flush air data sensing systems. By integrating machine learning algorithms with fundamental principles of physics, this research endeavors to enhance the understanding and efficiency of aerodynamic processes. Through the application of these methods, advancements in aerodynamic modeling and problem-solving are achieved, offering innovative solutions to complex aerodynamic problems. This interdisciplinary approach not only enriches the field of aerodynamics but also demonstrates the potential synergy between machine learning and traditional physics-based methodologies in engineering applications.

Author dc:creator

• LIAO YANGTIANCHUN

• Physics informed neural networks, Neural networks, airfoil optimization, air data sensing, tail-sitter UAVs