Machine learning based modeling and designing of cyclorotor for an eVTOL aircraft

The rise of electric vertical take-off and landing (eVTOL) vehicles has received significant attention in recent years because of their potential to transform transportation and solve key challenges in urban air mobility. These vehicles offer a variety of potential advantages, including lower environmental impact, improved energy efficiency, reduced noise, and operational flexibility and accessibility. However, majority of the existing eVTOL platforms face limitations in range, endurance, and payload capability. This thesis explores the use of a novel propulsion system, known as cycloidal rotor, that has the potential to overcome the limitations of existing eVTOL platforms. One of the key advantages of the cycloidal rotor is its instantaneous 360° thrust vectoring capability. This enables it to be used as a primary propulsion system for both vertical take-off landing (VTOL) and cruise conditions. A key contribution of the thesis is the development of a computationally efficient cycloidal rotor aerodynamic model that can be utilized for preliminary vehicle design and optimization purposes. The developed aerodynamic model utilizes a novel physics-guided neural network (PGNN) architecture that combines a physics-based unsteady aerodynamic model with a neural network (NN). Validations with existing experimental data show remarkable accuracy of the developed PGNN model. In the next step, this cycloidal rotor model is utilized for the conceptual design of an eVTOL aircraft propelled by a set of cycloidal rotors. After developing a baseline configuration, a coupled-trim model of the entire aircraft is developed by combining vehicle response equations with the trim equations of the aircraft. This model predicts the control inputs required to balance the forces and moments for a steady-level flight and also computes the cycloidal rotor performance corresponding to the predicted control inputs. This coupled trim model is further leveraged to perform systematic parametric study while designing key parameters of the cycloidal rotors and other aircraft components. The performance of the final designed cycloidal rotor is verified through high-fidelity CFD simulations. The present study along with the developed computational models forms an important foundation for developing a cycloidal rotor-based eVTOL aircraft.