Abstract
Electric vertical takeoff and landing (eVTOL) aircraft play a key role in urban air mobility (UAM), which aims to alleviate traffic congestion in urban areas. Despite their value, eVTOL aircraft suffer from battery energy consumption, which affects their range and endurance in real world flight tasks. Especially, the takeoff process has been identified for excessive power demands. Multidisciplinary analysis and optimization manage to discover optimal takeoff trajectories with minimum energy consumption while balancing multidisciplinary trade-offs, such as short distance takeoff and passengers' comfort. However, conventional parametrization methods (such as B-spline curves) leverage an empirically high-dimensional design space to include real optimal design, which incurs convergence difficulty in model analysis and design optimization. In this work, a twin-generator generative adversarial network (twinGAN) was developed and demonstrated for intelligent parametrization of takeoff trajectories. Specifically, the twinGAN model exhibited two-fold dimensionality reduction, i.e., implicit reduction by generating only realistic trajectories and explicit reduction by reducing the original 40-dimensional B-spline control points to three-dimensional twinGAN variables. Meanwhile, the twinGAN was verified by reconstructing existing, realistic trajectories via fitting optimization, which achieved 99% accuracy. Thus, the twinGAN presented a 92.5% reduction on the dimension of original design space while maintaining sufficient variability. This significantly facilitates the process of analysis model convergence, surrogate modeling, as well as design optimization in the future work and can be extended to other relevant engineering areas.
Recommended Citation
S. Sisk et al., "Generative Adversarial Networks for Dimensionality Reduction in eVTOL Aircraft Takeoff Trajectory Optimization," AIAA Science and Technology Forum and Exposition, AIAA SciTech Forum 2025, American Institute of Aeronautics and Astronautics, Jan 2025.
The definitive version is available at https://doi.org/10.2514/6.2025-0967
Department(s)
Mechanical and Aerospace Engineering
Publication Status
Full Access
International Standard Book Number (ISBN)
978-162410723-8
Document Type
Article - Conference proceedings
Document Version
Citation
File Type
text
Language(s)
English
Rights
© 2025 American Institute of Aeronautics and Astronautics, All rights reserved.
Publication Date
01 Jan 2025
