Shape Optimization of UCAVs using ML

In my research on designing unmanned combat aerial vehicles (UCAVs), shape optimization is a critical element for enhancing aerodynamic performance. The study employed multi-fidelity modeling to explore a wide range of airfoil shapes under different flight conditions. The shape optimization used key design variables, such as the leading-edge radius, airfoil thickness distribution, camber, and the position of maximum thickness. By adjusting these variables, I aimed to find the best combination that would optimize aerodynamic characteristics while maintaining structural integrity.

The objective functions in this optimization process included maximizing lift-to-drag ratio, improving flight stability, and enhancing overall maneuverability. This shape optimization extended beyond individual airfoils to cover the entire aerodynamic surface, ensuring the final design met multiple performance objectives across various flight conditions. To refine the optimization further, techniques like proper orthogonal decomposition (POD) were employed to visualize design rules and identify the most influential shape features.

The multi-fidelity model generated a diverse dataset, capturing the nuances of how different shape configurations affect aerodynamic behavior. This comprehensive approach allowed for extracting design shapes that satisfied operational requirements and improved the robustness of the aircraft's performance, providing a balanced solution across both low-speed stability and high-speed efficiency.

Click for more details

Shape Optimization of Manipulator using AI

Another study of shape optimization is for the design of a quasi-serial manipulator, where shape optimization was critical for improving both kinematic workspace and dynamic performance. The research adopted a deep learning-based surrogate model to guide the optimization of the linkage shapes, focusing on structural robustness and precision. The multi-objective framework aimed to identify shapes that minimized joint torques while maximizing reach and efficiency. This process facilitated the selection of optimal linkage shapes capable of performing specific tasks while handling high payloads effectively.

After the initial optimization, topology optimization was employed to further refine the manipulator's shape, improving its dynamic performance. By optimizing the shape based on the stress distribution and material properties, the final design achieved reduced mass and inertia without compromising structural integrity. This step was essential for realizing a feasible manipulator design, validated through physical testing, that met performance goals while maintaining a lightweight and robust structure.

Click for more details