A Coordinate Descent Method for Multidisciplinary Design Optimization of Electric-Powered Winged UAVs

Abstract

In this paper, we present an optimization framework for the conceptual design of electric-powered unmanned aerial vehicles (UAVs) with wings, to meet the community’s ever increasing interest in developing novel efficient winged UAVs. In our framework, the UAV design is formulated as an optimization problem with a user-defined objective. It also accepts various constraints, such as restricted aircraft size, weight, and preliminary wing (and fuselage) shape determined by industrial design, limited availability of propulsion systems, etc. Such a framework is particularly suitable for the design of small UAVs with many practical limitations, such as portability, size, cost, etc. In evaluating a given aircraft configuration (e.g., wing, fuselage, landing gears, etc.), we adopt various empirical aerodynamic models that have been commonly used in aviation history. We also retrieve hundreds of propeller and motor data from their manufacturers and fit them to constitute a high-fidelity propulsion system database. Wind tunnel testing on existing airframe data shows that our aerodynamic models fit the measurements very well. Propeller testing is also carried out to validate the fitted propeller model. With the ability to evaluate a given aircraft and propulsion, we propose a coordinate descent method that nicely decouples the optimization for the aircraft configuration, which involves continuous variables, and the propulsion system, which involves discrete variables (e.g., motor index, propeller index). With the presented optimization framework and coordinate descent method, a quadrotor tail-sitter vertical takeoff and landing (VTOL) UAV is designed, manufactured and tested.