Modular modeling and control of a hybrid unmanned aerial vehicle’s powertrain

The emerging trend of vehicle electrification is revolutionizing the transportation industry by replacing traditional mechanical and hydraulic components with higher performing, more reliable, and more efficient electrical components. However, the introduction of a complex electrical network onboard mobile systems poses significant challenges for control design engineers. The most notable challenge is the coordination of multi-domain and multi-timescale system dynamics. This thesis seeks to address the challenge of coordination between the slow battery state of charge dynamic and faster electro-mechanical dynamics for a hybrid unmanned aerial vehicle. The graph-based modeling framework for multi-domain systems is leveraged to capture the interactions between relevant energy domains. Additionally, the modularity and scalability of this modeling approach is used to develop a dynamic model for a hybrid unmanned aerial vehicle. The system model facilities the design and development of three control architectures of varying complexity. A baseline controller is developed for sake of comparison. A battery state of charge bounding algorithm in integrated into a centralized model predictive controller to provide system coordination across timescales. Lastly, an alternative model predictive hierarchical controller is designed to provide real-time planning of the slow battery state of charge dynamics. The proposed models and controllers are experimentally validated on a novel hybrid electric UAV powertrain testbed. The controllers are evaluated on three core figures of merit: performance, reliability, and efficiency. Both simulation and experimental results show that the advanced controllers outperform the baseline in all figures of merit with a 9-12.5% reduction in fuel usage