Towards Automated Timescale Decomposition for Graph-based Hierarchical Control of Electrified Aircraft Systems

The electrification of aircraft has brought new paradigms in performance, safety, sustain- ability, and lifetime costs. However, it also brings new challenges in energy management due to inherently involving complex nonlinear dynamics with tight coupling between the electrical, thermal, and mechanical energy domains. Model-based hierarchical control is an efficient and promising solution that employs a hierarchy of communicating control algorithms to coordinate decision-making across different energy domains and dynamic timescales. However, the design of a hierarchical control framework for a given application currently requires exhaustive manual iteration to determine an appropriate hierarchical network and reduce complex models to simpler forms that can be optimized online within each level of the hierarchy. This paper addresses that gap by developing an algorithm that leverages Koopman theory for systematic timescale decomposition of a networked dynamical system, which informs the design of a hierarchical control framework. The efficacy of the proposed timescale decomposition approach is demonstrated via application of hierarchical control in simulation, first for a simple nonlinear system with two dynamic states and then for an electro-thermo-mechanical system representative of an unmanned aerial vehicle.