LQG Optimal Control Applied to On-Board Energy Management System of All-Electric Vehicles
Résumé
This paper proposes a general frequency-separation-based strategy of coordinating power sources within off-grid applications. The application chosen to illustrate this strategy is an electric vehicle equipped with two power sources---a battery and an ultracapacitor (UC)---for which coordination problem can be formulated and solved as a linear quadratic Gaussian (LQG) optimal control problem. The two power sources are controlled to share the stochastically variable load according to their respective frequency range of specialization: low-frequency variations of the required power are supplied by the main source, the battery, whereas high-frequency variations are provided by the UC. The studied system is a bilinear one; it can be modeled as a linear parameter varying system. An LQG-based optimal control structure is designed and coupled with a gain-scheduling structure to cover the entire operating range. In this way, load regulation performance and the variations of battery current are conveniently traded off to preserve battery reliability and lifetime. Real-time experiments on a dedicated test rig---based on employing a real UC---validate the proposed optimal power flow management approach.