Key points are not available for this paper at this time.
This paper presents a novel battery health management system for electric UAVs (unmanned aerial vehicles) based on a Bayesian inference driven prognostic framework. The aim is to be able to predict the end-of-discharge (EOD) event that indicates that the battery pack has run out of charge for any given flight of an electric UAV platform. The amount of usable charge of a battery for a given discharge profile is not only dependent on the starting state-of-charge (SOC), but also other factors like battery health and the discharge or load profile imposed. This problem is more pronounced in battery powered electric UAVs since different flight regimes like takeoff/landing and cruise have different power requirements and a dead stick condition (battery shut off in flight) can have catastrophic consequences. Since UAVs deployments are relatively new, there is a lack of statistically significant flight data to motivate data-driven approaches. Consequently, we have developed a detailed discharge model for the batteries used and used it in a Bayesian inference based filtering (Particle Filtering) technique to generate remaining useful life (RUL) distributions for a given discharge. The results section presents the validation of this approach in hardware-in-the-loop tests.
Saha et al. (Tue,) studied this question.