面向eVTOL无人机应用的电池组观测器监控算法开发与硬件在环验证

Abstract Battery management systems significantly impact the performance of electric Vertical Take Off and Landing Unmanned Aerial Vehicles (eVTOL UAVs). In UAVs powered by multiple battery packs, optimising energy and power to extend cycle life and flight time requires monitoring of state of charge and temperature with a limited number of sensors. To achieve this, the battery pack electro-thermal model is first identified using a dedicated test and validated with experimental data from a current load profile based on a flight mission simulation of the reference UAV. A comprehensive analysis based on convergence speed, computational burden, and disturbance rejection is then conducted to determine the most suitable monitoring algorithm among three widely used observers: the Luenberger Observer (LO), Extended Kalman Filter Observer (EKFO), and Sliding Mode Observer (SMO). The comparison shows that the LO offers the best balance. To mitigate ageing sensitivity in state of temperature estimation, the observer uses the actual terminal voltage to compute heat generation rather than relying solely on model-predicted voltage. This accounts for resistive degradation and enhances robustness, though severe ageing needs a dedicated state of health estimator. The selected observer is validated in real time through a Hardware in the Loop test.