VARIABLE C-RATE IN-OPERANDO BATTERY RUL PREDICTION VIA EDGE-CLOUD ENABLED DEEP LEARNING IN AGNOSTIC BMS

Applications of Lithium-ion batteries (LIBs) are so widely spread from transportation like electric vehicles to portable storage devices. This is mainly due to their lighter weight and smaller size with higher energy density when compared to Lead-acid, Nickel Cadmium (Ni-Cd), and other batteries. One of the applications of LIB includes electric propulsion in-air like quadcopters. These electrically-propelled vehicles have diverse applications including risky jobs like wildlife management, search and rescue, and jobs that can be automated such as delivery of smaller packages, urban planning, and so on. These electrically-propelled vehicles produce heat around the LIB which leads to thermal abuse of the battery. Also, there are often cases where LIB undergoes different abuse conditions in-air when operating these vehicles. Present battery BMSs are highly accurate but require edge and cloud with a deep learning model to safely operate quadcopters in the air. In the work, we present a BMS capable of edge-cloud data transfer with a deep-learning model to predict the RUL of the battery. Benchmark differences between data collected on-ground and in-air are presented for comparison. It turns out that the temperature collected in the air is less than the temperature on the ground when different current profiles are experimented on different batteries used in quadcopters. This study helps in the improvement of BMS with edge-cloud and deep-learning models and helps in understanding the behavior of battery in-air.