Modern battery management systems, as part of Battery Digital Twin, include cloud-based predictive analytics algorithms. These algorithms predicts critical parameters like Thermal runaway events, state of health (SOH), state of charge (SOC), remaining useful life (RUL), etc. However, relying only on cloud-based computations adds significant latency to time-sensitive procedures such as thermal runaway monitoring. This is a very critical and safety function and delay is not acceptable, but automobiles operate in various areas throughout the intended path of travel, internet connectivity varies, resulting in a delay in data delivery to the cloud and similarly delay in return of the detected warning to the driver back in the vehicle.
As a result, the inherent lag in data transfer between the cloud and vehicles challenges the present deployment of cloud-based real-time monitoring solutions. This study proposes application of Federated Learning and applying to a thermal runaway model in low-cost microcontroller as a strategy to reduce transmission and processing costs and delays. Furthermore, this will ensure safety assured, rapid and efficient client experience, and other long term, history and huge data based algorithms running in cloud, giving OEMs a competitive advantage in the digital technology arena.