Automated Battery Power Fade Estimation for Fast Charge and Discharge Operations.

Pervasive devices are now part of daily lives for a multitude of human beings, due to their ability to perform simple to more complex tasks. Scenarios like Industry 4.0 and drone delivery are only few of the several ones which benefit from autonomous and modern smart devices. Due to their tasks, almost all of these devices are battery powered, with some of them for which it is hard to preventively maintain it. Most of the works which tackles this problem rely on processes which could be unpractical in the real world due to complexity, time or cost constraints. In this paper we propose a novel methodology which leverages data obtained from normal charge and discharge cycles to diagnose the current battery for power fade faults and possibly perform maintenance before service interruption occurs. Tests performed on a real dataset demonstrate the feasibility of our approach.