Data-Driven Capacity Estimation of Li-Ion Batteries Using Constant Current Charging at Various Ambient Temperatures

Abstract

The use of mobile devices, such as drones and electric vehicles, has rapidly increased in recent years. This has necessitated estimating the capacity of lithium-ion batteries in various situations. In this study, a capacity estimation algorithm using multilayer perceptron under different aging states and ambient temperature is proposed. The proposed algorithm estimates the capacity using charging time, voltage, and surface temperature, which can be measured during constant-current charging. Particularly, the surface temperature represents the state of battery differently than voltage. Therefore, the problem that the existing algorithm required a state of charge estimation was addressed, and the size of the neural network was significantly reduced. Using experimental data to validate the proposed algorithm, it was confirmed that the capacity was well estimated with a mean absolute error of 0.38% and a maximum error of 0.83%.