Real-time Battery Capacity Estimation Based on Opportunistic Measurements

Document Type

Conference Proceeding

Publication Date


Publication Title

1st IEEE Industrial Electronics Society Annual On-Line Conference, ONCON 2022


Battery management system, OCV-SOC curve, opportunistic measurements, real-time capacity estimation


This paper presents an approach to real-time battery capacity estimation by combining the advantages of the opportunistic zero-current states in the dynamic current profile of the battery and the knowledge of the open circuit voltage (OCV)-state of charge (SOC) curve of the battery. With the knowledge of OCV parameters, the SOC can be estimated through OCV lookup using the OCV-SOC curve. The difference in SOC between two different points is the change in Coulombs normalized by the battery capacity - this relationship is exploited to estimate the battery capacity. In the capacity estimation using the OCV-SOC curve, there are two existing approaches to OCV estimation. In the first approach, the battery is completely rested and the terminal voltage is measured; in a rested battery, the terminal voltage is treated as the OCV. In the second approach, the voltage drop is computed by estimating the equivalent circuit model (ECM) parameters of the battery; the OCV is then computed by subtracting the voltage drop from the measured terminal voltage. Both of these approaches have limitations: it takes a long time to fully rest a battery and ECM parameter estimation problem suffers form non-linearities and sub-optimal solutions as a result of that. In this paper, we propose an approach to estimate the battery capacity without the wait for complete rest of the battery or for the estimation of ECM parameters. Rather than waiting for battery rests, it is proposed to make OCV measurements whenever the current through the battery is zero. It is hypothesized in this paper that, the resulting OCV error, due to both the hysteresis and relaxation effect, can be considered zero-mean when sufficient number of measurements are taken. The proposed approach, when tested using real world battery data, show significantly accurate estimation of battery capacity. Further, it is observed that the amount of rest time before taking the OCV measurement positively correlated with capacity estimation accuracy. The standard deviation of the computed capacities immediately after zero current and after one hour of rest, relative to true capacity is 0. 3Ah and 0. 2Ah respectively.