Abstract

State-of-Charge (SoC) is commonly defined as a ratio between the available charge and the battery capacity. Unfortunately, SoC has been misguidedly interpreted as an indicator of autonomy. Despite its widespread use, this concept has two critical flaws that affect the quality of estimates for energy availability in lithium ion batteries: (i) SoC is typically computed as a percentage, but the corresponding normalization constant is difficult to compute, and (ii) SoC estimates do not acknowledge the fact that the total amount of energy that can be extracted from the battery depends significantly on the future battery usage profile (i.e., discharge current). To overcome these issues, this article introduces a novel metric for battery energy autonomy: State of-Latent-Energy (SoLE) [kWh]. Unlike SoC-based approaches, SoLE estimates directly the amount of available useful energy at the battery, without requiring a normalization constant, and thus limiting the uncertainty associated with the estimation problem. The implementation of SoLE estimators requires two major steps: (i) to estimate the amount of energy delivered previously, and (ii) to prognosticate future battery voltage trajectories to infer the End-of-Discharge (EoD) time and calculate the remaining useful energy. Therefore, SoLE not only incorporates information on how the battery was discharged so far, but also a probabilistic characterization of how it may be operated. The proposed SoLE concept is validated using data from actual usage of an electric bike, with promising results.