Contents lists available at ScienceDirect Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser Overview of model-based online state-of-charge estimation using Kalman flter family for lithium-ion batteries Prashant Shrivastava a , Tey Kok Soon a,∗ , Mohd Yamani Idna Bin Idris a , Saad Mekhilef b a Department of Computer System and Technology, University of Malaya, Malaysia b Department of Electrical Engineering, University of Malaya, Malaysia ARTICLEINFO Keywords: Lithium-ion battery Battery model State-of-charge Kalman flter Electric vehicle Battery management system Estimation ABSTRACT Carbon impression and the growing reliance on fossil fuels are two unique concerns for world emission reg- ulatory agencies. These issues have placed electric vehicles (EVs) powered by lithium-ion batteries (LIBs) on the forefront as alternative vehicles. The LIB has noticeable features, including high energy and power density, compared with other accessible electrochemical energy storage systems. However, LIB is exceedingly nonlinear and dynamic; therefore, it generally requires an accurate online state-of-charge (SOC) estimation algorithm for real-time applications. Accurate battery modelling is an essential and primary requirement of online SOC esti- mation to simulate the dynamics. In this paper, diferent modelling methods suitable for online SOC estimation are discussed, and four groups of available online SOC estimation approaches are reviewed. After the general survey, the study explores the available Kalman flter (KF) family algorithms suitable for model-based online SOC estimation. The mathematical process and limitations of diferent KF family algorithms are analysed in depth and discussed. Moreover, challenging steps in the implementation of KF family algorithms in model-based online SOC estimation processes, such as selection of battery model, initial SOC and flter tuning, are elaborated for the efcient development of a battery management system, especially for EV application. The on-going research is propelled by KF-based online SOC estimation approaches distinctly emphasised through reviewing various studies for future research progression. 1. Introduction The efcient deployment of electric vehicles (EVs) in the world transportation system is an appealing solution to decrease greenhouse gas emission and boost energy efciency. According to the Global EV outlook 2018 [1], the worldwide automobile market will successfully deploy 117.6 million EVs on the road by 2030, which will contribute to reducing 262 Mt of CO 2 emission. The potential advantages of lithium- ion batteries (LIBs) inspire EV manufacturer companies to use them as the primary source of energy storage systems [2,3]. However, high cost, low production capacity and a highly dynamic nature limit their suc- cessive application in EV [4]. Eforts have been made to address these issues. For instance, the production cost of LIBs exponentially declined from 599 USD/Kwh in 2013 to 209 USD/kWh in 2017, while produc- tion capacity signifcantly improved from 8.3 GWh to 94.2 GWh in the same period [5]. Other studies forecasted that LIB cost will drop to 125 USD/kWh by 2022–2025 [6]. Owing to the highly nonlinear and dynamic nature of LIBs, an ef- fective battery management system (BMS) is continuously required to operate them in a safe operating area, which can be obtained through the continuous monitoring of diferent parameters, such as internal impedance, open-circuit voltage (OCV), charge transfer constant, dif- fusion and polarisation constants [7]. These parameters are also used to determine the diferent states of LIBs, namely state-of-charge (SOC) [7–11], state-of-health (SOH) [12–15], state-of-energy [16–19], state- of-power [20–23] and state-of-function [24]. An accurate SOC estima- tion is a signifcant state for the optimal charging and discharging op- eration of LIBs used in electronic gadgets and a wide range of EVs. Generally, the SOC of the battery pack in EVs is used as a fuel gauge indicator, but it cannot be measured directly using electronic sensors. Therefore, SOC estimation is performed with the help of other directly measurable quantities, such as battery terminal voltage and current. SOC estimation must be quick, reliable and accurate to ensure the high performance of BMS. However, the accurate online estimation of bat- tery SOC is a challenging task due to the high afectability and com- plicated internal chemical dynamics of the battery. Fig. 1 presents the research statistics in the feld of battery online SOC estimation in the last decade (2009–2018). The data were taken from the Web of Science https://doi.org/10.1016/j.rser.2019.06.040 Received 24 December 2018; Received in revised form 15 June 2019; Accepted 18 June 2019 ∗ Corresponding author. E-mail addresses: prashant@um.edu.my (P. Shrivastava), koksoon@um.edu.my (T.K. Soon). Renewable and Sustainable Energy Reviews 113 (2019) 109233 1364-0321/ © 2019 Elsevier Ltd. All rights reserved. T