Particle filter-based state-of-charge estimation and remaining-dischargeable-time prediction method for lithium-ion batteries

Abstract Battery state of charge is a crucial indicator of battery management systems since an accurate estimated state of charge is critical to ensure the safety and reliability of the battery. However, polarization during the discharge process can affect the dischargeable capacity in the state-of-charge definition. Moreover, a nonlinear drop of the state-of-charge may lead to misjudgment of the remaining-dischargeable-time. To address these issues, the voltage-based state of charge is defined to reduce the effects of polarization and reflect the upper bound of the remaining capacity of the battery. This paper proposes a particle filter based open circuit voltage online estimation method to achieve the voltage-based state of charge. On this basis, an open-loop remaining-dischargeable-time prediction algorithm using the voltage-based state of charge is introduced. Static and dynamic tests are presented to identify battery model parameters as well as the relationship between available capacity and voltage-based state of charge. Two definitions of the state of charge definition are applied in the prognostics architecture. Results are compared and evaluated concerning the accuracy of the voltage tracking and the relative error of the remaining-dischargeable-time prediction. The comparison results show that prognostics via voltage-based state of charge has a lower prediction relative error under different current and temperature conditions. Therefore, the voltage-based state of charge is more suitable for the remaining-dischargeable-time forecast.