State-of-charge estimation for supercapacitors based on salp swarm algorithm-optimized high and low degree cubature Kalman filters considering temperature uncertainty

Abstract

The accurate estimation of state-of-charge (SOC) is crucial for the supercapacitor (SC) management system of electric vehicles. In this paper, a salp swarm algorithm (SSA)-optimized high and low degree cubature Kalman filters (CKFs) for SC SOC estimation is proposed. Firstly, the key parameters are identified for the Thevenin model by SSA under SC test data. Secondly, a temperature-varied model with temperature uncertainty is constructed by polynomial fitting. Thirdly, an adaptive rule based on SSA is designed to regulate the noise covariance matrix information in real time. Combining this rule and CKFs, SSA-based generalized high-degree CKF (SSA-GHCKF) and SSA-based low-degree CKF (SSA-CKF) are developed with the minimum prediction error as the objective. Finally, the effectiveness of improved CKFs is evaluated by the urban dynamometer driving schedule in varying temperatures. The simulation results show that the root mean square errors of SOC estimation by CKF, GHCKF, and SSA-CKF are kept within 1.4553 %, 1.2880 % and 1.2152 %, while for SSA-GHCKF, this metric is merely 1.1933 %. The proposed SSA-GHCKF enables superior accuracy and robustness for SOC estimation at extreme temperatures.