Generalized Mixture Minimum Error Entropy-Based Square Root Cubature Kalman Filter for State of Charge Estimation of Lithium-Ion Batteries ​

Abstract

The data obtained from a lithium-ion battery may be affected by the complex non-Gaussian noise, which may significantly impact the efficacy of the cubature KF. The minimum error entropy based CKF has recently been successfully applied to state of charge (SOC) estimation. However, the MEE criterion in MEE-CKF fails to address more complex noise environments. By utilizing the advantage of the generalized minimum mixture error entropy criterion in handling non-Gaussian noises encountered in the SOC estimation, this article proposes a novel GMMEE criterion-based square-root cubature Kalman filter (SRCKF) to estimate the SOC. The proposed GMMEE-SRCKF with two flexible kernels provides a more flexible method for fitting non-Gaussian noise, constituting a significant advantage of SRCKF in addressing numerical problems. Furthermore, this paper presents a discussion of the parameters in the GMMEE-SRCKF, indicating that the proposed method will degenerate into the mixture minimum error entropy SRCKF (MMEE-SRCKF) and the generalized minimum error entropy SRCKF (GMEE-SRCKF). The evaluation indicators presented in this paper further demonstrate that the proposed method is superior to other robust algorithms in estimating the SOC under non-Gaussian noise. Finally, the elapsed time for the proposed method is within the acceptable range for SOC estimation.

Keywords: Complex non-Gaussian noise, State of Charge, Generalized mixture minimum error entropy (GMMEE), Square-root cubature Kalman filter