Improved Backward Smoothing Square Root Cubature Kalman Filtering and Fractional Order—Battery Equivalent Modeling for Adaptive State of Charge Estimation of Lithium‐Ion Batteries in Electric Vehicles

Abstract

The accuracy of lithium‐ion battery state of charge (SOC) estimation affects the driving distance, battery life, and safety performance of electric vehicles. Herein, the polarization reaction inside the battery is modeled using a second‐order fractional‐order equivalent circuit model and uses an adaptive genetic algorithm for model parameter identification. Then, an improved adaptive fractional‐order backward smoothing square root cubature Kalman filtering algorithm (AFOBS‐SRCKF) is proposed by integrating Sage Husa adaptive filtering and backward smoothing processes to optimize the square root cubature Kalman filter for improving the accuracy and adaptability of real‐time estimation of SOC in a complex environment. Finally, the algorithm is compared with the integer‐order SRCKF, fractional‐order SRCKF through simulation, and fractional‐order backward smoothing SRCKF through simulation. Under complex operating conditions, the error sum of SOC estimation of the AFOBS‐SRCKF algorithm is controlled within 1.0% and the convergence speed is improved by at least 30%. The results show that the AFOBS‐SRCKF algorithm effectively improves the accuracy, stability, and convergence of SOC estimation.