Improved Fixed Range Forgetting Factor-Adaptive Extended Kalman Filtering (FRFF-AEKF) Algorithm for the State of Charge Estimation of High-Power Lithium-Ion Batteries

Abstract

The lithium-ion battery is perhaps the most powerful energy storage media available today and is used in virtually all electronic devices, especially electric and hybrid electric vehicles. The battery industry is growing rapidly in battery technology, development, and production to meet future demands. The difficulty in estimating battery states such as the state of charge (SOC) has led to the discovery of several methods and techniques. The use of improved algorithms coupled with a combination of methods and models has contributed immensely toward the accurate estimation of battery states. In this paper, the state of charge of the high-power lithium-ion battery is estimated based on an improved Fixed Range Forgetting Factor-Adaptive Extended Kalman filtering (FRFF-AEKF) algorithm. The interference of system noise is overcome with the use of the fixed range forgetting factor and the Saga-Husa adaptive filter (SHAF) to calculate the SOC more accurately. The experiments performed for the acquisition of data, parameterization, and verification of results, the methods employed and the use of the improved algorithm were all done to accurately estimate the SOC. Two other algorithms, the Adaptive extended Kalman filtering (AEKF) algorithm, and the Adaptive Unscented Kalman filtering (AUKF) algorithm are used as benchmarks for verifying the performance of the improved FRFF-AEKF algorithm. The improved FRFF-AEKF algorithm achieved 99.74 % estimation accuracy under Hybrid Pulse Power Characterization (HPPC) test working conditions and 99.44 % under Beijing Bus Dynamic stress test (BBDST) working conditions. The estimation accuracy of the AEKF algorithm under HPPC and BBDST conditions was 98.37% and 99.27% respectively, and the estimation accuracy of the AUKF algorithm under HPPC and BBDST conditions was 97.97% and 99.07% respectively. The verification experiment proved that the method was successful and can accurately estimate the state of charge of the high-power lithium-ion battery