基于PF-RLS的电池组SOC估算.

To solve the problems of ampere-hour integration method for battery packs with inaccurate state of charge (SOC) estimation due to sensor accuracy, battery aging, integration error and initial value, the parameters of SOC initial value, nominal capacity and integration period of the traditional ampere-hour integration were improved. The first-order RC equivalent circuit model was established based on the internal relationship between the SOC and the open circuit voltage (Uo). In the circuit model, the Uo was extracted and updated in real time by the recursive parameter-free least squares (PF-RLS) with forgetting factor. The temperature with crucial effect on Uo was considered, and a three-dimensional model of SOC-Uo-θ was established to provide an accurate initial value for the improved ampere-time integration method. Based on the inconsistency of battery pack, the maximum voltage and the minimum voltage fusion algorithm were proposed， and the federal urban driving schedule（FUDS) working condition detection and the real vehicle working condition verification were carried out. The results show that the accuracies of the PF-RLS online extraction of Uo, the single cell SOC and the battery pack SOC algorithm are respective 2.55%, 3.20% and 4.00%, which meets the technical requirements of QC/T 897—2011 battery management system for electric vehicles claim. [ABSTRACT FROM AUTHOR]