关键词: 动力电池, 等效电路模型, 在线参数辨识, SOC估计

Abstract: Traditional joint estimation algorithms for the model parameters and state of charge (SOC) of the battery usually employ a two-layer architecture: one recursive estimator identifies all the model parameters and the other infers SOC. Due to the large number of parameters to be identified, these algorithms often have the problems of tedious tuning and poor robustness. In order to solve these problems, an online joint estimation algorithm of impedance parameters and SOC is proposed based on a three-layer architecture. This algorithm identifies Ohmic internal resistance and polarization parameters separately for reducing the complexity of the problem. In addition, by analyzing the dynamic characteristics of the modeling error of first order RC model, a lumped error model based on first order inertial link is introduced, with the accuracy of the first order RC model improved. The results of verification on two set of real vehicle operation condition data show that compared with traditional algorithm, the algorithm proposed has significantly higher robustness with improved accuracy; the SOC estimation error can quickly converge to less than 2% and 3% at 25 and -20 ℃, respectively. Meanwhile, the results of sensitivity analysis show that the algorithm also has good robustness to initial errors as well.

Key words: traction battery, equivalent circuit model, on-line parameter identification, SOC estimation