Error Analysis of the Model-Based State-of-Charge Observer for Lithium-Ion Batteries

Abstract

Model-based observers are widely applied in state-of-charge (SOC) estimation. The existing model-based observers can achieve high precision in theory, but the estimation precision is influenced by many factors in practical application. The impact of the possible error sources on the SOC estimation needs to be clarified in order to enhance the practical estimation accuracy. In this paper, the effect of five error sources in the model-based SOC observer is analyzed, including the initial error, the capacity error, the current measurement error, the voltage measurement error, and the model prediction error. Both theoretical derivation and simulation analysis are implemented. The results show that the initial SOC error converges to zero during the iteration, while the errors of the capacity, current, voltage, and model cause the SOC estimation error in different extents. Quantitative comparisons of these different error sources are carried out in order to determine the critical factors of the SOC estimation. The SOC estimation accuracy is mainly limited by the model precision under the current hardware and software conditions of the battery management system, indicating a possible improving direction of the SOC estimation.