Fast State-of-Charge Estimation for Lithium-Ion Batteries Using a Simplified Electrochemical Model Without Initial State Restrictions

To effectively assess retired batteries from electric vehicles (EVs), it is indispensable to estimate the battery state-of-charge (SoC), as many aging characteristic parameters are sensitive to it. This study aims to develop a rapid SoC estimation approach based on the simplified electrochemical model (SEM) and genetic algorithms (GAs) without restricting the initial equilibrium state of batteries. Specifically, the full-order pseudo-2-D (P2D) model is simplified with Padé approximation and further improved by introducing the initial state parameters from the perspective of overpotentials. Next, a novel fast SoC estimation scheme is proposed using the constrained optimization method with a current pulse of 60 s instead of recursive estimation with a time-consuming relaxation procedure from 1 to 3 h in conventional methods. The parameter sensitivity analysis is conducted to determine the key variables to be optimized. In addition, the GA is improved by applying a repair operator to handle the constraints caused by battery operating conditions. Finally, three test cycles are designed to diversify the battery initial states for experimental verification. The results indicate that the mean absolute deviations (MADs) among the entire SoC range are no more than 4% with different battery initial states, which are reduced by 3.5% compared to the adaptive extended Kalman filter (AEKF) algorithm with an equivalent circuit model (ECM).