Addressing accuracy limitations in state of charge estimation for lithium-ion batteries: model parameter,plateau period gain,and end calibration

This study focuses on the research of three key technical difficulties in the field of modeling and state of charge (SOC) estimation of lithium iron phosphate (LFP) batteries, which are interrelated and synergistically affect the estimation accuracy, and constructs optimization and improvement strategies. The first one is that in the parameter identification process of the second-order Thevenin model, the ohmic resistance R₀ is obtained through the voltage rise (or drop) of the terminal voltage. Due to sampling errors and the randomness of manually selecting the starting and ending points of the voltage rise (or drop), the accuracy of R₀ can be affected, thereby affecting the accuracy of the battery model. This work proposes the viewpoint of optimizing the R₀ value to weaken its impact on the accuracy of the battery model. The second one is to address the issue of SOC estimation bias caused by the flat plateau characteristics of open circuit voltage (OCV), plateau period gain is introduced to reduce the weight of measurement value with low reliability during the plateau period and increase the weight of state estimation value with higher reliability. The whale optimization algorithm (WOA) is used to obtain the optimal plateau period gain, which can minimize the SOC estimation error. And the third one is to address the problem of cumulative error caused by ampere-hour integration in SOC estimation, this work proposes an end calibration method that introduces a correction coefficient to compensate for errors, enabling end calibration for charging and discharging. The cumulative correction is also introduced to accurately determine the calibration inflection point to prevent overcorrection. The verification results have demonstrated the availability of the proposed solutions. This work provides a critical support for battery management systems (BMSs).