Lithium-Ion Battery State of Charge Estimation Using Least Squares Support Vector Machine

Electric vehicles (EVs) are being developed in response to the decline of fossil fuels and growing concerns about the environment. To power these vehicles, lithium-ion (Li-ion) batteries are commonly utilized due to their high energy density and long cycle life. The design of the Battery Management System (BMS) relies heavily on maintaining the State of Charge (SOC) of the Li-ion battery. However, there are several areas that require further attention, including temperature regulation, cell balancing within the battery pack, and internal state estimation. This study proposes a novel framework based on machine learning called Least Squares Support Vector Machine (LSSVM) to enhance the accuracy of battery SOC estimation. A comparison is made between the LSSVM method and a conventional AH method in terms of accuracy and performance of their respective outputs.