REGRESSION BASED SOC PREDICTION IN ELECTRIC CAR VEHICLE

State of Charge (SOC) estimation is essential for electric vehicles (EV) lithium-ion battery
safety and management system optimisation. Vehicle reliability, user trust, battery performance and
lifespan rely on accurate SOC measurements. To assess SOC, most research has employed machinebased classification techniques. Due to battery nonlinearity during vehicle operation and battery
behavior changes, these techniques are limited. This research provides a regression-based machine
learning technique for EV SOC prediction to challenge these problems. TheilSen Regressor, linear
regression, and polynomial regression are compared for SOC prediction accuracy. The performance
metrics where TheilSen Regressor consistently outperforms other models. Even in various conditions,
it predicts SOC well. Evaluation of regression models may help explain their relevance and efficiency
in EV settings. This system regression-based approaches increase SOC estimation accuracy. This
enhances battery management and advances EV technology. Researchers, engineers, and practitioners
working on EV battery management systems and energy system machine learning utilise this system.