Comparative Analysis of State of Charge Estimation Methods for Li-Ion Batteries Using Kalman Filter Variants and Machine Learning Techniques for Electric Vehicle Applications

State of Charge (SoC) measurement accuracy stands as a fundamental requirement to boost both safety and power performance together with energy efficiency in Battery Management Systems (BMS) of Electric Vehicles (EVs). This research evaluates SoC estimation techniques for lithium- ion batteries using Kalman Filter (KF), Extended Kalman Filter (EKF), and Unscented Kalman Filter (UKF) with Equivalent Circuit Models (ECMs) from 1RC to 5RC.These filters undergo independent assessment as well as evaluation when paired with established Machine Learning (ML) technologies that include Adaptive Neuro-Fuzzy Inference System (ANFIS) and Artificial Neural Networks (ANN). Performance evaluation utilizes Root Mean Square Error(RMSE) and Mean Absolute Error(MAE) alongside RMAE and Coefficient of Determination (R² score) and Average Mean Square Error (AMSE) to derive quantitative results. Test results using actual battery and simulated datasets demonstrate that UKF produces accurate and stable performance in combination with ANFIS as a system under dynamic battery conditions. Research data demonstrates that this combined hybrid system configuration offers proper effectiveness when used for real-time SoC estimation in EV applications. MATLAB/Simulink was used to develop and simulate the KF-based and hybrid SoC estimation models, while Python was used to implement standalone Machine Learning models on the Raspberry Pi to measure their computational performance and ability to deploy them in embedded systems.