An Integrated EV Battery Charging and Health Optimization Framework Using Advanced Charging Techniques and Adaptive Kalman Filter–Based SOH Estimation

Abstract

Electric vehicle (EV) battery performance and lifespan are strongly dependent on charging strategies and accurate health monitoring. Conventional EV chargers emphasize fast charging using continuous current profiles, often accelerating battery degradation. This paper proposes a novel integrated EV battery charging and health optimization framework that combines advanced charging techniques with an Adaptive Kalman Filter (AKF)–based State of Health (SOH) estimator. A CC–CV controlled buck–boost converter is employed for EV battery charging, while continuous, pulse, burp, and taper charging techniques are implemented. Simultaneously, battery SOC and SOH are estimated in real time using the AKF based on voltage, current, and temperature feedback. MATLAB/Simulink results demonstrate that burp charging, when coupled with AKF-based health estimation, significantly improves SOH and reduces internal resistance growth compared to other charging methods under identical operating conditions. The proposed framework enables health-aware EV charging, making it suitable for next-generation smart and fast charging infrastructures.

Introduction: The transition to electric vehicles is driven by sustainability goals, yet battery degradation remains a major challenge affecting performance and cost. Conventional CC–CV charging and standalone health monitoring accelerate battery ageing and limit reliability. This work proposes an integrated charging and battery health optimization framework using advanced charging strategies with an Adaptive Kalman Filter–based estimator. Accurate SOC estimation is addressed by comparing Coulomb Counting and Extended Kalman Filter methods using a Thevenin battery model. MATLAB simulations demonstrate improved estimation accuracy and support reliable battery management for EV applications.

Objectives: This paper aims to develop an integrated EV battery charging and health optimization framework that jointly supports fast charging and long-term battery reliability. It emphasizes the integration of advanced charging strategies with real-time battery health estimation to enable intelligent, health-aware charging. The study further compares continuous, pulse, burp, and taper charging techniques under identical conditions using a CC–CV controlled buck–boost converter to evaluate their impact on battery SOH and degradation. Finally, an Adaptive Kalman Filter–based SOC and SOH estimation algorithm is designed and validated in MATLAB/Simulink to demonstrate improved battery lifespan, safety, and reliability for future intelligent EV charging systems

Methods: This work employs a first-order Thevenin equivalent circuit model to represent battery dynamics, as it effectively captures polarization effects and provides an accurate balance between simplicity and performance. Battery SOC is initially estimated using the Coulomb Counting method due to its simplicity and suitability for short-term estimation, despite its sensitivity to initial SOC errors and current measurement noise. To address battery nonlinearity and ageing effects, an Extended Kalman Filter (EKF) is implemented using the discretized state-space equations of the Thevenin model. The EKF operates through prediction, correction, and noise update stages to provide robust SOC estimation under dynamic operating conditions. In addition, battery SOH is evaluated using coulomb-based capacity estimation and direct internal resistance estimation, enabling reliable assessment of battery degradation and health evolution.

Results: iteration-based EKF provides more accurate SOC estimation than the Coulomb Counting method, with the EKF curve (cyan) closely tracking the true state compared to the CC curve (brown). This improved accuracy is achieved through iterative state correction and adaptive gain updates using current and previous system states.

Conclusions: The proposed system effectively estimates battery SOC using an Extended Kalman Filter implemented in Simulink and benchmarked against the conventional Coulomb Counting method. The EKF achieves a maximum SOC error of 3% and an average error of 2%, demonstrating superior accuracy compared to Coulomb Counting. Although dynamic discharge current variations are considered, factors such as temperature effects, sensor noise, and measurement inaccuracies are not included, indicating scope for further model refinement. Achieving real-time SOC estimation with minimal error remains challenging without increasing system cost or complexity. This study therefore emphasizes a reliable, low-complexity, and cost-effective filter-based SOC estimation approach suitable for modern lithium-ion battery applications.