A Practical Approach for Estimating State of Health of Li-Ion Batteries in Electric Vehicles

Accurately estimating the state-of-health (SOH) of rechargeable batteries is crucial to ensure the safe and depend-able performance of electric vehicles (EVs), consumer electronic products, and other systems powered by batteries. We propose an optimized and practical solution for accurate capacity-based SOH estimation that does not necessarily require full charge and discharge. The battery parameters related to SOH are identified in runtime using the forgetting factor recursive least square algorithm (FFRLS), followed by capacity estimation based on the identified parameters. The online update of capacity into an extended Kalman filter-based state-of-charge (SOC) estimation has improved the accuracy of the estimation. Using only partial discharge during normal driving conditions, less than 1.98 % maximum absolute error for the test scenarios has been achieved for SOH estimation. The proposed method has increased the accuracy of SOC and SOH estimation, and can maintain the computational complexity from an embedded constraint point of view by making use of relations among the states. The scheme is validated on real-time aging data to show effectiveness and practicability.