An Adaptive Moving Window Total Least Square Method Considering Load Current Condition Aimed at Online Lithium-Ion Battery State-of-Health Estimation

Accurate and online state-of-health estimation is critical to lithium-ion batteries in electrified vehicles to ensure their high performance over time. Relevant analytical methods perform online SOH estimation requiring inputs of the SOC variation and the charge stored in the battery in a time interval. However, SOC estimation errors and unclear time interval selection are two technical challenges for analytical methods. To overcome the issues and fulfill the online SOH estimation objective, this paper proposes an adaptive moving window total least square method. The performance of the proposed algorithm is examined using the Stanford battery degradation dataset, which cycles the battery using the urban dynamometer driving schedule (UDDS) profile. Compared with relevant methods in the literature, the proposed algorithm achieves more accurate SOH estimation, achieving a -0.01% mean relative error and 1.96% mean absolute error, averaged from fourteen benchmark checkpoints, in the presence of slowly-varying 2% SOC estimation error; the proposed algorithm also renders a constrained estimation with a 2.60% standard deviation. Also, the implementation procedures and their decision-making are well explained, allowing future implementation in electric vehicles.