Electric vehicles (EVs) are becoming increasingly mainstream since sustainability and environmental concerns are growing. The majority of EVs are powered by using the Lithium-ion batteries (LiBs) batteries, hence it is important to know the LiBs State-of-Health (SOH) to ensure their safety, longevity and efficiency for a reliable Evs’ operation. Traditional SOH estimation methods require complex calculations, elaborate experiments and long-time; hence data driven techniques that relieve historical data to predict SOH is becoming more popular. In brief, the data driven SOH estimation methods for batteries leverage machine learning, deep learning, and statistical techniques and which they are typically being categorized into sequential and non-sequential approaches. Sequential approaches are well-suited for SOH estimation due to their ability to capture temporal dependencies in battery degradation patterns. Meanwhile, the non-sequential approaches do not explicitly account for temporal dependencies but are often computationally more efficient and interpretable. Sequential models are often preferred due to their ability to capture temporal dependencies and model degradation trends effectively. In this paper both sequential and non-sequential methods are discussed in addition to the different extraction feature and selection methods that are employed to enhance predictive accuracy and computational efficiency.

State of Health, EV Battery, LiB, SOH