State of Health Estimation and Remaining Useful Life Prediction of Electric Vehicles Based on Real W

Lithium-Ion Batteries (LIBs) are among the most used batteries in consumer electronics or electric vehicles due to various properties ranging from higher charge density to longer charge-discharge cycles. As the batteries will be the primary power source in these applications, maintaining the battery health and knowing the accurate State of Charge (SoC) is crucial. The model-based approaches may not produce accurate results in all scenarios because of the non-linear relation between the Remaining Useful Life (RUL) and SoC of LIBs. To overcome this, data-driven deep learning approaches are used. Long Short-Term Memory (LSTM) is one of the widely used algorithms to estimate SoC or RUL independently using different datasets. This paper presents a unified framework to predict both SoC and RUL using the same LSTM architecture and the dataset. The Center for Advanced Life Cycle Engineering (CALCE) battery datasets are used to train and test the proposed system. The validation accuracy obtained for both SoC and RUL is around 97%. It is observed that the results obtained are on par or slightly improved when compared to the existing algorithms that use a similar dataset to predict either SoC or RUL independently. This unified approach addresses any dependencies in data for practical applications.