BATTERY PARAMETER ESTIMATION OF LITHIUM-ION BATTERIES USING TEMPORAL CONVOLUTIONAL NETWORK

Abstract

For electric vehicles, lithium-ion (Li-ion) batteries serve as the primary energy source. Accurately estimating and predicting the state of health (SOH) and remaining useful life (RUL) of lithium-ion batteries play an increasingly crucial role in intelligent battery health management systems. Additionally, it serves as a battery failure early warning system. Model-based and data-driven approaches can be broadly classified as the two types of ap proaches that have been documented in recent years for SOH estimation and RUL assessment of lithium-ion batteries. Among these, data-driven techniques can produce more accurate predictions. This work focuses on developing a deep learning model for battery parameter estimation in electric vehicles. A Temporal Convolutional Network (TCN) model is proposed for SOH monitoring and RUL assessment in Lithium-ion batteries before the failure of the battery. Three analytical indices: RMSE, MAE and R SQUARE are chosen to evaluate the prediction results numerically. The proposed model is experimented and tested using NASA lithium-ion battery health dataset. The prediction results of both SOH and RUL showed good accuracy of 99% indicating that the proposed model has high robustness and good performance.