Deep Learning-Based Approach for State-of-Health Estimation of Lithium-Ion Battery in the Electric Vehicles

Abstract

State-of-health (SoH) of the battery is crucial for ensuring the long-term safe, reliable and robust operation of electric vehicles. The health condition of the battery cannot be quantified directly using some measurement tools; therefore, estimation has to be done based on measurable quantities readily obtained from the battery management system. SoH estimation is possible either by analyzing the electrochemical process of the battery, which is highly nonlinear and unpredictable or by using data-driven techniques to trace the behavior pattern of the battery with aging. The latter stated approach is adopted here because it depends on historic data and does not require specific knowledge of material properties. The main objective is to build an accurate state-of-health estimation approach for lithium-ion batteries using the following algorithms and compare the proposed model’s performance. Thus, the three most widely used data-driven approaches, i.e., back propagation neural network (BPNN), support vector regression (SVR) and deep long short-term memory (LSTM) are applied for SoH estimation. The proposed algorithm can be used for onboard applications concerning processing and memory restrictions or used remotely by utilizing cloud data technology.