State of Health Estimation and Remaining Useful Life Prediction of Electric Vehicles Based on Real-World Driving and Charging Data

Abstract

As the dominant choice for powering the Electric Vehicles (EVs), it is crucial to estimate its state of health (SOH) and predict its remaining useful life (RUL). This article proposes a novel machine learning-based prognostic method for lithium-ion batteries with real-world driving and charging data. A SOH evaluation system and a cluster interpolation correction method are applied to address the various data problems. Based on the capacity estimation method, select the voltage ranges through Dynamic Non-dominated Sorting Genetic Algorithm II (D-NSGA-II), which can dynamically capture the optimal ranges in different environments. A multi-dimensional input fusion model (GM-LSTM) is proposed to predict RUL, overcoming the problem of limited data. Additionally, several experiments based on EVs are implemented to verify the proposed method. The experimental results demonstrate the effectiveness of the proposed methodology, with the average relative error for SOH estimates and RUL forecasts are 1.53% and 1.34%.