State-of-health Estimation of Power Battery regarding Capacity and Internal Resistance from Field Data

Given the rapid growth of the electric vehicle (EV) industry, the investigation into state-of-health (SOH) estimation of power batteries becomes increasingly important. This study introduces a new approach for estimating power battery capacity and internal resistance using field data. Data preprocessing with EV operational status classification are carried out. Health factors are constructed based on the collected battery charge and discharge data. Power battery capacity is calculated using the ampere-hour integration method during charging, and internal resistance is determined using the Thevenin's equivalent circuit model when discharging. A multi-task convolutional neural network (MT-CNN) model is developed to simultaneously estimate the capacity and the internal resistance of power battery. The results demonstrate that the model employed in this paper exhibits higher SOH estimation accuracy compared to other machine learning methods, including the traditional convolutional neural network (CNN) method.