Title:Electric Vehicle Battery Remaining Charging Time Estimation Considering Charging Accuracy and Charging Profile Prediction

Abstract:Electric vehicles (EVs) have been growing rapidly in popularity in recent years and have become a future trend. It is an important aspect of user experience to know the Remaining Charging Time (RCT) of an EV with confidence. However, it is difficult to find an algorithm that accurately estimates the RCT for vehicles in the current EV market. The maximum RCT estimation error of the Tesla Model X can be as high as 60 minutes from a 10 % to 99 % state-of-charge (SOC) while charging at direct current (DC). A highly accurate RCT estimation algorithm for electric vehicles is in high demand and will continue to be as EVs become more popular. There are currently two challenges to arriving at an accurate RCT estimate. First, most commercial chargers cannot provide requested charging currents during a constant current (CC) stage. Second, it is hard to predict the charging current profile in a constant voltage (CV) stage. To address the first issue, this study proposes an RCT algorithm that updates the charging accuracy online in the CC stage by considering the confidence interval between the historical charging accuracy and real-time charging accuracy data. To solve the second issue, this study proposes a battery resistance prediction model to predict charging current profiles in the CV stage, using a Radial Basis Function (RBF) neural network (NN). The test results demonstrate that the RCT algorithm proposed in this study achieves an error rate improvement of 73.6 % and 84.4 % over the traditional method in the CC and CV stages, respectively.