A Method of SOC Estimation for Electric Vehicle Based on Limited Information

In this work, an estimation model of state of charge (SOC) based on machine learning algorithm is proposed for the real-time back cloud driving data of electric vehicle (EV). The features of driving data transmitted to the cloud is too few to use traditional SOC estimation methods based on power battery models. We process and reconstruct the online-data combined with the characteristics of EV and power battery before training the model. Subsequently, two kinds of methods summarized for processing such cloud data, namely SOC-Interpolation based Regression Algorithm and Driving-Accumulate based Classification Algorithm. Experimental results of various machine learning algorithms show that the model is able to accurately predict SOC stocks. Experiments using various machine learning algorithms show that the model is able to accurately estimate the SOC Stock of electric vehicles in motion. Among the trained models, the SOC-Interpolation based LGB model achieves the best performance.