An Optimized Method for Battery Swapping Demand Prediction based on Random Forest Regression

Battery swapping mode (BSM) is a feasible pattern for replenishing new energy vehicle (NEV), where its operation is directly affected by the battery swapping demand. As the demand for battery swapping is varied with time, this demand can be regarded as time series. Recently, as an important part of machine learning, random forest (RF) approach has shown its capability of regression analysis. Thereby, a battery swapping demand prediction method based on RF regression approach is proposed in this paper. For applying this method, we design features on the basis of seasonal and trend-cyclicity decomposition using loess (STL). Numeric simulation is conducted to assess the performance of the RF method. The results show that the proposed method can provide relatively accurate and robust prediction for battery swapping demand.