In order to solve the problem of low specific power of the single-power battery that unable to respond to transient power demand in plug-in hybrid electric vehicle,a hybrid energy storage system with parallel connection between battery and supercapacitor was designed.To the problem of lack real time when using dynamic programming(DP)method to optimize load current distribution,a training set composed of DP optimization results under different driving conditions data sets was used for training,and a multi-model fusion energy distribute strategy under Stacking integrated learning framework was proposed by integrating GRU network and XGBoost algorithm.The simulation results show that compared with the energy storage system using only a single battery,the real-time energy distribution system based on Stacking fusion model can reduce the peak current by 48.7％and 50.8％under UDDS and US06 cycles,respectively,effectively reduces the peak current of the battery and improves the overall performance of the battery.
electrical measurementhybrid energy storage systemsplug-in hybrid electric vehicledynamic programmingXGBoostStacking fusion model
维普
万方数据
