Robust Energy Management Strategy Design for PHEB Based on DFSS and LSTM

This paper presents a robust energy management strategy framework combining Design for Six Sigma (DFSS) and long short-term memory (LSTM). Various driving uncertainties affecting plug-in hybrid electric buses (PHEBs) are explicitly addressed. First, the segmented equivalent fuel consumption minimum strategy (ECMS) is developed as the foundational energy management framework. In the offline phase, a robust equivalent factor (EF) sequence for a single driving cycle is designed using the DFSS methodology to account for uncertainties in resistance coefficients and passenger loads. Then, the LSTM is trained with road segment speed clustering labels alongside the robust EF sequence, enabling the online generation of these robust parameters for new driving cycles. By referencing the actual historical speed data, the EF sequence for the entire driving cycle can be proactively designed to mitigate the impact of speed uncertainties. The simulation results show that the robust energy management strategy achieves lower fuel consumption expectations and standard deviation compared to the deterministic design. The results of HIL test substantiate the validity and effectiveness of the proposed robust design method.