Personalized Behavior-Aware Energy Management for Plug-In Hybrid Electric Vehicles Using Real-World Large-Scale User Data

Dealing with complicated personalized driving behaviors and finding a global energy management strategy (EMS) for plug-in hybrid electric vehicles (PHEVs) still remains a technical issue. Considering only external driving conditions without integrating driving behavior in EMS often results in energy waste and shortened power source lifespan. To address this issue, this study proposes a Personalized Behavior-Aware Energy Management Optimization (PBA-ECMS) framework based on large-scale real-world user data. Firstly, trip data from 300 PHEVs with relatively long average travel distances between charges are collected via a cloud platform. The Growing Neural Gas (GNG) algorithm is then employed to perform unsupervised clustering on the travel and charging behaviors of users, resulting in four representative user clusters. Secondly, for each cluster, the Equivalent Consumption Minimization Strategy (ECMS) is optimized by calibrating the equivalent factor using the bisection method, in order to reflect user-specific behavioral characteristics. Thirdly, Gamma distribution is fitted to model the driving distance between charges. Based on this model, Monte Carlo simulations are conducted to predict the users’ future travel for strategy evaluation. The results indicate that, compared to the conventional charge depletion-charge sustaining (CD-CS) strategy, the proposed PBA-ECMS significantly reduces energy cost by approximately 5.5% and battery aging cost by over 35% on average. The improvements are particularly evident for users who frequently undertake long-distance travels between charging events, for whom the reductions reach 8.5% in energy cost and 46% in battery aging cost. Furthermore, compared to ECMS with a uniform equivalent factor, the PBA-ECMS demonstrates better adaptability to behavioral diversity and achieves lower energy costs.