并联式混合动力汽车的BP网络实时能量管理  

Real-time energy management of parallel hybrid electric vehicle based on BP neural network

作　　者：吴剑[1] 张承慧[1] 崔纳新[1] 

机构地区：[1]山东大学控制科学与工程学院,山东济南250061

出　　处：《电机与控制学报》2008年第5期610-614,共5页Electric Machines and Control

基　　金：国家自然科学基金(50477042);教育部高等学校博士点基金(20040422052);山东省自然科学基金重点项目(Z2004G04)

摘　　要：为了提高并联式混合动力汽车(PHEV)的燃油经济性,提出了一种基于BP神经网络的并联式混合动力汽车实时能量管理策略。采用瞬时优化能量管理策略结合多种路况离线仿真得到能量管理规则,利用模糊C-均值聚类对能量管理规则进行分类并提取部分规则作为神经网络的训练样本。训练得到的BP神经网络控制器根据车辆实时工况控制混合动力系统的转矩分配,以实现最优的能量分配。基于ADVISOR的仿真研究表明,与瞬时优化能量管理策略相比,该能量管理策略不仅能够保证车辆的燃油经济性,而且明显提高了能量管理的实时性。In order to improve fuel economy of parallel hybrid electric vehicle （ PHEV）, a real-time energy management strategy （EMS） is proposed for PHEV based on BP neural network. Firstly, the energy management rules are got by offline simulation using instantaneous optimization EMS based on many kinds of drive cycles. Then the energy management rules are classified by fuzzy C-mean cluster and selected as training sample of neural network. The BP neural network controller is used to control the torque distribution of hybrid powertain for the sake of optimizing energy distribution. Finally, the energy management strategy is implemented on a PHEV prototype in ADVISOR. And the simulation results demonstrate that, compared with instantaneous optimization EMS, the proposed EMS not only satisfies the fuel economy, but also increases real-time performance of energy management effectively.

关 键 词：电动汽车 并联式混合动力汽车 神经网络 能量管理策略 模糊C-均值聚类 

分 类 号：U469.72[机械工程—车辆工程]