袁永明,王贵勇,王伟超,等.基于可控多层感知器的混合动力汽车能量管理策略[J].内燃机工程,2025,46(6):82-94.
基于可控多层感知器的混合动力汽车能量管理策略
Energy Management Strategies for Hybrid Electric Vehicles Based on Controllable Multilayer Perceptrons
DOI：10.13949/j.cnki.nrjgc.2025.06.009
关键词:混合动力汽车  能量管理策略  多层感知器  射击多层感知器  自适应多层感知器
Key Words:hybrid electric vehicle  energy management strategy（EMS）  multilayer perceptron（MLP）  shooting MLP  adaptive MLP
基金项目:云南省重大科技专项计划项目（202402AE090009）
作者单位E-mail
袁永明* 昆明理工大学 云南省内燃机重点实验室，昆明 650500 2305763601@qq.com 
王贵勇* 昆明理工大学 云南省内燃机重点实验室，昆明 650500 wangguiyong@kust.edu.cn 
王伟超 昆明理工大学 云南省内燃机重点实验室，昆明 650500 287692950@qq.com 
李绕强 昆明理工大学 云南省内燃机重点实验室，昆明 650500 1727368978@qq.com 
摘要点击次数: 265
全文下载次数: 136
摘要:针对混合动力汽车神经网络能量管理策略存在难以控制动力电池工作模式和最终电池荷电状态（state of charge， SOC）的问题，提出一种可控多层感知器（multilayer perceptron， MLP）能量管理策略。将目标SOC作为MLP输入，并将其类比为等效最小燃油消耗策略（equivalent consumption minimization strategy， ECMS）中的等效因子E，引入射击法与自适应法调节目标SOC进而控制最终SOC和电池工作模式。仿真结果表明：提出的射击MLP能够实现控制电池工作模式和最终SOC，并且对比动态规划，射击MLP方法能够实现燃油消耗量仅增加0.93%，计算时间下降91.60%。对比射击ECMS，射击MLP计算时间减少72.80%，燃油消耗量降低1.22%。提出的自适应MLP回归器（MLP-R）在各种工作模式下最终的SOC偏差均在可接受范围内，相比于自适应ECMS所需计算时间减少79.00%，等效燃油消耗量下降0.15%。
Abstract:To address the issue of controlling the operating mode of the power battery and the final state of charge（SOC） in the neural network-based energy management strategy for hybrid electric vehicles， a controllable multilayer perceptron（MLP） energy management strategy was proposed. The target SOC was used as an input to the MLP and was analogized to the equivalent factor （E） in the equivalent consumption minimization strategy（ECMS）. The shooting method and adaptive method were introduced to adjust the target SOC， thereby controlling the final SOC and battery operating mode. Simulation results show that the proposed shooting MLP can effectively control the battery’s operating mode and final SOC. Compared with dynamic programming， the shooting MLP method increased fuel consumption by only 0.93% while reducing computation time by 91.60%. Compared with the shooting ECMS， the shooting MLP reduced computation time by 72.80% and decreased fuel consumption by 1.22%. The proposed adaptive MLP regressor maintained the final SOC deviation within an acceptable range across various operating modes， reduced computation time by 79.00%， and lowered equivalent fuel consumption by 0.15%， compared to the adaptive ECMS.
查看全文  HTML   查看/发表评论