基于插电式燃料电池公交车能量管理策略

doi:10.19562/j.chinasae.qcgc.2025.07.008

摘要/Abstract

摘要：

针对庞特里亚金极小值原理（Pontryagin's minimum principle， PMP）仅适用于离线计算且难以实车应用的问题，提出了一种基于麻雀搜索优化密度聚类（density-based spatial clustering of applications with noise， DBSCAN）的工况在线识别能量管理策略。该策略结合离线训练与在线控制，并充分利用公交车线路固定性与片段性特征，以公交站为节点，将线路划分为多个驾驶片段。在车辆靠站期间，对上一个驾驶片段的电机输出功率进行工况识别，从而计算下一个驾驶片段的协态变量（co-state）；在车辆开始运行时，将计算好的co-state应用于PMP算法中完成功率的实时分配。最后，通过构建基于实车数据的仿真实验，将所提出的策略移植到整车控制器中。结果显示，与当前实车运行的规则能量管理策略相比，该策略可减少等效氢耗17.6%，并能有效维持动力电池荷电状态（state of charge， SOC）。且每步计算均在60 ms以内，具有良好的实时性，能够满足燃料电池公交车实际运行中对能量管理策略的应用要求。

关键词: 能量管理, 工况识别, 密度聚类, 庞特里亚金极值, 硬件在环

Abstract:

For the problem that Pontryagin's Minimum Principle （PMP） is only applicable to offline calculation and difficult to be applied in real vehicles， an energy management strategy for online identification of working conditions based on density clustering （DBSCAN） is proposed. This strategy combines offline training with online control， and makes full use of the fixed and fragmentary characteristics of bus routes， using the bus stops as nodes to divide the routes into multiple driving segments. During the vehicle's stop， the motor output power of the previous driving segment is identified to calculate the co-state of the next driving segment. When the vehicle starts running， the calculated co-state is applied to the PMP algorithm to complete the real-time power distribution. Finally， by constructing the simulation experiment based on real vehicle data， the proposed strategy is transplanted into the vehicle controller. The results show that compared with the current regular energy management strategy for real vehicle operation， the proposed strategy can reduce the equivalent hydrogen consumption by 17.6% and effectively maintain the state of charge （SOC） of the power battery. Moreover， each calculation step is within 60 ms， which has good real-time performance and can meet the application requirements of energy management strategies in the actual operation of fuel cell buses.

Key words: energy management, condition identification, density clustering, Pontryagin’s maximum principle, hardware in loop

连静,杨鹏,李琳辉,周雅夫,孙雪松. 基于插电式燃料电池公交车能量管理策略[J]. 汽车工程, 2025, 47(7): 1305-1316.

Jing Lian,Peng Yang,Linhui Li,Yafu Zhou,Xuesong Sun. Energy Management Strategy Based on Plug-In Fuel Cell Buses[J]. Automotive Engineering, 2025, 47(7): 1305-1316.

图/表 21

图1

图2

表1

车辆参数"

参数	数值	单位
动力电池组容量 $Q b a t$	173	A·h
动力电池标称电压 $U S$	608.58	V
动力电池单体标称电压 $U 0$	3.22	V
动力电池单体电池数 $N b a t$	189
电机峰值功率 $P m o t o r m a x$	200	kW
燃料电池系统峰值功率 $P f c s m a x$	80	kW
燃料电池单体数 $N c e l l$	300
DCDC转化器效率 $η d c d c$	0.95

表1

图3

图4

图5

表2

不同条件下燃料电池电压衰减率"

工况	电压衰减率
启停操作 $δ o n - o f f$	$13.79 ? μ V / c y c l e$
高功率负荷 $δ h i g h$	$8.66 ? μ V / h$
怠速运行 $δ i d l e$	$10.00 ? μ V / h$
变载 $δ l o a d$	$0.044 ? 1 ? μ V / k W$

表2

图6

图7

图8

图9

表3

图10

图11

表4

图12

图13

表5

图14

图15

图16

参考文献 24

[1]	方川，袁殿，邵扬斌，等.面向冬奥示范的新一代燃料电池系统技术突破［J］.汽车工程，2022，44（4）：535-544.
	FANG Chuan， YUAN Dian， SHAO Yangbin， et al. Technological breakthrough of a new generation of fuel cell system for Winter Olympics demonstration［J］. Automotive Engineering， 2022， 44（4）： 535-544.
[2]	张瑞亮，陈准，刘森海，等.基于低通滤波的大功率型氢燃料电池重型货车自适应能量管理策略［J］.汽车工程，2021，43（11）：1693-1701，1729.
	ZHANG Ruiliang， CHEN Zhun， LIU Senhai， et al. Adaptive energy management strategy for high-power hydrogen fuel cell heavy trucks based on low-pass filtering［J］. Automotive Engineering， 2021， 43（11）： 1693-1701，1729.
[3]	LIAN J， LI D， LI L. Real‐time energy management strategy for fuel cell/battery plug‐in hybrid electric buses based on deep reinforcement learning and state of charge descent curve trajectory control［J］. Energy Technology， 2401696.
[4]	MA M， XU E， ZHENG W， et al. The optimized real-time energy management strategy for fuel-cell hybrid trucks through dynamic programming［J］. International Journal of Hydrogen Energy， 2024， 59： 10-21.
[5]	CHEN Z， LIU Y， ZHANG Y， et al. A neural network-based ECMS for optimized energy management of plug-in hybrid electric vehicles［J］. Energy， 2022， 243： 122727.
[6]	YAO Z， SHAO R， ZHAN S， et al. Energy management strategy for fuel cell hybrid electric vehicles using Pontryagin’s minimum principle and dynamic SOC planning［J］. Energy Sources， Part A： Recovery， Utilization， and Environmental Effects， 2024， 46（1）： 5112-5132.
[7]	LUO W， ZHANG G， ZOU K， et al. MLD modeling and MPC-based energy management strategy for hydrogen fuel cell/supercapacitor hybrid electric vehicles［J］. World Electric Vehicle Journal， 2024， 15（4）： 151.
[8]	LI S， CHU L， HU J， et al. A novel A-ECMS energy management strategy based on dragonfly algorithm for plug-in FCEVs［J］. Sensors， 2023， 23（3）： 1192.
[9]	WU P， PARTRIDGE J， ANDERLINI E， et al. Near-optimal energy management for plug-in hybrid fuel cell and battery propulsion using deep reinforcement learning［J］. International Journal of Hydrogen Energy， 2021， 46（80）： 40022-40040.
[10]	JIA C， HE H， ZHOU J， et al. A novel health-aware deep reinforcement learning energy management for fuel cell bus incorporating offline high-quality experience［J］. Energy， 2023， 282： 128928.
[11]	CHEN D， WANG T， QIAO T， et al. Driving cycle recognition based adaptive equivalent consumption minimization strategy for hybrid electric vehicles［J］. IEEE Access， 2022， 10： 77732-77743.
[12]	SHI J， XU B， SHEN Y， et al. Energy management strategy for battery/supercapacitor hybrid electric city bus based on driving pattern recognition［J］. Energy， 2022， 243： 122752.
[13]	ZHENG C， LI W， LI W， et al. A deep reinforcement learning-based energy management strategy for fuel cell hybrid buses［J］. International Journal of Precision Engineering and Manufacturing-Green Technology， 2022， 9（3）： 885-897.
[14]	GHADERI R， KANDIDAYENI M， SOLEYMANI M， et al. Online health-conscious energy management strategy for a hybrid multi-stack fuel cell vehicle based on game theory［J］. IEEE Transactions on Vehicular Technology， 2022， 71（6）： 5704-5714.
[15]	HUANG R， HE H， SU Q. Towards a fossil-free urban transport system： an intelligent cross-type transferable energy management framework based on deep transfer reinforcement learning［J］. Applied Energy， 2024， 363： 123080.
[16]	LI H， RAVEY A， N’DIAYE A， et al. Online adaptive equivalent consumption minimization strategy for fuel cell hybrid electric vehicle considering power sources degradation［J］. Energy Conversion and Management， 2019， 192： 133-149.
[17]	ZHOU Y， RAVEY A， PÉRA M C. Real-time cost-minimization power-allocating strategy via model predictive control for fuel cell hybrid electric vehicles［J］. Energy Conversion and Management， 2021， 229： 113721.
[18]	HAMDIPOUR A， BASIRI A， ZAARE M， et al. Artificial rabbits optimization algorithm with automatically DBSCAN clustering algorithm to similarity agent update for features selection problems［J］. The Journal of Supercomputing， 2025， 81（1）： 150.
[19]	XUE J， SHEN B. A novel swarm intelligence optimization approach： sparrow search algorithm［J］. Systems Science & Control Engineering， 2020， 8（1）： 22-34.
[20]	黄鹤，李文龙，杨澜，等.跳跃跟踪SSA交叉迭代AP聚类算法［J］.电子学报，2024，52（3）：977-990.
	HUANG He， LI Wenlong， YANG Lan， et al. Jump tracking SSA cross-iteration AP clustering algorithm［J］. Journal of Electronics， 2024， 52（3）： 977-990.
[21]	翁嘉诚，周晓杰，叶蓓蕾，等.基于改进麻雀搜索算法的K-means聚类［J］.数学的实践与认识，2024，54（2）：152-166.
	WENG Jiacheng， ZHOU Xiaojie， YE Beilei， et al. K-means clustering based on improved sparrow search algorithm［J］. Mathematics in Practice and Theory， 2024， 54（2）： 152-166.
[22]	洪志湖，李奇，陈维荣.基于PMP的机车用燃料电池混合动力系统能量管理策略［J］.中国电机工程学报，2019，39（13）：3867-3879.
	HONG Zhihu， LI Qi， CHEN Weirong. Energy management strategy of fuel cell hybrid power system for locomotive based on PMP［J］. Proceedings of the CSEE， 2019， 39（13）： 3867-3879.
[23]	周雅夫，孙雪松，连静，等.基于工况识别的燃料电池公交车能量管理策略［J］.哈尔滨工业大学学报，2023，55（8）：97-105.
	ZHOU Yafu， SUN Xuesong， LIAN Jing， et al. Energy management strategy of fuel cell bus based on operating condition identification［J］. Journal of Harbin Institute of Technology， 2023， 55（8）： 97-105.
[24]	SUN X， ZHOU Y， ZHANG X， et al. Real-time optimal EMS of adaptive charge sustenance for fuel cell hybrid buses based on driving characteristics recognition［J］. Energy Conversion and Management， 2022， 254： 115192.

实验编号	工况已知/未知	是否加SOC 惩罚项	计算的协态变量
实验1	已知	是	最优co-state （PMP）
实验2	已知	否	最优co-state （PMP）
实验3	未知	是	最优识别co-state （SSA-DBSCAN）
实验4	未知	是	次优识别co-state （DBSCAN）
实验5	未知	否	最优识别co-state （SSA-DBSCAN）
实验6	未知	否	次优识别co-state （DBSCAN）

驾驶循环策略	DC1	DC2	DC3	DC4	DC5	平均节省
DBCSCN	16.0	16.3	16.2	15.5	13.9	15.6
SSA-DBCSCN	16.8	18.2	17.5	18.6	17.1	17.6
硬件在环	16.8	18.2	17.5	18.6	17.1	17.6
工况已知	18.4	19.7	19.0	20.2	19.2	19.2

驾驶循环	终止SOC/%		等效氢气消耗/g
驾驶循环	MATLAB	硬件在环	MATLAB	硬件在环
DC1	45.00	45.01	1 034.8	1 034.8
DC2	44.95	44.95	1 056.3	1 056.4
DC3	45.07	45.07	1 050.9	1 050.9
DC4	45.02	45.02	988.6	988.5
DC5	45.18	45.18	1 003.7	1 003.7