【作者】 杨鹏；

【导师】 连静； 李康；

【作者基本信息】 大连理工大学 ， 车辆工程， 2025， 硕士

【摘要】 在全球“双碳”战略和氢能政策的推动下,氢能的开发与应用正日益受到广泛关注。质子交换膜燃料电池作为氢能的主要利用形式之一,因其输出功率不易突变的原因,通常需要与其他能源系统构成混合动力系统,以满足交通运输的动态需求。在此背景下,能量管理策略的设计对提升整车经济性与运行效率具有重要意义。因此,本文以大连市氢能示范项目中的插电式燃料电池公交车为研究对象,结合工况识别方法,构建包含纯电优先与电池荷电状态(State of Charge,SOC)维持的双模式能量管理策略框架。旨在降低整车等效氢耗并延缓燃料电池寿命衰减。研究工作主要包括以下四个方面:(1)为实现能量管理策略的验证与优化,在MATLAB/Simulink环境中搭建整车前向仿真平台。构建的模型包括整车能量流动模型、燃料电池模型、动力电池模型、车辆纵向动力学模型及整车等效氢气消耗模型。为分析能量管理策略对系统寿命的影响,分别在燃料电池和动力电池模型中引入相应的退化模型,以反映其性能衰减特性。(2)针对公交车运行过程中工况多变的问题,提出一种融合麻雀搜索(Sparrow Search Algorithm,SSA)优化密度聚类(Density-Based Spatial Clustering of Applications with Noise,DBSCAN)的工况在线识别方法。首先对实车采集的数据进行预处理和片段划分。随后通过分析以车速表征工况对公交车的不适用性,选取功率相关参数来表征工况。从选取的15维特征参数中,通过核主成分分析法将特征维度降至4～5维,以有效降低计算复杂度。最后分别采用DBSCAN算法与SSA-DBSCAN算法进行工况聚类识别,通过对比两者的轮廓系数值,验证SSA-DBSCAN方法在工况识别精度方面的优势。(3)针对庞特里亚金极小值原理(Pontryagin’s Minimum Principle,PMP)在能量管理策略中仅适用于离线优化、难以实际应用的问题,本文基于实时工况识别结果,动态调整协态变量,实现PMP策略的在线应用。为进一步提升燃料电池系统的耐久性,在哈密顿函数中引入怠速与变载惩罚项;同时加入SOC惩罚项以增强SOC维持能力。仿真结果表明,与当前实车应用的规则策略相比,所提策略可降低等效氢耗17.6%,燃料电池电压的平均衰减减少11.4%,并显著改善SOC维持效果。(4)搭建HiL仿真平台,用于验证所提能量管理策略的可行性。通过构建与实车相同的控制策略,并与实车采集数据对比,验证了所建立整车模型的准确性。同时,通过对比HiL实验结果与MATLAB仿真结果,分析表明两者在SOC变化趋势和燃料电池功率输出方面高度一致,展示出良好的工程可实现性与实际应用潜力。更多还原

【Abstract】 Driven by the global"dual carbon"strategy and hydrogen energy policy,the development and application of hydrogen energy are increasingly attracting widespread attention.As one of the main forms of hydrogen energy utilization,proton exchange membrane fuel cells usually need to form a hybrid system with other energy systems to meet the dynamic needs of transportation because their output power is not easy to change suddenly.In this context,the design of energy management strategy is of great significance to improving the economy and operation efficiency of the whole vehicle.Therefore,this paper takes the plug-in fuel cell bus in the Dalian hydrogen energy demonstration project as the research object,and combines the working condition identification method to construct a dual-mode energy management strategy framework including pure electric priority and battery state of charge(SOC)maintenance.It aims to reduce the equivalent hydrogen consumption of the whole vehicle and delay the life decay of the fuel cell.The research work mainly includes the following four aspects:(1)In order to realize the verification and optimization of the energy management strategy,a whole vehicle forward simulation platform is built in the MATLAB/Simulink environment.The constructed models include the whole vehicle energy flow model,fuel cell model,power battery model,vehicle longitudinal dynamics model and whole vehicle equivalent hydrogen consumption model.In order to analyze the impact of energy management strategy on system life,corresponding degradation models are introduced into the fuel cell and power battery models to reflect their performance attenuation characteristics.(2)In view of the problem of variable working conditions during bus operation,an online working condition identification method integrating Sparrow Search Algorithm(SSA)optimized density clustering(DBSCAN)is proposed.First,the data collected from the real vehicle are preprocessed and segmented.Then,by analyzing the inapplicability of using vehicle speed to characterize the working condition for buses,power-related parameters are selected to characterize the working condition.From the selected 15-dimensional feature parameters,the kernel principal component analysis method is used to reduce the feature dimension to 4-5dimensions to effectively reduce the computational complexity.Finally,the DBSCAN algorithm and the SSA-DBSCAN algorithm are used for working condition clustering identification.By comparing the silhouette coefficient values of the two,the advantages of the SSA-DBSCAN method in working condition identification accuracy are verified.(3)In view of the problem that Pontryagin’s Minimum Principle(PMP)is only applicable to offline optimization and difficult to apply in practice in energy management strategies,this paper dynamically adjusts the co-state variables based on the real-time working condition identification results to realize the online application of the PMP strategy.In order to further improve the durability of the fuel cell system,idle and load penalty terms are introduced into the Hamiltonian function;at the same time,the SOC penalty term is added to enhance the SOC maintenance capability.The simulation results show that compared with the current rule strategy applied in the actual vehicle,the proposed strategy can reduce the equivalent hydrogen consumption by 17.6%,reduce the average attenuation of the fuel cell voltage by 11.4%,and significantly improve the SOC maintenance effect.(4)A HiL simulation platform is built to verify the feasibility of the proposed energy management strategy.By constructing the same control strategy as the actual vehicle and comparing it with the data collected from the actual vehicle,the accuracy of the established vehicle model is verified.At the same time,by comparing the HiL experimental results with the MATLAB simulation results,the analysis shows that the two are highly consistent in terms of SOC change trend and fuel cell power output,showing good engineering feasibility and practical application potential.更多还原