【作者】 郭海龙；

【导师】 李礼夫；

【作者基本信息】 华南理工大学 ， 机械电子工程， 2013， 博士

【摘要】 混合动力电动汽车(Hybrid electric vehicle，HEV)整车驱动能量控制策略是HEV核心技术之一，一直以来都是HEV的研究热点，但目前相关研究因为没有很好地处理以下三个问题，导致HEV油耗和尾气排放依然偏高。第一，控制策略没有考虑行驶状况(整车载荷、道路坡度)动态变化对于能量控制策略的反作用；第二，控制策略主要基于动力型驾驶模式，当驾驶人对动力性要求过高时，会导致HEV的油耗和排放变差；最后，控制策略优化模型中的部件特性忽略了车况变化的影响。鉴于此，论文进行了基于行驶状况识别的混联式HEV多模式能量控制策略研究。具体如下：进行了基于行驶状况识别的驾驶人意图辨识模型研究，首先，以一款混联式HEV作为研究对象，基于汽车纵向动力学原理，利用实车动态采集的车速、加速度等汽车行驶动力学参数，运用最小二乘方法构建了HEV整车载荷和道路坡度求解模型，运用粒子群智能优化算法，研究了载荷和坡度在线求解机理，并开展了18组实车载荷和坡度识别验证实验，验证方法可行性；然后，为探析驾驶人的驾车意图，对180名驾驶人进行了调查，得出“驾驶人对于加速踏板的操作更多的体现为驾驶人对车速的需求”结论，由此利用人机工程学舒适度与疲劳原理，建立了“加速踏板——车速需求”非线性关系模型，用于解析驾驶人的车速需求，基于此，研究了基于行驶状况识别的HEV驾驶人需求转矩意图辨识模型，可为HEV能量控制策略优化提供驾驶人的需求转矩信息；最后，为了分析驾驶人意图辨识结果对HEV驱动能耗的影响，进行了不同载荷和坡度下的实车实验，得出基于载荷和坡度识别的驾驶人意图辨识模型，较原车可节约驱动能耗3.08%的结论。进行了基于行驶状况识别的混联式HEV建模研究，运用MATLAB/SIMULINK软件，建立了HEV能量控制策略整车模型，包括整车动力学模块、行驶状况识别模块、驾驶意图辨识模块、发动机、发电机、电动机模块和动力电池模块等，并利用实车实验数据对模型进行了有效性验证。其中，在发动机模型方面，研究了发动机平均值模型的基础理论，基于此，结合行星齿轮动力耦合机构的动力学模型和效率模型，提出了用发电机转矩来间接测算发动机输出转矩的方法，并进行了实验验证，解决了发动机平均值模型中发动机输出转矩难以获得的问题；此外，从270组发动机实车实验(实验道路包括城市、城郊、高速、山区道路，实验车况为空载、半载和满载，实验路况为平路和坡道)数据中，选取其中47组典型的发动机运行参数实验数据，运用最小二乘辨识方法和粒子群算法及遗传算法，对发动机平均值模型的37个待定系数进行了辨识，最终建立了1NZ—FXE4缸16气门发动机平均值模型，基于此，提出了利用发动机平均值模型构建发动机效率模型的方法，可解决HEV在用过程中，因发动机效率模型变化，进而导致HEV节能减排效果恶化的问题；最后，为解决动力总成效率等特性参数查表模型存在插值误差的问题，基于查表模型数据，经过训练，建立了发动机、发电机、电动机和动力电池的特性参数神经网络模型。进行了HEV多工况多模式能量控制策略研究，首先，由HEV能量控制模型和基于行驶状况识别的驾驶意图辨识模型，给出驾驶人需求驱动转矩意图，作为HEV能量控制策略输入参数之一；然后，结合遗传粒子群智能仿生理论的全局最优和快速收敛优点，研究了HEV多工况多模式能量控制策略，控制策略涵盖5种工况(停车充电、起步、起步后、制动、倒车)、3种驾驶模式(动力型、经济型、平衡型)、4种能量流模式(电驱动、油驱动、油驱电充、油驱电放)，共37种具体优化策略。控制策略以综合考虑能耗和排放的HEV综合性能指标为优化目标，共计优化了179584个HEV工作状态点，其中动力型能量控制策略体现了动力性强的特点，经济型策略体现了能耗特性优的特点，而平衡型策略介于二者之间，通过研究还发现了在SOC、车速相同条件下，HEV能量转化率随驾驶人需求驱动转矩的增加呈现“U形抛物线”规律；实验结果表明，所研究的控制策略不论能耗还是排放特性相对原车均有改善；最后，对PRIUS原车倒车工况下，混动驱动模式的能量控制策略进行了理论分析和实车实验，发现原车因发动机转向与驱动轮转向相反，而导致能量浪费的问题，并提出改进方案。最后，对基于行驶状况识别的混联式HEV多模式能量控制策略进行了仿真和实车实验研究。更多还原

【Abstract】 Energy control strategy is one of the core technology of Hybrid electric vehicle (HEV),and has been a research hot spot all the time. But the existent research didn’t deal with thethree following problems very well caused high fuel consumption and emission of HEVrelatively. Firstly, the existent strategy didn’t take the reaction of dynamic variation ofdriving conditions (the carload, road slope) to the energy control strategy into account.Secondly, because the control strategy based on the power driving mode, as the requirementof power is too high, it will cause HEV worse fuel consumption and emission. Thirdly, thecomponents’ character of control strategy optimization model ignores the influence ofvehicle condition varies. In view of this, this dissertation did study on series-parallel HEV’smultimode energy control strategy based on driving condition recognition.In the research of identification model of driver’s intention based on driving conditionrecognition, Firstly, taking a serial-parallel HEV as research object, based on vehiclelongitudinal dynamics and making use of vehicle speed, acceleration speed etc which werecollected by real vehicle tests, the carload and road slope solving model was built by leastsquare method. Applying particle swarm optimization method, the solving theory of carloadand road slope was studied. Eighteen groups of identification experiments of real vehicledynamical carload and road slope were collected to verify the theory, which turns out to befeasible. Secondly, in order to analyze driver’s intentions,180drivers were investigated,and the conclusion of ‘driver’s operation on accelerate pedal is much more like therequirement of vehicle’s speed’ was obtained. Since then, taking advantage of comfortdegree and fatigue theory in ergonomics, the nonlinear model called ‘acceleratorpedal-speed need’ was built to identify the speed requirement of drivers. Based on that, theidentification model of driver intention about torque requirement was studied, which canprovide information of the driver’s torque requirement for energy control strategyoptimization. Thirdly, for analyzing the influence of the result of driver’s intentionidentification on driving power consumption of HEV, varies of real vehicle tests of carloadand road slope was proceeded. The results showed that driver’s intention recognition model based on carload and road slope identification can save driving energy3.08%than theoriginal vehicle.In the research of model building of the serial-parallel HEV based on driving conditionrecognition, The full vehicle model of HEV energy control strategy was built byMATLAB/SIMULINK, including full vehicle dynamic module, driving conditionidentification module, driver intention recognition module, engine module, generatormodule, motor module, and power battery module, and the models’ effectiveness wasidentified with the real vehicle tests data. In aspect of engine module, the basic theory ofthe engine mean value model was studied. Based on this, in combination of the dynamicmodel and efficiency model of planetary gear coupling mechanism, an approach wasproposed which making use of the torque of generator to calculate the torque of engineoutput indirectly. And also an experiment was proceeded to prove it. It solved the problemof acquiring the engine output torque in the mean value model of engine. Besides,47groups of typical engine operation parameters were chosen from270groups of real vehicletests data (the test road type included city, suburb, freeway, and mountain area, the vehiclecondition included no-load, half load and full load, the road condition included flat andslope) to identify the37undetermined coefficient of the engine mean value model with amethod of least square identification and algorithms of particle swarm and genetic. And amean value model of1NZ—FXE engine which has4-Cylinder and16-Valve was founded.On base of this, an approach of taking advantage of engine mean value model to build theefficiency model of engine was proposed, which could solve the problem of deterioration ofsaving fuel and emission of HEV because of the changing of engine model during operatingof the vehicle. At last, in order to solve the problem of interpolation precision of the MAPmodels of power components efficiency and other parameters, on the base of the MAPmodel’s data, the neural network model of characteristic parameters of engine, generator,electric motor and power battery were founded.In the research of energy control strategy of multi driving cycles and multi modes,Firstly, based on the whole vehicle model of HEV energy control and driver’s intentionidentification model, the driver’s intention of torque acquirement was given as one of theinput parameters of energy control strategy of HEV. Secondly, taking advantage of global optimal and fast convergence of the genetic particle swarm intelligent bionic theory, multidriving condition and modes of HEV energy control strategy were studied. The strategyincludes5driving conditions (such as Ready, Starting, After starting, Braking, Reversing),and3kinds of driving modes (such as Sport, Economic, Balance), and4kinds of energyflow modes (Battery drive, Fuel drive, Fuel drive and battery charging, Fuel and batterydrive), and37groups of specific energy optimal strategy. The strategy takes integratedperformance of fuel consumption and emission of HEV as optimization target, and179584working points were optimized in all. The sport mode energy control strategy shows thecharacteristic of power, and the economic strategy shows the characteristic of low fuelconsumption, and the balance strategy is between the two. It was also found that HEV’senergy transform rate presents ‘U’ style rule with driver’s torque requirement increasing.The test shows that the strategies those were studied either in fuel consumption or emissionwere better than original vehicle. Thirdly, aiming at the reversing condition of PRIUS, thestrategy of hybrid drive mode was analyzed and real vehicle was tested. It turns out thatthere is energy waste of the original vehicle because of the opposite rotational direction ofengine and the driving wheels. And a scheme was proposed to improve it。At last, simulations tests and real vehicle tests were proceeded to verify the drivingcondition identification based multi modes energy control strategy of serial-parallel HEV.更多还原