【作者】 田硕；

【导师】 欧阳明高；

【作者基本信息】 清华大学 ， 动力工程及工程热物理， 2008， 博士

【摘要】 集成启动发电机(ISG)型式的并联混合动力系统是汽车动力系统发展方向之一。目前大部分对ISG混合动力技术的研究结合汽油机进行,并主要围绕降低油耗开展。但对于比汽油机热效率高的柴油机,降低其排放,特别是降低其瞬态过程排放在日益严格的排放法规下具有更重要的意义。本文以降低瞬态过程排放为目标,对柴油机ISG混合动力系统的优化控制问题展开研究。所研究的动力系统由增压(无EGR)共轨柴油机、ISG电机和超级电容器组成。以降低瞬态过程排放为目标的优化控制算法是本文的核心研究内容,柴油机和电机的转矩协调控制是降低排放的基础。本文首先建立了面向实时控制的增压柴油机的瞬态排放模型,采用了多层前馈神经网络对排放模型的非线性静态特征进行描述,利用稳态试验数据对模型进行训练,采用线性自回归(ARX)模型对稳态模型进行动态修正,对瞬态过程烟度及NOx排放的模型估计准确度进行了试验验证。而后基于平均值模型结构建立了柴油机转矩估计模型,对估计结果的准确度进行了试验验证。基于非线性自回归(NARX)动态神经网络模型结构建立了总体转矩需求计算模型。基于上述模型,完整的提出了一套ISG混合动力系统转矩协调控制算法,并考虑了电机输出转矩动态约束和转矩输出延时问题。控制算法的有效性得到了试验验证。对于瞬态过程的多状态动态全局优化问题,首先提出了一种基于移动局部域原理的求解方法,进行了优化算法仿真分析。而后将算法进行简化,设计了自适应SOC维持能量管理算法,在单片机上实现了实时控制。发动机台架试验结果表明,本优化控制算法在保持发动机总体转矩输出特性不变和SOC自适应维持的前提下,成功地降低了瞬态排放,抑制了碳烟排放的瞬态恶化。在ETC瞬态循环中,排气烟度峰值下降约77%,微粒排放降低约15%,其中碳烟降低约20%。油耗和NOx等排放基本保持不变。针对等效中型客车驾驶循环,通过制动能量回收等功能可降低油耗约9%。更多还原

【Abstract】 Integrated Starter Generator (ISG) parallel hybrid system set a new trend of vehicle powertrain systems. Most of the studies of ISG hybrid technology are conducted on gasoline engines and the focus is put on reducing fuel consumption. Diesel engines have the advantage of fuel efficiency over gasoline engines, but reduce emission especially in transient state has become more significant with the more stringent regulation on exhaust emission. For the purposes of reducing emission in transient state, optimal control of ISG hybrid system with diesel engines has been studied in this dissertation. The powertrain system being studied is composed of a turbocharged (non-EGR) common rail diesel engine, an ISG motor and ultracapacitors. The optimal control algorithm targeting at reducing emission in transient state is the core of the study, and the torque coordination control between the engine and the motor serves as a basis.First, a control oriented emission model for turbocharged diesel engines is built. The nonlinear static behavior of emission is described by a multi-layer neural network, which is trained by steady state experimental data. A linear auto-regression with exogenous input (ARX) model is implemented for dynamical correction of the steady model. The accuracy of the estimated smoke and NOx value by the model is validated by experiments.Then a mean value model based torque estimation model of the diesel engine is built. The accuracy of the estimation is examined by validation experiments. The total torque requirement is calculated by a nonlinear auto-regression with exogenous input (NARX) dynamic neural network model. Based on the above models, a complete torque coordination algorithm for ISG hybrid system is proposed, in which the dynamic limit on motor torque output and the delay in torque output is considered. The effectivness of the algorithm is validated by experiments. For the dynamic global optimization problem in transient state with multi-states, a receding local horizon solving algorithm is proposed. Firstly simulation analyse of the optimal control algorithm is carried out on computer. Then the algorithm is simplified, and an adaptive SOC sustaining energy management algorithm is designed. The algorithm is implemented on a microcontroller in real-time. The test results on engine dynamometer show that, the total output torque characteristic is kept unchanged and the adaptive sustaining of SOC is ensured. The deterioration of soot emission in transient state is successfully suppressed. In ETC test cycle, the peak smoke is reduced by about 77%. The total PM emission is reduced by about 15%, in which the total soot emission is reduced by about 20%. The fuel consumption, NOx and other emission compositions are kept unchanged. For the test cycle simulating the driving cycle of a middle-sized passenger car, with the help of regenerative braking, the fuel consumption can be reduced by about 9%.更多还原