【作者】 张亮；

【导师】 陈虹；

【作者基本信息】 吉林大学 ， 控制理论与控制工程， 2019， 硕士

【摘要】 为了顺应世界范围内日益严格的汽车排放和燃油经济性要求,进一步实现柴油机低污染、低油耗的目标,一系列柴油机新技术应运而生。其中高压燃油共轨电喷技术,以其较高的喷射压力和燃油的多次喷射,保证了燃烧过程中油和气的良好混合,在改善发动机性能方面发挥巨大的优势。但是由于高压泵和喷油器的不断开启关闭,导致喷射压力存在较大波动问题,直接影响了燃油的精确喷射。且燃油的多次喷射过程中,前一次喷射针阀关闭时的水击效应,必然会给下次喷射的油量计算带来偏差,从而直接造成混合气配比改变,燃烧不充分,发动机性能下降的缺点。针对上述问题,完善有效的控制策略是实现多次喷射中精确喷油的重点和核心。为此,本课题以高压燃油共轨系统为被控对象,设计相应的喷油量控制策略。整体控制框架包括两部分,喷射压力控制和喷油量补偿算法设计。首先对系统的组成和动力学进行分析,在仿真平台上搭建系统模型,并进行参数匹配和模型合理性验证。选取滑模控制理论设计具有Lyapunov稳定性的喷射压力控制器,保证了较高且稳定的喷射压力,为后续多次喷射的实现和喷油量的补偿奠定基础。然后分析多次喷射喷油量波动的影响因素,通过一定的激励数据样本构建喷油脉宽修正值的自学习系统,从而对多次喷射喷油量偏差进行补偿。其中自学习系统采用BP神经网络实现。但是传统BP神经网络设计的喷油量补偿算法,以梯度下降法进行偏差的反向学习传播,有收敛速度慢、多参数输入时易陷入局部最优解的缺点。为了保证输入参数增多时喷油量波动有较好的补偿效果,针对喷油量补偿算法进行一系列改进。首先采用有二次收敛效果的LM算法对反向传播算法进行设计,加快算法的收敛速度。为了提升网络的拟合和预测精度,采用遗传算法对LM-BP神经网络全局优化。最后将改进后的喷油量波动补偿算法放入整体控制框架中,验证了所设计的喷油量控制策略对喷油量波动有较好的抑制效果。本文从控制多次喷射过程喷油量精准的目的出发,为高压燃油共轨系统喷油量控制策略的设计提供一种设计思路。虽然在搭建的仿真平台中,验证了整体喷油量控制效果,但仍有工作需要进一步完成。如控制器的实物台架实现,围绕其工程实现问题做更深入的研究。更多还原

【Abstract】 In order to comply with the increasingly stringent worldwide requirements of automobile emissions and fuel economy,and further achieve the goal of low pollution and fuel consumption of diesel engines,a series of new diesel engine technologies emerged at the historic moment.High pressure common rail system,with its high injection pressure and multiple injections technology,ensures the better mixing of oil and gas in the combustion process.It helps improve the engine performance.However,the unstable injection pressure caused by the constant opening and closing of high-pressure pumps and injectors,which directly affects the accurate quantity of fuel injection.In the process of multiple injections,the water hammer effect during the closing of the previous injection needle valve will inevitably lead to deviation in the calculation of fuel quantity for the next injection,thus directly causing the change of mixture ratio,insufficient combustion and decline diesel engine performance.In view of the above problems,it is the key and core to achieve accurate fuel injection quantity in multiple injections with perfect and effective control strategy.For this reason,this subject takes the high pressure common rail system as the controlled object,and designs reasonable and effective control strategy to precisely control the fuel quantity in multiple injections.The composition and dynamics of the system are analyzed,and the corresponding system model is built on the simulation platform.Sliding mode control theory is selected to design the injection pressure controller with Lyapunov stability,which ensures a high and stable injection pressure and lays a foundation for the realization of subsequent multiple injections and the compensation of injection volume.Then,the influence factors of multiple injections volume fluctuation are analyzed,and a self-learning system of injection pulse width correction value is constructed through a certain sample of stimulation data,so as to compensate the deviation of multiple injection volume.The selflearning system is implemented by gradient descent method BP neural network method.However,the traditional BP neural network design of fuel injection compensation algorithm has the shortcomings of slow convergence speed and easy to fall into the local optimal solution.In order to ensure a better compensation effect for fuel injection fluctuation when input parameters increase,a series of improvements are made to the fuel injection compensation algorithm.Firstly,the back propagation algorithm is designed by using the LM algorithm with quadratic convergence effect to accelerate the convergence speed of the algorithm.In order to improve the fitting and prediction accuracy of the network,the genetic algorithm is used to optimize the LM-BP neural network globally.Finally,the improved algorithm of fuel injection fluctuation compensation is put into the overall framework,which verifies that the designed injection quantity control strategy has better suppression effect on fuel injection fluctuation.In order to control the injection quantity accurately in multiple injections process,this paper provides a design idea for injection quantity control in the high pressure common rail system.Although the overall fuel injection control effect is verified in the simulation platform,there is still work to be done.For example,the realization of the physical platform of the controller will be further studied around its engineering implementation.更多还原