【作者】 袁荣棣；

【导师】 黄席樾；

【作者基本信息】 重庆大学 ， 控制理论与控制工程， 2012， 博士

【摘要】 压缩天然气(Crushed Natural Gas，CNG)发动机因其低排放、低成本被广泛应用做代用燃料汽车动力总成。发动机电控技术通过精确控制进气流量、燃料喷射和点火，大幅提升发动机综合性能，是动力总成技术的核心和热点之一。目前国内燃气发动机电控核心技术缺失，主要产品及平台均被国外公司垄断，严重制约了自主品牌及核心产业的发展。本文针对CNG发动机电控中的建模、控制、排放优化及开发平台等关键技术进行了全面、深入的分析研究。根据发动机多输入多输出非线性混杂系统特征，建立了面向发动机控制、仿真、排放标定的高集成度CNG发动机平均值模型；基于模型和线性反馈技术，采用一种改进的非线性变换架构进行解耦；针对转速控制回路，提出基于非线性回退方法(Backstepping)的CNG发动机转速控制算法；基于离散滑模非线性观测器，提出一种离散变结构空燃比控制方法。建立了基于V模式和虚拟仪器技术的CNG发动机电控单元开发平台，基于平台完成建模仿真、代码生成、虚拟标定和排放优化，开发了控制器快速原型。本文主要工作如下：(1)针对离散变结构时滞系统，提出一种基于变指数趋近律调节因子的趋近律，使系统运动快速收敛且抖振幅度较小，为离散滑模应用于非线性发动机控制奠定良好基础。通过合理选择指数收敛项的变调节因子和平移项幅值关联因子，系统运动到达准滑动模态区前符合指数趋近律；进入滑动模态区后，抖振幅值遵循指数规律收敛。证明了采取双曲正切及反双曲正切函数作为趋近律调节因子的可行性，提出参数估值方法。将此趋近律应用于离散状态时滞系统，实验结果表明具有较好的收敛特性及低抖振频率。(2)针对目前CNG发动机实时模型价格昂贵且底层接口不开放，建立了高集成度、面向电控开发用CNG发动机平均值模型。模型集成了涡轮增压、电子节气门、进气歧管、动力输出及排放等子系统，能够描述EGR、中冷器、缸内工作过程、CNG供给、减压阀及传感器组等辅助环节。基于温差气体定容混合过程的CNG发动机进气歧管方程和燃料喷射延时计算方法建模，具有较高精度及实时性，大幅降低开发成本，接口开放，满足二次开发需要。(3)针对CNG发动机的转速与空燃比控制进行了研究，提出一种基于离散滑模进气流量观测器的空燃比学习控制算法。首先采用一种基于李导数的解耦方法对CNG发动机进行转速与空燃比解耦控制。对CNG发动机仿射非线性模型进行变换，满足相对阶数向量及输出方程李括号矩阵非奇异的条件，通过线性反馈实现解耦。其次，提出一种基于非线性模型和Backstepping方法的滑模变结构转速控制。选择转速误差与进气流量误差组成二维滑模面，利用虚拟控制变量和分层子系统，保证系统条件稳定性的前提下设计控制律。实验结果表明，相比自适应PID控制，具有较高度响应速度及鲁棒性。采用离散滑模进气流量观测器和进气流量延时补偿，设计CNG喷射控制律，实现空燃比瞬态过程进气流量的补偿与学习，与量产ECU广泛采用的PI控制相比，具有更高控制精度和鲁棒性。(4)针对目前排放优化与控制参数标定主要依靠台架实验的现状，提出一种基于混沌粒子群优化的CNG发动机控制参数虚拟标定和排放优化方法。通过混沌的随机不确定性和粒子群空间寻优快速收敛的特性，第一轮优化基于台架数据，校验排放模型。第二轮优化基于开发平台和排放模型进行虚拟标定，优化控制参数，对发动机工况点进行了控制参数优化。与传统的13点工况法相比，排放优化效率大幅提高，有效降低优化费用。(5)建立了一个基于虚拟仪器技术和V模式的CNG发动机电控ECU开发专用平台。平台支持Matlab/Simulink建模、硬件在环仿真、代码生成、标定和ECU功能测试功能，提供可定制化的界面及测试流程，实现了逻辑层与驱动层分离。利用平台完成了32位ECU原型的建模仿真、HIL测试、功能测试、虚拟标定和代码生成工作。建立了一套CNG发动机ECU控制软件的实时逻辑。利用Matlab/Simulink软件建立了进气，CNG喷射和点火系统控制逻辑，对控制算法进行实时性工程化处理。逻辑具备诊断、跛行回家等功能。通过自动代码生成工具，支持第三方开发的ECU原型，实现了逻辑层与任务调度层、驱动层及硬件分离的结构。平台大幅度提升了CNG发动机ECU的开发效率，降低了开发成本。论文的主要创新点：1、提出一种新的面向电控开发的CNG发动机平均值模型。2、提出一种基于CNG发动机进气流量离散滑模观测器和线性反馈的空燃比控制方法。将仿射非线性的精确线性化反馈解耦和基于模型的CNG发动机转速非线性算法应用于CNG发动机控制。3、建立了CNG发动机电控ECU开发平台。基于专用平台进行建模、仿真、虚拟标定及代码生成，开发了ECU快速原型。将一种改进的混沌粒子群算法应用于CNG发动机虚拟标定和排放优化。综上所述，论文对CNG发动机的建模、控制、优化和ECU开发平台理论及应用进行了深入研究，控制模型精度较高，控制算法鲁棒性及实时性优于现有产品，建模方法、控制算法及开发平台为CNG发动机ECU产业化奠定良好基础。针对开发过程中遇到的实际问题和不足，提出了未来改进的方向。更多还原

【Abstract】 CNG (Crushed Natural Gas) is an effective alternative fuel due to its highcombustion efficiency, low emissions and minimum reconstruction to traditional SparkIngnition engines. Combined with signal processing, modern control and optimizationtechnology, ECU (Electronic Control Unit) greatly promoted comphrehensive engineperformance. Precise fuel injection and spark ignition timing control is the milestone oflow emission and fuel consumption. Although it is well developed and applied to manyengines in Europe and American, it has more technology and industry potential in China.Domestic enterprises need to carry out research on the nature of engine control schemeand ECU development platform.This paper focuses on the aspects such as engine modeling, simulation, controlalgorithm, emissions optimization, platform building and applicaitions of CNG enginecontrol. CNG engine is analyzed as a continuous and discrete hybrid, nonlinear andtime-delay system. First, a MVEM (Mean Value Engine Models) of CNG engineorienting for control and diagnostics is built and verified. Secondly, a nonlineardecoupling algorithm is applied to facilitate fast, stable speed and A/F (air/fuel) ratiocontrol. Based on Backstepping method, a nonlinear engine speed control algorithm isproposed to acquire excellent disturbance rejection performance. An A/F Ratio controlscheme aiming for high control frequency bandwidth as well as accuracy is presented,in which a DVSC (Discrete Variable Structure Control) observer is built. Finally, anintegrated ECU development platform is introduced as a virtual instrument. A prototypeECU is developed to demonstrate the efficiency and effectiveness of the platform,which is deployed according to V mode flow: modeling, simulation, calibration andoptimization. The main innovative works are as follows:(1) A new reaching law of the DVSC control is proposed, in which alternativeregulate factor is adopted. Two variable regulate factors are added in exponentialconvergence part and correlation term of amplitude. Its trajectory is convergenced byexponential law before it reaches the quasi-sliding mode area. After it enters theboundary layer, the chattering amplitude is convergenced according to exponential law.It is proved that the hyperbolic tangent function is feasible to act as the regulate factor.Furthermore, a factor estimation method is presented. A simple performance analysis iscarried out to prove that it achieves good balance of convergence speed and chatteringamplitude. (2) A MVEM of CNG egine is constructed which integrates Electronic Throttle，Turbo, Exhaust Gas Recycler, Inter-cooler, emission subsyst. This scheme overcomesthe disadvantages of commercial CNG engines models, such as expensive costs，poorinterface and hardware support. A new manifold pressure model is presented that isbased on an adiabatic assumption rather than an isothermal one. Furthermore, a newtime delay estimation method is proposed to calculate the time-delay of fuel injectionand prevent sudden deteriorations of transient A/F ratio.(3) Sevral algorithms are prsented to promote the stability of CNG engine speedand A/F ratio control. A nonlinear transformation decouple scheme is propsed, which isbased on nonlinear mean value model linearization policy through Lie derivative. Acontrol law for CNG engine speed control is derived from nonlinear model. Enginespeed error and intake air mass flow error are selected as2-D sliding surface, a virtualcontrol variable subsystem is used to prove the system stability by the Backsteppingmethod. The experiment performance data shows that it achieves the goals of goodresponse time and robustness. A DSMC observer is adopted to estimate the intake portair mass flow, intake manifold pressure and mass folw through throttle. Working withfeedforward and feedback control channel, a self-study control scheme is verified byreal-time contol logics to prove its high bandwidth and control accuracy in A/F controlthan PI control, which is widely used in ECU product.(4) An emission optimization policy based on Chaos particle swarm method isproposed. Firstly, the parameters of emission MVEM are verified and refined toappropriate value. The second round optimization is carried out on the ECU developplatform along with Virtual Calibration. The effectiveness of the optimization policy isproved through a CNG engine emission optimization procedure, which saved muchmore time and reduced expenses than traditional bench policy.(5) A CNG engine ECU develop platform is constructed to provide developmentflexibility through virtual instrument technology. A prototype ECU is developed on theplatform according to V mode workflow: modeling, simulation, hardware-in-loop test,code generation and ECU function test. Basic control logic is built to verify the realtimeperformance of engine control algorithm, which is independent of task scheduling,driver layer and hardware layer. The top layer models cover regular work conditions aswell as fundmental fault tolerant functions. Based on Matlab/Simulink, automatic Ccode generation is supported; therefore it can be easily modified to evaluate the codereliatility, resource consumption, scheme effectiveness and model sharpness.The main creative points focus on belows:1、A new mean value models of CNG engine is presented for ECU development.2、An algorithm based on DSMC Observerand linear feedback for A/F Ratio control is proposed. A decoupling method based onLie derivative precise linearization and Backstepping methods is adopted for enginespeed control.3、A virtual instrument develop platform for CNG engine ECU isconstructed. It consists of modeling, simulation, virtual calibration and codeauto-generation subsystem. A swarm particles algorithm is proposed for CNG enginevirtual calibration and emission optimization.The performance of MVEM, algorithm and development platform is showed bythe data of simulation and experiment analysis. The realtime quality, stability,roboustness and low development expense of proposed scheme make it possible to bewidely used for ECU industrialization. In the end of this paper, future technologytrends of the CNG engine control are summarized in the fields of algorithms, platformperformance and practice requirements.更多还原