【作者】 郭春梅；

【导师】 刘鲁源；

【作者基本信息】 天津大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 快速发展的电子技术和计算机技术越来越多的应用到车用发动机控制系统，使得发动机朝着提高动力性和节能的目标发展。日益严格的排放法规对发动机控制系统提出了更高的要求，而提高空燃比的控制精度是改善发动机性能的关键。本文分析了点燃式发动机的物理特性，给出了发动机平均值模型，该模型主要包括气路子模型、油路子模型和动力输出子模型。在Matlab/Simulink的环境中,建立了基于平均值模型的点燃式发动机的仿真模型。该模型的建立节约了研究成本，减少了开发周期，能对发动机的相关内特性进行分析及控制策略的研究与检验。发动机进气通路的非线性环节主要包括节气门有效面积和容积效率两部分，本文分别采用了改进的BP网络和RBF网络对二者进行辨识。采用这种基于神经网络的模型对进气空气流进行估计，仿真结果表明该方法提高了相应环节的辨识精度。 发动机是一个高度非线性和多扰动性的系统，空燃比对发动机动力性、经济性和排放产生重大的影响。本文综合分析了车用汽油机的空燃比控制策略。分别介绍了基于经典理论的PID控制及其与非线性控制相结合的方法、现代控制技术和智能控制理论在发动机精确空燃比控制中的应用，研究了它们的优点与不足，并分析了其发展趋势与应用前景。 考虑到被控对象本身的非线性和时间延迟，针对发动机瞬态工况特性，本文提出了一种基于神经网络的空燃比预测控制方法，给出了离散时间非线性发动机模型，用三层前馈神经网络对空燃比进行了预测和控制，仿真结果表明该控制方法提高了瞬态工况的空燃比控制精度，其控制效果优于车用ECU。更多还原

【Abstract】 The fast developing electronic techniques and computer techniques are increasingly applied to the control system of automobile engine. These techniques lead the engine along the direction of increasing driveability and reducing energy consumption. With the daily tightening standard of exhaust emission, higher requirements are raised for the engine control system, while increasing the control accuracy of air-fuel ratio is crucial to enhancing the engine performance.In view of the physical characters of spark-ignition engine, this thesis presents a mean value engine model(MVEM), which mainly consists of three subsystems - the air-flow dynamics, the fuel-flow dynamics and the crankshaft dynamics. Under the environment of Matlab/Simulink, a simulation model of MVEM is built up, which has reduced the research cost and saved the developing time, with which some inside characters and control strategies can be analyzed and testified.As for the nonlinear part of intake manifold subsystem, such as effective throttle area and volume efficiency, modified BP neural network and RBF neural network are used for identification respectively. Estimation of intake air mass flow is made by use of the model based on neural network, and simulation results indicate that the method proposed in this thesis has increased the corresponding identification precision.Automotive engine is a highly nonlinear and multi-disturb system. Air-fuel ratio has a great influence on the vehicle driveability, fuel economy and emission levels. The comprehensive analysis of air-fuel ratio control strategy is given in the paper. Several methods applied to the air-fuel ratio control are presented respectively, such as the PID control based on classical theory and its combination with nonlinear control methods, modern control technology, intelligent control theory and so on, with which the application and developing trend of air-fuel ratio is also analyzed.Considering the nonlinear characters and time delay of the plant, as far as the transient dynamics of engine is concerned, in this thesis a predictive air-fuel ratio control method is given based on neural network. In this paper a discrete nonlinear engine model is also presented, using forward neural network of three layers to <WP=5>predict and control air-fuel ratio. Simulation results illustrate that the control method proposed in the paper has increased the air-fuel ratio control accuracy under the transient conditions, which is better than that of automotive ECU.更多还原