New Control Methodology for the Next Generation of Engine Management Systems

尽管电动汽车正在开发中，并且有可能为短距离轻型运输提供替代方案，但内燃机将在未来几十年内继续成为车辆中的主要动力单元。与燃烧和后处理系统的广泛研究相比，发动机系统控制优化方面的工作几乎没有完成，这为提高燃油经济性和降低排放留下了相当大的空间。当前的发动机校准过程依赖于导出静态表格关系和每个校准的发动机操作点之间的对应值，利用闭环反馈控制来相应地调整用于真实的发动机操作中的空燃比控制的设置，以便满足性能目标和排放法规。然而，由于基于发动机台架的校准的持续时间和成本的限制，这种广泛采用的方法在实现车辆的最佳燃料经济性方面不是有效的。环境条件变化、闭环控制响应所需的时间、逐循环变化以及气缸间变化使得当前的发动机控制不可能处理发动机功能的优化。未来发动机的发展趋势是朝向机载智能控制和校准，并在文献中报道了一些基于模型的控制系统的发展的研究活动。然而，逐个循环和逐个气缸控制发动机运行的可行策略尚未可用。该项目的总体目标是利用三个研究小组的共同努力，利用发动机技术，控制技术和计算算法的专业知识，开发和测试一种新的发动机控制和校准方法，并内置在线智能。这一总体目标将通过实现以下目标来实现：（1）开发用于气缸解析控制目的的全实时多缸发动机模型（2）开发一种新型发动机控制策略，包括控制点和控制点位置的优化以及测试循环性能的多目标评估（3）开发用于在线发动机控制优化的动态多目标进化算法（4）首先在硬件在环（HIL）dSPACE系统上演示发动机控制模型的实施，然后在测试发动机上进一步快速原型化。(5)比较使用新技术与传统标定和控制方法的发动机性能，并证明在发动机输出、油耗和排放方面的改进。新的发动机控制方法将在新的捷豹汽油直喷（GDI）发动机模型上进行评估。

Despite of the fact that electrical cars are under development and have the potential to provide alternatives for short distance light duty transport, the internal combustion engine will continue to be the main power unit in vehicles for several decades to come. Compared with extensive research on combustion and after-treatment systems, little work has been completed with respect to engine system control optimisation, leaving considerable room to improve fuel economy and lower emissions. Current engine calibration process relies on deriving static tabular relationships and the corresponding values between each calibrated engine operating point, with closed-loop feedback control to adjust the settings accordingly for air-fuel ratio control in real engine operation so as to meet the performance targets and emissions legislation. Such a widely adopted method, however, is not efficient in achieving the best fuel economy of the vehicle due to the constraints in the time duration and cost of engine-bed based calibration. Environmental conditions changes, the time required for the closed-loop control to respond, cycle-by-cycle variations, and cylinder-to-cylinder variations make the current engine control impossible to handle the the optimisation of the engine functionalities. The development trend for future engines is towards an on-board intelligence for control and calibration and some research activities for the development of model based control systems are reported in literature. However, feasible strategies to control the engine operation cycle-by-cycle and cylinder-by-cylinder are not yet available. Expanding the work of the applicants in the related areas for many years, the overall Goal of this project is to use a combination of joint efforts from 3 research groups with expertise of engine technology, control technology and computing algorithm in order to develop and test a new engine control and calibration methodology with on-line intelligence built in. This overall goal will be achieved through realising the following objectives:(1) To develop a full real-time multi-cylinder engine model for cylinder-resolved-control purpose(2) To develop a novel engine control strategy involving optimization of control points and control point locations, and multi-objective evaluation of test cycle performance(3) To develop dynamic multi-objective evolutionary algorithms for online engine control optimization(4) To demonstrate the implementation of the engine control models initially on Hardware-in-the-Loop (HIL) dSPACE system and then further rapid prototyping on a test engine.(5) To compare the engine performance using the new techniques with traditional calibration and control approaches, and demonstrate improvements in terms of engine output, fuel consumption, and emissions. The new engine control methodology will be evaluated on a new Jaguar gasoline direct injection (GDI) engine model.