Modelling of combustion, fuel consumption and emission formation for design, calibration and control of diesel engines

The current PhD thesis aims to describe the work during three year PhD career in a detailed and comprehensive way. As indicated by the title of thesis, the work has been concentrated on the engine modelling of combustion, fuel consumption and pollutant emission with different approaches, and the related application for the on-board engine control and offline engine calibration. The thesis has been organized as follows:  Chapter 1: a general introduction of the thesis content.  Chapter 2: the description of the engine setup, the experimental activity and the research background  Chapter 3: a low throughput predictive physical engine model has been presented. The model includes several sub models to simulate the heat release process, the in-cylinder pressure evolution, the in-cylinder temperature distribution and the engine-out level NOx emission.  Chapter 4: a fast running engine model based on the machine learning technique has been introduced. This model focuses on the prediction of combustion phasing and the engine torque. In particular, an innovative neural network training strategy on the basis of virtual engine has been discussed in detail.  Chapter 5: a semi-empirical model for the prediction of significant pressure metrics has been described. This model features simple mathematical expression and maintains physical consistency simultaneously.  Chapter 6: the physical engine model has been coupled with kinetic vehicle model and has been integrated in an offline optimization framework, so as to identify the optimal engine control strategy for the reduction of fuel consumption and pollutant emission. This methodology has the potential to realize the offline engine calibration in order to reduce the calibration efforts and test bench activity.  Chapter 7: the physical engine model has been used for the on board open loop control of combustion phasing. In parallel, an alternative closed loop control based on pressure measured has been evaluated. These two controllers have been tested in Model in the Loop, Hardware in the Loop and Rapid Prototyping. It should be noticed that although the current thesis has been divided into several chapters, the content of each chapter is not independent from each other. In fact, the topic of the thesis can be summarized as engine modelling and its application for engine control in brief, and different chapters describe various aspects of this topic.