Mathematical modeling and embedded nonlinear control of automotive powertrains

Automotive powertrains are becoming increasingly sophisticated, due to consumer demand for better fuel economy and power response, as well as extremely tight government regulations for exhaust emissions. To satisfy these requirements simultaneously, additional actuators as well as improved sensing are incorporated in modern powertrains. Interacting subsystems more and more commonly integrated into modern vehicles such as traction control and dynamic stability control emphasize the need for more rigorous validation of the entire complex vehicle control system.;In this thesis, mathematical modeling of powertrains and powertrain control systems is considered, along with model-based development of control system algorithms and embedded control software.;First an overview of powertrain control technology and challenges is presented. Second considered is the control-oriented mathematical modeling of powertrains. Models of automotive powertrain subsystems are presented and discussed, with some emphasis on models relevant to the cold start emissions control problem. These models represent the system behavior averaged over the course of the engine cycle. The use of these models as the basis for the design of individual control algorithms is made concrete with the development of a nonlinear adaptive fueling control which accounts for plant time delays in the adaptation scheme.;Next discussed are models of engine cylinder processes which are used in estimation schemes with cylinder-based sensing. An estimator for air-fuel ratio using cylinder pressure measurements is designed, and its performance evaluated experimentally. Challenges related to estimation during cold start are discussed.;Lastly, the modeling of powertrain controllers is considered. In the development of control systems software, it is desirable to create specifications for desired software behavior in the early stages of the process, and to assure that these behaviors are carried through to the finished product, embedded application software. The MoBIES program aimed toward facilitating the use of models at all stages of the software development process. A baseline embedded software development process utilizing commercial tools in conjunction with software models is presented. The state of development of tools for model-based design and analysis of embedded software is discussed. The thesis concludes with an automotive industry perspective regarding the advancement of model-based processes.