An automotive engine model for air-fuel ratio control using cylinder pressure information

Abstract

Increasingly strict emission standards require very accurate and fast air-fuel ratio (AFR) control in combustion engines. This thesis addresses the design methodology currently used for synthesizing a control system for an automotive internal combustion (IC) engine and proposes an engine model which can be used to design a "better" AFR control system. First, an averaging engine model was derived in the time-domain, using continuous differential equations. H2-optimal control theories for robust design were applied to the linearized engine model and a classical feedback-feedforward controller was designed, which strives to achieve robust non-overshooting response. This design was based on the current state-of-the-art exhaust gas oxygen (EGO) sensor which is located in the exhaust manifold. The analysis, results and shortcomings of this design methodology are presented. Second, a non-averaging, cyclic, crank angle domain model was derived. The model was linearized and discretized to an event-based model which requires only one value of air mass and fuel mass input into the combustion cylinder each cycle and predicts only one value of AFR in the cylinder for each cycle. From a controls viewpoint, this is more useful. The model was analyzed and proven to have enough information for a feedback controller to be designed to close the loop so that the response of the system meets the specified objectives. A proposed control scheme based on the in-cylinder pressure information is presented. Analysis, simulations and results are presented. The achievable performance with the pressure-based controller is believed to be superior to that with the H2optimal controller. Therefore, with the pressure information, design for a "better" non-overshooting response while at the same time achieving other feedback objectives is feasible.