On-Engine Validation of Mean Value Models for IC Engine Air-Path Control and Evaluation

Abstract

Mean value engine models (MVEM) are well established for the study of IC engine air-path performance. Closed-loop control of the air-path is used to ensure the tracking of reference signals subject to constraints. To meet these requirements model predictive control (MPC) has been applied successfully to several air-path applications. This paper reports on the development of an MVEM and suitable MPC controller for transient performance evaluation which is then applied to the engine. A linearised prediction model is evaluated at each control time step, based on the current operating conditions, which ensures the model remains valid across a wide range of speeds and loads. This linearisation of the nonlinear model allows for the formulation of quadratic programming (QP) problems, which are efficiently solved as part of the proposed MPC algorithm. Special treatment in the measurement and calculation of the states and their derivatives is demonstrated to provide excellent tracking performance and a good match between the model and real-time experimental results. The systematic process of modelling coupled with MPC is shown to closely match test-bed performance, which verifies the methodology for studying air-path hardware alternatives.