Real-time predictive control for SI engines using linear parameter-varying models

Majecki, Pawel and Molen, Gerrit M Van Der and Grimble, Michael J. and Haskara, Ibrahim and Hu, Yiran and Chang, Chen Fang (2015) Real-time predictive control for SI engines using linear parameter-varying models. IFAC-PapersOnLine, 48 (23). pp. 94-101. ISSN 1474-6670 (https://doi.org/10.1016/j.ifacol.2015.11.267)

[thumbnail of Majecki-etal-IFAC2015-real-time-predictive-control-for-si-engines-using-linear-parameter]
Preview
 Text. Filename: Majecki_etal_IFAC2015_real_time_predictive_control_for_si_engines_using_linear_parameter.pdf
Accepted Author Manuscript
Download (277kB)| Preview

Abstract

As a response to the ever more stringent emission standards, automotive engines have become more complex with more actuators. The traditional approach of using many single-input single output controllers has become more difficult to design, due to complex system interactions and constraints. Model predictive control offers an attractive solution to this problem because of its ability to handle multi-input multi-output systems with constraints on inputs and outputs. The application of model based predictive control to automotive engines is explored below and a multivariable engine torque and air-fuel ratio controller is described using a quasi-LPV model predictive control methodology. Compared with the traditional approach of using SISO controllers to control air fuel ratio and torque separately, an advantage is that the interactions between the air and fuel paths are handled explicitly. Furthermore, the quasi-LPV model-based approach is capable of capturing the model nonlinearities within a tractable linear structure, and it has the potential of handling hard actuator constraints. The control design approach was applied to a 2010 Chevy Equinox with a 2.4L gasoline engine and simulation results are presented. Since computational complexity has been the main limiting factor for fast real time applications of MPC, we present various simplifications to reduce computational requirements. A benchmark comparison of estimated computational speed is included.

CORE (COnnecting REpositories)