In this article, model-based calibration of a spark-ignition (SI) engine for reducing fuel consumption and pollutant emissions is performed with a new perspective. Instead of conventional local modeling, a global model with high accuracy and fast response is employed for optimization. As will be explained, this relieves the need for harmful harsh smoothening required for local models. A driving cycle-oriented calibration is performed with after-catalyst emissions and knock limits taken into consideration, which is a notable achievement in SI engine calibration. Both part-load and full-load modes of an engine in the widest ranges of a fully warmed-up operation will be covered. To eliminate variance errors resulted from meta-heuristic optimization, committee machines will be used, which, unlike other approaches, does not distort the original results. In the end, mid-level controller calibration maps for spark advance, throttle angle, air-fuel ratio (for full-load mode), and Variable Valve Timing (VVT) are presented in the widest ranges of engine operation.
Engine Model-Based Pre-calibration and Optimization for Mid-level Hierarchical Control Design
Sae Int. J. Engines
Sae International Journal of Engines ; 14 , 5 ; 651-669
2021-04-13
19 pages
Conference paper
English
Advanced Math-Based Engine Calibration and Control System Design
British Library Conference Proceedings | 2006
|Learning Based Approaches to Engine Mapping and Calibration Optimization
Tema Archive | 2014
|