Transient engine models for optimization and implementation of model predictive control

• Autorzy: Šika Z. , Valášek M. , Florián M. , Macek J. , Polášek M. , Dolecek V. , Bozek J.
• Konferencja: Kongres Silników Spalinowych (20-23.05.2007; Kraków, Poland)
• Języki publikacji: EN

Abstrakty

EN

The paper deals with the investigation of model based predictive control of combustion engines. Three levels of dynamic models have been used. Firstly the reference1-D model with 3-D extensions calibrated by the experimental data to interpolate/extrapolate the data for identification of simplified fast models. Secondly the nearly real-time physical simulation model which allows for very fast simulation of engine transients and control algorithm design. Finaly the neuro-fuzzy predictive models implemented directly in controller. The predictive models are based on LOLIMOT approach. It is a description of generally nonlinear dynamic system by a sequence of linear dynamic systems valid in particular in one subregion of the whole state space. The decomposition of the state space into subregions is provided using the validity of linear dynamic models. The predictive models are used for the prediction of future engine states in dependence on actual measured states and possible control inputs. The future control inputs are optimized on-line based on locally linearized models. The problem of attainable prediction horizons in context of different speed of engine response to different control inputs is discussed.

Słowa kluczowe

EN

combustion engine; transient engine model; model identification; neuro-fuzzy predictive model; model predictive contro

PL

silnik spalinowy; identyfikacja modelu

• Wydawca: Polskie Towarzystwo Naukowe Silników Spalinowych
• Czasopismo: Silniki Spalinowe
• Rocznik: 2007
• Tom: R. 46, nr SC1
• Strony: 234--241
• Opis fizyczny: Bibliogr. 9 poz.

Twórcy

autor

Šika Z.

autor

Valášek M.

autor

Florián M.

autor

Macek J.

autor

Polášek M.

autor

Dolecek V.

autor

Bozek J.

• Czech Technical University in Prague Faculty of Mechanical Engineering

Bibliografia

• [1] Pischinger, S. et al.: Investigation of Predictive Models for Application in Engine Cold-Start Behaviour, SAE Paper 2004 01-0994, 2004.
• [2] Winsel, T. et al.: Hil-Calibration of SI Engine Cold Start and Warm-Up Using Neural Real-Time Model, SAE Paper 2004-01-1362, 2004.
• [3] Morel, T., et al.: Manual for GT Power version 6.1. Gamma Technologies, 2004.
• [4] Macek, J., et al.: Transient Engine Model as a Tool for Predictive Control, SAE Paper 2006-01-0659, 2006.
• [5] Nelles, O.: Nonlinear system identification with local linear fuzzy-neuro models. Automatisierungs technik. Shaker Verlag, Aachen 1999.
• [6] Štefan, M.: CTU-LOLIMOT, Research Report U2052-02-45, FME CTU in Prague, Prague 2002.
• [7] Camacho, E.F., Bordons, C.: Model Predictive Control. Springer Verlag, Berlin, 1999.
• [8] Valášek, M., et al.: Model Based Predictive Control of Combustion Engine with Constraints, Review of Automotive Engineering of Japan SAE 2005, vol. 26, no. 3, pp. 349-356.
• [9] Štefan, M., Šika, Z., Valášek, M.: MIMO Neuro-fuzzy Model for Dynamic System Prediction, In: Proc. of Interaction and Feedbacks 2005, IT AS CR, Praha 2005, pp. 135-142.