0
TECHNICAL BRIEFS

Exact and Linearized Neural Predictive Control: A Turbocharged SI Engine Example

[+] Author and Article Information
G. Colin

Laboratoire de Mécanique et d’Energétique (LME, EA 1206,  Orleans University), 8, rue Léonard de Vinci 45072 Orléans Cedex 2 Franceguillaume.colin@univ-orleans.fr

Y. Chamaillard, A. Charlet

Laboratoire de Mécanique et d’Energétique (LME, EA 1206,  Orleans University), 8, rue Léonard de Vinci 45072 Orléans Cedex 2 France

G. Bloch

Centre en Recherche en Automatique de Nancy (CRAN, UMR 7039,  Nancy University, CNRS) CRAN-ESSTIN rue Jean Lamour 54519 Vandoeuvre les Nancy cedex France

J. Dyn. Sys., Meas., Control 129(4), 527-533 (Feb 12, 2007) (7 pages) doi:10.1115/1.2745881 History: Received January 10, 2005; Revised February 12, 2007

This paper describes a real-time control method for non-linear systems based on model predictive control. The model used for the prediction is a neural network because of its ability to represent non-linear systems, its ability to be differentiated, and its simplicity of use. The feasibility and the performance of the method, based on on-line linearization, are demonstrated on a turbocharged spark-ignited engine application, where the simulation models used are very accurate and complex. The results, first in simulation and then on a test bench, show the implementation of the proposed control scheme in real time.

FIGURES IN THIS ARTICLE
<>
Copyright © 2007 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 1

Air Intake of a turbocharged SI engine

Grahic Jump Location
Figure 2

Static nonlinearities of the turbocharger: supercharging pressure Psup (Pa) versus in-cylinder air mass (mg) and wastegate closing WG (%) at a fixed engine speed Ne

Grahic Jump Location
Figure 3

Sum of squared error (SSE, top), final prediction error criterion (FPE, bottom) versus number of hidden nodes of the neural model. The selected number corresponds to the minimal value of FPE.

Grahic Jump Location
Figure 4

Example of actuator opening in the test data. Wastegate closing (top in%) and throttle opening (bottom in%) versus time (s)

Grahic Jump Location
Figure 5

Example of neural model output for the test data. Supercharging pressure Psup and neural model output for the test data (top in Pa) and estimation error (bottom in Pa) versus time (s).

Grahic Jump Location
Figure 6

Autocorrelation function of the prediction error versus lag

Grahic Jump Location
Figure 7

Comparison of supercharging pressure (bar) control, for different prediction horizon, simulation results: Set point (blue solid line), supercharging pressure with Np=1 (red dotted line), Np=2 (dark green dash-dotted line), Np=3 (green solid line), Np=10 (purple dotted line) versus time (s)

Grahic Jump Location
Figure 8

Comparisons among constrained exact neural predictive control (CENPC), constrained linearized neural predictive control (CLNPC), and saturated linearized neural predictive control (SLNPC), simulation results: Set point (dark blue dash-dotted line in Pa), Supercharging pressure with CENPC (green solid line in Pa), CLNPC (red solid line in Pa), and SLNPC (blue dashed line in Pa) versus time (s).

Grahic Jump Location
Figure 9

Pressure results for the saturated neural predictive control applied to the turbocharged SI engine, simulation results: manifold pressure set point (blue in bar), supercharging pressure (green in bar), and manifold pressure (red in bar) versus time (s)

Grahic Jump Location
Figure 10

Actuator opening for the saturated neural predictive control applied to the turbocharged SI engine, simulation results: Wastegate closing (top in%) and throttle opening (bottom in%) versus time (s)

Grahic Jump Location
Figure 11

Indicated torque (Nm) versus time (s) for the saturated neural predictive control applied to the turbocharged SI engine, simulation results

Grahic Jump Location
Figure 12

Pressure results for the saturated neural predictive control applied to the turbocharged SI engine, test bench results: Manifold pressure set point (dash-dotted line in Pa), supercharging pressure with SLNPC (green solid line in Pa), supercharging pressure with PID (black dashed line in Pa), and manifold pressure (red dotted line in Pa) versus time (s).

Grahic Jump Location
Figure 13

Actuator opening for the saturated neural predictive control applied to the turbocharged SI engine, test bench results: Wastegate closing (top in%), throttle opening measurement (bottom, red, in%), and set point (bottom, blue, in%).

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In