D-optimization based model calibration for air mass flow in engines

Abstract

The aim of this paper is to explore the model calibration method consisting of the effective design of experiment (DoE) and the model selection for DoE. D-optimality is exploited as the cost function, and the optimal experiment design for the model is calculated by Fedorov's exchange method. An actual implementation example for air-mass flow model of gasoline engines is carried out. The model is chosen in the fashion of minimum cAIC criteria with the preliminary experiment for the purpose of the application of DoE. The experiment result with a full-scale engine test bench is validated to confirm that the proposed method can improve the accuracy of the calibrated model. The comparison of the proposed method with the conventional mesh experiment design by the calibration result is shown.

Keywords

• automobile
• internal combustion engine
• system identification
• modeling
• optimal design
• d-optimal design of experiment