Keivan Kalantari
Mohammad Ghafuri

Implementation of offline calibration methods which leads to the reduction of expensive test bench measurements gets more serious due to introduction of more complex engine controls. Design of experiments (DoE) is a convenient solution that is typically applied for the steady state desktop calibration. Although such methods are advantageous but there are still some limitations which have to be eliminated. Determination the correlation between the engine characteristics and vehicle behavior – which can be tightly limited by legislations - is a problematic task. Introducing a method based on the simulation of the entire vehicle coupled with the modeling some ECU functions can lead to favorite results. One of the most interesting features is the possibility of calibration of engine maps, curves and constants directly on the vehicle in the desktop area in order to figure out the effect of modifications in various respects. Optimization may be in terms of reduction of emissions and fuel consumption or any other engine property which can be modeled.

Simulation, DoE, Modeling, Neural networks