Engine transient calibration for real driving conditions: a holistic statistical approach

摘要

Real Driving Emissions (RDE) regulation has dramatically expanded the operating domain under which the engine performance needs to be optimised: higher dynamics, increased speed and load coverage and higher diversity of atmospheric conditions. To meet this engine calibration challenge, a number of technical solutions have been proposed. On the engine control side, physically motivated transient corrections have been introduced, adding new degrees of freedom. On the calibration methods side, it is proposed to use Markov chain models for driving patterns to obtain an estimation of the emissions statistical distribution under real driving conditions for validation purposes. A new calibration method is presented, uniting both aspects: a simulation environment able to predict the effect of transient corrections on emissions and an algorithmic optimisation method considering the population of driving patterns, in order to reach goals and constraints with respect to statistical measures of emissions distributions. A first analysis of the interactions between emissions distributions and the calibration is made. The calibration obtained using the proposed method is compared to results obtained with a conventional drive-cycle approach on worst case profiles. This approach opens the way to more efficient calibration processes and algorithms which are better able to optimise average performance and also improve robustness under real driving conditions.