Efficient real driving emissions calibration of automotive powertrains under operating uncertainties

• Published in: Engineering optimization 2023-01, Vol.55 (1), p.140-157
• Main Authors: Wasserburger, A., Didcock, N., Hametner, C.
• Format: Article
• Language: English
• Subjects: Calibration; Controllability; distribution estimation; engine calibration; Optimization; Powertrain; Real driving emissions; Statistical analysis; stochastic optimization; Uncertainty
• ISSN: 0305-215X; 1029-0273
• DOI: 10.1080/0305215X.2021.1989589

The steady-state calibration of automotive powertrains is typically based on the assumption of one specific drive cycle and perfectly controllable operating conditions. During real operation, however, these assumptions are violated, which implies that the calibration might in fact not be optimal. Therefore, in order to achieve reliable performance in a real-world setting, these uncertainties have to have been considered already during the calibration process. In this article, a stochastic optimization approach that takes the mentioned operating uncertainties into account by including probability distributions of the disturbances, is suggested. Furthermore, an approximation is derived of the distribution of the optimization performance criterion that greatly reduces the computational load during optimization compared with Monte Carlo sampling. Simulation results show that the proposed probabilistic approach leads to lower expected values of emissions and consumption when compared with deterministic optimization approaches ignoring the stochastic influences.