Automated Generation of Real Driving Emissions Compliant Drive Cycles Using Conditional Probability Modeling

Car manufacturers are required by the European Union to carry out real driving emissions tests as part of the type approval process. Comprehensive regulations define numerous requirements regarding, for example, velocity, acceleration and cycle dynamics in various environments for a real driving emission test to be valid. In order to reduce the need of extensive real world testing, drive cycles can be used during the development, calibration and testing phase of the engine or powertrain. This, however, requires the availability of a broad variety of drive cycles that not only meet the legislative specifications, but also represent naturalistic driving behavior. In this work, an entirely new method for constructing such drive cycles is presented. It is based on simulating conditional probabilities of acceleration based on logistic regression models that are estimated using real driving data. The presented approach produces diverse cycles, which meet all requirements while resembling naturalistic driving profiles.