Enabling a Privacy-Preserving Synthesis of Representative Driving Cycles from Fleet Data using Data Aggregation

Driving cycles are of fundamental relevance in the design of vehicle components, in the optimization of control strategies for different drivetrain topologies and the identification of vehicle properties. Ideally, a high quantity of real fleet driving data including varying operation conditions is used to generate representative driving cycles that are the basis for further investigations. Traditionally, a specific testing fleet is employed to gather the driving data. Nevertheless, driving data can nowadays also be gathered from regular production cars, as they are already equipped with the required sensors. This approach would be more real-driving representative and cost efficient, but on the other side imposes new challenges. In particular, gathered driving data has to be handled efficiently and the privacy of individuals must be guaranteed. In this work, an approach to synthesize representative driving cycles using data aggregation is presented. It is shown that the approach is efficient and generates driving cycles with excellent quality when compared to classical approaches, thus acting as an enabler for privacy-preserving techniques.