A Study of Personalized Shift Strategy based on Styled Driver Modeling

Abstract

Shift schedule is an important parameter for automatic transmissionvehicle calibrations, which directly affects the vehicle's power andfuel economy. This paper tries to establish a system to optimize shiftperformance based on the styled driver model. First researchers proposedto imitate the real-world drivers' characters based on a locallydesigned neural network, which divides the driving style into threecatagories, aggressive, moderate and mild. Second, the personalizeddriver models were tested for simulation under the standard speedprofile, e.g. FTP-75. Third, the times series of shift and fuelconsumption were analyzed under different driving styles. Finally, theshift map for different driving styles are corrected and the testresults demonstrate the effectiveness of this method. The correctedshift schedule can effectively reduce fuel consumption and improveshifting comfort, which is useful for Eco-driving.