Abstract: Aiming at the problem that there is a large gap between the actual fuel consumption of vehicle optimized and calibrated according to the European Driving Cycle Standard (NEDC) and the regulatory certification results, and to provide a theoretical reference for the study of vehicle fuel consumption and pollutant emission control, the construction method of typical driving cycle of passenger car was researched in this paper. First, according to the driving data of passenger cars in Fuzhou City, the reduced-dimension driving characteristics are obtained by cluster analysis. As the BIRCH clustering algorithm is sensitive to the insertion order of the data set, based on K-means clustering algorithm optimization BIRCH clustering algorithm (KM-BIRCH algorithm)is proposed; Then, Markov chain principle were applied to construct the typical driving cycle; Finally, a comparative analysis was made between the constructed typical driving cycle and the existing driving conditions. The results show that the average deviation of the overall characteristics of the constructed typical driving conditions and the actual operating conditions sample database is only 3.21%, which meets the development accuracy requirements with a deviation of less than 5%, which verifies the accuracy and effectiveness of the typical driving cycle.