Research on the Construction Method of Passenger Car Driving Conditions Driven by Big Data
YANG Yang
WEI Qian
YU Qian
Abstract：To accurately characterize urban passenger vehicle driving patterns and address limitations in traditional driving scenario studies—such as limited sample sizes,poor clustering stability,and the omission of low-probability events during scenario synthesis—a big data-driven method for constructing driving scenarios is proposed.Using one year of OBD data from 200 passenger vehicles in Xi'an,a sample repository was built through data preprocessing and short-trip segmentation.Principal Component Analysis(PCA)was employed to reduce dimensionality to 16 feature parameters.The K-means++algorithm enhances clustering stability and accuracy.Markov Chain Monte Carlo(MCMC)optimizes cycle synthesis while preserving low-probability events.Results show the constructed candidate cycles exhibit an average relative error of 3.63%compared to raw data,significantly outperforming traditional methods:cluster-splicing(4.80%)the Markov chain method(5.60%),and standard driving cycles CLTC-P(8.74%)and WLTC(22.70%).
Keywords：big datashort tripkinematic segmentsK-means++cluster analysisMarkov chain Monte Carlo
Publication Date：2025-12-25
Online Publishing Date：2026-01-05(First online date of this platform, not the publication date of the document)
Pages：11( 38-48 )