A Novel Approach to Extract Speed Pattern from VTDR Data

Abstract

This paper describes a novel technique to analyze newly available Vehicle Traveling Data Recorder (VTDR) data from Passenger Transport Company. To help us be more conducive to study, we preprocessed the VTDR data firstly. 1) Algorithm that detects highway range in the VTDR data is proposed. 2) We transformed VTDR data from their original Time based Speed Series (TSS) to Location based Speed Series (LSS), which makes a link between road location and speed. 3) Then, we applied a self-training technique to distinguish two route directions of the VTDR data. After preprocessing, we extracted general speed pattern from the VTDR data. This information greatly improves our understanding on the road traffic influence to speed. In addiction, based on this, novel driver behavior analysis and outlier analysis are given.The technique introduced in this paper utilizes the huge amount of VTDR data to extract useful information, so that enhances transportation safety management.