Method for Identifying Truck Traffic Site Clustering Using Weigh-in-Motion (WIM) Data

Dan Liu,Wang Yinhai,Evangelos I Kaisar,Zhenghong Deng

doi:10.1109/access.2020.3011433

Abstract

The increasingly growing truck traffic volume data while limited truck weigh-in-motion weight data has posed great challenges for transport agencies to access the freight tonnage of all the truck traffic sites. By mapping a group of traffic sites with similar traffic patterns to a weigh-in-motion site, the clustered truck traffic data is expected to be smaller than the sum of all data from all traffic sites, and the cluster can be fully utilized in a period of time by transport agencies to evaluate the freight tonnage. This study developed a novel and implementable approach of integrating two complementary data, Weigh-in-Motion (WIM) weigh data and Telemetric Traffic Monitoring Sites (TTMSs) volume data, to produce truck traffic sites clustering. An improved k-means clustering with three attributes is fitted to the TTMS, which are the distances to the WIM sites (WIMSs), truck volumes in TTMS, and vehicle class distribution. The aforementioned methodology was tested in a case study in Florida using WIM data in 2012 and 2017. The proposed model might shed light on the statewide performance evaluation of freight traffic with low computing cost.

Highlights

With the globalization and economic restructuring, freight volume has increased substantially for all transport modes
Average annual daily truck traffic (AADTT) is the total truck traffic volume divided by 365 days (366 days for leap years) collected by Telemetric Traffic Monitoring Sites (TTMSs)
Based on the presented methods, here, three attributes are considered when TTMSs are allocated to the nearest cluster-AADTTof the TTMSs, the distance between the TTMS and the WIM, and the vehicle class distribution of the TTMS

Summary

INTRODUCTION

With the globalization and economic restructuring, freight volume has increased substantially for all transport modes. Inter-county movements of commodities by trucks, but not intra-county movements are investigated by CFS, for example, vehicle class 5 accounts for a larger percentage of the total volume, it is not included in the investigation Considering the limitations, another type of vehicle-based data extracted from traffic-actuated sensing systems is adopted more and more by transport agencies to acquire updated truck traffic information. Papagiannakis et al adopted clustering techniques to establish similarities in axle-load distributions and vehicle classification between traffic data collection sites to evaluate pavement performance. Based on the presented methods, here, three attributes are considered when TTMSs are allocated to the nearest cluster-AADTTof the TTMSs, the distance between the TTMS and the WIM, and the vehicle class distribution of the TTMS. Step 5: Repeat step three and four until the assignment has not changed

IMPROVEMENT OF K-MEANS CLUSTERING

CASE STUDY

CONCLUSION