Understanding the Operating Speed Profile Patterns Using Unsupervised Machine Learning Approach: Short-Term Naturalistic Driving Study

Pawar, Digvijay S. (2023) Understanding the Operating Speed Profile Patterns Using Unsupervised Machine Learning Approach: Short-Term Naturalistic Driving Study. Journal of Transportation Engineering, Part A: Systems, 149 (2). 04022151. ISSN 2473-2907

	Text JTEPBS.TEENG-7440 (1).pdf Download (1MB)
	Text JTEPBS.TEENG-7440 (1).pdf - Published Version Download (1MB)

Abstract

Several studies have measured the minimum operating speed on horizontal curves to model the operating speed to assess the geometric design consistency. Most of these studies approximated equal lengths of deceleration and acceleration in the operating speed profiles for the curves and assumed the minimum operating speed position at the midpoint of the curve. In contrast, a few recent studies showed different percentages of deceleration lengths on the curve and measured the minimum operating speed at the deceleration end on the curve to model the operating speed. A defined pattern of the operating speed profile on the horizontal curve was not reported in the previous studies and therefore presents opportunities to determine the patterns of the operating speed profiles on curves. In this study, the operating speed profiles of different drivers for the given features of the horizontal curve were studied, and the clustering technique was used to categorize the different patterns in the operating speed profiles on horizontal curves. The optimal number of clusters was determined using four methods: silhouette, elbow, gap statistic, and NbClust function. The different patterns observed from the clustering results are as follows: (1) complete deceleration on the curve, (2) complete acceleration on the curve, (3) deceleration length slightly greater or lower than acceleration length, and (4) longer deceleration/acceleration lengths followed by shorter acceleration/deceleration lengths, respectively. The study results imply that all operating speed profiles are not symmetric around the midpoint of the curve (MC), and the group of drivers exhibited defined patterns of the operating speed profiles on the curves. This study helps in understanding the different patterns of operating speed profiles exhibited by the drivers and the measurement of the minimum operating speed at the deceleration end to model the operating speed to assess the geometric design consistency.

[error in script]

IITH Creators:

IITH Creators	ORCiD
Pawar, Digvijay S.	https://orcid.org/0000-0003-4228-3283

Item Type:

Article

Uncontrolled Keywords:

Clustering; Design consistency; Midpoint of the curve; Minimum operating speed; Deceleration; cluster analysis; design; machine learning; Acceleration length; Clusterings; Design consistency; Geometric design; Horizontal curves; Midpoint of the curve; Minimum operating speed; Speed profile; Unsupervised machine learning; Acceleration; Clustering algorithms; Machine learning

Subjects:

Civil Engineering

Divisions:

Department of Civil Engineering

Depositing User:

Mr Nigam Prasad Bisoyi

Date Deposited:

22 Aug 2023 05:25

Last Modified:

22 Aug 2023 05:25

URI:

http://raiithold.iith.ac.in/id/eprint/11579