Sequential Interpretation and Prediction of Secondary Incident Probability in Real Time
Temporal disruptions that take away part of the roadway from use is a major issue, because speed reduction and rubbernecking caused by primary incidents provoke additional incidents, which are referred to secondary incidents. It is important to sequentially predict the probability of secondary incidents and develop a counter measure to reduce the risk. Advanced computing techniques are used to easily understand and reliably predict secondary incident occurrences that have low sample mean and a small sample size. The quality of predictions improves as new information becomes available. The prediction performance of the principled Bayesian learning to neural networks (BNN) is compared to the Stochastic Gradient Boosted Decision Trees (GDBT). A pedagogical rule extraction, TREPAN, which extracts comprehensible rules from the neural networks, improves the ability to understand secondary incidents with simpler forms. With an acceptable accuracy, GDBT is a useful tool that simply presents relative importance of the predictor variables. Unexpected traffic congestion incurred by an incident is a dominating factor to cause a secondary incident at different stages of incident clearance. This symbolic description represents a series of decisions to assist emergency operators and improve their decision capabilities. Analyzing causes and effects of traffic incidents helps traffic operators develop strategic plans for prompt emergency response and clearance. Application of the model in connected vehicle environment will help drivers receive proactive corrective feedback before a crash. The proposed methodology can be used to alert drivers about potential highway conditions and increase awareness of potential events when no re-routing is possible.
-
Supplemental Notes:
- This paper was sponsored by TRB committee ANB20 Standing Committee on Safety Data, Analysis and Evaluation.
-
Corporate Authors:
500 Fifth Street, NW
Washington, DC United States 20001 -
Authors:
- Park, Hyoshin
- Gao, Song
- Haghani, Ali
-
Conference:
- Transportation Research Board 96th Annual Meeting
- Location: Washington DC, United States
- Date: 2017-1-8 to 2017-1-12
- Date: 2017
Language
- English
Media Info
- Media Type: Digital/other
- Features: Figures; Maps; Photos; References; Tables;
- Pagination: 17p
- Monograph Title: TRB 96th Annual Meeting Compendium of Papers
Subject/Index Terms
- TRT Terms: Crash causes; Emergency response time; Hazards and emergency operations; Mathematical models; Mobile communication systems; Secondary crashes; Traffic congestion; Traffic incidents
- Subject Areas: Highways; Safety and Human Factors;
Filing Info
- Accession Number: 01628178
- Record Type: Publication
- Report/Paper Numbers: 17-06254
- Files: TRIS, TRB, ATRI
- Created Date: Mar 7 2017 10:25AM