Development of a crash risk index to identify real time crash risks on freeways: Fid-Bau Portal

Development of a crash risk index to identify real time crash risks on freeways

Xu, Chengcheng / Liu, Pan / Wang, Wei / Jiang, Xuan

Abstract The primary objective of this study is to develop a quantitative indicator that can be directly used to identify hazardous traffic states on freeway mainlines. Using data obtained from a 22-mile freeway segment on the I-880N freeway in San Francisco Bay Area in the United States, Fisher discriminant analysis was conducted to derive a linear combination of traffic flow parameters which can be used to distinguish traffic states which may lead to crashes from normal traffic conditions which were potentially safe. A new variable termed “crash risk index” was defined based on the discriminant function. A crash risk index which is smaller than 0 represents a hazardous traffic condition potentially leading to a crash. Accordingly, a crash risk index which is greater than 0 represents normal traffic conditions which, theoretically, will not lead to crash occurrences. The research team tested the prediction performance of using the crash risk index to identify hazardous traffic conditions on freeway mainlines. An overall prediction accuracy of about 65.7% was achieved. Conditional logistic regression analysis was then conducted to evaluate the impacts of crash risk indexes on the likelihood of crash occurrences. It was found that the likelihood of freeway crashes increased with the decreases in crash risk index. One unit decrease in crash risk index increased the risk of crash occurrence by 180.3%. The research team also looked extensively at the impacts of crash risk indexes at different time slices. It was found that the impacts of crash risk index decrease with time and the crash risk index at the time slice most immediately before crashes had the greatest impact on crash occurrences.