Abstract
A significant portion of nonrecurrent freeway congestion is caused by incidents. This nonrecurrent congestion negatively impacts safety and mobility. It produces enormous travel delay and results in secondary incidents, which cause additional delay and injury. The simulation-based secondary incident filtering (SBSIF) method is proposed for identifying secondary incidents from archived incident data. The proposed methodology is computationally efficient and overcomes deficiencies of existing techniques. It is evaluated on incident data from a 6-month period along a segment of I-287 in the New York state. Results of the evaluation show that the SBSIF method has a significantly reduced misclassification rate (e.g., a reduction of 58 percentage points and greater) as compared with static methods commonly used in practice, despite that, it requires comparable computational effort.

This publication has 3 references indexed in Scilit: