摘要
Surrogate safety measures （SSM） are suitable tools to detect dangerous situations. These indicators can be applied as a warning strategy in collision avoidance systems （CAS）. Timeto-collision （TTC） and post-encroachment time （PET） are two important time-based SSM that identify the probability of a rear-end collision. TIC refers to the imminent danger, and PET implies the potential danger. However, sometimes the results from each indicator are inconsistent. An appropriate warning strategy for CAS can be developed using a new index that combines the properties of both TTC and PET. For this purpose, a new mixed index （MI） is proposed. In order to develop this MI, three main microscopic parameters, clearance, speed and the relative speed, are simultaneously applied to the leading vehicle. To cali- brate MI, based on a fuzzy inference system （FIS）, a value would be determined by a combination of TTC and PET at each instant and then by regression analysis the model parameters would be determined. Finally, MI, TIC and PET values for real car-following scenarios on the 1-80 freeway are determined and compared. The results show that MI may be more suitable in detecting the rear-end collision risk within the proper time and with less errors.
Surrogate safety measures （SSM） are suitable tools to detect dangerous situations. These indicators can be applied as a warning strategy in collision avoidance systems （CAS）. Timeto-collision （TTC） and post-encroachment time （PET） are two important time-based SSM that identify the probability of a rear-end collision. TIC refers to the imminent danger, and PET implies the potential danger. However, sometimes the results from each indicator are inconsistent. An appropriate warning strategy for CAS can be developed using a new index that combines the properties of both TTC and PET. For this purpose, a new mixed index （MI） is proposed. In order to develop this MI, three main microscopic parameters, clearance, speed and the relative speed, are simultaneously applied to the leading vehicle. To cali- brate MI, based on a fuzzy inference system （FIS）, a value would be determined by a combination of TTC and PET at each instant and then by regression analysis the model parameters would be determined. Finally, MI, TIC and PET values for real car-following scenarios on the 1-80 freeway are determined and compared. The results show that MI may be more suitable in detecting the rear-end collision risk within the proper time and with less errors.
