Analysis of speed-related crashes using event data recorder data.

摘要

Safety in general is defined as "the absence of unintended harm to living creatures or inanimate objects" (Evans, 2004). Traffic safety is measured in terms of motor vehicle crashes. A crash is an unplanned event which results in either fatality, or injury or property damage. Of all the factors that contribute to a crash, speed is a major issue, as more than 31 percent of all fatal crashes involve speeding (NHTSA, 2007). The speed-safety relationship can be determined by knowing the effect of speed on the frequency and severity of crashes. The relationship between speed and frequency of crashes is much more complicated than that between speed and severity of crashes which is based on principle of physics. Many researchers have studied of these relationships, but there is a risk of validation of the results so obtained, as the data used for most of these studies was obtained from police reports, self reports, case-control studies etc which may be erroneous. As a result, there is a need to revisit these relationships using more accurate and reliable data, and to determine the consistency between police reported speed and actual travel speed. If these two sources have a high level of correlation, there would be more trust placed in findings of older research. Event Data Recorders (EDR), installed in most current vehicles with airbags are capable of capturing and storing pre-crash, crash and post-crash information is proved to be one of the best sources available today.;This research focuses on analyzing the crashes for the years 2002 and 2003 using the supplemental EDR data from the National Automotive Sampling System/Crashworthiness Data System (NASS-NASS/CDS) database collected/maintained by the National Highway Traffic Safety Administration (NHTSA). All crashes in database were filtered to identify a set of records with complete EDR data. From this selection set, analysis was undertaken to include descriptive statistics and comparisons of speed vs type of crash, severity of crash, age and sex of driver, etc. It is observed from the descriptive statistics that EDR subset has some bias that is not apparent in overall NASS sample. It is also observed that the speeding crashes showed the same trend as earlier. For example, high involvement of young drivers, high amount of speeding in case of DUI, etc. Finally, it is found from the statistical tests that the speed five seconds prior to the crash is a better indicator of chosen speed, and that the difference in the average of police reported speed and speed at-5 seconds is practically significant.