The prediction of accidents on digital networks: Characteristics and issues related to the application of accident prediction models.

摘要

Transportation planning models are used to estimate, as accurately as possible, future traffic patterns, peak periods, travel time, and various environmental or other related traffic flow by-products. Unfortunately, traffic safety is seldom, if ever, explicitly analyzed during the transportation planning process. The non-evaluation of safety is attributed to various factors, including the lack of available tools needed to estimate the number of accidents on digital networks or urban transportation networks. Thus, the primary objective of this work was to develop a series of models that would allow the estimation of traffic accidents on digital networks; that is, before a physical transportation facility is built or upgraded. The secondary objective consisted of describing all the issues surrounding their application on digital networks. To accomplish this goal, several accident prediction models that include trend were developed to predict accidents at nodes and on links. As part of this work, a new method to estimate the coefficients of models with trend is explained. A few illustrative applications of the models are also presented. The models were applied to three sample digital networks and the simulation of traffic was performed with either EMME/2 or Paramics. The results showed that it is possible to predict accidents on digital networks, but the accuracy is directly related to the precision of transportation planning software programs. Hence, inaccurate traffic flow prediction leads to incorrect accident prediction. Thus, efforts should be made in trying to find better flow estimates. Some proposed models are also sensitive to how the digital network is coded and the predicted number of accidents should be adjusted accordingly. Finally, several issues and limitations related to the application of accident prediction models to determine the safest paths on digital networks and evaluate the safety effects of dynamic route guidance systems are described in this thesis.