Relevance of reactive traffic management strategies such as freeway incident detection has been diminishing with advancements in mobile phone usageand video surveillance technology. On the other hand, capacity to collect, store, and analyze traffic data from underground loop detectors has witnessed enormous growth in the recent past. These two facts together provideus with motivation as well as the means to shift the focus of freeway traffic management toward proactive strategies that would involve anticipating incidents such as crashes. This report provides details of the research effort for developing a proactive traffic management system for a 36.25-mile instrumented corridor of Interstate-4 in Orlando metropolitan area. Complete research effort is documented in two separate volumes. This volume (Volume I) documents the work to develop real-time crash risk assessment models. Among available historical crashes used in the analysis; the rear-end crashes are analyzed first. They are the single most frequent type of crash on the freeway. It was found that rear-end crashes may be separated into two groups based on the speeds prevailing at surrounding stations 5-10 minutes before the crash. The authors have developed separate models for the rear-end crashes that occur during congested speed regime and those which occur under moderate to high speed regime. A classification tree modelfor regime identification is followed up with neural network models for each of the two groups of rear-end crashes. It was found that the proposed approach for real-time identification of crash prone conditions can predict 3/4 of the rear-end crashes with reasonable false alarm rate. The analysis is then followed up by artificial intelligence based neural network models for lane-change related crashes. However, the authors did not find any groupings in these crashes analogous to the rear-end crashes belonging to two distinct regimes. These models for rear-end and lane-change related crashes are then put together in the form of a framework for real-time crash risk assessment. (Author/publisher)
Abstract