Modelos de previsão de acidentes de trânsito em interseções semaforizadas de Fortaleza

Abstract

Urban road intersections concentrate a significant portion of traffic accidents due to the high degree of interaction between users, vehicles and road. Accident prediction models are regression models that establish a relationship between the frequency of traffic accidents and geometric and operational attributes of the road, dealing, with relative success, with the high degree of randomness of this phenomenon. The objective of this work isto develop traffic accident prediction models for signalized intersections located in the city of Fortaleza, with emphasis on the methodology for the development of simple and reliable models. The models were estimated based on a sample of 101 intersections, having as prediction variables annual average daily traffic (AADT), number of lanes, number of approaches, and type of central median. The model containing AADT and number of lanes showed satisfactory performance for the es-timation of the total number of traffic accidents at the evaluated intersections.