Impact analysis and behavioral modeling of spatial-temporal crossing violations by elderly pedestrians at signalized intersections

Studying the crossing violation behaviors of elderly pedestrians is of great significance to traffic safety. Thus, this paper proposed to model the crossing violation behaviors of elderly pedestrians using the Bayesian network method. Firstly, five types of crossing violation behaviors of elderly pedestrians were temporally and spatially analyzed. Considering the selecting behavior of pedestrians crossing the street, the influencing factors of elderly pedestrian crossing violations were divided into elderly attributes, dynamic traffic information and intersection facilities. A total of 748 data samples of the elderly crossing the street were collected through the survey. Combined with the mechanism of elderly pedestrian crossing violations, a greedy search algorithm was used to learn the Bayesian network structure. Then, the Bayesian network models of the temporal and spatial violations of elderly pedestrians were constructed, respectively. The model parameters were learned by the expectation-maximization algorithm. The key parameters were analyzed by the Bayesian inference method. The temporal analysis shows that travel purpose, pedestrian flow, traffic flow, green light duration and number of lanes were the main influences on the temporal violation behavior of elderly pedestrians. The spatial analysis shows that age, number of children, waiting time, pedestrian flow and number of lanes were the main influences on the spatial violation behavior of elderly pedestrians. Finally, the established temporal and spatial violation models for elderly pedestrians were verified by case studies, with an accuracy of 83.33% and 74.29%, respectively. The results of this study can provide some suggestions for the traffic safety management of the elderly: pay attention to the traffic safety education of the elderly who are lonely; Control the traffic flow at intersections, and add age-appropriate designs at large intersections with long green light duration and many lanes. The research results can provide certain suggestions for traffic safety management for elderly people, including broadcasting traffic safety education for the elderly people, adding age-friendly designs at signalized intersections and so on.