Abstract: The paper was aimed at analyzing quantitatively the safety of pedestrian crossing signalized section.Firstly,the pedestrian crossing characteristics were analyzed through field investigation.Then a concept of safety entropy was proposed,and a safety entropy model was built based on the modified traditional COX proportional hazard regression model.In order to solve the model,the principal component analysis was used to extract the main components of pedestrian crossing risk.The safety grades were classified by fuzzy theory.The model was applied to evaluation of safety of pedestrians crossing in a signalized section by the Nanjing Children's Hospital.The research shows that green light inertial psychology and red light tailing psychology are the main psychology dominating pedestrians running red lights at signalized sections,that the higher the horizontal cutoff α,the more conservative the level of interception is,and entropy value collection of α=9 can be taken as a basis for safety classitacation,and that the risk evaluation result obtained by using the model conforms to the reality.Therefore,the model constructed can be used to evaluate the safety of pedestrian crossing in the signal control section.

Key words: safety entropy, signalized section, survival analysis, COX regression model, risk evaluation