engleski

Predicting state of traffic signs using logistic regression

Traffic signs as part of the overall traffic signalization system convey a message to road users using shapes, colours, text and symbols. They inform road users about regulations, warnings, directions, and guidance in traffic systems, in order to ensure safe traffic flow. In conditions of low visibility drivers receive less visual information in traffic which makes the perception of the surroundings, and thus the driving, significantly more difficult. In order to overcome the mentioned problems traffic signs must have satisfactory retroreflection properties and be properly positioned and maintained. Given the number of traffic signs on the roads, it is necessary to optimize their maintenance activities. The aim of this study is to develop a model for predicting the state of traffic signs regarding their retroreflective values based on their age. The study included 21, 467 traffic signs on 30 state roads throughout the Republic of Croatia. Linear models for predicting state of signs were developed using binary logistic regression for each class of retroreflective material. Even though the models very accurately predict when the signs meet minimal prescribed retroreflection values for all the three classes of retroreflective material, they have certain downsides when predicting when the signs are not valid, i.e. do not meet minimal prescribed retroreflection values. Although the developed models did not show satisfactory accuracy, they represent unique prediction models of traffic signs functional service life, enabling prediction without conducting previous retroreflection measurements which, considering the number of in-service traffic signs, thus enables the optimization of the traffic signs maintenance system.

traffic signs ; retroreflection ; logistic regression ; road safety ; traffic signs maintenance ;

nije evidentirano