Safety improvement of traffic at railway crossings

[S. Kasalica]

PDF/A (149 listova)

Mašinstvo - Železničko mašinstvo – bezbednost u saobraćaju / Mechanical engineering - Railway engineering – Traffic safety

U ovom radu su istraženi statistički modeli za procenu frekvencije nesreća, težine posledica nesreća i empirijskog rizika koji najviše odgovaraju lokalnim karakteristikama putnih prelaza Srbiji u cilju identifikacije mesta visokog rizika na mreži. Uzorak putnih prelaza lociranih na železničkoj mreži u Srbiji je korišćen za istraživanje sa istorijskim podacima o nesrećama od pet godina. Korišćeni su modeli regresione analize (Puasonova regresija, Puasonov model sa viškom nula – ZIP, negativni binomni model – NB, NB model sa viškom nula – ZINB). Za modeliranje frekvencije nesreća najbolje se pokazao ZIP model. Kod modeliranja i procene težine posledica nesreća korišćen je multinomni logit model. Za modeliranje empirijskog rizika podacima najviše odgovara ZINB model. Identifikovane su promenljive koje su značajno povezane sa brojem i posledicama nesreća. Konačno, kalibrisani modeli zajedno sa dva kriterijuma za rangiranje koji su primenjeni u ovom radu su poslužili za identifikaciju putnih prelaza visokog rizika na mreži Železnica Srbije. Prvi kriterijum je srednji totalni rizik na putnom prelazu, a drugi kriterijum je procena rizika koja je zasnovana na srednjem empirijskom riziku. Prihvaćeni modeli frekvencije i posledica nesreća su poslužili za procenu redukcije nesreća na putnim prelazima primenom odgovarajućih tehničkih mera za podizanje nivoa bezbednosti. U tu svrhu urađena je sveobuhvatna analiza tehnologija i mera iz drugih regiona, a koje se danas primenjuju ili se ispituje njihova efikasnost. Koristeći njihova iskustva data je ocena efikasnosti koju bi te mere imale na putnim prelazima u Srbiji. U ovom radu vršeno je i neposredno istraživanje ponašanja vozača na putnom prelazu osiguranom drumskom signalizacijom, kao i nakon osiguranja istog putnog prelaza aktivnim sistemom. Ovaj prelazak sa jednog sistema osiguranja na drugi sistem bio je pogodan da bi se sprovelo istraživanje u cilju sagledavanja ponašanja vozača u dva različita sistema, kao i bezbednosnih efekata koji su proizašli iz toga.

In this paper there were developed statistical models for estimation of accident frequency, accident severity and empirical risk, that correspond to local characteristics of crossings in Serbia. The goal was to identify high risk locations. The data used were from five year period. Several regression models were tried (Poisson regression, Zeroinflated Poisson – ZIP, negative binomial model – NB, Zero-inflated NB model – ZINB). The most suitable model for modeling accident frequency was ZIP model. For modeling accident severity multinomial logit model was used, whereas the best model for empirical risk was ZINB model. The variables significantly linked to accident frequency and severity were identified. Finally, the calibrated models together with two ranking criteria were used in order to indentify high risk crossings in Serbian railway network. The first criterion was mean total risk at a crossing, and the second one was based on mean empirical risk. The accepted models for accident frequency and severity were used for estimating the reduction of accidents at railway crossings applying the corresponding technical procedures for safety improvement. In order to accomplish this, there has been performed a complete analysis of technologies and procedures from various regions, which are used nowadays or are in phase of examining there efficency. Their results and experience were used to estimate the efficency that those procedures would have in Serbia. In this paper the drivers' behaviour on railway crossing was also studied, both with road signs and with active warining devices. This upgrade from one system to another was useful for examining drivers' behaviour in two different systems, as well as the safety effects that concequently appeared.

bezbednost, putni prelazi, frekvencije nesreća, posledice nesreća

safety, railway crossings, accident frequency, accident severity

656

Železnički saobraćaj

Tekst