The road user's travel time considerably increases when an incident occurs in a section of the road due to the reduction of the offered capacity. To inform the road users about the possible delay they will suffer travelling the partially or completely interrupted section of the road a prediction model of the Incident Resolution Time (IRT) has been developed. This time must be added to the normal travel time of the road section and to the time required to get over the formed queue to evaluate the new increased travel time to be displayed of the traffic information devices. The paper describes the prediction model developed to estimate the IRT values of motorway crash. The model was developed analyzing the resolution time of incidents occurred in a 5 year time period on the Italian A1 Motorway from Bologna to Firenze according to the CART (Classification and Regression Tree) statistical procedure. The incident data were included in the newly implemented Incident Duration Data Base (IDDB) and were analyzed to identify the IRT influencing variables. The variables retained included the crash type, the type of involved vehicles, the crash severity and the hour of the day in which the incident occurs. The constructed classification tree brings to 24 IRT classes of incidents, each one described by the average value and standard deviation of IRT and by the probabilities of that a given IRT value can be over passed. The developed model has been validated using new incident data with promising results. The developed IRT prediction model is a dynamic model able to increase the reliability of the information offered provided the IDDB on which the model operates is continuously updated with the new crash data occurring along the motorway network considered. The continuous enrichment of the IDDB will also allow an off line periodic analysis of the data to evaluate if new descriptive variables should be included in the developed classification tree.

Prediction of incident resolution time on motorways / L. Domenichini; F. Fanfani; M. Bacchi; A. Braccini. - In: ADVANCES IN TRANSPORTATION STUDIES. - ISSN 1824-5463. - STAMPA. - 7:(2013), pp. 5-22.

Prediction of incident resolution time on motorways

DOMENICHINI, LORENZO;FANFANI, FRANCESCO;
2013

Abstract

The road user's travel time considerably increases when an incident occurs in a section of the road due to the reduction of the offered capacity. To inform the road users about the possible delay they will suffer travelling the partially or completely interrupted section of the road a prediction model of the Incident Resolution Time (IRT) has been developed. This time must be added to the normal travel time of the road section and to the time required to get over the formed queue to evaluate the new increased travel time to be displayed of the traffic information devices. The paper describes the prediction model developed to estimate the IRT values of motorway crash. The model was developed analyzing the resolution time of incidents occurred in a 5 year time period on the Italian A1 Motorway from Bologna to Firenze according to the CART (Classification and Regression Tree) statistical procedure. The incident data were included in the newly implemented Incident Duration Data Base (IDDB) and were analyzed to identify the IRT influencing variables. The variables retained included the crash type, the type of involved vehicles, the crash severity and the hour of the day in which the incident occurs. The constructed classification tree brings to 24 IRT classes of incidents, each one described by the average value and standard deviation of IRT and by the probabilities of that a given IRT value can be over passed. The developed model has been validated using new incident data with promising results. The developed IRT prediction model is a dynamic model able to increase the reliability of the information offered provided the IDDB on which the model operates is continuously updated with the new crash data occurring along the motorway network considered. The continuous enrichment of the IDDB will also allow an off line periodic analysis of the data to evaluate if new descriptive variables should be included in the developed classification tree.
2013
7
5
22
L. Domenichini; F. Fanfani; M. Bacchi; A. Braccini
File in questo prodotto:
File Dimensione Formato  
Domenichini_5-22.pdf

Accesso chiuso

Descrizione: Articolo pubblicato
Tipologia: Pdf editoriale (Version of record)
Licenza: Tutti i diritti riservati
Dimensione 843.69 kB
Formato Adobe PDF
843.69 kB Adobe PDF   Richiedi una copia

I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificatore per citare o creare un link a questa risorsa: https://hdl.handle.net/2158/821286
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? ND
social impact