Feature selection with genetic algorithms for accident duration forecasting on freeway
This study develops two artificial neural network-based models to providea sequential forecast of accident duration from the accident notificationto the accident site clearance. With these two models, the estimated duration time can be provided by plugging in relevant traffic data as soon as an accident is notified. To select suitable data features, genetic algorithm is employed to decrease the number of model inputs while preserving relevant traffic characteristics with fewer inputs. This study shows that theproposed models are feasible in the intelligent transportation systems (ITS) context. For the covering abstract see ITRD E140665.
-
Authors:
- LEE, YING
- WEI, CHIEN-HUNG
- Publication Date: 2007-10
Language
- English
Media Info
- Serial:
Subject/Index Terms
- TRT Terms: Calculation; Conferences; Crashes; Forecasting; Incident detection; Intelligent transportation systems; Mathematical models; Methodology; Traffic delays; Travel time
- ITRD Terms: 1643: Accident; 6464: Calculation; 8525: Conference; 9010: Delay; 132: Forecast; 1632: Incident detection; 8735: Intelligent transport system; 697: Journey time; 6473: Mathematical model; 9102: Method
- Subject Areas: Operations and Traffic Management; Safety and Human Factors; I73: Traffic Control;
Filing Info
- Accession Number: 01148868
- Record Type: Publication
- Source Agency: Transport Research Laboratory
- Files: ITRD
- Created Date: Jan 25 2010 8:56AM