摘要: |
This report describes the development of operational surveillance data processing algorithms and software for application to urban freeway systems, conforming to a framework in which data processing is performed in stages: sensor malfunction detection, data repair, calibration, qualitative modeling of traffic conditions, and, finally, discrimination of incidents from recurrent congestion. Development and testing used real data obtained from freeway systems in Oakland, CA, San Diego, CA and the Twin Cities in Minnesota.Statistical pattern recognition techniques, including optimal decisions trees and neural nets were used. The algorithms and software produced are designed for integration into real-time traffic management systems. This report focuses on the development and testing of the surveillance data processing algorithms. This document is intended for the technical staffs of local traffic management authorities responsible for operating and/or planning freeway incident and response capabilities. |