摘要: The main goal of this research is to evaluate the performance of incidentdecision algorithms under different highway geometric conditions on I-4. A particular problem that faces traffic operators at the Orlando Freeway Management ... 展开 The main goal of this research is to evaluate the performance of incidentdecision algorithms under different highway geometric conditions on I-4. A particular problem that faces traffic operators at the Orlando Freeway Management Center (FMC) is how to differentiate between incident and non-incident congestion on I-4 segments that are located at grades and/or at horizontal curves. Traffic flow and incident data on I-4 phase I surveillance system have been collected for a period of 15 months. A set of incident detection algorithms was selected for off-line testing using a total of 909 confirmed incidents (with valid loop data). The set of algorithms includes California Algorithm 7, 8, and 10 and the modified McMaster Algorithm developed by G. Forbes. Three measures of effectiveness were used to evaluate these algorithms: detection rate, false alarm rate, and mean detection time. 收起