详情

原文传递 Development & Testing of the I-77 CARAT Incident Detection Algorithm using CORSIM

题名：	Development & Testing of the I-77 CARAT Incident Detection Algorithm using CORSIM
作者：	Sudhir Murthy, P.E
关键词：	Development; Testing; I-77 CARAT; Incident Detection Algorithm; CORSIM
摘要：	This paper describes the methodology for the development and testing of an on-line incident detection algorithm project in Charlotte, NC. The automatic incident detection algorithm was required as part of the design-build ATMS project, currently led by AlliedSignal Technical Services Corporation, Inc(ATSC) for the North Carolina Department Of Transportation (NCDOT), for 16 miles of I-77 passing through Charlotte, NC. Louis Berger & Associates, Inc. (Berger) was selected by ATSC to identify, evaluate and recommend an incident detection algorithm that would meet a set of pre-determined criteria as per their requirement for the selected algorithm. The development of an incident detection algorithm was first carried out by performing an extensive literature review of all existing and promising incident algorithms. A comparison of various algorithms was performed with respect to certain criteria, such as, average incident detection times, false alarm rates, incident detection rates, number of detector stations used , number of parameters used by algorithm and ease of implementation. Finally, due to the uniqueness and specific requirements of the project, Berger developed a customized incident detection algorithm for this project. The algorithm, essentially a modified version of basic California algorithm using elements of the All Purpose Incident Detection (APID) algorithm, used occupancy data from detector stations for incident detection due to lane blockages, and speed for detection of compression waves. Two sets of data were generated from the CORSIM model. One set was used to calibrate the thresholds of the parameters where as the other set was used to actually to test the algorithm based on the calibrated thresholds. The test cases were a combination of different types of incidents, different volumes levels and different random seeds of simulation. Each test case consisted of 30 minutes of simulation time with 10 minutes of lane blockage due to an incident. Each detection station was classified into three categories based on the geometric location of the station such as, whether the station was in a normal section or it was influenced by off ramps or on ramps. Results of the off-line testing showed that algorithm was able to detect almost all the incidents with an average time of detection of less than 3 minutes and a false alarm rate of nearly 0.05%. The off-line testing of the algorithm is very useful as it gave not only a good estimation of the thresholds for the parameters of the algorithm, but also lot of insights in what to anticipate and expect in the on-line implementation of the algorithm. Several other important findings of the study were the type and effect of traffic patterns generated by CORSIM during incidents, evaluation of the algorithm based on detector data by lane versus combination of all lanes at a detector station and several key characteristics of the incident detector algorithm itself. Currently, the algorithm is under on-line implementation and, the thresholds are being fine tuned to reflect the real life traffic conditions.
总页数：	ITS America. Meeting (9th : 1999 : Washington, D.C..). New thinking in transportation : conference proceedings. 1999. pp12
报告类型：	科技报告