Calibration and Validation Procedure for Intersection Safety Simulation Using SSAM and VISSIM

摘要

Traffic safety simulation is a preferable way to evaluate safety through computing conflict indices between every two vehicles during their running process. Compared to operational efficiency simulation, lots of parameters need to be calibrated, and high precision is required and a little change of parameters may result in great difference. Moreover, commonly used calibrations based on intelligent algorithms prefer result precision to behavioral parameters, and may not reflect real traffic conditions for stochastic heuristic algorithm. Thus, a two stage calibration methodology based on experimental optimization is put forward. To depict its effectiveness, 6 group data of 3 intersections are surveyed and imposed for calibration and validation through above method by linking Surrogate Safety Analysis Model (SSAM) using vehicle trajectories exported from VISSIM. The result is quite promising that the errors of conflict number, conflict velocity and delay are 7%~13%, -7%~5%, -ll%~-6% respectively with a decreasing of 39%~43%,-6%~-2%,-17%~-10%, which objectively demonstrates feasibility and validity of the method.