Off-Line Calibration of Dynamic Traffic Assignment- Analysis Algorithms and Applications

Dynamic Traffic Assignment (DTA) models estimate and predict the evolutionof congestion through detailed models and algorithms that capture traveldemand network supply and their complex interactions. The availability ofrich time-varying traffic data spanning multiple days collected by automaticsurveillance technology provides the opportunity to calibrate such a DTAmodel's many inputs and parameters so that its outputs reflect field conditions.DTA models are generally calibrated sequentially: supply model calibration(assuming known demand inputs) is followed by demand calibrationwith fixed supply parameters. This book develops an off-line DTA modelcalibration methodology for the simultaneous estimation of all demand andsupply inputs and parameters using sensor data. A complex non-linearstochastic optimization problem is solved using any general traffic data. Casestudies with DynaMIT a DTA model with traffic estimation and predictioncapabilities indicate that the simultaneous approach significantly outperformsthe sequential state of the art. This book is addressed to professionalsand researchers who apply large-scale transportation models.