A technique to calibrate nonlinear car-following laws for traffic oscillation estimation

Traffic congestion is one of the chief transportation problems in the world today. The total cost of transportation congestion in the United States in 2011 was estimated at 121 billion dollars. Congested traffic rarely flows smoothly; instead, traffic goes through cyclical slow and fast movements that result in what is commonly known as traffic oscillations. There are numerous negative effects associated with these oscillations, including extra fuel consumption and greenhouse gas emissions. Therefore, transportation professionals would like to find a way to reduce traffic oscillations. Hence, researchers have developed numerous models to help reproduce traffic oscillations, in the hopes of better understanding how oscillations form and propagate through a platoon of vehicles. Many simulation and experimental studies have been performed to study oscillations, but as data collection technologies have improved, they have provided the opportunity to study oscillations as they occur in the real world. By calibrating the parameters of the theoretical models with empirical data, it is possible to more accurately reproduce driver behavior in oscillations and consequently to evaluate the impacts of various potential traffic control strategies on traffic oscillations. This thesis proposes a calibration technique that is useful in calibrating nonlinear car-following laws to accurately model traffic oscillations from field trajectory data in both the time domain and the frequency domain. The base of the technique is maximum likelihood estimation, as calculated using the speed-spacing diagram. Time-domain and frequency-domain feedback is then added to this base to achieve accuracy in both domains. Numerical examples using NGSIM data (Next Generation SIMulation) are provided to verify the proposed method. Further analysis of the model parameters and potential traffic control strategies to mitigate the traffic oscillation problem are then discussed