Reliability Prediction Method of Arterial Travel Time in Connected Mixed Traffic Environment

Travel time, as an important parameter of traffic flow, provides the basic information for supporting route planning for travelers and route guidance for managers. Meanwhile, travel time reliability enables one to fully characterize the relationship between traffic supply and demand, describes the state of traffic flow, and plays an important role in route optimization, traffic demand assessment and prediction, traffic management and control, etc. By focusing on signalized urban arterials, this study uses traffic simulation to build the mixed traffic environment with Connected autonomous vehicle (CAV) and Hand-driven vehicle (HDV). Based on a large amount of simulation data, the distribution characteristics of arterial travel time were analyzed, the key influencing factors of travel time reliability were identified, and a fixed-coefficient mixed distribution model of arterial travel time was established. The travel time reliability was evaluated in terms of a series of metrics such as mean, variance, coefficient of variation (CV), buffer index (BI), planning time index (PTI), and misery index (MI). Moreover, by fully considering the dynamics, randomness, and complexity of the mixed traffic flow on urban arterials, this study proposed a travel time distribution prediction method based on the mixed distribution model with variable weights. Specifically, the mixing coefficients and components of the mixture distribution model were modeled as the function of the key parameters such as CAV penetration and traffic flow, which helps to enhance the interpretability of the mixed traffic flow characteristics. Last, the prediction model was verified under the typical scenarios in simulation. The results show that the proposed model achieves desirable prediction accuracy of arterial travel time distributions under mixed traffic conditions.