Accurate prediction of incident duration is not only important information of Traffic Incident Management System, but also an effective input for travel time prediction. In this paper, the hazard based prediction models are developed for both incident clearance time and arrival time. The data are obtained from the Queensland Department of Transport and Main Roads’ STREAMS Incident Management System (SIMS) for one year ending in November 2010. The best fitting distributions are drawn for both clearance and arrival time for 3 types of incident: crash, stationary vehicle, and hazard. The results show that Gamma, Log‐logistic, and Weibull are the best fit for crash, stationary vehicle, and hazard incident, respectively. The obvious impact factors are given for crash clearance time and arrival time. The quantitative influences for crash and hazard incident are presented for both clearance and arrival. The model accuracy is analyzed at the end.

Deep learning models for short-term travel speed prediction on urban expressways, such as the convolutional neural network (CNN), still present several limitations in multiscale spatiotemporal feature extraction. Hence, in this paper, three hybrid CNN models are proposed to improve the basic CNN model with regard to three target aspects for short-term (i.e., 5 min) travel speed prediction on urban expressways. More specifically, long short-term memory (LSTM), AutoEncoder (AE), and Inception module are incorporated into the basic CNN model to capture multiscale spatiotemporal features of travel speed data effectively and improve the accuracy and robustness of the basic CNN model. Based on loop detector data collected on the Yan'an expressway in Shanghai, the proposed hybrid CNN models are trained and tuned. To validate the improvements on the target aspects, a comprehensive comparison is conducted using a classical statistical model (i.e., autoregressive integrated moving average), a typical shallow neural network model (i.e., artificial neural network), and two basic deep learning models (i.e., recurrent neural network and CNN). Results show that the prediction accuracies of all the proposed hybrid CNN models exceed 96% and the mean absolute errors are less than 2.5 km/h, which are superior to other models. In terms of target improving aspects, two new metrics were introduced, and the proposed models, especially the AE-CNN model, showed better robustness under various input data structures and traffic states. The LSTM-CNN model outperformed the other models in learning time-series features, and the Inception-CNN model is superior in reproducing the dynamics of traffic congestion patterns on urban expressways.

Due to full record of discharging vehicle headway, License Plate Recognition (LPR) is used as an ideal source in most existing queue length estimation methods through a double-section detection using shock-wave models or input-output models. However, the impacts of heavy vehicles and miss detection by LPR detectors are mostly ignored. Therefore, this paper proposes a lane-based queue length estimation method using single-section LPR detection, considering miss detection and heavy vehicles. The queue length estimation problem is transformed to a change-point identification problem for discharging headways time-series, using E-Divisive with Medians (EDM) method. The maximal queue length is identified as the change-point of the discharging headways with the maximal differences between queued and non-queued vehicles, considering the queuing homogeneity of a lane group and the miss detection rate of LPR. The proposed method is validated using simulation and empirical cases with promising performance and good robustness under various conditions.

At high-speed intersections in many Chinese cities, a traffic-light warning sequence at the end of the green phase—three seconds of flashing green followed by three seconds of yellow—is commonly implemented. Such a long phase transition time leads to heterogeneous decision-making by approaching drivers as to whether to pass the signal or stop. Therefore, risky driving behaviors such as red-light running, abrupt stop, and aggressive pass are more likely to occur at these intersections. Proactive identification of risky behaviors can facilitate mitigation of the dilemma zone and development of on-board safety altering strategies. In this study, a real-time vehicle trajectory prediction method is proposed to help identify risky behaviors during the signal phase transition. Two cases are considered and treated differently in the proposed method: a single vehicle case and a following vehicle case. The adaptive Kalman filter (KF) model and the K-nearest neighbor model are integrated to predict vehicle trajectories. The adaptive KF model and intelligent driver model are fused to predict the following vehicles’ trajectories. The proposed models are calibrated and validated using 1,281 vehicle trajectories collected at three high-speed intersections in Shanghai. Results indicate that the root mean square error between the predicted trajectories and the actual trajectories is 5.02 m for single vehicles and 2.33 m for following vehicles. The proposed method is further applied to predict risky behaviors, including red-light running, abrupt stop, aggressive pass, speeding pass, and aggressive following. The overall prediction accuracy is 95.1% for the single vehicle case and 96.2% for the following vehicle case.

Shared parking is an effective means to alleviate the shortage of parking spaces in urban central areas during the morning peak hours. Meanwhile, walking time cost and parking fees are two critical factors affecting shared parking efficiency. Firstly, based on the classic bottleneck model, walking time cost and shared parking fees are added to the model, then the user equilibrium solutions are obtained considering two choices of parking lots: accessorial parking and shared parking. Next, taking the minimum total social cost and the minimum total queue time as the management goals, respectively, the quantitative relationship between parking fee as well as the dispersion degree of the shared parking spaces with the optimal travel pattern is proved. Besides, the rules and methods for the reasonable setting of shared parking fee and the dispersion degree of shared parking spaces layout are proposed. Through the research, it was demonstrated that: (1) differentiated shared parking fee based on the capacity of the accessorial parking lot can reduce both the total social cost and road congestion, while total social cost increases with the dispersion of the layout of shared parking spaces and road congestion decreases with the dispersion of the layout of shared parking spaces; and (2) when derived the optimal travel pattern, it is found that there is an inert zone of shared parking fee, i.e., regardless of adjusting shared parking fee, it had no impact on the determination of the optimal travel pattern. Finally, we put forward policy recommendations combining the numerical examples. Considering the total social cost and congestion, managers can improve the utilization efficiency of shared parking spaces while taking into account social benefits by reasonably setting shared parking fee and the dispersion degree of shared parking spaces layout.