A graph deep learning method for short‐term traffic forecasting on large road networks

Short‐term traffic flow prediction on a large‐scale road network is challenging due to the complex spatial–temporal dependencies, the directed network topology, and the high computational cost. To address the challenges, this article develops a graph deep learning framework to predict large‐scale network traffic flow with high accuracy and efficiency. Specifically, we model the dynamics of the traffic flow on a road network as an irreducible and aperiodic Markov chain on a directed graph. Based on the representation, a novel spatial–temporal graph inception residual network (STGI‐ResNet) is developed for network‐based traffic prediction. This model integrates multiple spatial–temporal graph convolution (STGC) operators, residual learning, and the inception structure. The proposed STGC operators can adaptively extract spatial–temporal features from multiple traffic periodicities while preserving the topology information of the road network. The proposed STGI‐ResNet inherits the advantages of residual learning and inception structure to improve prediction accuracy, accelerate the model training process, and reduce difficult parameter tuning efforts. The computational complexity is linearly related to the number of road links, which enables citywide short‐term traffic prediction. Experiments using a car‐hailing traffic data set at 10‐, 30‐, and 60‐min intervals for a large road network in a Chinese city shows that the proposed model outperformed various state‐of‐the‐art baselines for short‐term network traffic flow prediction.     

Wavelet‐Based Denoising for Traffic Volume Time Series Forecasting with Self‐Organizing Neural Networks

Abstract: In their goal to effectively manage the use of existing infrastructures, intelligent transportation systems require precise forecasting of near‐term traffic volumes to feed real‐time analytical models and traffic surveillance tools that alert of network links reaching their capacity. This article proposes a new methodological approach for short‐term predictions of time series of volume data at isolated cross sections. The originality in the computational modeling stems from the fit of threshold values used in the stationary wavelet‐based denoising process applied on the time series, and from the determination of patterns that characterize the evolution of its samples over a fixed prediction horizon. A self‐organizing fuzzy neural network is optimized in its configuration parameters for learning and recognition of these patterns. Four real‐world data sets from three interstate roads are considered for evaluating the performance of the proposed model. A quantitative comparison made with the results obtained by four other relevant prediction models shows a favorable outcome.     

Kernel Smoothing Method Applicable to the Dynamic Calibration of Traffic Flow Models

Abstract: Real time traffic flow simulation models are used to provide traffic information for dynamic traffic management systems. Those simulation models are supplied by traffic data in order to estimate and predict traffic conditions in unobserved sections of a traffic network. In general, most of recent real time traffic simulators are based on the macroscopic model because the macroscopic model replicates the average traffic behavior in terms of observable variables such as (time–space) flow and speed at a relatively fast computational time. Like other simulation models, an important aspect of the real time macroscopic simulator is to calibrate the model parameters online. The most conventional way of the online calibration is to add a random walk to the parameters to constitute an augmentation of the traffic variables and the model parameters to be estimated. Actually, this method allows the parameters to vary at every time step and, therefore, describes the adaptation of the model to the prevailing traffic conditions. However, it has been reported that the use of the random walk results in a loss of information and an increase of the covariance of parameters, which consequently leads to posteriors that are far more diffuse than the theoretical posteriors for the true parameters. To this end, this article puts forward a Kernel density estimation technique in the calibration process to handle the covariance issue and to avoid the information loss. The Kernel density estimation technique is embedded in the particle filter algorithm, which is extended to the calibration problems. The proposed framework is investigated using real‐life data collected in a freeway in England.     

Increasing the Capacity of the Intersection Downstream of the Freeway Off‐Ramp Using Presignals

High volume from urban freeway off‐ramps coupled with extensive traffic weaving and limited capacity at downstream intersections create major bottlenecks in urban road networks. This article presents an integrated design model to eliminate traffic weaving and to maximize the section's overall capacity by using the presignal and sorting area concept. The selection of movements controlled by the presignal, the layout of the section, and the signal timing are optimized in a uniform framework by a mixed‐integer nonlinear program model. The mathematical model was linearized and solved using the standard branch‐and‐bound technique. Extensive numerical analysis and a case study validate the effectiveness of the proposed integrated model in improving capacity with the comparison of conventional design under various geometric configuration and traffic demand pattern scenarios. The proposed model has promising application at locations where the queuing space is long enough and the number of exit lanes is enough to receive the traffic stream from the sorting area.     

Methodology for spatio‐temporal predictions of traffic counts across an urban road network and generation of an on‐road greenhouse gas emission inventory

On‐road emission inventories in urban areas have typically been developed using traffic data derived from travel demand models. These approaches tend to underestimate emissions because they often only incorporate data on household travel, not including commercial vehicle movements, taxis, ride hailing services, and other trips typically underreported within travel surveys. In contrast, traffic counts embed all types of on‐road vehicles; however, they are only conducted at selected locations in an urban area. Traffic counts are typically spatially correlated, which enables the development of methods that can interpolate traffic data at selected monitoring stations across an urban road network and in turn develop emission estimates. This paper presents a new and universal methodology designed to use traffic count data for the prediction of periodic and annual volumes as well as greenhouse gas emissions at the level of each individual roadway and for multiple years across a large road network. The methodology relies on the data collected and the spatio‐temporal relationships between traffic counts at various stations; it recognizes patterns in the data and identifies locations with similar trends. Traffic volumes and emissions prediction can be made even on roads where no count data exist. Data from the City of Toronto traffic count program were used to validate the output of various algorithms, indicating robust model performance, even in areas with limited data.     

Evaluation of Information Applications of a Self‐Organizing Distributed Traffic Information System for a Large‐Scale Real‐World Traffic Network

Abstract: This article presents an evaluation of the system performance of a proposed self‐organizing, distributed traffic information system based on vehicle‐to‐vehicle information‐sharing architecture. Using microsimulation, several information applications derived from this system are analyzed relative to the effectiveness and efficiency of the system to estimate traffic conditions along each individual path in the network, to identify possible incidents in the traffic network, and to provide rerouting strategies for vehicles to escape congested spots in the network. A subset of vehicles in the traffic network is equipped with specific intervehicle communication devices capable of autonomous traffic surveillance, peer‐to‐peer information sharing, and self‐data processing. A self‐organizing traffic information overlay on the existing vehicular roadway network assists their independent evaluation of route information, detection of traffic incidents, and dynamic rerouting in the network based both on historical information stored in an in‐vehicle database and on real‐time information disseminated through intervehicle communications. A path‐based microsimulation model is developed for these information applications and the proposed distributed traffic information system is tested in a large‐scale real‐world network. Based on simulation study results, potential benefits both for travelers with such equipment as well as for the traffic system as a whole are demonstrated.     

An Approach to Dynamical Classification of Daily Traffic Patterns

This article proposes a prototype of an urban traffic control system based on a prediction‐after‐classification approach. In an off‐line phase, a repository of traffic control strategies for a set of (dynamic) traffic patterns is constructed. The core of this stage is the k‐means algorithm for daily traffic pattern identification. The clustering method uses the input attributes flow, speed, and occupancy and it transforms the dynamic traffic data at network level in a pseudo‐covariance matrix, which collects the dynamic correlations between the road links. A desirable number of traffic patterns is provided by Bayesian Information Criterion and the ratio of change in dispersion measurements. In an on‐line phase, the current daily traffic pattern is predicted within the repository and its associated control strategy is implemented in the traffic network. The dynamic prediction scheme is constructed on the basis of an existing static prediction method by accumulating the trials on set of patterns in the repository. This proposal has been assessed in synthetic and real networks testing its effectiveness as a data mining tool for the analysis of traffic patterns. The approach promises to effectively detect the current daily traffic pattern and is open to being used in intelligent traffic management systems.     

Collecting comprehensive traffic information using pavement vibration monitoring data

Traffic data is essential for intelligent traffic management and road maintenance. However, the enormous effort used for data collection and analysis, combined with conventional approaches for traffic monitoring, is inefficient due to its high energy consumption, high cost, and the nonlinear relationships among various factors. This article proposes a new approach to obtain traffic information by processing raw data on pavement vibration. A large amount of raw data was collected in real time by deploying a vibration‐based in‐field pavement monitoring system. The data was processed with an efficient algorithm to achieve the monitoring of the vehicle speed, axle spacing, driving direction, location of the vehicle, and traffic volume. The vehicle speed and axle spacing were back‐calculated from the collected data and verified with actual measurements. The verification indicated that a reasonable precision could be achieved using the developed methods. Vehicle types and vehicles with an abnormal weight were identified by a three‐layer artificial neural network and the k‐means++ cluster analysis, respectively, which may help law enforcement in determining on an overweight penalty. A cost and energy consumption estimation of an acceleration sensing node is discussed. An upgraded system with low cost, low energy consumption, and self‐powered monitoring is also discussed for enabling future distributed computing and wireless application. The upgraded system might enhance integrated pavement performance and traffic monitoring.     

Network‐wide traffic speed forecasting: 3D convolutional neural network with ensemble empirical mode decomposition

Accurate traffic speed forecasting is one of the most critical tasks in proactive traffic management and the deployment of advanced traveler information systems. This paper proposes a hybrid forecasting approach named DeepEnsemble by integrating the three‐dimensional convolutional neural network (3D CNN) with ensemble empirical mode decomposition (EEMD). There are four steps in this hybrid approach. First, EEMD is adopted to decompose the complex traffic speed time series data with noise into several intrinsic mode functions (IMFs) and a residue. Second, a three‐dimensional tensor is established and fed into 3D CNN for prediction. Third, the output of 3D CNN prediction is obtained by a linear combination of the results of all components. Finally, the 3D CNN prediction output, external features, and historical features are fused to predict the network‐wide traffic speed simultaneously. The proposed DeepEnsemble approach is tested on the three‐month traffic speed series data of a real‐world large‐scale urban expressway network with 308 traffic flow detectors in Beijing, China. The experimental results indicate that DeepEnsemble outperforms the state‐of‐the‐art network‐wide traffic speed forecasting models. 3D CNN learns temporal, spatial, and depth information better than 2D CNN. Moreover, forecasting accuracy can be improved by employing EEMD. DeepEnsemble is a promising model with scalability and portability for network‐wide traffic speed prediction and can be further extended to conduct traffic status monitoring and congestion mitigation strategies.     

Increasing Signalized Intersection Capacity with Unconventional Use of Special Width Approach Lanes

Heavy traffic volume coupled with insufficient capacity due to limited space cause most of traffic congestion at urban signalized intersections. This article presents an innovative design to increase the capacity of heavily congested intersections by using the special width approach lane (SWAL), which consists of two narrow approach lanes that are dynamically utilized by either two passenger cars or a heavy vehicle (e.g., buses or trucks) depending on the composition of traffic. The impact of the SWAL on the saturation flow rate is quantified and validated, followed by an optimization model for best geometric layout and signal timing design with the presence of the SWAL. The optimization model is formulated as a mixed‐integer‐linear‐program for intersection capacity maximization which can be efficiently solved by the standard branch‐and‐bound technique. Results of extensive numerical analyses and case studies show the effectiveness of SWAL to increase intersection capacity, indicating its promising application at intersections with very limited space that prevents the addition of separate lanes.     

Passing Rates to Measure Relaxation and Impact of Lane‐Changing in Congestion

Abstract: Passing rate measurements of backward‐moving kinematic waves in congestion are applied to quantify two traffic features; a relaxation phenomenon of vehicle lane‐changing and impact of lane‐changing in traffic streams after the relaxation process is complete. The relaxation phenomenon occurs when either a lane‐changer or its immediate follower accepts a short spacing upon insertion and gradually resumes a larger spacing. A simple existing model describes this process with few observable parameters. In this study, the existing model is reformulated to estimate its parameter using passing rate measurements. Calibration results based on vehicle trajectories from two freeway locations indicate that the revised relaxation model matches the observation well. The results also indicate that the relaxation occurs in about 15 seconds and that the shoulder lane exhibits a longer relaxation duration. The passing rate measurements were also employed to quantify the postrelaxation impact of multiple lane‐changing maneuvers within a platoon of 10 or more vehicles in queued traffic stream. The analysis of the same data sets shows that lane‐changing activities do not induce a long‐term change in traffic states; traffic streams are perturbed temporarily by lane‐changing maneuvers but return to the initial states after relaxations.     

Reconstructing the Traffic State by Fusion of Heterogeneous Data

Abstract: We present an advanced interpolation method for estimating smooth spatiotemporal profiles for local highway traffic variables such as flow, speed and density. The method is based on the “adaptive smoothing method” which takes as input stationary detector data as typically collected by traffic control centers. We generalize this method to allow for fusion with floating car data or other traffic information. The resulting profiles display transitions between free and congested traffic in great detail, as well as fine structures such as stop‐and‐go waves. We establish the accuracy and robustness of the method and demonstrate three potential applications: (1) compensation for gaps in data caused by detector failure; (2) separation of noise from dynamic traffic information; and (3) the fusion of floating car data with stationary detector data.     

An Approach to Dynamic Classification of Traffic Flow Patterns

Abstract: Analyses on the dynamics of traffic flow, ranging from intersection flows to network‐wide flow propagation, require accurate information on time‐varying local traffic flows. To effectively determine the flow performance measures and consequently the congestion indicators of segmented road pieces, the ability to process such data in real time is out of the question. In this article, a dynamic approach to specify flow pattern variations is proposed mainly concentrating on the incorporation of neural network theory to provide real‐time mapping for traffic density simultaneously in conjunction with a macroscopic traffic flow model. To deal with the noise and the wide scatter of raw flow measures, a filtering is applied prior to modeling processes. Filtered data are dynamically and simultaneously input to processes of neural density mapping and traffic flow modeling. The classification of flow patterns over the fundamental diagram, which is dynamically plotted with the outputs of the flow modeling subprocess, is obtained by considering the density measure as a pattern indicator. Densities are mapped by selected neural approximation method for each simulation time step considering explicitly the flow conservation principle. Simultaneously, mapped densities are matched over the fundamental diagram to specify the current corresponding flow pattern. The approach is promising in capturing sudden changes on flow patterns and is open to be utilized within a series of intelligent management strategies including especially nonrecurrent congestion effect detection and control.     

A Dantzig‐Wolfe Decomposition‐Based Heuristic for Off‐line Capacity Calibration of Dynamic Traffic Assignment

Abstract: One of the critical elements in considering any real‐time traffic management strategy requires assessing network traffic dynamics. Traffic is inherently dynamic, since it features congestion patterns that evolve over time and queues that form and dissipate over a planning horizon. Dynamic traffic assignment (DTA) is therefore gaining wider acceptance among agencies and practitioners as a more realistic representation of traffic phenomena than static traffic assignment. Though it is imperative to calibrate the DTA model such that it can accurately reproduce field observations and avoid erroneous flow predictions when evaluating traffic management strategies, DTA calibration is an onerous task due to the large number of variables that can be modified and the intensive computational resources required. To compliment other research on behavioral and trip table issues, this work focuses on DTA capacity calibration and presents an efficient Dantzig‐Wolfe decomposition‐based heuristic that decomposes the problem into a restricted master problem and a series of pricing problems. The restricted master problem is a capacity manipulation problem, which can be solved by a linear programming solver. The pricing problem is the user optimal DTA which can be optimally solved by an existing combinatorial algorithm. In addition, the proposed set of dual variable approximation techniques is one of a very limited number of approaches that can be used to estimate network‐wide dual information in facilitating algorithmic designs while maintaining scalability. Two networks of various sizes are empirically tested to demonstrate the efficiency and efficacy of the proposed heuristic. Based on the results, the proposed heuristic can calibrate the network capacity and match the counts within a 1% optimality gap.     

基于网格的道路网交通量统计算法及其在VISUM中的实现

道路交通流量数据的记录、统计一般是基于路段的。在城市道路规划、建设与管理工作中,需要从区域的角度对道路交通相关数据进行统计,就需要一种简便的工具将基于路段的数据纳入不同区域范围的对象中。由此提出一种基于网格的道路网交通量统计算法,并在VISUM软件中通过编写脚本文件实现该算法的应用。该方法为路段数据提供更多的空间统计信息,可辅助用于城市道路网规划、建设、管理相关工作的决策。     

Exploring a Local Linear Model Tree Approach to Car‐Following

Because car‐following (CF) models are fundamental to replicating traffic flow they have received considerable attention over the last 50 years. They are in a continuous state of improvement due to their significant role in traffic microsimulations, intelligent transportation systems, and safety engineering models. This article uses the local linear model tree (LOLIMOT) approach to model driver's CF behavior to incorporate human perceptual imperfections into a CF model. This model defines some localities in the input space. These localities are fuzzy and have overlaps with each other. Specific models for each of the localities are then defined and combined in a fuzzy manner to predict the final output. The model was developed using real world dynamic data sets. Three different data sets were used for training, testing, and validating the model. The performance of the model was compared to a number of existing CF models. The results showed very close agreement between the real data and the LOLIMOT outputs.     

A Bayesian mixture model for short-term average link travel time estimation using large-scale limited information trip-based data

Accurate estimation and prediction of urban link travel times are important for urban traffic operations and management. This paper develops a Bayesian mixture model to estimate short-term average urban link travel times using large-scale trip-based data with partial information. Unlike typical GPS trajectory data, trip-based data from taxies or other sources provide limited trip level information, which only contains the trip origin and destination locations, trip travel times and distances, etc. The focus of this study is to develop a robust probabilistic short-term average link travel time estimation model and demonstrate the feasibility of estimating network conditions using large-scale trip level information. In the model, the path taken by each trip is considered as latent and modeled using a multinomial logit distribution. The observed trip data given the possible path set and the mean and variance of the average link travel times can thus be characterized using a finite mixture distribution. A transition model is also introduced to serve as an informative prior that captures the temporal and spatial dependencies of link travel times. A solution approach based on the expectation–maximization (EM) algorithm is proposed to solve the problem. The model is tested on estimating the mean and variance of the average link travel times for 30 min time intervals using a large-scale taxi trip dataset from New York City. More robust estimation results are obtained owing to the adoption of the Bayesian framework.     

Graph‐Based Modeling of Information Flow Evolution and Propagation under V2V Communications‐Based Advanced Traveler Information Systems

A vehicle equipped with a vehicle‐to‐vehicle (V2V) communications capability can continuously update its knowledge on traffic conditions using its own experience and anonymously obtained travel experience data from other such equipped vehicles without any central coordination. In such a V2V communications‐based advanced traveler information system (ATIS), the dynamics of traffic flow and intervehicle communication lead to the time‐dependent vehicle knowledge on the traffic network conditions. In this context, this study proposes a graph‐based multilayer network framework to model the V2V‐based ATIS as a complex system which is composed of three coupled network layers: a physical traffic flow network, and virtual intervehicle communication and information flow networks. To determine the occurrence of V2V communication, the intervehicle communication layer is first constructed using the time‐dependent locations of vehicles in the traffic flow layer and intervehicle communication‐related constraints. Then an information flow network is constructed based on events in the traffic and intervehicle communication networks. The graph structure of this information flow network enables the efficient tracking of the time‐dependent vehicle knowledge of the traffic network conditions using a simple graph‐based reverse search algorithm and the storage of the information flow network as a single graph database. Further, the proposed framework provides a retrospective modeling capability to articulate explicitly how information flow evolves and propagates. These capabilities are critical to develop strategies for the rapid flow of useful information and traffic routing to enhance network performance. It also serves as a basic building block for the design of V2V‐based route guidance strategies to manage traffic conditions in congested networks. Synthetic experiments are used to compare the graph‐based approach to a simulation‐based approach, and illustrate both memory usage and computational time efficiencies.     

Dynamic origin‐destination flow estimation using automatic vehicle identification data: A 3D convolutional neural network approach

Dynamic origin‐destination (OD) flow estimation is one of the most fundamental problems in traffic engineering. Despite numerous existing studies, the OD flow estimation problem remains challenging, as there is large dimensional difference between the unknown values to be estimated and the known traffic observations. To meet the needs of active traffic management and control, accurate time‐dependent OD flows are required to understand time‐of‐day traffic flow patterns. In this work, we propose a three‐dimensional (3D) convolution‐based deep neural network, “Res3D,” to learn the high‐dimensional correlations between local traffic patterns presented by automatic vehicle identification observations and OD flows. In this paper, a practical framework combining simulation‐based model training and few‐shot transfer learning is introduced to enhance the applicability of the proposed model, as continuously observing OD flows could be expensive. The proposed model is extensively tested based on a realistic road network, and the results show that for significant OD flows, the relative errors are stable around 5%, outperforming several other models, including prevalent neural networks as well as existing estimation models. Meanwhile, corrupted and out‐of‐distribution samples are generated as real‐world samples to validate Res3D's transferability, and the results indicated a 60% improvement with few‐shot transfer learning. Therefore, this proposed framework could help to bridge the gaps between traffic simulations and empirical cases.     

Predicting Motorway Traffic Performance by Data Fusion of Local Sensor Data and Electronic Toll Collection Data

Abstract: This article proposes data fusion from different sources to improve estimation and prediction accuracy of traffic states on motorways. This is demonstrated in two case studies on an intraurban and an interurban motorway section in Austria. Data fusion in this case combines local detector data and speed data from the Electronic Toll Collection (ETC) system for heavy goods vehicles (HGV). A macroscopic model for open motorway sections has been used to estimate passenger car and HGV density, applying a standard state‐space model and a linear Kalman filter. The resulting historical database of 4 months of speed‐density patterns has been used as a basis for pattern recognition. A nonparametric kernel predictor with memory length of 9 and 18 hours has been used to predict HGV speed for a prediction horizon of 15 minutes to 2 hours. Results show good overall prediction accuracy. Correlation analysis showed little bias of predicted speed for free flow and congested time intervals, whereas transition states between free flow and congestion were frequently biased. Prediction accuracy can be improved by applying a combination of different prediction methods. On the other hand, computational performance of the prediction has to be further improved prior to implementation in a traffic management center.