A Methodology for estimating traffic safety improvements at intersections

Intersection accidents represent a significant proportion of overall motor vehicle accidents. More accurate estimates of the actual effectiveness of intersection safety improvements are required. This study develops an improved methodology for post-implementation evaluation of safety countermeasures at intersections. Accidents are random, rarely occurring events. For a given time period, this leads to random fluctuations in accident frequencies, which suggests that statistical analysis employing confidence intervals, rather than point estimates, is required. Two technical problems complicate this treatment of accident occurrence as a random variable. The first problem is that identifying of hazardous locations is generally based on above-average accident frequency during the most recent period(s) for which data is available. The second problem arises from changes in external factors such as traffic volume, motor vehicle safety standards, etc., during the period of analysis, which may also affect traffic safety. A “combined” approach which addresses these technical issues is developed. Empirical Bayesian methodology is combined with regression techniques to derive a more accurate measure of the effect of safety treatments. An important consideration is the derivation of the variance of this measure, so that appropriate confidence intervals may be constructed. The approach is then applied to a sample of locations that underwent treatment by the Massachusetts Department of Public Works (MDPW). We compare our results to those which might be obtained using alternative methodologies that correct for neither or only one of the technical problems. We also illustrate how preliminary conclusions may be drawn regarding the effectiveness of broad categories of treatments, and how individual sites requiring further investigation may be identified.     

Modelling traffic flows and estimating road travel times in transportation network under dynamic disturbances

Transportation - Traffic congestion is a common phenomenon in road transportation networks, especially during peak hours. More accurate prediction of dynamic traffic flows is very important for...     

Comparing traffic state estimators for mixed human and automated traffic flows

This article addresses the problem of modeling and estimating traffic streams with mixed human operated and automated vehicles. A connection between the generalized Aw Rascle Zhang model and two class traffic flow motivates the choice to model mixed traffic streams with a second order traffic flow model. The traffic state is estimated via a fully nonlinear particle filtering approach, and results are compared to estimates obtained from a particle filter applied to a scalar conservation law. Numerical studies are conducted using the Aimsun micro simulation software to generate the true state to be estimated. The experiments indicate that when the penetration rate of automated vehicles in the traffic stream is variable, the second order model based estimator offers improved accuracy compared to a scalar modeling abstraction. When the variability of the penetration rate decreases, the first order model based filters offer similar performance.     

Bayesian inference for day-to-day dynamic traffic models

There is significant current interest in the development of models to describe the day-to-day evolution of traffic flows over a network. We consider the problem of statistical inference for such models based on daily observations of traffic counts on a subset of network links. Like other inference problems for network-based models, the critical difficulty lies in the underdetermined nature of the linear system of equations that relates link flows to the latent path flows. In particular, Bayesian inference implemented using Markov chain Monte Carlo methods requires that we sample from the set of route flows consistent with the observed link flows, but enumeration of this set is usually computationally infeasible.We show how two existing conditional route flow samplers can be adapted and extended for use with day-to-day dynamic traffic. The first sampler employs an iterative route-by-route acceptance–rejection algorithm for path flows, while the second employs a simple Markov model for traveller behaviour to generate candidate entire route flow patterns when the network has a tree structure. We illustrate the application of these methods for estimation of parameters that describe traveller behaviour based on daily link count data alone.     

Extreme value methods for estimating road traffic capacity

Common sense suggests that, at any point on a road network, there is an absolute limit to the volume of traffic which can be carried. But previous attempts to measure this “limiting capacity” have met with difficulties. First, there may not be enough vehicles to saturate the section of road under observation. Second, the flow may be constrained by a bottleneck upstream or downstream. Third, even under favourable conditions, the flows actually observed at saturation point tend to vary over a wide range, giving little clear indication as to what the value of the limiting capacity might be. In this paper, consideration is given to the variations in flow which occur over a time during normal traffic conditions, and to the characteristics of the extreme values which occur from time to time under these conditions. Two distinct types of statistical theory can be applied to extreme values. First, one can apply straight- forward probability theory, to predict the largest flows likely to be observed during a given period, assuming an idealised traffic stream with a known flow counting distribution. Second, one can attempt to deduce an upper limit from observed flow data using asymptotic methods of the kind which are frequently used in connection with meteorological and flood defense problems. Both methods are applied to a sample of 9000 flow values recorded at a site in London. Both methods are shown to fit the data reasonably well, but only the asymptotic method reveals a clear upper limit. Possible applications of the method are briefly discussed.     

A novel variable selection method based on frequent pattern tree for real-time traffic accident risk prediction

With the availability of large volumes of real-time traffic flow data along with traffic accident information, there is a renewed interest in the development of models for the real-time prediction of traffic accident risk. One challenge, however, is that the available data are usually complex, noisy, and even misleading. This raises the question of how to select the most important explanatory variables to achieve an acceptable level of accuracy for real-time traffic accident risk prediction. To address this, the present paper proposes a novel Frequent Pattern tree (FP tree) based variable selection method. The method works by first identifying all the frequent patterns in the traffic accident dataset. Next, for each frequent pattern, we introduce a new metric, herein referred to as the Relative Object Purity Ratio (ROPR). The ROPR is then used to calculate the importance score of each explanatory variable which in turn can be used for ranking and selecting the variables that contribute most to explaining the accident patterns. To demonstrate the advantages of the proposed variable selection method, the study develops two traffic accident risk prediction models, based on accident data collected on interstate highway I-64 in Virginia, namely a k-nearest neighbor model and a Bayesian network. Prior to model development, two variable selection methods are utilized: (1) the FP tree based method proposed in this paper; and (2) the random forest method, a widely used variable selection method, which is used as the base case for comparison. The results show that the FP tree based accident risk prediction models perform better than the random forest based models, regardless of the type of prediction models (i.e. k-nearest neighbor or Bayesian network), the settings of their parameters, and the types of datasets used for model training and testing. The best model found is a FP tree based Bayesian network model that can predict 61.11% of accidents while having a false alarm rate of 38.16%. These results compare very favorably with other accident prediction models reported in the literature.     

Estimating link travel time functions for heterogeneous traffic flows on freeways

Oversized vehicles, such as trucks, significantly contribute to traffic delays on freeways. Heterogeneous traffic populations, that is, those consisting of multiple vehicles types, can exhibit more complicated travel behaviors in the operating speed and performance, depending on the traffic volume as well as the proportions of vehicle types. In order to estimate the component travel time functions for heterogeneous traffic flows on a freeway, this study develops a microscopic traffic‐simulation based four‐step method. A piecewise continuous function is proposed for each vehicle type and its parameters are estimated using the traffic data generated by a microscopic traffic simulation model. The illustrated experiments based on VISSIM model indicate that (i) in addition to traffic volume, traffic composition has significant influence on the travel time of vehicles and (ii) the respective estimations for travel time of heterogeneous flows could greatly improve their estimation accuracy. Copyright © 2016 John Wiley & Sons, Ltd.     

Real-time network-wide traffic signal optimization considering long-term green ratios based on expected route flows

We propose a novel real-time network-wide traffic signal control scheme which is (1) applicable under modern data technologies, (2) flexible in response to variations of traffic flows due to its non-cyclic feature, (3) operable on a network-wide and real-time basis, and (4) capable of considering expected route flows in the form of long-term green time ratios for intersection movement. The proposed system has a two-level hierarchical architecture: (1) strategy level and (2) control level. Considering the optimal states for a long-term period found in the strategy level, the optimal signal timings for a short-term period are calculated in the control level which consists of two steps: (1) queue weight update and (2) signal optimization. Based on the ratio of the cumulative green time to the desired green time is the first step to update the queue weights, which are then used in the optimization to find signal timings for minimum total delay. A parametric queue weight function is developed, discussed and evaluated. Two numerical experiments were given. The first demonstrated that the proposed system performs effectively, and the second shows its capability in a real-world network.     

A framework for estimating traffic emissions: The development of Passenger Car Emission Unit

In this study, we develop a Passenger Car Emission Unit (PCEU) framework for estimating traffic emissions. The idea is analogous to the use of Passenger Car Unit (PCU) for modeling the congestion effect of different vehicle types. In this approach, we integrate emission modeling and cost evaluation. Different emissions, typically speed-dependent, are integrated as an overall cost via their corresponding external costs. We then develop a normalization procedure to obtain a general trend that is applicable for all vehicle types, which is used to derive a standard cost curve. Different vehicle types with different emission standards are then mapped to this standard cost curve through their corresponding PCEUs that are to be calibrated. Once the standard cost curve and PCEUs have been calibrated, to estimate the overall cost of emission for a particular vehicle, we only need to multiply the corresponding PCEU of that vehicle type to the standard cost curve. We apply this PCEU approach to Hong Kong and obtain promising results. Compared with the results obtained by the full-blown emission model COPERT, the approach achieves high accuracy but obviates tedious inputs typically required for emission estimation.     

A unified framework for estimating or updating origin/destination matrices from traffic counts

This paper presents an in-depth study of the methodology for estimating or updating origin-to-destination trip matrices from traffic counts. Following an analysis of the statistical foundation of the estimation and updating problems, various basic approaches are reviewed using a generic traffic assignment map. Computational issues related to specific assignment maps and estimation models for both road and transit networks are then discussed. Finally, additional insight into the relative performance of several estimators is provided by a set of test problems with varying input data.     

A hybrid deep learning based traffic flow prediction method and its understanding

Deep neural networks (DNNs) have recently demonstrated the capability to predict traffic flow with big data. While existing DNN models can provide better performance than shallow models, it is still an open issue of making full use of spatial-temporal characteristics of the traffic flow to improve their performance. In addition, our understanding of them on traffic data remains limited. This paper proposes a DNN based traffic flow prediction model (DNN-BTF) to improve the prediction accuracy. The DNN-BTF model makes full use of weekly/daily periodicity and spatial-temporal characteristics of traffic flow. Inspired by recent work in machine learning, an attention based model was introduced that automatically learns to determine the importance of past traffic flow. The convolutional neural network was also used to mine the spatial features and the recurrent neural network to mine the temporal features of traffic flow. We also showed through visualization how DNN-BTF model understands traffic flow data and presents a challenge to conventional thinking about neural networks in the transportation field that neural networks is purely a “black-box” model. Data from open-access database PeMS was used to validate the proposed DNN-BTF model on a long-term horizon prediction task. Experimental results demonstrated that our method outperforms the state-of-the-art approaches.     

A clustering based traffic flow prediction method with dynamic spatiotemporal correlation analysis

Transportation - There are significant spatiotemporal correlations among the traffic flows of neighboring road sections in the road network. Correctly identifying such correlations makes an...     

A dynamic Bayesian network model for real-time crash prediction using traffic speed conditions data

Traffic crashes occurring on freeways/expressways are considered to relate closely to previous traffic conditions, which are time-varying. Meanwhile, most studies use volume/occupancy/speed parameters to predict the likelihood of crashes, which are invalid for roads where the traffic conditions are estimated using speed data extracted from sampled floating cars or smart phones. Therefore, a dynamic Bayesian network (DBN) model of time sequence traffic data has been proposed to investigate the relationship between crash occurrence and dynamic speed condition data. Moreover, the traffic conditions near the crash site were identified as several state combinations according to the level of congestion and included in the DBN model. Based on 551 crashes and corresponding speed information collected on expressways in Shanghai, China, DBN models were built with time series speed condition data and different state combinations. A comparative analysis of the DBN model using flow detector data and a static Bayesian network model was also conducted. The results show that, with only speed condition data and nine traffic state combinations, the DBN model can achieve a crash prediction accuracy of 76.4% with a false alarm rate of 23.7%. In addition, the results of transferability testing imply that the DBN models are applicable to other similar expressways with 67.0% crash prediction accuracy.     

A computationally efficient two-stage method for short-term traffic prediction on urban roads

Abstract

Short-term traffic prediction plays an important role in intelligent transport systems. This paper presents a novel two-stage prediction structure using the technique of Singular Spectrum Analysis (SSA) as a data smoothing stage to improve the prediction accuracy. Moreover, a novel prediction method named Grey System Model (GM) is introduced to reduce the dependency on method training and parameter optimisation. To demonstrate the effects of these improvements, this paper compares the prediction accuracies of SSA and non-SSA model structures using both a GM and a more conventional Seasonal Auto-Regressive Integrated Moving Average (SARIMA) prediction model. These methods were calibrated and evaluated using traffic flow data from a corridor in Central London under both normal and incident traffic conditions. The prediction accuracy comparisons show that the SSA method as a data smoothing step before the application of machine learning or statistical prediction methods can improve the final traffic prediction accuracy. In addition, the results indicate that the relatively novel GM method outperforms SARIMA under both normal and incident traffic conditions on urban roads.     

A generic class of first order node models for dynamic macroscopic simulation of traffic flows

Node models for macroscopic simulation have attracted relatively little attention in the literature. Nevertheless, in dynamic network loading (DNL) models for congested road networks, node models are as important as the extensively studied link models. This paper provides an overview of macroscopic node models found in the literature, explaining both their contributions and shortcomings. A formulation defining a generic class of first order macroscopic node models is presented, satisfying a list of requirements necessary to produce node models with realistic, consistent results. Defining a specific node model instance of this class requires the specification of a supply constraint interaction rule and (optionally) node supply constraints. Following this theoretical discussion, specific macroscopic node model instances for unsignalized and signalized intersections are proposed. These models apply an oriented capacity proportional distribution of the available supply over the incoming links of a node. A computationally efficient algorithm to solve the node models exactly is included.     

A method for measuring accurate traffic density by aerial photography

Traffic density can be accurately measured by counting the number of vehicles within 1 km; however, it is often calculated between macroscopic traffic parameters using the fundamental equation because of difficulty of observing traffic density directly in the field. Measuring density in this way may be inaccurate and may bias the analysis because the relationship between these traffic parameters can vary across the study sites. The purpose of this study is to find a method for measuring traffic density from aerial photography that is easy and accurate, and for this purpose, we investigated whether the measuring length (i.e., the length of a section of roadway from which observations of traffic are simultaneously collected) can be shorter than 1 km and yet retain the same measured traffic density. We divided an aerial photograph into several 20‐m unit sections, counted the number of vehicles manually, and examined measured traffic density according to central limit theory. According to the results of this study, with the number of 20‐m unit sections for observing traffic density at 15 (the measuring length is 300 m), the measured traffic density was almost the same as the density of a representative section of 1 km. Copyright © 2014 John Wiley & Sons, Ltd.     

A queueing based traffic flow model

The assessment of uninterrupted traffic flow is traditionally based on empirical methods. We develop some analytic queueing models based on traffic counts and we model the behavior of traffic flows as a function of some of the most relevant determinants. These analytic models allow for parameterized experiments, which pave the way towards our research objectives: assessing what-if scenario’s and sensitivity analysis for traffic management, congestion control, traffic design and the environmental impact of road traffic (e.g. emission models). The impact of some crucial modeling parameters is studied in detail and links with the broader research objectives are given. We illustrate our results for a highway, based on counted traffic flows in Flanders.     

A robust method for estimating transit passenger trajectories using automated data

Development of an origin-destination demand matrix is crucial for transit planning. The development process is facilitated by automated transit smart card data, making it possible to mine boarding and alighting patterns on an individual basis. This research proposes a novel trip chaining method which uses Automatic Fare Collection (AFC) and General Transit Feed Specification (GTFS) data to infer the most likely trajectory of individual transit passengers. The method relaxes the assumptions on various parameters used in the existing trip chaining algorithms such as transfer walking distance threshold, buffer distance for selecting the boarding location, time window for selecting the vehicle trip, etc. The method also resolves issues related to errors in GPS location recorded by AFC systems or selection of incorrect sub-route from GTFS data. The proposed trip chaining method generates a set of candidate trajectories for each AFC tag to reach the next tag, calculates the probability of each trajectory, and selects the most likely trajectory to infer the boarding and alighting stops. The method is applied to transit data from the Twin Cities, MN, which has an open transit system where passengers tap smart cards only once when boarding (or when alighting on pay-exit buses). Based on the consecutive tags of the passenger, the proposed algorithm is also modified for pay-exit cases. The method is compared to previous methods developed by the researchers and shows improvement in the number of inferred cases. Finally, results are visualized to understand the route ridership and geographical pattern of trips.     

Platoon-based traffic flow model for estimating breakdown probability at single-lane expressway bottlenecks

This study investigates the mechanism of traffic breakdown and establishes a traffic flow model that precisely simulates the stochastic and dynamic processes of traffic flow at a bottleneck. The proposed model contains two models of stochastic processes associated with traffic flow dynamics: a model of platoon formation behind a bottleneck and a model of speed transitions within a platoon. After these proposed models are validated, they are applied to a simple one-way, one-lane expressway section containing a bottleneck, and the stochastic nature of traffic breakdown is demonstrated through theoretical exercises.