Competing risk mixture model and text analysis for sequential incident duration prediction

Predicting the duration of traffic incidents sequentially during the incident clearance period is helpful in deploying efficient measures and minimizing traffic congestion related to such incidents. This study proposes a competing risk mixture hazard-based model to analyze the effect of various factors on traffic incident duration and predict the duration sequentially. First, topic modeling, a text analysis technique, is used to process the textual features of the traffic incident to extract time-dependent topics. Given four specific clearance methods and the uncertainty of these methods when used during traffic incidents, the proposed mixture model uses the multinomial logistic model and parametric hazard-based model to assess the influence of covariates on the probability of clearance methods and on the duration of the incident. Subsequently, the performance of estimated mixture model in sequentially predicting the incident duration is compared with that of the non-mixture model. The prediction results show that the presented mixture model outperforms the non-mixture model.     

An optimal variable speed limits system to ameliorate traffic safety risk

Active Traffic Management (ATM) systems have been emerging in recent years in the US and Europe. They provide control strategies to improve traffic flow and reduce congestion on freeways. This study investigates the feasibility of utilizing a Variable Speed Limits (VSL) system, one key part of ATM, to improve traffic safety on freeways. A proactive traffic safety improvement VSL control algorithm is proposed. First, an extension of the METANET (METANET: A macroscopic simulation program for motorway networks) traffic flow model is employed to analyze VSL’s impact on traffic flow. Then, a real-time crash risk evaluation model is estimated for the purpose of quantifying crash risk. Finally, optimal VSL control strategies are achieved by employing an optimization technique to minimize the total crash risk along the VSL implementation corridor. Constraints are setup to limit the increase of average travel time and the differences of the posted speed limits temporarily and spatially. This novel VSL control algorithm can proactively reduce crash risk and therefore improve traffic safety. The proposed VSL control algorithm is implemented and tested for a mountainous freeway bottleneck area through the micro-simulation software VISSIM. Safety impacts of the VSL system are quantified as crash risk improvements and speed homogeneity improvements. Moreover, three different driver compliance levels are modeled in VISSIM to monitor the sensitivity of VSL effects on driver compliance. Conclusions demonstrated that the proposed VSL system could improve traffic safety by decreasing crash risk and enhancing speed homogeneity under both the high and moderate compliance levels; while the VSL system fails to significantly enhance traffic safety under the low compliance scenario. Finally, future implementation suggestions of the VSL control strategies and related research topics are also discussed.     

Identification of occupant posture using a Bayesian classification methodology to reduce the risk of injury in a collision

Recent studies have shown that smart restraint systems, which will recognize and then adapt to a specific collision and occupant combination, have a strong opportunity to significantly reduce occupant injuries during a traffic accident. As a step toward the development of these adaptive restraint systems, this study proposes a novel methodology for the classification of pre-crash occupant posture. Various occupant postures were simulated with a human model and the corresponding data was recorded using sensor models implemented in a mid-size car interior. The sensor data was then used to train two Bayesian classifiers which categorized an unknown occupant posture as one of nine predefined classes. The posture classifiers and a look-up table which contained optimized restraint laws for each class were combined to form catalog controllers for the restraint systems. The benefit of these restraint systems with catalog controllers vs. a restraint system optimized at a nominal posture was estimated by analyzing crash simulations with the occupant in 200 different postures. While the minimum error rate classifier showed the highest correct classification rate (90%), the Bayesian minimum risk classifier estimated the highest average injury reduction (21%). As expected, the highest injury reduction (up to 45%) was recorded for the posture classes closest to the windshield, whereas the lowest injury reduction was found for the classes closest to the nominal position. While the proposed restraint system with a catalog controller requires considerable “offline” computational effort, it is more versatile in terms of using complex human models and injury criteria and is much faster during the brief decision window available than recent “online” controllers proposed previously in literature.     

Prediction of travel time variability for cost-benefit analysis

Unreliable travel times cause substantial costs to travelers. Nevertheless, they are often not taken into account in cost-benefit analyses (CBA), or only in very rough ways. This paper aims at providing simple rules to predict variability, based on travel time data from Dutch highways. Two different concepts of travel time variability are used, which differ in their assumptions on information availability to drivers. The first measure is based on the assumption that, for a given road link and given time of day, the expected travel time is constant across all working days (rough information: RI). In the second case, expected travel times are assumed to reflect day-specific factors such as weather conditions or weekdays (fine information: FI). For both definitions of variability, we find that the mean travel time is a good predictor. On average, longer delays are associated with higher variability. However, the derivative of variability with respect to delays is decreasing in delays. It can be shown that this result relates to differences in the relative shares of observed traffic ‘regimes’ (free-flow, congested, hyper-congested) in the mean delay. For most CBAs, no information on the relative shares of the traffic regimes is available. A non-linear model based on mean travel times can then be used as an approximation.     

Contrasting the direct use of data from traffic radars and video-cameras with traffic simulation in the estimation of road emissions and PM hotspot analysis

This study investigates the effect of traffic volume and speed data on the simulation of vehicle emissions and hotspot analysis. Data from a microwave radar and video cameras were first used directly for emission modelling. They were then used as input to a traffic simulation model whereby vehicle drive cycles were extracted to estimate emissions. To reach this objective, hourly traffic data were collected from three periods including morning peak (6–9 am), midday (11–2 pm), and afternoon peak (3–6 pm) on a weekday (June 23, 2016) along a high-volume corridor in Toronto, Canada. Traffic volumes were detected by a single radar and two video cameras operated by the Southern Ontario Centre for Atmospheric Aerosol Research. Traffic volume and composition derived from the radar had lower accuracy than the video camera data and the radar performance varied by lane exhibiting poorer performance in the remote lanes. Radar speeds collected at a single point on the corridor had higher variability than simulated traffic speeds, and average speeds were closer after model calibration. Traffic emissions of nitrogen oxides (NOx) and particulate matter (PM₁₀ and PM_2.5) were estimated using radar data as well as using simulated traffic based on various speed aggregation methods. Our results illustrate the range of emission estimates (NO_x: 4.0–27.0 g; PM₁₀: 0.3–4.8 g; PM_2.5: 0.2–1.3 g) for the corridor. The estimates based on radar speeds were at least three times lower than emissions derived from simulated vehicle trajectories. Finally, the PM₁₀ and PM_2.5 near-road concentrations derived from emissions based on simulated speeds were two or three times higher than concentrations based on emissions derived using radar data. Our findings are relevant for project-level emission inventories and PM hot-spot analysis; caution must be exercised when using raw radar data for emission modeling purposes.     

A commuting model for the analysis of the impacts of a tramway project: application to the Lens area

Road congestion is not only an issue for major European urban agglomerations, but also for smaller ones. It is also the case of the Lens urban area, where car use is much higher than the average for medium-sized urban agglomerations in France. Local authorities put forward tramway projects to deal with the strongly negative externalities of congestion: travel time losses and pollution. To analyse its medium-term impact, we have developed a commuting with congestion model, inspired by the four-step traditional model, but with data made available from an origin–destination matrix. The results are encouraging but insufficient, and it is necessary to adopt supplementary measures in order to retrieve and justify the sizeable investments needed. Some measures prove to be very efficient, such as parking fees and urban tolls. Other measures, such as the subsidization of public transport, are partially efficient since they have an impact mainly on intra-urban commutes.     

Embedding risk attitude and decision weights in non-linear logit to accommodate time variability in the value of expected travel time savings

In recent years we have seen important extensions of logit models in behavioural research such as incorporation of preference and scale heterogeneity, attribute processing heuristics, and estimation of willingness to pay (WTP) in WTP space. With rare exception, however, a non-linear treatment of the parameter set to allow for behavioural reality, such as embedded risk attitude and perceptual conditioning of occurrence probabilities attached to specific attributes, is absent. This is especially relevant to the recent focus in travel behaviour research on identifying the willingness to pay for reduced travel time variability, which is the source of estimates of the value of trip reliability that has been shown to take on an increasingly important role in project appraisal. This paper incorporates, in a generalised non-linear (in parameters) logit model, alternative functional forms for perceptual conditioning (known as probability weighting) and risk attitude in the utility function to account for travel time variability, and then derives an empirical estimate of the willingness to pay for trip time variability-embedded travel time savings as an alternative to separate estimates of time savings and trip time reliability. We illustrate the richness of the approach using a stated choice data set for commuter choice between unlabelled attribute packages. Statistically significant risk attitude parameters and parameters underlying decision weights are estimated for multinomial logit and mixed multinomial logit models, along with values of expected travel time savings.     

Using the Bayesian updating approach to improve the spatial and temporal transferability of real-time crash risk prediction models

This study aimed to improve the spatial and temporal transferability of the real-time crash risk prediction models by using the Bayesian updating approach. Data from California’s I-880N freeway in 2002 and 2009 and the I-5N freeway in 2009 were used. The crash risk models for these three datasets are quite different from each other. The model parameters do not remain stable over time or space. The transferability evaluation results show that the crash risk models cannot be directly transferred across time and space. The updating results indicate that the Bayesian updating approach is effective in improving both spatial and temporal transferability even when new data are limited. The predictive performance of the updated model increases with an increase in the sample size of the new data. In addition, when limited new data are available, updating an existing model is better than developing a model using the limited new data.     

Estimating the willingness to pay and value of risk reduction for car occupants in the road environment

In recent years there has been a re-focus on the valuation of a statistical life from the ex post or human capital method to an ex ante willingness to pay (WTP) approach. This is in part a recognition that we may have been undervaluing the cost of fatalities and injuries to society associated with crashes, but also a strong belief in the need to focus on establishing the amount, ex ante, that individuals are willing to pay to reduce the risk of exposure to circumstances that might lead to death or degree of injury on the road network. This study has developed a framework in which to identify the degree of preference heterogeneity in willingness to pay by individuals who are drivers or passengers in cars to avoid being killed or injured. A stated choice experiment approach is developed. The empirical setting is a choice of route for a particular trip that a sample of individuals periodically undertakes in Australia. The particular trip is described in enough detail to provide the respondent with a familiar market environment, providing all the relevant background information required for making a decision. Mixed logit models are estimated to obtain the marginal (dis)utilities associated with each influence on the choice amongst the attribute packages offered in the stated choice scenarios. These estimates are used to obtain the WTP distributions for fatality and injury avoidance, which are then aggregated to obtain estimates of the value of risk reduction (VRR), of which the fatality class is also known as the value of a statistical life (VSL).     

Probabilistic analysis of the release of liquefied natural gas (LNG) tenders due to freight-train derailments

Liquefied natural gas (LNG) has emerged as a possible alternative fuel for freight railroads in the United States, due to the availability of cheap domestic natural gas and continued pursuit of environmental and energy sustainability. A safety concern regarding the deployment of LNG-powered trains is the risk of breaching the LNG tender car (a special type of hazardous materials car that stores fuel for adjacent locomotives) in a train accident. When a train is derailed, an LNG tender car might be derailed or damaged, causing a release and possible fire. This paper describes the first study that focuses on modeling the probability of an LNG tender car release incident due to a freight train derailment on a mainline. The model accounts for a number of factors such as FRA track class, method of operation, annual traffic density level, train length, the point of derailment, accident speed, the position(s) of the LNG tender(s) in a train, and LNG tender car design. The model can be applied to any specified route or network with LNG-fueled trains. The implementation of the model can be undertaken by the railroad industry to develop proactive risk management solutions when using LNG as an alternative railroad fuel.