Journal of Transportation Engineering, Part A: Systems

Train wheel failures account for disruptions of train operations and even a large portion of train derailments. Remaining useful life (RUL) of a wheelset measures how soon the next failure will arrive, and the failure type reveals how severe the failure will be. RUL prediction is a regression task, whereas failure type is a classification task. In this paper, the authors propose a multitask learning approach to jointly accomplish these two tasks by using a common input space to achieve more desirable results. A convex optimization formulation is developed to integrate least-squares loss and negative maximum likelihood of logistic regression as well as model the joint sparsity as the L2/L1 norm of the model parameters to couple feature selection across tasks. The experiment results show that the multitask learning method outperforms both the single-task learning method and Random Forest.

Because the important task of traffic signs is to provide drivers with valuable information, the replacement of ineffective signs leads to a safer and more efficient environment for road users. Previously, many researchers studied traffic signs from the perspective of the road user. However, research regarding the identification of factors contributing to sign degradation is far from complete. To fill this gap, this study examines a large number of possible explanatory variables that may affect a sign’s visual condition. A data integration strategy is proposed to combine a large traffic sign data set with location and climate information. The Random Forests model and Odds ratio were applied to analyze the mobile light detection and ranging (LiDAR) and digital photolog data and rank all of the contributing factors based on their importance to the sign visual condition. The results showed that the odds of sign failure for signs with mount height less than or equal to 2 m were between 1.55 and 1.72 times those of signs placed higher than 2 m. These findings may reflect the importance of snow frequency and vandalism factors. The findings also revealed that air pollutants were among the most important contributing factors to traffic sign deterioration. Based on the results, a sign inspection schedule is also proposed. The findings of this study provide transportation agencies with useful information in identifying traffic signs that are more likely to be degraded. This study also provides a basis for employing advanced data collection and integration methods to assess the performance of transportation systems with greater consistency and establish asset tracking and risk analysis plans, and thus improve the efficiency of the surface transportation systems by making informed decisions.

This paper proposes a bilevel optimization model for multiclass origin–destination (O–D) estimation using various types of data. The multiclass character of the model, a new feature and major contribution to the literature, is important because of increasing interest in simultaneous estimation of O–D tables for various classes of trucks and automobiles. The upper-level optimization is used to derive O–D table entries by minimizing the sum of squared differences between observations from different data sources and the predictions of those values. A probit model is assumed in the lower-level stochastic user equilibrium problem for flow prediction. Extensive experiments have been performed on a test network with different types of link count sensors and turning movements. The tests verify the problem formulation and solution algorithm and offer important insights into the multiclass O–D estimation process with the different types of available data. Adding turning movement data can improve O–D estimation by 71%. Furthermore, classification information is interchangeable among different types of sensors.

The real-time monitoring of visibility for expressway traffic safety cannot be widely achieved due to the high cost of the traditional method. This paper proposes a novel method to monitor expressway visibility using two types of cameras, one normal and one infrared. The former is affected by visibility conditions when the latter is not. The concept of visibility level (VL) is proposed correspondingly, based on a comparison model of normal and infrared images. The mapping relation of the ratio of two kinds of image edge pixels to the decrease of visual distance is determined, establishing the VL model. Camera correction coefficients and temperature correction coefficients are proposed to make the model more reasonable and accurate. The model is validated in experiments under foggy conditions. The results show that the average error of the VL model is 10%. The research may enable practical implementation on roadsides in the near future to determine visibility for the benefit of expressway safety.

This study focuses on both vehicle kinematic parameters (speed and acceleration) and behavior parameters (critical interval and follow-up time) of drivers at turbo-roundabouts. Empirical evaluations of such parameters can be helpful in calibrating traffic microsimulation models or assigning behavior parameters to closed-form capacity models in turbo-roundabouts (gap-acceptance capacity models) and are also related to evaluation of vehicles pollutant emissions. The research was based on traffic process observed in the first turbo-roundabout implemented in the city of Maribor in Slovenia. In 2016 a great number of traffic samples were taken with high-frame-rate video recordings [$>50$ frames per second (fps)]. All vehicle trajectories were obtained with the methods and algorithms typical of the digital imageprocessing technique (DIP) by filtering the discrete signal of vehicle trajectories $f(t{)}_i$ with wavelet analysis. The research results showed that vehicle speeds on entry lanes are rather moderate (below $25\mathrm{km}/\mathrm{h}$, 15 m prior to the Yield line), whereas accelerations usually have values inferior to $2\mathrm{m}/{\mathrm{s}}^2$ on arm lanes and to $1.5\mathrm{m}/{\mathrm{s}}^2$ on ring lanes. The critical intervals $t_c$ [distributed according to an Erlang probability distribution function (PDF)] and follow-up headways $t_f$ (distributed according to an inverse Gaussian PDF) have, on the other hand, values in ranges of $t_c=4.03–5.48\mathrm{s}$ and $t_f=2.52–2.71\mathrm{s}$, respectively, according to the right- or left-turn lane and to the major or minor entry in question.

Public infrastructure is facing many challenges resulting in capacity reduction capped with funding limitation. Thus, not only is significant funding required to maintain the infrastructure under the current challenges, but infrastructure maintenance investment decision-making practices also require assessment to plan for future improvements. The objective of this study is to assess the state of the practice of the decision-making entities involved in transportation infrastructure management in the United States at the district DOT level. The covered aspects include current decision-making considerations, evaluation of investment practices, and performance assessment practices. Sixty-three out of 244 approached districts/regions across 26 states responded to the survey. Major findings revealed that physical condition is typically a main decision parameter, whereas a majority of the districts are currently working under budgetary constraints having (on an average) 30% budget deficit. Alignments drawn between certain federal laws and findings identified few gaps in adoption of asset management concepts, consideration of lifecycle approach, and performance measurement with business and management perspective, recommended to be covered through awareness and training programs in collaboration with academia.

Level of service (LOS) is a measure recommended to evaluate the operational performance of roadway facilities. It is generally agreed that LOS is related to traffic safety at signalized intersections; however, this relationship has not been fully investigated. This study sought to examine the relationship between LOS and safety by looking at types of crashes on approaches to signalized intersections. Crash data were acquired from 164 four-legged signalized intersections and categorized by rear-end, left-turn, and total crashes based on vehicle maneuvers. It was observed that intersection approach LOS and crash type frequencies differed for morning peak, midday, and afternoon peak hours. This led to the development of two models: a grouped random parameter negative binomial model to estimate the LOS–safety relationship for total crashes across different time periods, and a bivariate grouped random parameter negative binomial model to estimate this relationship for rear-end and left-turn crashes. Considering that geometric design, signal control, traffic flow, and crash types vary across different approaches within any given intersection, the models were developed at the approach level. The frequency of total and rear-end crashes (most to least) were LOSs D, C, B, and A on signalized intersection approaches. Left-turn crashes were not significantly associated with LOS. The relationship of a specific LOS to rear-end and total crashes varied across different time periods. This study determined that an understanding of the LOS–safety relationship at the intersection approach level requires examining different crash types and different time periods.

The construction of exclusive bus lanes is a strategy for giving priority to buses and improving the efficiency of the urban transportation system, and for that purpose, center transit lanes are an important type of transit lane. Along the center transit lanes, the operation of intersections with bus stops becomes more complex because there are four kinds of travelers that need to be organized: buses, private vehicles, crossing pedestrians, and bus passengers. This paper presents a signal timing model for intersections with center transit lanes and bus stops to balance the trade-off between the operations of various users. The optimization model is formulated as a two-objective programming problem. In each subobjective, the weighted average person delay is used. The Pareto frontier of solutions is obtained by iterating all possible combinations of weights for the two objectives. For each combination of weights, the optimization problem becomes a single-objective programming problem. The performance of the proposed model is evaluated using a case study and extensive numerical analyses. A shorter cycle length is recommended to balance the trade-off between vehicles and pedestrians. For the geometric design, a near-side bus stop and larger central refuge island are recommended.

Unsignalized intersections in India essentially function as uncontrolled ones due to the high prevalence of indisciplined traffic maneuverability. Right-turning vehicles from major and minor roads undertake risky crossing maneuvers in front of right-of-way vehicles by disobeying the priority rules. At such situations, nonprioritized vehicles hinder the movements of the right-of-way vehicles. Because of this, a through moving vehicle plying on the major road needs to take any evasive action to avoid an imminent collision. When the through traffic on the major road has high conflicting speed, the criticality of conflict increases, resulting into a potential crash. This study is an attempt to develop a new proximal indicator by incorporating the conflicting speed of through moving vehicle with a conventional time-proximity conflict indicator postencroachment time. Four unsignalized intersections have been selected from the National Capital Region of India. A speed parameter termed as critical speed is calculated based on the concept of braking distance and used to identify the critical conflicts. More critical conflicts are observed when the conflicting speed of through vehicles is high. Among various crossing situations, the maximum proportions of critical conflicts are found between through moving powered two wheelers (PTWs) along the major road and right-turning heavy vehicles. The conflicting speed of through moving PTW is observed to be the highest among all vehicle categories. The appropriateness of the proposed method is verified using 5 years’ crash data for right-turn right angle and right-turn head-on collisions and finding a statistically significant relationship between these right-turn-related crashes and critical crossing conflicts.