Journal of Transportation Engineering, Part A: Systems

The prediction of future traffic conditions represents a main building block for traffic management. With the advent of multiple traffic and environmental sensors, diverse data for predictions are available and models may incorporate not only traffic data but also additional aspects such as local weather conditions. A review of the state-of-the art methods in short-term traffic forecasting presented in this paper reveals that machine learning (ML) algorithms from the field of deep learning are occasionally used for forecasts based on historical traffic data, but not for traffic predictions including exogenous factors. Weather conditions represent an exogenous factor, which, for example, may affect travel times. This paper investigates how short-term travel time predictions may be improved by applying deep recurrent neural networks that incorporate weather data. Therefore, two hypotheses are formulated. Hypothesis 1 tests the prediction quality of the recurrent neural network (RNN) models, long short-term memory (LSTM), and gated recurrent unit (GRU) compared to the autoregressive moving average (ARMA) prediction method. The respective results indicate that the RNN models using historical traffic data and weather data show significant improvement compared to the ARMA model using only historical traffic data. Hypothesis 2 tests the prediction quality of LSTM and GRU compared to ML-based forecast models already in place in the field of traffic predictions, namely k-nearest neighbor (kNN), support vector regression (SVR), and neural networks (NN). In this context, the LSTM and GRU models using historical traffic data and weather data show significant improvement compared to the models kNN, SVR, and NN that also consider weather data. Despite the results presented in this work, there is still further potential for improvement. Thus, further research focusing on hyperparameter tuning of the RNN algorithms and the optimized selection of (additional) input variables with significant influence on travel times can contribute to further improvements of the forecast quality.

In the context of Industry 4.0, the building information modeling (BIM) methodology is a powerful tool, that has gained its place in Architecture, Engineering and Construction. The use of BIM for transportation infrastructure has been increasing, although research has mainly been focusing on highways and bridges. Considering the advances in this technology, the application in infrastructure has a lot of advantages as well as reduction of design time, reduction of costs, more quality and precision, and clash detection. This paper presents some research gaps in this area a case study of a Brazilian railway design considering the integration of photogrammetry and conventional topography data with other disciplines of projects in BIM. The results show a good agreement in the interaction of the different disciplines and furthermore, a discussion of the challenges and suggestions for BIM deliverables in topographic studies for railway projects using BIM is presented.

Analyzing road risks and developing targeted countermeasures are essential for a safe and orderly traffic flow. However, previous intersection safety analyses were conducted based on crash data. Little research has been conducted on surrogate safety measures based on risky driving behavior. In this study, categorical boosting (CatBoost) and Shapley additive explanation (SHAP) were used to analyze the impact of features on traffic order using a set of multisource data that include roadway geometry, signal control, and land use. The traffic data for intersection entrances in Beijing were collected from navigation systems, field investigations, and application programming interfaces. The model results showed that CatBoost exhibits a prediction accuracy of 83.5%, a recall of 83.5%, and an $F_1$ score of 81.1%. Moreover, the importance, total effects, main effects, and interaction effects of influence factors were analyzed by using SHAP. It was found that the congestion index (CI) has significant negative effects on traffic order. A larger number of lanes and more electronic traffic control were found to have a positive effect on traffic order. Intersection entrances with three-phase signals or off-peak intersection entrances helped increase traffic order. Moreover, a high green ratio for through vehicles can reduce the positive impact of CI on traffic order when the value of CI is 1.1–1.4, and the signal control scheme with a high left-turn green ratio would result in a safe and orderly traffic flow. The results from this study can be used for further studies on improving traffic safety at signalized intersections.

To improve the operational service capability of flex-route transit under the condition of uncertain and inhomogeneous passenger travel demand, a dynamic meeting points optimization solution is proposed. This solution can set dynamic pickup and drop-off meeting points according to real-time travel demand. The flex-route transit operation mode is described in detail and is divided into a two-stage optimization problem. Mixed integer programming is employed to formulate the problem with a twofold objective—(1) serve as many requests as possible; and (2) minimize the time cost of the accepted passengers. An improved memetic solution algorithm is proposed for the model constructed. Computational results based on real-world cases show that the dynamic clustering meeting points solution breaks the space constraints of the original pickup and drop-off points of passengers and can significantly reduce the proportion of passenger reservation requests rejected without increasing operation cost.

This paper presents the results of a comprehensive experimental study on the flexural performance of ballastless discrete precast slab tracks with two alternative reinforcement types. These are the precast slab tracks produced using ordinary reinforcement [nonpost-tensioned (NPT)] and the precast slab tracks produced using post-tensioned (PT) reinforcement. The study aims to compare the flexural performances of NPT- and PT-type slab tracks using three-point bending tests, even though they continuously sit on the ground with different rigidities in real-life applications. The tests performed on four full-scale slab tracks in the longitudinal and transversal directions resulted in the first cracking and the ultimate loads were greater in PT specimens. Also, the maximum crack widths and residual displacements of PT specimens were smaller. Although the aforementioned performance parameters of NPT-type slab tracks are relatively behind those of PT-type slab tracks, their overall response is still acceptable, with the advantages of easier production and lower cost.

With the ambitious goal of zero deaths but growing traffic and congestion at highway–rail grade crossings (HRGCs), grade separation of these conflicting points has become an increasing strategy that improves railway safety but also increases network capacity and enhances livability attributes. Meanwhile, the engineering costs of grade separation are expensive and can vary considerably, which results in difficulties in the rational selection of such projects. This paper implemented the data envelopment analysis (DEA) for the benchmarking of high-risk HRGCs for grade separation considerations in Oklahoma. The top 1% of HRGCs were identified based on the New Hampshire Hazard Index (NHI) rankings and used as the decision-making units (DMUs) for DEA. The outputs of the DEA process included safety benefits, travel time savings, vehicle operating cost savings, and environmental benefits, and the input was the estimated cost of a grade separation project. The DEA efficiency scores of the DMUs were calculated, and nine HRGCs were identified on the frontiers with efficiency scores of 1.0. To further discriminate the DEA-efficient HRGCs, the enhanced DEA (E-DEA), which combines self-evaluation and cross-evaluation methods, was performed and the most efficient crossing was decreased to four with the E-DEA scores of 1.0. This study established a hierarchical framework combining the NHI, DEA, and E-DEA approaches, available data sets, and decision-making tools for the selection of the most efficient investment alternatives for highway–rail grade separation.

Drones have recently become a new tool in railway inspection and monitoring (RIM) worldwide, but there is still a lack of information about the specific benefits and costs. This study conducts a systematic literature review (SLR) of the applications, opportunities, and challenges of using drones for RIM. The SLR technique yielded 47 articles filtered from 7,900 publications from 2014 to 2022. The SLR found that key motivations for using drones in RIM are to reduce costs, improve safety, save time, improve mobility, increase flexibility, and enhance reliability. Nearly all the applications fit into the categories of defect identification, situation assessment, rail network mapping, infrastructure asset monitoring, track condition monitoring, and obstruction detection. The authors assess the open technical, safety, and regulatory challenges. The authors also contribute a cost analysis framework, identify factors that affect drone performance in RIM, and offer implications for new theories, management, and impacts to society.

Sensor deployment is important to the detection performance of traffic accidents on freeways. However, the temporally and spatially varying effects of accident risks are seldom considered in sensor deployment, which can induce a decrease in accident detection accuracy and detection timeliness. To address this problem, this paper proposes an optimal deployment method of sensors according to the temporal-spatial effect of accidents on traffic flow. Considering the uncertainty of traffic accidents, the temporal-spatial distribution of accident risk is estimated according to historical accident data. Then the kinematic wave model and Van Aerder model are introduced to analyze the temporal-spatial effect of accidents on traffic flow. To obtain the optimal locations of sensors, the sensor deployment problem is converted into a coverage problem on accident risk in time and space, and an optimal deployment model of sensors is built by considering the temporal-spatial effect of accident risk, the coverage rate on accident risk, the detection timeliness of the accident, and deployment cost. The particle swarm optimization algorithm is introduced to solve it. A case study is carried out to validate the proposed method using consistency, sensitivity, reliability, and comparison experiments. The results show that the proposed method is effective and reliable under different traffic volumes, and it outperforms the uniform spacing method in traffic accident detection.

Recently, roadside lidar sensors have been adopted as a reliable measure to extract high-resolution vehicle trajectory data from the field. The trajectory-level data can be extracted from roadside lidar detection using a series of data processing algorithms such as background filtering, object clustering, object classification, and object tracking. However, the results from current methods are associated with trajectory fluctuations, indicating that vehicle trajectory optimization remains a challenge. Previous studies used traditional Kalman filters to optimize vehicle trajectories, but there remains room for improvement in the smoothing effect. This paper addresses the issue by presenting an effective method for trajectory smoothing using the adaptive Kalman filter. The proposed method demonstrates two significant highlights: a multipoint matching algorithm for velocity estimation, and a robust adaptive strategy for Kalman filter. The performance of the proposed method was found to be satisfactory after evaluation using field data collected at a site in Reno, Nevada. Additionally, the method can be implemented under simple prior assumptions, making it user-friendly for real-world applications.

Abstract

Bus rapid transit (BRT) systems have generated a tremendous positive change in road safety and public transport operations. However, the literature reveals the need for a quantitative preventive risk evaluation from a microscopic approach, integrated with outcomes from accident-based and behavioral research, which is the primary purpose of this paper. Its scope includes five components: (1) a novel risk diagnosis and mapping for three selected signalized intersections of the BRT system of (Columbia) City, using an event-based exposure definition of encounters, the Swedish Traffic Conflict Technique, and T-Analyst. (2) Cross-comparison of risk outcomes between intersections, in search for risk factors. (3) Validation of encounters against traffic volumes. (4) Exploration of the relationship between the number of observed conflicts and the number of reported accidents, using an American predictive model. (5) Discussion of the microscopic findings against the behavior declared locally and the limitations of the literature. As a result, the method provides detailed risk verdicts from 61 h of video data. BRT turning buses, and risky pedestrian behavior, are risk factors. Many pedestrians crossed intersections, taking risks without evident cause. The findings provide evidence for the hypothesis that such behavior is caused by several sources of overstimulation. In conclusion, encounters define the conflict frequency, overcoming the limitations of aggregate indicators. Traffic volumes do not fully explain the risk because most events were due to disobedience. The number of reported accidents was much higher than expected, considering the number of conflicts observed as evidence of aggressiveness. Cars and taxicabs drivers declared risky attitudes coinciding with our observations. BRT drivers’ behavior was acquiescent in most of the events observed, because the characteristics of BRT infrastructure. The findings comparison between signalized intersections and roundabouts allows for validating the methodology transferability.

Practical Applications

The preventive approach herein proposed for microscopic traffic risk assessment, based on the traffic conflicts/encounters ratio, objectively provides detailed risk outcomes for an entire intersection or particular areas. The preventive risk ratio overcomes the limitations of the traditional indicator (crashes/exposure) by using a traffic conflicts technique, i.e., conflicts as surrogates for crashes, and an event-based exposure definition of encounters. This methodology allows practitioners to hypothesize the observed events’ causality and formulate countermeasures based on the exposure or the risk. A number of encounters and conflicts involved pedestrians crossing intersections at risk, without an apparent cause. Hence, this situation allows for identifying possible overstimulation sources for pedestrians, as a causality hypothesis. The proposed method helps to reduce traffic risk by verifying and rescheduling clearance times at signalized intersections according to the lengths of the BRT system vehicles. Moreover, traffic signs and sound signals reinforce the information about the imminent danger in case of recurrent distracted pedestrians at traffic lights. The process is also valid in roundabouts with mixed traffic, applying to any intersection.

The estimation of the number of passenger boardings at every bus stop along an itinerary was solved using a Bayesian method. The method employs probability distributions of Wi-Fi probe requests. Any available technique capable of uniquely identifying probe requests made by onboard Wi-Fi devices can be used for obtaining the samples needed for the estimation. Probability distributions of time intervals between consecutive probe requests provide the information needed for generating the probability density function of boarding amounts at each stop. A case study of synthetic probe requests randomly generated for a bus line comprised of 17 stops indicated the adequacy of the method. Its efficacy increases with the grouping of nearby stops in boarding zones. It is straightforward to modify the method for the estimation of alightings, origin–destination matrixes, and bus loads.

Existing studies estimate the economic value (EcV) of network reliability and robustness separately. In this study, a model is proposed for estimating the EcV of reserve capacity in a rail transit network (RTN), which consists of: (1) the reduction of passengers’ total generalized travel cost (GTC) in normal operations, (2) the EcV of reliability enhancement in normal operations, and (3) the EcV of robustness enhancement when disturbances occur. The standard deviation of travel time (SDT) is used to measure travel time reliability (TTR). The reserve capacity helps to reduce passengers’ total GTC and the SDT during normal operations. The EcV of reliability enhancement equals the EcV converted from the reduction in SDT when the RTN operates normally. The EcV of robustness enhancement is the difference of passengers’ total GTC with and without reserve capacity when disturbances occur. To optimize reserve capacity, a model is developed to maximize its net benefit, equaling its EcV minus its cost, which is then solved with a quantum genetic algorithm. A case study of Chengdu’s RTN shows the effectiveness of the model with promising numerical results. The research findings show that the value of reserve capacity is severely underestimated without considering the EcV of reliability and robustness. The study can guide policymakers and operators in quantifying the EcV of reliability and robustness enhancement when expanding an RTN’s capacity.

Defects present in the wheel-rail system will seriously affect ride comfort and endanger driving safety. This study aims to identify slight changes in the vibrations caused by the geometric defects of the wheel-track system from the time-frequency characteristics of rail vibration signals, in order to analyze the wheel-track dynamic system for the purpose of defect testing. The fast kurtogram (FK) method is introduced to describe the vibration characteristics in different frequency bands caused by wheel-rail interaction, and capture abnormal vibrations from the characteristics of periodic pulse vibrations. To verify the effectiveness of the proposed method, in situ measured data were analyzed and the similarities and differences of vibration signals of the wheel-rail system when wheels passed over different track structural defects were compared. A coupled vehicle-track dynamic model of a train formation was established, and the abnormal vibration behavior of different wheelsets with polygonal wear when passing over a joint was simulated. It was found that this method can locate a faulty wheel with mixed defects. After processing the field data, the abnormal vibration signals were identified as coming from back-of-flange contact, demonstrating that the proposed method was capable of identifying abnormal wheel-rail dynamic responses. It was found that the FK method can be used to reveal the similarities and differences of vibration responses of different track defects, and can be used to evaluate the transient vibration of wheel-rail contact to locate faulty wheels with mixed defects.