Hualiang Teng

APPLICATION OF WAVELET TECHNIQUE TO FREEWAY INCIDENT DETECTION

This paper presents an application of the wavelet technique to freeway incident detection because wavelet techniques have demonstrated superior performance in detecting changes in signals in electrical engineering. Unlike the existing wavelet incident detection algorithm, where the wavelet technique is utilized to denoise data before the data is input into an algorithm, this paper presents a different approach in the application of the wavelet technique to incident detection. In this approach, the features that are extracted from traffic measurements by using wavelet transformation are directly utilized in detecting changes in traffic flow. It is shown in the paper that the extracted features from traffic measurements in incident conditions are significantly different from those in normal conditions. This characteristic of the wavelet technique was used in developing the wavelet incident detection algorithm in this study. The algorithm was evaluated in comparison with the multi-layer feed-forward neural network, probabilistic neural network, radial basis function neural network, California and low-pass filtering algorithms. The test results indicate that the wavelet incident detection algorithm performs better than other algorithms, demonstrating its potential for practical application.

Optimization of railway operations using neural networks

Railroad operations involve complex switching and classification decisions that must be made in short periods of time. Optimization with respect to these decisions can be quite difficult due to the discrete and non-linear characteristics of the problem. The train formation plan is one of the important elements of railroad system operations. While mathematical programming formulations and algorithms are available for solving the train formation problem, the CPU time required for their convergence is excessive. At the same time, shorter decision intervals are becoming necessary given the highly competitive operating climates of the railroad industry. The field of Artificial Intelligence (AI) offers promising alternatives to conventional optimization approaches. In this paper, neural networks (an empirically-based AI approach) are examined for obtaining good solutions in short time periods for the train formation problem (TFP). Following an overview, and formulation of railroad operations, a neural network formulation and solution to the problem are presented. First a training process for neural network development is conducted followed by a testing process that indicates that the neural network model will probably be both sufficiently fast, and accurate, in producing train formation plans.

Detection-delay-based freeway incident detection algorithms

Even though incident detection algorithms are designed and implemented for quickly detecting incidents, the criterion of mean detection delay has hardly been well defined and utilized in developing and evaluating incident detection algorithms. In addition, most incident detection algorithms do not have an optimal property in terms of detection delay with respect to false alarm rate. In the study presented in this paper, the incident detection problem was formulated as an optimization problem. To implement the algorithm, called the CUSUM algorithm that was derived from the optimization formulation of the incident detection problem, a simplified procedure was developed. Based on this procedure, three varieties of the CUSUM algorithm were developed and tested based on real incident data against a newly defined criterion for mean detection delay. Selected incident detection algorithms were also compared with the CUSUM algorithms. The comparison demonstrates the superiority of the CUSUM algorithms against other selected algorithms in reducing detection delay while maintaining an acceptable detection rate.

Analysis of the Frequency and Severity of Rear-End Crashes in Work Zones

Objective: The objective of this study was to identify the factors that influence the frequency and severity of rear-end crashes in work zones because rear-end crashes represent a significant proportion of crashes that occur in work zones. Methods: Truncated count data models were developed to identify influencing factors on the frequency of read-end crashes in work zones and ordered probit models were developed to evaluate influencing factors on the severity of rear-end crashes in work zones. Results: Most of the variables identified in this study for these 2 models were significant at the 95 percent level. The statistics for models indicate that the 2 developed models are appropriate compared to alternative models. Conclusions: Major findings related to the frequency of rear-end crashes include the following: (1) work zones for capacity and pavement improvements have the highest frequency compared to other types of work zones; (2) work zones controlled by flaggers are associated with more rear-end crashes compared to those controlled by arrow boards; and (3) work zones with alternating one-way traffic tended to have more rear-end crashes compared to those with lane shifts. Major findings related to the severity of the rear-end crashes include the following: (1) rear-end crashes associated with alcohol, night, pedestrians, and roadway defects are more severe, and those associated with careless backing, stalled vehicles, slippery roadways, and misunderstanding flagging signals are less severe; (2) truck involvement and a large number of vehicles in a crash are both associated with increased severity, and (3) rear-end crashes that happened in work zones for bridge, capacity, and pavement are likely to be more severe than others.

Comparative Study of EMME/2 and QRS II for Modeling a Small Community

The availability of so many computer-based travel-demand forecasting models provides transportation planners with powerful and flexible tools in the modeling phase of their planning or traffic-impact studies, but it has confused users in the selection of an appropriate model for a particular study. It is commonly recognized that none of the existing travel-demand models is perfectly suited for all network scenarios and traffic conditions. A particular model that is strong in one application scenario may be weak in a different application scenario. A comparative study is presented of two widely used travel-demand forecasting models, EMME/2 and QRS II, for applications to a small community. A structural comparison is performed, and a real-world small network is modeled by EMME/2 and QRS II to identify specific features and limitations of each model. Areas for comparison include model structure, drawing of the network, data input, network modification, parameter calibration, and modeling output. The study does not recommend either model to transportation planners for a practical application to a small community. Instead, the study identifies the major differences and common features of two models, which can help planners understand what they can expect from a certain model when they choose to use it.

A Novel Description Method for Track Irregularity Evolution

Track Irregularity has a significant influence on the safety of train operation. Due to the fact that the extremely large number of factors affect track irregularity,it is challenging to find a concise yet effective mathematical method to describe the evolution of track irregularity. In this paper, inspection data generated by GJ-4 track inspection cars from Jinan Railway Bureau in China were analyzed to identify the characteristics of track irregularity changes common to different mileage points. Based on these characteristics, a multi-stage linear fitting model to describe the pattern of track irregularity evolution over time was developed. The availability of new inspection data will make the model revise itself. In this sense, the model is a machine learning model. Finally, inspection data from the Beijing-Shanghai Railway Line (Jing-Hu Line) were used to verify the model.

Modeling Signalized-Intersection Safety with Corner Clearance

Signalized intersections next to each other on the same arterial share some unobservable information, such as traffic flow and roadway characteristics. This study investigated the impact of access management techniques on crash counts at signalized intersections. The analysis was performed using crash data from 275 signalized intersections in southern Nevada. The panel data random-effect model was used to account for the unobserved factors for each unique arterial. It was found that the negative binomial (NB) regression models were the best in reflecting the dispersion in the crash data. Therefore, the random-effects negative binomial model (RENB) was applied to investigate the relationship between crash occurrence and access-management techniques. The results of the panel data RENB models were compared with those from the pooled NB models, which did not account for the panel data structure. Evaluation of the goodness-of-fit of the models developed indicated that the random-effect negative binomial model was the best-fit for the data at hand. The results from the panel data RENB showed that nine variables significantly affecting the safety at signalized intersections were the average length of corner clearance, traffic flow, land-use types, number of left-turn lanes for main streets, number of through lanes for main and minor streets, posted speed limit on main and minor streets, and grades of legs.

Identifying Access Management Factors Associated With Safety of Urban Arterials Mid-Blocks: A Panel Data Simultaneous Equation Models Approach

Objective: The objective of this study was to evaluate the safety impact of selected access management techniques in urban areas because access management techniques play an important role in urban roadway safety on the roadway network. Methods: In order to correct the interdependency between safety and mobility for heterogeneous mid-block segments, simultaneous equation models were adopted. The panel data structure of the model was used to address the heterogeneity issue for mid-block segments along a corridor. The integrated random coefficient simultaneous equation models were proposed to interpret both issues. Results: The models developed were used to identify influential factors. The length of mid-block segments, driveway density, and median opening density were among the significant factors found to be associated with crash rate on mid-block segments. Conclusions: From the results it can be concluded that the access management techniques, mid-block segment length, driveway density, and median opening density are significant factors that influence safety on mid-block segments. The longer the distance between signals, driveways, and median openings, the fewer the potential crashes are. In addition to these access management techniques, land use, especially the commercial land type, influences the safety on mid-block segments.

Evaluation of Speed Monitoring Displays for Work Zones in Las Vegas, Nevada:

Speed monitoring displays (i.e., speed trailers) have been evaluated in many states for reducing vehicular speeds in work zones. This study evaluated the enhancements to speed trailers regarding message size, the use of flashing, and the presence of more than one speed trailer in work zones. Field tests were conducted at two sites in the Las Vegas, Nevada, area, and traffic data were collected for statistical analysis. Regression models were developed to estimate the speeding likelihood and vehicle speeds on the basis of the free-flow speed data. The results indicated that the size of displayed messages and the use of flashing did show significant impact on speeding likelihood and speed reduction for vehicles in work zones. The extent of the impact varied for vehicle classification, the lanes they were operating in, and the day or night in which they were deployed. This study recommended larger message size and the use of flashing signs for speed trailers. More than one speed trailer was recommended for additional speed reduction.

Investigation of the AP-42 Sampling Method

Abstract The AP-42 method has been recommended by the U.S. Environment Protection Agency to collect dust emission data. According to this method, the number of sampling sites needs to be determined first. At these sites, the dust will be collected based on plots drawn on the road surface. Apparently, there has been no systematic rule to follow to determine the number of sampling sites. In addition, it is not known whether the required number of plots and their sizes are validated based on real data. Mobile sampling technology can collect dust emission data at very close space intervals, which to some extent can be viewed as being close to actual dust emission data continuously distributed over roadway segments. With such data available, this study investigated the number of sampling sites and the number of plots and their sizes based on the optimal allocation sampling method and the Monte Carlo simulation method. The results from the optimal allocation method indicated that most of the sampling sites should be drawn from the local roads because the variance of emission and proportion of road segments of this roadway classification are significantly bigger than other roadway classifications. This observation may lead to the application of other cost-effective sampling approaches. The results from the Monte Carlo simulation method imply that clear patterns of improved estimation of emission factors versus plot number and size can be observed only for three roadway classifications, not for other classifications. This result indicates that the AP-42 method may not be applicable to some roadway classifications, and thus different data collection methods such as the mobile sampling technology may be necessary.