Sherif Ishak

Optimization of Dynamic Neural Network Performance for Short-Term Traffic Prediction

An approach is presented for optimizing short-term traffic-prediction performance by using multiple topologies of dynamic neural networks and various network-related and traffic-related settings. The conducted study emphasized the potential benefit of optimizing the prediction performance by deploying multimodel approaches under parameters and traffic-condition settings. Emphasis was placed on the application of temporal-processing topologies in short-term speed predictions in the range of 5-min to 20-min horizons. Three network topologies were used: Jordan–Elman networks, partially recurrent networks, and time-lagged feedforward networks. The input patterns were constructed from data collected at the target location and at upstream and downstream locations. However, various combinations were also considered. To encourage the networks to associate with historical information on recurrent conditions, a time factor was attached to the input patterns to introduce time-recognition capabilities, in addition to information encoded in the recent past data. The optimal prediction settings (type of topology and input settings) were determined so that performance was maximized under different traffic conditions at the target and adjacent locations. The optimized performance of the dynamic neural networks was compared to that of a statistical nonlinear time series approach, which was outperformed in most cases. The study showed that no single topology consistently outperformed the others for all prediction horizons considered. However, the results showed that the significance of introducing the time factor was more pronounced under longer prediction horizons. A comparative evaluation of performance of optimal and nonoptimal settings showed substantial improvement in most cases. The applied procedure can also be used to identify the prediction reliability of information-dissemination systems.

FUZZY ART NEURAL NETWORK MODEL FOR AUTOMATED DETECTION OF FREEWAY INCIDENTS

Pattern recognition techniques such as artificial neural networks continue to offer potential solutions to many of the existing problems associated with freeway incident-detection algorithms. This study focuses on the application of Fuzzy ART neural networks to incident detection on freeways. Unlike back-propagation models, Fuzzy ART is capable of fast, stable learning of recognition categories. It is an incremental approach that has the potential for on-line implementation. Fuzzy ART is trained with traffic patterns that are represented by 30-s loop-detector data of occupancy, speed, or a combination of both. Traffic patterns observed at the incident time and location are mapped to a group of categories. Each incident category maps incidents with similar traffic pattern characteristics, which are affected by the type and severity of the incident and the prevailing traffic conditions. Detection rate and false alarm rate are used to measure the performance of the Fuzzy ART algorithm. To reduce the false alarm rate that results from occasional misclassification of traffic patterns, a persistence time period of 3 min was arbitrarily selected. The algorithm performance improves when the temporal size of traffic patterns increases from one to two 30-s periods for all traffic parameters. An interesting finding is that the speed patterns produced better results than did the occupancy patterns. However, when combined, occupancy–speed patterns produced the best results. When compared with California algorithms 7 and 8, the Fuzzy ART model produced better performance.

Statistical Evaluation of Interstate 4 Traffic Prediction System

Short-term traffic prediction systems have received considerable attention in the past few years as a means to support advanced traveler information and traffic management systems. Predictive information allows transportation system users to make better trip decisions at the pretrip planning stage and en route. A comprehensive statistical analysis of the traffic prediction system performance implemented on the 40-mi corridor of Interstate 4 in Orlando, Florida, is presented. The system was evaluated under a wide range of traffic conditions and various model parameters. The prediction performance in terms of prediction errors was examined with both link-based and path-based approaches.

Safety Evaluation of Joint and Conventional Lane Merge Configurations for Freeway Work Zones

Inefficient operation of traffic in work zone areas not only leads to an increase in travel time delays, queue length, and fuel consumption but also increases the number of forced merges and roadway accidents. This study evaluated the safety performance of work zones with a conventional lane merge (CLM) configuration in Louisiana. Analysis of variance (ANOVA) was used to compare the crash rates for accidents involving fatalities, injuries, and property damage only (PDO) in each of the following 4 areas: (1) advance warning area, (2) transition area, (3) work area, and (4) termination area. The analysis showed that the advance warning area had higher fatality, injury, and PDO crash rates when compared to the transition area, work area, and termination area. This finding confirmed the need to make improvements in the advance warning area where merging maneuvers take place. Therefore, a new lane merge configuration, called joint lane merge (JLM), was proposed and its safety performance was examined and compared to the conventional lane merge configuration using a microscopic simulation model (VISSIM), which was calibrated with real-world data from an existing work zone on I-55 and used to simulate a total of 25 different scenarios with different levels of demand and traffic composition. Safety performance was evaluated using 2 surrogate measures: uncomfortable decelerations and speed variance. Statistical analysis was conducted to determine whether the differences in safety performance between both configurations were significant. The safety analysis indicated that JLM outperformed CLM in most cases with low to moderate flow rates and that the percentage of trucks did not have a significant impact on the safety performance of either configuration. Though the safety analysis did not clearly indicate which lane merge configuration is safer for the overall work zone area, it was able to identify the possibly associated safety changes within the work zone area under different traffic conditions.

Improvement and Evaluation of Cell-Transmission Model for Operational Analysis of Traffic Networks: Freeway Case Study

The cell-transmission model (CTM), developed by Daganzo in 1994, has not been fully exploited as an operations model for analysis of largescale traffic networks. Because of its macroscopic and mesoscopic features, CTM offers calibration and computational advantages over microscopic models. This paper demonstrates specific improvements to the original form of CTM to increase its accuracy and realism of traffic flow representation. These improvements include modifications to provide flexibility in selection of cell lengths, noninteger movements of vehicles between cells, and algorithmic enhancements in the merging and diverging logics. The effect of such improvements on the performance of CTM was evaluated independently and comparatively. A sample freeway network of the I-10 corridor in Baton Rouge, Louisiana, was used to evaluate and compare the performance of the improved version of CTM versus CORSIM under heavily congested traffic conditions. The results showed comparable performance of both platforms in...

Fuzzy-Clustering Approach to Quantify Uncertainties of Freeway Detector Observations

Information is a key component of today’s surface transportation systems. Yet the quality of information is often determined by the quality of raw data from which it is extracted. A plethora of data is currently being compiled in real time from hundreds of miles of freeway sections nationwide. A large proportion of the data is collected via inductive loop detectors and is vital to the successful implementation of transportation data warehouses and decision support systems. Little effort has been made to establish procedures that quantify the amount of uncertainties in the traffic observations and enhance data screening algorithms. This study presents an approach derived from a fuzzy-clustering concept to measure the level of uncertainties associated with dual loop detector observations. The developed algorithm does not rely on a specific mathematical model and avoids estimating the effective vehicle length because of the limitations explained here. Based on a divide-and-conquer approach, the algorithm clusters the input space of the three traffic parameters (speed, occupancy, and volume) into regions of highly concentrated observations. The level of uncertainty in each observation can then be measured with one parameter that is derived from the membership grade and a decaying function of the normalized Euclidean distance. The parameter can thus be used for data screening as well as detector maintenance and recalibration purposes. A data screening algorithm was developed to identify erroneous observations in four sequential stages. The results obtained from screening 50,000 observations indicate that most of the noise cluttering the input space was significantly reduced by setting the uncertainty measure to 0.9.

Stochastic Approach for Short-Term Freeway Traffic Prediction During Peak Periods

Using a stochastic approach, this paper explores and models the basic stochastic characteristics of freeway traffic behavior under a wide range of traffic conditions during peak periods and then applies the models to short-term traffic speed prediction. The speed transition probabilities are estimated from real-world 30-s speed data over a six-year period at three different locations along the 38-mi corridor of Interstate 4 (I-4) in Orlando, FL. The cumulative negative/positive transition probabilities and expected values are derived from the transition probabilities and fitted using logistic and exponential models, respectively. The expected values associated with the most likely transition of speed are then derived from the fitted models and used for predicting speed. Each predicted speed is also associated with a probability value, indicating the chance of observing the occurrence of such transition. The prediction performance was compared for three methods using the root mean square errors (RMSEs). The weighted average method was very close to the higher probability method in most cases. For the two probabilistic methods, the performance was slightly better for the morning peak periods than the evening peak period or all data combined. While the prediction performance of the probabilistic models was comparable with those of other methods found in the literature, the probabilistic approach based on the higher probability provides estimates of the associated probability with each prediction. This provides a measure of confidence in the predicted values before such information is disseminated to the public by traffic agencies.

Impact of traffic diversion with ATIS on travelers' safety

This study looks at the potential impact of diverting traffic from freeways under incident conditions to arterials with the use of Advanced Traveler Information Systems (ATIS). Accident predictions models were developed for freeways and arterials. A variety of incident scenarios were also simulated using a traffic diversion model, and regression techniques were used to model the network safety as a function of network and incident parameters.

Freeway Truck Lane Restriction and Differential Speed Limits: Crash Analysis and Traffic Characteristics

Over the past decade, several fatal truck-related crashes have occurred on the elevated freeway over the Atchafalaya Basin segment of Interstate 10 in southern Louisiana. In an attempt to reduce the crash rates, the Louisiana Department of Transportation and Development has implemented two policies to regulate the truck traffic on this rural section of freeway. These policies restrict truck traffic to the right lane and reduce the maximum truck speed limit to 55 mph while maintaining passenger car speed limits at 60 mph. To investigate potential relationships between compliance with these policies and crash rates, traffic and crash data were collected for the segment while the policies were in force. Relationships between hourly observations of crash rates and compliance rates were sought at the .05 significance level with multiple linear regression. The traffic characteristics that might affect such a relationship were also incorporated into the regression models. These characteristics included difference between truck and car speeds, speed variance, truck volume, and lane occupancy. The regression models were performed with SAS software. Confidence intervals on the means of the explanatory variables were constructed to understand the variability in the values of the traffic characteristics over different days of the data collection period. The results showed that violation of the lane restriction and truck speed limits, truck speed variance, differences between car and truck mean speeds, and lane occupancy were positively correlated with crash rates. The findings suggested that prohibiting trucks from traveling in the left lane and setting a truck speed limit of 55 mph and a car speed limit of 60 mph on a four-lane elevated rural freeway can offer traffic safety benefits.

Post and During Event Effect of Cell Phone Talking and Texting on Driving Performance—A Driving Simulator Study

Objective: A number of studies have been done in the field of driver distraction, specifically on the use of cell phone for either conversation or texting while driving. Researchers have focused on the driving performance of drivers when they were actually engaged in the task; that is, during the texting or phone conversation event. However, it is still unknown whether the impact of cell phone usages ceases immediately after the end of task. The primary objective of this article is to analyze the post-event effect of cell phone usage (texting and conversation) in order to verify whether the distracting effect lingers after the actual event has ceased. Methods: This study utilizes a driving simulator study of 36 participants to test whether a significant decrease in driver performance occurs during cell phone usage and after usage. Surrogate measures used to represent lateral and longitudinal control of the vehicle were standard deviation (SD) of lane position and mean velocity, respectively. Results: Results suggest that there was no significant decrease in driver performance (both lateral and longitudinal control) during and after the cell phone conversation. For the texting event, there were significant decreases in driver performance in both the longitudinal and lateral control of the vehicle during the actual texting task. The diminished longitudinal control ceased immediately after the texting event but the diminished lateral control lingered for an average of 3.38 s. The number of text messages exchanged did not affect the magnitude or duration of the diminished lateral control. Conclusion: The result indicates that the distraction and subsequent elevated crash risk of texting while driving linger even after the texting event has ceased. This finding has safety and policy implications in reducing distracted driving.