This paper studies the detection of traffic anomalies and their recovery times using travel time data. Travel time anomalies—caused by congestion, incidents, and other disruptive events—are detected using an algorithm inspired by the California algorithm. Detection focuses on how frequently anomalies occur, that is, how often travel times deviate from normal conditions. Recovery quantifies how quickly traffic conditions return to normal after an anomaly has occurred. The algorithm is used to detect the timing and location of anomalies and to analyze spatial and temporal variations in recovery times. A numerical case study using travel time data from the E20/E4 freeway between Södertälje and Stockholm over a 3-year period is conducted. Findings reveal links that are more prone to anomalies, and when anomalies that may have a large negative impact on accessibility. The results contribute to understanding network resilience and support improved incident management.

Accurate travel time estimations are essential for traffic analysis and enable modern applications such as dynamic route guidance and traffic control. With the growing availability of high-resolution traffic data from GPS-enabled devices and probe vehicles, advanced models have been developed to estimate travel times more precisely. This paper proposes a sequential link estimation method for trip-level travel time estimation. The method exploits how travel times on one link are influenced by the preceding link and influence the subsequent link along a route. The method uses a chain of regression estimation models where each link's estimated travel time depends on the travel time of the adjacent link. Each estimated value is passed as input to the model for the next link, creating a chain of conditional estimates that extends from an arbitrary link to both the beginning and end of a freeway. We evaluate the proposed travel time estimation method using real-world traffic data from freeways in Sweden. The results show an average percentage error as low as 2.38 percent with a standard deviation of 1.88 percent, indicating highly accurate travel time estimates.

This study presents a statistical analysis of train delays in the Swedish railway system. The focus of the study is to identify the best-fitting probability distributions for train arrival times across different stations and travel directions. Using the Kolmogorov-Smirnov (K-S) test, we evaluate the goodness of fit for common distributions—gamma, log-normal, and inverse Gaussian—to capture delay patterns at ten stations. Our findings reveal significant variability across stations, with the log-normal distribution providing the best fit for 70% of cases. However, some stations exhibited direction-specific deviations, emphasizing the need for localized analysis. Traditionally, train delays in Sweden have been assumed to be uniformly distributed across the network, an oversimplification frequently used in generating synthetic datasets for AI-based timetable rescheduling systems. This study challenges that assumption, demonstrating that delay distributions vary by station and direction. By incorporating station- and direction-specific modeling, our results contribute to the development of more accurate synthetic datasets. These insights support data-driven approaches to predictive modeling, operational efficiency improvements, and increased reliability in railway networks. Based on the best-fitting distributions identified through statistical testing, we generate synthetic data using maximum likelihood estimates and direct sampling. Our study systematically assesses the distributional characteristics of train arrivals across stations and directions in the southern Swedish railway network, aiming both to understand operational variability and to generate realistic synthetic data for AI-based rescheduling. Building on this analysis, our method produces datasets that preserve the statistical characteristics of real train delays, ensuring they are more suitable for training and evaluating AI-based rescheduling algorithms.

This paper presents a methodology for predicting remaining train arrival times along a predefined railway path using sectionally sampled GPS-based observations collected between consecutive stations. Station-level remaining arrival times are derived from multiple train trips and analyzed using a linear regression model, enabling assessment of both prediction accuracy and temporal consistency. The approach provides continuous predictions of remaining arrival times at each station along the path rather than focusing solely on the final destination. Results demonstrate that median remaining arrival time errors decrease with increasing sampling density, an effect attributed to greater sample availability that captures a larger proportion of the underlying train dynamics and timetable adherence. Under these conditions, the proposed methodology achieves a station-based average median remaining arrival time error of less than 91 seconds. Across 100 repeated out-of-sample evaluations, the results indicate reproducible performance and acceptable coarseness even when the sampling density is substantially reduced. This continuous, station-level prediction capability enables early detection of delays and potential timetable conflicts, providing minute-level predictive accuracy suitable for decision support for train dispatchers