Prediction of travel time for railway traffic management by using the AdaBoost algorithm

AKÇAY, Mehmet Taciddin; Akgündoğdu, Abdurrahim; Tiryaki, Hasan · 2022 · OpenAlex-citations

DOI: 10.25092/baunfbed.937333

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the challenge of accurately predicting travel times between stations in intracity metro systems, a critical parameter for railway traffic management and signalization system design. Precise travel time estimation is essential for ensuring defined headways, optimizing vehicle schedules, and maintaining operational safety. The authors propose using the Adaptive Boosting (AdaBoost) machine learning algorithm to predict these values, aiming to improve upon existing methods that often rely on complex simulations or less accurate predictive models. The research utilizes a dataset of 500 samples generated through operational simulations, incorporating eight input parameters: waiting time, motion resistance, slope, curve, traction force, maximum speed, vehicle mass, and distance between stations. The output variable is the travel time. The AdaBoost algorithm was implemented and compared against five other well-known machine learning methods: Random Forest, Neural Network, k-Nearest Neighbors (kNN), Decision Trees, and Support Vector Machines (SVM). To evaluate performance and prevent overfitting, the authors employed two validation techniques: 10-fold cross-validation (CV) and a random sampling hold-out (RSHO) method, where 66% of the data was used for training and 34% for testing. Performance was assessed using Mean Squared Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R²). The results demonstrate that the AdaBoost algorithm outperformed all other tested methods in both validation scenarios. Under the 10-fold CV method, AdaBoost achieved an R² of 0.9972, an MSE of 4.52, an RMSE of 2.13, and an MAE of 1.55. In the RSHO evaluation, it recorded an R² of 0.9964, an MSE of 5.71, an RMSE of 2.39, and an MAE of 1.74. While other models like Random Forest and Neural Networks also yielded high R² values, AdaBoost consistently produced lower error metrics. Specifically, the study notes that AdaBoost outperformed the next best method by approximately 15.67% in MSE, 7.51% in RMSE, and 9.68% in MAE. The significance of this work lies in its demonstration that AdaBoost provides a highly accurate and robust tool for predicting railway travel times, which is vital for designing efficient signalization systems and managing operational traffic. By achieving high precision even under varying operational conditions, the proposed method facilitates better schedule adherence and system optimization. The findings support the broader adoption of machine learning applications in railway traffic engineering, offering a reliable alternative to traditional simulation-based approaches for determining vehicle movement characteristics.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• speed choice