مريم شفيعي علويجه

This research investigates and predicts the Track Quality Index (TQI) in the railway network using machine learning algorithms. The Track Quality Index, which is obtained through geometric measurements of the track by special machines, is a key criterion in assessing the condition and health of railway lines. Given the limitations in the number of these machines and the necessity of blocking the lines during data collection, the aim of this research is to develop a predictive model for the Track Quality Index using machine learning algorithms. In this regard, Track Quality Index (CTR) data along with other effective variables such as the kilometer of the beginning and end of the block, the kilometer of the station at the beginning and end of the block, the block length, the line capacity, the line demand, the rail type, the sleeper type, the type of fastening, the type of welding, the minimum curve radius, the maximum gradient and ascent, the maximum fleet speed, etc. have been used. Five machine learning algorithms including Random Forest, Decision Tree, k-Nearest Neighbors, CatBoost and XGBoost have been used to model and predict the line quality index. The results of the model eva‎luation show that the CatBoost algorithm performs better than other algorithms in all eva‎luation indicators and is recognized as the best model in this study with a prediction accuracy of more than 90%. These findings indicate the high potential of machine learning algorithms in accurately predicting the line quality index and, as a result, facilitating maintenance planning and reducing costs associated with periodic inspections of railway lines.

شاخص كيفيت خط , يادگيري ماشين , ماشين اندازه‌گيري خط , پيش‌بيني , الگوريتم CatBoost , نگهداري و تعميرات

Line quality index , machine learning , line measuring machine , prediction , CatBoost algorithm , maintenance and repairs

Maryam Shafiee

Dr. Saeed Mohammad zadeh, Dr. Jabbar ali zakeri