LIU Weiqing, ZHANG Hao, XU Zhifen
Journal of Railway Engineering Society. 2026, 43(3): 99-104.
Research purposes:In the context of digital transformation of railway engineering construction, intelligent prediction of construction duration indicators is an inevitable requirement for the development of intelligent construction of railway engineering. This study applies machine learning methods to the prediction of railway engineering construction duration indicators. By preprocessing input features such as railway tunnel construction date, surrounding rock grade, and duration, five prediction models, including linear regression, random forest, LightGBM, XGBoost, and CatBoost, are constructed, and the prediction errors of each model are compared. Based on the results of model performance comparison, the model with the best comprehensive performance is selected, and the importance of each influencing factor is further analyzed to provide a decision-making basis for railway engineering project management.
Research conclusions:(1) Based on expert experience and field survey data, the data missing of influencing factors such as railway tunnel construction date, surrounding rock grade, duration, designed volume and actual volume were preprocessed. By comparing the training performance of five models, namely linear regression, random forest, XGBoost, LightGBM and CatBoost, it was found that the LightGBM algorithm had small prediction error and strong generalization ability. (2) The importance of influencing factors was analyzed using the LightGBM model. The results showed that in case A,construction date, duration and actual volume ranked in the top three in importance; in case B, actual volume, construction date and maintenance ranked in the top three in importance. The study showed that during the construction process, dynamic factors such as daily personnel status, resource usage and environmental changes have a significant impact on the construction schedule.(3) The results of this study can provide reference for the construction organization design of railway engineering.