Predicting Delay in Dam Construction Projects Using Machine Learning: A Case Study of the Balaroud Dam

Dam construction projects are among the most complex infrastructure developments and are highly vulnerable to schedule delays due to technical, managerial, environmental, and resource-related uncertainties. Early prediction of delays can improve planning quality, reduce time-cost overruns, and support proactive project control. This study develops and evaluates a machine learning-based framework for delay prediction in dam construction projects, with a case study of the Balaroud Dam. Data were collected from ۴۷ expert questionnaires and project documents, preprocessed, and modeled using three algorithms: Random Forest (RF), Support Vector Machine (SVM), and a Genetic Algorithm-optimized Random Forest (GA-RF). The results show that GA-RF achieved the best standalone performance, while a weighted ensemble of the models further improved prediction accuracy. Variable importance analysis identified four dominant delay drivers: material supply delays, contractor performance, design changes, and shortage of skilled labor. The Balaroud Dam case study confirmed a high level of consistency between the model outputs and real project conditions. Based on these findings, the study proposes an operational early warning framework for monitoring activity-level delay risks and prioritizing corrective actions. The proposed approach can support proactive decision-making, reduce delay risk, and improve time-cost efficiency in large-scale dam construction projects.

کلیدواژه ها:

Dam construction ، Project delay prediction ، Machine learning ، Random Forest ، Genetic Algorithm ، Early warning system