Development of a Hybrid Model for Risk Assessment and Management in Complex Road Infrastructure Projects

During the execution of road infrastructure projects, project managers face significant challenges, including financial, technical, regulatory, and operational risks. More than 90% of infrastructure projects have incurred costs exceeding initial estimates, impacting both completion timelines and the operational efficiency of road infrastructure. Effectively assessing and managing these risks is crucial for improving project outcomes and ensuring the sustainability of infrastructure investments. To address these challenges, this study developed a hybrid model for risk assessment and management in road infrastructure projects. The model quantifies risks across seven key categories: Design, External, Resource, Employer, Contractor, Engineer, and Project, based on three primary input factors: Environment coefficient, Contractual coefficient, and Design coefficient. Initially, various machine learning models, including linear regression, Random Forest, Gradient Boosting, Stacking Models, and neural networks, were applied to assess risk predictions. However, due to the specific nature of the dataset, these models did not achieve satisfactory predictive accuracy. As a result, fuzzy logic systems (Mamdani and Sugeno) were employed, demonstrating superior performance in modeling risk occurrence probabilities. Comparative analysis between these two fuzzy logic approaches revealed that the Sugeno model provided the most accurate predictions. The findings highlight the benefits of applying fuzzy logic for risk assessment in complex infrastructure projects, providing a structured framework for enhancing decision-making processes. This study provides a structured methodology for accurately predicting risks and enhancing project safety, efficiency, and long-term sustainability.