Hybrid Fuzzy-NSGA-II Decision-Support Framework for Multi-Objective Risk Assessment of Construction

Abstract

Credit risk assessment of construction contractors is a process that predicts the likelihood of project default by analyzing contractors’ past performance. Existing methods are mostly single-objective and often neglect key project dimensions such as time, quality, and cost, limiting their accuracy and effectiveness. This study proposes a multi-objective approach for contractor credit risk assessment, simultaneously optimizing two conflicting goals: enhancing contractor quality and minimizing the financial–temporal burden of projects. A Fuzzy Logic Controller (FLC) is employed for its interpretability and alignment with human decision-making processes. The design of the knowledge base and membership functions is optimized using a Genetic Algorithm integrated with the Non-dominated Sorting Genetic Algorithm II (NSGA-II). The proposed model is evaluated on a dataset of 540 real construction contractor records. Experimental results demonstrate that the FLC-NSGA-II framework outperforms comparative methods including group regression, non-group regression, MOPSO, and SPEA-II in terms of predictive accuracy, achieving performance indices of R² = 0.1000, MSE = 0.0005, RMSE = 0.0224, and MAE = 0.0183. This high accuracy improves credit risk prediction, prevents project defaults, and reduces financial losses for construction firms. The proposed framework provides a novel, precise, and generalizable tool for credit risk assessment and supports decision-making in complex construction projects.