Neuro-Fuzzy System Dynamics for Effective Risk Management in Off-Site Construction Supply Chains

Abstract This research presents a novel neuro-fuzzy system dynamics (NN-Fuzzy-SD) model designed to manage risks in off-site construction supply chains (OCSCs). By combining the predictive power of neural networks (NNs) with the uncertainty-handling capabilities of fuzzy logic, the model enables dynamic, real-time risk assessment and mitigation, addressing challenges such as material shortages, transportation delays, and labor disruptions. The model integrates real-time data to optimize inventory buffers, resource allocation, and scheduling strategies, improving project efficiency and resilience. An empirical case study across six off-site construction projects in Beijing and Shanghai demonstrates the model’s effectiveness in adapting mitigation strategies according to varying risk profiles, yielding measurable improvements in cost efficiency and operational performance. This study makes four primary contributions to construction engineering and management. First, it introduces the first integrated NN-Fuzzy-SD framework for real-time risk management in off-site construction. Second, it develops the dynamic risk mitigation strategy optimization (DRMSO) and cross-domain neuro-fuzzy decision support system (C-DNF-DSS) for adaptive, cross-domain mitigation. Third, it empirically validates the model with real-world projects, providing quantitative evidence of improved resource allocation. Finally, it identifies dominant risk drivers and delivers a scalable, industry-ready framework that enhances project resilience, operational efficiency, and informed decision-making, contributing significantly to the global construction management body of knowledge.