Human–Machine System Modeling and Safety Risk Assessment in Construction Operations Incorporating Workers’ Performance Variability

The increasing involvement of diverse construction machines has significantly transformed construction operations. Within the nonlinear interaction of multiple worker and machine functions, workers’ performance exhibits complex variability due to cognitive uncertainty and contextual correlation. These factors have a profound impact on the construction operation process, leading to the emergence of new and intricate risk patterns. Adopting a human–machine system perspective, investigating the influence of workers’ performance variability (WPV) on construction operation processes holds great potential for enhancing system safety. This study introduces an integrated method merging a Bayesian network (BN)-based cognitive model with the functional resonance analysis method (FRAM). The BN model estimates the probability distribution of WPV manifestations. Subsequently, Bayesian inference and FRAM-based quantitative analysis are consolidated within a Monte Carlo simulation framework to evaluate the WPV’s influence on system safety. The efficacy of this approach was evaluated by conducting a shield tunneling construction case study, which revealed that the proposed method could identify critical types of WPV, assess their impacts, and explain their causes. The contribution of this study lies in providing an appropriate modeling and analysis method for the intricately functionally coupled worker–machine collaborative construction operation process and its associated risks. The proposed approach has the potential to support the development of practical tools that offer quantitative evidence for managing WPV and mitigating its adverse effects on system safety.