Risk assessment in sociotechnical systems based on functional resonance analysis method and hierarchical fuzzy inference tree.

Effective risk assessment is crucial for ensuring safety and preventing severe consequences in complex socio-technical systems. Traditional probabilistic risk assessment methods have limitations in capturing the systemic complexities and the role of human operators, necessitating the advancement of methodologies. The present study aims to develop a comprehensive risk assessment methodology tailored for complex socio-technical systems, integrating the functional resonance analysis method, soft computing, natural language processing, and multi-criteria decision-making. Novel approaches, including expressing timing and precision variability through Z-numbers, employing Z-TOPSIS for their integration, utilizing a hierarchical fuzzy inference systems to model variability propagation, and text classification to compute the amplification factor using near-miss data, were introduced in the study. Further analyses were also conducted to identify critical couplings and paths. Additionally, the concept of degree centrality was utilized to identify functions that are impacted by multiple upstream functions and those that exert influence on several downstream functions. The methodology's application to an anode change operation in an aluminum smelter highlighted its effectiveness. For practitioners, the framework offers a structured, data-driven tool for risk assessment. Its implementation can improve safety, reduce accidents, and prevent economic losses from disruptions and injuries.