Análise da confiabilidade humana via redes Bayesianas: uma aplicação à manutenção de linhas de transmissão

Human Reliability Analysis (HRA) has been used to study the execution of human actions and their interactions with a system, taking into account its limitations and factors influencing the human performance. The HRA methods can be categorized into the first and second generation models. These models present some shortcomings such as unrealistic independence assumptions and simple binary representation of events. Therefore, they present a significant deficiency in modeling human actions. In this paper it is described a methodology for modeling human actions through Bayesian belief networks (BBN). The methodology provides a greater flexibility as not it only allows for a more realistic representation of the dynamic nature of man-system, but also allows for representation of the relationship of dependence among the events and performance shaping factors. This methodology is demonstrated in the context of human reliability in activities of maintenance of transmission line.