摘要
It is important to monitor operators' mental workload during the operation phase. Physiological measurement approaches could record the operator's mental data continuously, and might be less interruptive on the work activities. However, these methods often require the attachment of physical sensors, which are not unobtrusive in the physical sense. Furthermore, the individual difference makes calibrating to each individual tedious and requires trained persons to use. Often high noise-to-signal ratio data are hard to analyze. Due to these factors, physiological workload measurements are hardly widely applied in practical fields. In this study, an objective, non-intrusive and performance-based mental workload predictive model was proposed with high validity (R
2
= 0.51), which can be applied during the operation phrase. This model, developed based on the Rasmussen's skill-rule-knowledge framework, is comprised of two novel cognitive indices, the attention required index and uncertainty index. It can be used as the basis for establishing an early online warning system automatically. Furthermore, this model also predicts the types of error-prone tasks. This kind of information is expected to provide managers and supervisors with opportunities to intervene and improve tasks before error occurred. Finally, the predictive model proposed in this paper requires more practical application in fields to be completed.