Validating an abnormal situation prediction model for smart manufacturing in the oil refining industry

Human beings play an important role in a smart manufacturing economy. The repetitive and cognitive demanding task operations of smart manufacturing require the development of system models for measuring and predicting human performance, including oil refinery monitoring tasks. The main objective of this research was to validate the generalizability of a mathematical model for the prediction of refinery operators' detection of abnormal events. Moreover, we examined operators' visual behaviors in response to abnormal situations at different ages and with different task loads, task complexities, and input devices. We found that participants had lower mean fixation durations, total fixation numbers, and fixation/saccade ratios when they were in the condition of a touchscreen device. Moreover, we found that older adults had higher mean saccade durations and saccade amplitudes when they were in the condition of a touchscreen device. Finally, the statistical model borrowed from our prior paper was found to be generalizable to different task loads and age groups for the prediction of operators’ detection of abnormal events. Our results showed that visual behaviors can indicate specific internal states of participants, including their cognitive workload, attention, and situation awareness in a real-time manner. The findings provide additional support for the value of using visual behavior to predict responsiveness of oil refinery operators and for future applications of smart manufacturing monitoring systems.