You are here:

Supporting human performance in dynamic process control: An information needs approach to training and decision aiding
DISSERTATION

, Georgia Institute of Technology, United States

Georgia Institute of Technology . Awarded

Abstract

This study examined the effects of displaying alternative types and formats of information on learning and fault diagnosis in a dynamic process control task. Two experiments were conducted. In the first experiment, participants were trained with a computer simulation of a dynamic process control environment using one of five alternative displays and then were transferred to new conditions. Four different formats of displaying the procedural rules on the training interface, rule-based training, were compared to a control condition in which the rules were not displayed. During transfer, participants were shown a display without rules and tested on both the familiar training scenarios and a different, more complex set of disturbance scenarios. In general, rule-based training led to more accurate process control as measured by significantly reduced amounts of simulation damage and control deviation during training and transfer. In addition, rule-based trainees were quicker to respond to new, more complex disturbances. In the second experiment, the accuracy and time to diagnose novel faults were examined. After three blocks of simulation training were administered over two days, participants were placed in one of five display conditions and asked to diagnose novel faults. The diagnostic performance of participants viewing no additional data was compared to those viewing either a sequential or a simultaneous presentation format of the procedural rules or data describing the structure and functions of the component systems. Overall, participants were quicker to diagnose novel faults when viewing the procedural rules on the display regardless of format. Providing structural and functional data on the display did not significantly effect fault diagnosis in either format. The results of these experiments suggest that human performance in dynamic process control can be supported by furnishing the procedural rules particularly in conditions of low familiarity such as initial learning and novel fault diagnosis. These findings provide new insight into the design of training displays and decision aids in dynamic process control.

Citation

Crosland, M.B. Supporting human performance in dynamic process control: An information needs approach to training and decision aiding. Ph.D. thesis, Georgia Institute of Technology. Retrieved October 18, 2019 from .

This record was imported from ProQuest on October 22, 2013. [Original Record]

Citation reproduced with permission of ProQuest LLC.

For copies of dissertations and theses: (800) 521-0600/(734) 761-4700 or https://dissexpress.umi.com

Keywords