Evaluation of operator competence to identify system state in conditioning automated processes

This study investigates human works in conditioning an automated system in terms of the engineers' competence to identify the system state which enables them to intervene in the automated processes at the right time and in the appropriate fashion. For this purpose, the conditional entropy of the engineers' operation given the latent system state is utilized to evaluate one aspect of this ability from historical operational data. Computer-controlled multihead weigher (CCW) system is employed as the target domain for the skill analysis, and a comparative experiment was conducted in this work domain. Operational data acquired from engineers who have different levels of expertise are analyzed based on the hypothesis that the most proficient operator could successfully identify the actual system state from his available cues. The calculated indices indicate the skillful operator uses various operations for conditioning the system which are proper for the task requirements.