Supporting diagnostic functions in human supervisory control

Fault diagnosis tasks of human supervisors interacting with technical dynamic systems are discussed under different conceptual views, namely the task and the interaction orientations as well as the integrated automation concept. The paper presents techniques of cognitive task analyses for fault diagnosis. The relationships of these techniques to the different conceptual views are outlined. The possibilities of decision support for interactive fault diagnosis under human supervisory control are considered within the integrated automation architecture. The state-based fault diagnosis approach is further elaborated. It provides a detailed diagnosis based on information from the supervision and control system in cooperation with the human operator. The steps of the state-based diagnosis as well as the architecture of the knowledge-based system CAUSES using the state-based diagnosis approach are described. An example from the power plant domain is also presented in some more detail. A useful evaluation procedure is suggested for investigating the effectiveness of the supporting diagnostic functions.<
>