AN APPROACH OF DYNAMICAL ALLOCATION OF SUPERVISION TASKS BETWEEN MAN AND COMPUTER IN CONTROL ROOMS OF AUTOMATIZED PRODUCTION SYSTEMS

Abstract The variability of human operator's performances in the supervisory and control tasks of automatized, industrial systems may lead to serious errors endangering not only the production system but also himself, and the environment, as well. One possible method for reinserting the operator Into the control and supervisory loop of these systems is the dynamic allocation of decision tasks between man and computer. This article proposes a principle for dynamic allocation of information between the two decision makers, man and computer to the end of optimizing the decision-making team's performance under the constraint of the operator's maximum permissible workload during supervision of the process's disturbed functioning. Performances and workload are estimated on line from a decision model receiving the operator's information and commands. This model comprises two decision sets : one “strategical”, which defines which variable to survey, the other “tactical”, which selects what decision or action to take on this variable. These two sets comprise several stages : event detection, situation assessment, decision-making or action. A feasibility study defines the model's functioning, as well as the worlkoad indices it can provide on line