Publikationen

Human-machine interfaces for cooperative supervision and control by several human users, either in control rooms or in group meetings, are dealt with. The information flow between the different human users, and their overlapping information needs, are explained. The example of a cement plant illustrates this in more detail. Cognitive science concepts for supporting visual and mental coherence, as well as multi-media, hypertext and CSCW (computer supported cooperative work) technologies, are discussed with respect to their usage in multi-human machine interfaces. The design process for these interfaces is outlined, with emphasis on the different design stages, user participation and the possibility of knowledge-based design support. Some ideas for the conceptual structure of multi-human machine interfaces are also presented.

The centralised control rooms of large industrial plants have separated people from the processes they should control. Perception is restricted mainly to the visual sense. Only telephone or radio links provide narrow-band voice communication with maintenance personnel down in the plant. Multimedia equipment can perceptionally bring back the operator into the plant while bodily keeping him the comfortable and safe control room. This involves video and audio transmission from process components as well as sights and sounds artificially generated from measurements. Groupware systems support inter-action between operators, engineers, and managers in different plants. With support from the German government, the state of Hessen, and industrial companies the Laboratory for Systems Engineering and Human-Machine Systems at the University of Kassel establishes an Experimental Multimedia Process Control Room. Core of this set-up are two high-performance graphics workstations linked to one of several process or vehicle simulators. Multimedia periphery includes video and teleconferencing equipment and a vibration and sound generation system.

Abstract An application-oriented procedure for task and knowledge analyses was developed and applied in coal-fired power plants. After introducing operator dialogs in such plants, the proposed task and knowledge analysis procedure is described. The development and structure of state- and failure-oriented questionnaires as well as results from interviews with engineers and operators are explained.

A rule-based fault correction task in a decentralized hierarchical technological system supervised by a human operator was investigated. Different combinations of four types of failures had to be corrected by the human operator who could select the appropriate subsystems and their components, test and repair them, and switch temporarily to a central process computer. Two different display contents and formats as well as three levels of decision support were compared in experiments with four subjects. The results showed that the fault correction strategies and the system performance varied largely among the subjects. Generally, the performance increased with longer decision times. Significantly more simple execution errors were made with the enlarged display format. The system performance increased with full decision support.

Information needs of control room operators are derived from task- and knowledge analysis in coal-fired power plants. This leads to functional state diagrams for knowledge representation. Ergonomic knowledge for display designers is structured with perceptional and cognitive aspects. Issues such as colour, level of abstraction, and information format are exemplified. Some ideas about the implementation of an ergonomic knowledge support system are presented.

The methodological foundations of task and knowledge analyses are summarized and evaluated in this report. Based on these discussions, a combined task and knowledge analysis procedure is proposed. Its application-oriented form has been developed in more detail for investigations in power plants but can easily be translated for other application domains of dynamic systems. Particiularly, observation interview techniques based on several specifically designed questionnaires are used complemented by other task analysis techniques and knowledge elicitation tools. The basic process as well as fuel and waste treatment, automation concepts, and control room design of coal- fired power plants are briefly introduced only in order to supply some necessary information to those readers who are not familiar with power plants. On that basis, the application-oriented analysis procedure and, particularly, the results collected in two different power plants can be better understood. The results are derived from interviews with engineers and operators. Especially, basic systems knowledge, the start-up procedures, and some relevant failure situations are investigated. The need for refinements as well as for further investigations of additional failure situations is explained.