DEVELOPMENT OF SUPERVISORY TOOLS FOR ACTIVE HUMAN MONITORING

摘要

In this paper, an approach to extract "trend" features from measured variables in order to design human-machine interface improving active human monitoring and situation awareness is proposed. This approach is based on an "on line" linear segmentation of data and on a fuzzy classification of the shape of segments. Nine basic temporal shapes have been defined. Hence, for each measured variable, this approach generates high level information (or abstractions) characterizing the dynamic and the trend. An active monitoring interface example of three tanks systems is used to illustrating this approach.