Classification of Operator Behaviors Using a Self-Organizing Map and Ontology

概要

論文の詳細を見る
This paper proposes a new approach to the classification of operator behaviors using a self-organizing map (SOM) and ontology. In order to construct the behavior classificator, ontology and Petri net are used to analyze the types of operations. An SOM is used to establish relations between behavior modes and operational states. An experiment was performed using a virtual transportation task simulator to verify the presented approach, and the operator's behavior during machine manipulation was classified on the basis of the time-series data obtained through the experiment. The behaviors classified using the proposed technique were compared with those extracted by human analysis, and it was confirmed that the classification accuracy of the presented method ranged from 30% to 80% (M = 41.2%, SD = 18.9%), whereas average accuracy of human analysis was as low as 29.7%.