eduzhai > Applied Sciences > Engineering >

A Novel Approach for Recognizing Abnormal Activities of Operators of Complex Technical Systems: Three Non-Standard Metrics for Comparing Performance Patterns

  • Save

... pages left unread,continue reading

Document pages: 18 pages

Abstract: Presented is a novel approach for recognizing abnormal activities of operators of complex technical systems, which is based on comparing wavelet representations of current activity parameters and the previously collected performance patterns. The following three non-standard metrics are in use for these comparisons: the Euclidean metric in a wavelet coefficient space, the likelihood metric of eigenvalue trajectories for the activity parameters transforms and the Kohonen metric calculated by means of the self-organizing maps (or the Kohonen networks). The technique in question has demonstrated the capabilities for recognizing abnormal activities in various applications as well as for detecting parameters, which characterize operator mistakes, to reveal the causes of abnormality.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×