DETECTING ABNORMAL ACTIVITIES OF OPERATORS OF COMPLEX TECHNICAL SYSTEMS AND THEIR CAUSES BASING ON WAVELET REPRESENTATIONS

Presented is the approach for supporting the outcome grading for activities of operators of complex technical systems. It is based on comparisons of current exercises with the activity database patterns in the wavelet representation metric associated with observed parameters as well as on probabilistic assessments of skill class recognition using sample distribution functions of exercise distances to cluster centers in a scaling space and Bayesian likelihood estimations with the aid of probabilistic profile of staying in activity parameter ranges. These techniques have demonstrated the capabilities of recognizing sets of abnormal exercises in the scaling spaces with the wavelet coefficient metric and detection of parameters characterizing operator mistakes to reveal the causes of abnormality. The techniques presented overcome limitations of existing methods and provide advantages over manual data analysis since they greatly reduce the combinatorial enumeration of the options considered