(Man is the measure of all things)

Power quality, a recently introduced term, covers several phenomena related to the fact that voltage and current in an electric power system deviate from an ideal sinusoidal signal. The increased requirements on supervision, control, and performance in modern power systems make power quality monitoring necessary. However, the databases created by monitoring, are large and new tools are required for the fast and efficient use of the data. This thesis focuses on a number of events that occur during power system operation and presents the results of the analysis of actual recordings. New classes of events are defined and the distinctive characteristics (features) of each class are given. The guidelines for the development of a knowledge-based classification system, which utilises these features, are given. Two groups of power system events are considered in this thesis. The first group contains transients due to synchronised capacitor switching. An algorithm is proposed for the detection and classification of two types of synchronised switching. The second grou