Making implicit knowledge of distance protective relay operations and fault characteristics explicit via rough set based discernibility relationship.

This paper discusses the novel application of the discernibility concept inherent in r ough set theory in making explicit of t he implicit knowledge of d istance protective relay o perations a nd fault characteristics t hat are hidden away in t he reco rded relay event report. A rough-set-based data m ining strategy is formulated to analyze t he rel ay trip assertion, impedance element ac tivation, an d fault characteristics o f distance relay decision syst em. Using rough set theory, the uncertainty and vagueness in t he relay event report can be resolved using t he concepts of discernibility, elementary sets and s et approximations. Nowadays protection engineers are suffering from very complex implementations of protection s ystem analysis due t o massive q uantities of data co ming fro m diverse po ints of intelligent electronic devices (IEDs such as digital protective relays, digital fault recorders, SCADA's remote terminal units, sequence of event recorders, circuit breakers, fault locators and I EDs specially used for variety of monitoring and cont rol applications). To help the protection en gineers come to t erm wi th t he crucial necessity and b enefit of protection system a nalysis without the arduous dealing of overwhelming data, using record ed data res ident in d igital p rotective rel ays alone i n an automated approach cal led knowledge discovery i n d atabase (KDD) is certainly of an immense he lp in their protection o peration a nalysis t asks. Digital protective relay, instead of a host of other intelligent electronic devices, is the only device for analysis in this work because it sufficiently provides virtually most attributes needed for data mining process in KDD. Unlike some artificial intelligence aproaches like artificial nueral network and de cision tree i n whi ch t he d ata m ining anal ysis is "population-based" and single si nce it is common t o the entire population of t raining data set, the rough set ap proach adopts a n "individually-event-based" paradigm in which detailed time tracking analysis of relay operation has been successfully performed.