A Monte Carlo-rough based mixed-integer probablistic non-linear programming model for predicting and minimizing unavailability of power system components

Data mining processes were initially invented and developed for finding out meaningful information and deriving rules from more recent data sets. With growth in the size of power systems due to the increased requirements of industries and cities for uninterrupted power supply, data mining in electrical systems has emerged as a very efficient tool for continuous assessment of power systems. In this paper we propose Rough Set based knowledge discovery for predicting faults and probability of failure in high voltage equipment present in a real electrical systems. The approach has several different steps viz. analysis of real time data, creation and population of databases, pre-processing of data and use of rough sets based data mining algorithm to finally determine the set of rules for knowledge discovery. The methodology has been validated by presenting a case study and application of the algorithm on real time data.