On Line Steady State Security Assessment and Enhancement of an Energy System Using Pattern Recognition Techniques

Abstract

Pattern recognition technique (PRT) are now accepted tools for handling situations demanding dimensionality reduction, proper identification of states of system and for arriving at a reliable decision. On line control functions fall in domain of such situations. This report gives a systematic approach to cater to such problems using PRT. A three class, linear classifier is designed. Further, an interactive classifier is described using principal component method, and is compared with other available techniquesTwo separate classifiers are reported for peak and off-peak loading conditions using repetative projection technique. Classifiers are tested on sample system and results are reported. If the decision is mapped into area which confirm need for a corrective control action, the security enhancement algorithm is activated. Revised values of control variables are obtained using gradient projection matrix