Artificial Intelligence Applied to On-line Transient Stability Assessment of Electric Power Systems

Abstract

A novel approach to fast transient stability assessment of power systems is presented. It aims at building decision rules, expressed in terms of static, directly or indirectly, controllable variables, appropriate for both on-line analysis and preventive control. The decision rules are constructed by using inductive inference in conjunction with large bodies of off-line simulation data. The proposed inductive inference procedure derives from ID3 by Quinlan, appropriately modified so as to handle the large number of continuous variables characterizing power system operating points. This modification is achieved by developing a new quantization procedure, based upon the entropy concept. The method is illustrated by means of an academic example of a one-machine-infinite-bus system.