Fast contingency screening based on data mining

This paper presents a framework for real-time contingency screening using a situational awareness oriented tool. While some contingencies might have a significant impact on power-system stability, the impact of others is negligible. The consequences of a contingency depend on network conditions. Therefore, fast and effective tools for rapid recognition of hazardous operating conditions from the stability point of view are required. For that purpose, a fast method is proposed, which carries out on-line stability assessment with respect to first swing transient stability. The proposed method is segmented into two steps. In the first step, a database is constructed, which consists of a large spectrum of pre-fault operating points, as well as the critical clearing times of several contingencies. In the second step, a pattern recognition method called Principal Components Analysis is applied, in order to develop a metric, which evaluates the similarity of operating conditions with respect to the first-swing stability problem. The IEEE New England system is used to test the proposed method regarding the assessment accuracy and computation time. However, practical implementation of this concept would be closely linked to the development of WAMPAC systems.