[C64] Semitekos D. Avouris N., Contingency Analysis Based on a Hybrid Machine Learning Approach, Proc. IEEE ISAP 2003, Lemnos, July 2003. (pdf)
In this paper we present an innovative approach for power systems - contingency risk assessment. A number of stand-alone and hybrid machine learning tools are proposed for off-line steady state network operation, making the use of classical load flow studies not necessary. This paper presents an outline of the proposed methodology, the network indices used as well as an interpretation of the experimental results. A hybrid solution that combines a number of machine learning approaches, is demonstrated. It produces contingency predictions with a higher accuracy than the other stand-alone machine learning tools.
Contingency Analysis, Decision Trees, Neural Networks, k-Nearest Neighbors and Hybrid Machine Learning Tools.