[J39] D. Semitekos, N.Avouris, G. Giannakopoulos, A Toolkit for Power systems Security Assessment Based on Hybrid Machine-Learning Techniques, Int. Journal of Engineering Intelligent Systems,2004. (pdf)
In this paper, we present a flexible software environment that facilitates the use of machine learning techniques in power system contingency studies as an alternative to traditional power flow analysis. The architecture of this toolkit, which includes the database repository, and a number of machine learning tools are described. The toolkit approach enables the user to experiment with the predictive powers of various machine-learning tools over various network operating points. The paper covers the findings of a case study performing a sensitivity analysis using the presented software environment.
Power Systems, Contingency Analysis, Operating points repository, Steady State Security Assessment, Machine-Learning, neural networks, decision trees, nearest neighbors.