Artificial neural networks for power system static security assessment

An artificial neural network (ANN) is used to assess the static security of a test system. It is demonstrated that an ANN can be a useful tool for static security assessment of power systems. It is shown that ANNs perform significantly better than a nearest-neighbor search in terms of classification, recall time, and data storage requirements. The ANN, however, requires a great deal of time for offline training. This problem is compounded as the system size increases. Learning complexity theory can be used to better understand this scaling problem. Alterations which may lead to better performance include accelerated learning algorithms and the use of oracle-based learning.<<ETX>>