Combined use of unsupervised and supervised learning for dynamic security assessment

It is desirable to assess the security and stability of electric power systems after exposure to large disturbances. In this connection, the critical clearing time (CCT) is an attribute which provides significant information about the quality of the postfault system behavior. It may be regarded as a complex mapping of the prefault, fault-on, and postfault system conditions into the time domain. High prediction and generalization capabilities of artificial neural networks provide the basis for synthesis of such a complex mapping carrying input pattern attributes into the single valued space of the CCT. The authors consider the possibility of using unsupervised and supervised learning programs to discover what combination of raw measurements is significant in determining CCT. Correlation analysis and a Euclidean metric are used to specify interfeature dependencies. An example of a four-machine power system is used to illustrate the suggested approach