A hierarchical clustering based pattern recognition scheme for power system security assessment

A pattern recognition method for power system security assessment is presented. The security status of the power system operation is recognized from the stored knowledge about similar operating condition patterns. The method not only classifies system operating conditions into a secure/insecure state but for an insecure operating state it also identifies the pertinent failure modes. The knowledge about the system operating conditions (patterns) are stored in a structured memory by grouping similar patterns into clusters which are then arranged into a hierarchical tree structure. This permits a fast two level search for the nearest neighbor. For patterns near the separating region of secure and insecure patterns a fuzzy estimation technique is used to provide an error-free security assessment