Application of Kohonen self-organising neural network to static security assessment

Static security analysis is one of the important control functions in modern energy management systems. Its main objective is to access the existing operating state of the system and if the state is secure it then performs a set of contingency assessment. The traditional methods of assessing a contingency is to use a full ac load flow. The paper presents a new method for static security assessment with a Kohonen self-organising neural network. The Kohonen neural network can identify similarities of system states in an unsupervised manner and form a self-organising feature map for the classification of security states of power systems with respect to contingency analysis. The voltage magnitude of each busbar and active power flow of transmission lines are chosen as input features which can represent the complete operating condition of a power system. Contingency selection and security evaluation can be achieved simultaneously in the new proposed method