Non-observability analysis by means of an evolutionary technique in measured Electric Power Systems

The observability characterization of an Electric Power System (EPS) from a topological point of view with respect to a given measurement acquisition system is equivalent to the existence of a certain spanning tree. In previous work, a genetic algorithm was developed in order to address this issue. In this paper the behavior of this evolutionary algorithm is studied by means of probabilistic methods. Although the main purpose of the algorithm is to find a tree, the determination of the non-existence of the tree due to the uncertainty inherent to evolutionary techniques was addressed using statistical hypothesis testing. This allows to characterize, with a given certainty level, the non-existence of such a solution and, therefore, the non-observability of the EPS. The techniques developed in this paper were tested over two benchmark systems: the IEEE networks with 118 and 300 nodes.