Cirio Diego*, Alberto Berizzi**, Cristian Bovo**, Maurizio Delfanti**, Marco Merlo** Massimo Pozzi*** Rivista: Electric Power Systems Research, Amstredam * CESI RICERCA ** POLITECNICO DI MILANO *** CESI SpA Many voltage stability indicators have been proposed in the past for the voltage collapse assessment. Almost all of them are determined through quite complex analytical tools; therefore, it is difficult for system operators to give them a physical meaning. In order to perform a simple and reliable evaluation of the security margins, it is necessary to make a synthesis of the information given by the various indices. The present work proposes an Artificial Intelligence-based tool for the evaluation of the voltage security. In particular, a Fuzzy Inference Engine is developed and optimized by two different approaches (Neural Networks and Genetic Algorithms). Starting from the state estimation, a given set of mathematical indices is computed to represent a snapshot of the current electric system operating point. The numerical values are then translated into a set of symbolic and linguistic quantities that are manipulated through a set of logical connectives and Inference Methods provided by the mathematical logic. As a result, the Fuzzy Logic gives a MW measure of the distance from the collapse limit, a metric usually appreciated by system operators. The Fuzzy System has been built and optimized by using, as a test system, a detailed model of the EHV Italian transmission network connected to an equivalent of the UCTE network (about 1700 buses).