A rule-based fuzzy logic approach for the voltage collapse risk classification

In recent years, an increasing number of voltage stability indicators have been proposed for voltage collapse assessment. A lot of them are determined by very complex analytical tools and are difficult to interpret by system operators. In the present work, a different direction has been followed: Artificial intelligence (AI) approaches have been exploited, based on fuzzy logic (FL) and artificial neural network (ANN) support. A decision model built on FL has been developed. It receives as input a given set of numerical variables, which are collected to represent a snapshot of the actual operating point for the power system. The set of numerical values is translated into a set of symbolic and linguistic quantities. These variables are manipulated by a set of logical connectives and inference methods provided by mathematical logic. As a final result, the FL approach gives a measure in a percent rate of the security level degradation with respect to the voltage collapse risk. The settled fuzzy inference engine has been built and optimised by utilizing, as a test system, an appropriate equivalent of the EHV Italian transmission network. The results obtained with the FL approach are compared with the ones given by a conventional analytical tool.