Predictive out-of-step relaying using fuzzy rule-based classification

This paper presents a new approach of out-of-step relaying based on fuzzy classification. The proposed scheme is equipped with one transducer which is fed by local measurements. The out-of-step relay is designed by selecting the appropriate features, partitioning the pattern space into fuzzy subspaces and inducing a fuzzy if-then rule for each subspace by applying the training set. Then, the designed relay is tested through numerical simulation for a typical study system. Compared to other artificial intelligence (AI)-based schemes, the developed scheme exhibits simpler design, faster on-line operation and better reliability. (C) 1999 Elsevier Science S.A. All rights reserved.