Information fusion method of multi-data resources and its application to fault diagnosis in power system

Abstract

Fast and accurate fault diagnosis is a necessary precondition of isolating fault components and restoring a faulted power system. The existent fault diagnosis methods based on the protective relay and circuit breaker acting information do not take full advantage of the electrical measurements. The electrical measurements can provide more direct information for fault diagnosis, but they have been rarely used in the existing fault diagnosis methods. A new fault diagnosis method using both two kinds of information and the D-S theory is presented. Following the idea of information fusion, PN fault degree obtained by fuzzy Petri net while wavelet energy variation degree and wavelet singularity variation degree by wavelet transform are extracted for power system fault diagnosis. Then fault diagnosis is conducted by using information fusion based on D-S evidence theory and the basic probability assignment is set up. While a decision-making method based on the basic probability number is used to diagnose the fault elements. Simulations of a practical system show that the proposed diagnosis method can effectively diagnose the fault elements, when errors occur in the protective relay and circuit-breaker alarm messages.