Intelligent decision support for diagnosis of incipient transformerfaults using self-organizing polynomial networks

摘要

To serve as an intelligent decision support for the transformernfault diagnosis, a new self-organizing polynomial networks (SOPNs)nmodeling technique is proposed and implemented in this paper. Thentechnique heuristically formulates the modeling problem into anhierarchical architecture with several layers of functional nodes ofnsimple low-order polynomials. The networks handle the numerical,ncomplicated, and uncertain relationships of dissolved gas contents ofnthe transformers to fault conditions. Verification of the proposednapproach has been accomplished through a number of experiments usingnpractical numerical diagnostic records of the transformers of Taiwannpower (Taipower) systems. In comparison to the results obtained from thenconventional dissolved gas analysis (DGA) and the artificial neuralnnetworks (ANNs) classification methods, the proposed method hits beennshown to possess far superior performances both in developing thendiagnosis system and in identifying the practical transformer faultncases