Neural network based transformer incipient fault detection

Abstract

The most common diagnosis method for power transformer faults is the dissolved gas analysis (DGA) of transformer oil. Various methods have been developed to interpret DGA results such as key gas method, and roger's ratio method. The present approach utilizes IEC 60599 ratio method to discriminate fault in transformers, which is having the advantage of usage of three gas ratios instead of four gas ratios used in other ratio methods. In some cases, the DGA results cannot be matched by the existing codes, making the diagnosis unsuccessful in multiple faults. To overcome this, the authors have proposed the use of neural networks to highlight their ability to detect the incipient faults in transformer.