Transformer oil diagnosis using fuzzy logic and neural networks

Abstract

Dissolved-gas analysis (DGA) is widely used for detection and diagnosis of incipient faults in large oil-filled transformers. Many factors contribute to extreme "noisiness" in the data and make early fault detection and diagnosis difficult. This paper shows how fuzzy logic and neural networks are being used to automate standard DGA methods and improve their usefulness for power transformer fault diagnosis. The use of neural networks for DGA-with or without fuzzy logic-is discussed, and some related work is described briefly.< >