Artificial intelligence in OLTC fault diagnosis using dissolved gas-in-oil information

Abstract

On-load tap changers (OLTCs) are one of the most problematic components of power transformers. Detecting incipient faults in OLTCs is one of the key challenges facing the power equipment predictive maintenance community. This paper addresses the issue with an artificial intelligence approach, where logistic analysis is used to find the principal gases related to the fault conditions and neural networks are used to improve the performance of the diagnosis. The developed techniques could be integrated into a power transformer incipient fault diagnosis systems.