Multiresolution signal decomposition: a new tool for fault detection in power transformers during impulse tests

Abstract

Detection of major faults in power transformers during impulse tests has never been an issue, but is rather difficult when only a minor fault, say a sparkover between adjacent coils or turns and lasting for a few microseconds, occurs. However, detection of such a type of fault is very important to avoid any catastrophic situation. In this paper, the authors propose a new and powerful method capable of detecting minor incipient faults. The approach is based on wavelet transform analysis, particularly the dyadic-orthonormal wavelet transform. The key idea underlying the approach is to decompose a given faulty neutral current response into other signals which represent a smoothed and detailed version of the original. The decomposition is performed by the multiresolution signal decomposition technique. Preliminary simulation work demonstrated here shows that the proposed method is robust and far superior to other existing methods to resolve such types of faults.