Dry-Type AMDT Fault Detection Based on Vibration Signal Analysis by Wavelet Packet

Abstract

Amorphous metal distribution transformers (AMDT) are widely used in power grids due to their low no-load loss. Many scholars have carried out research on the fault detection of transformer windings, tap changers and the other parts. However, due to the high magnetostriction of the amorphous alloy, the vibration generated by AMDT during operation will cause various mechanical failures. This paper studies the vibration characteristics of SCBH 15-200/100 AMDT through no-load tests to find some mechanical failures of AMDT. The installation position of the vibration sensor in AMDT are determined according to finite element analysis (FEA) of the magnetic flux density distribution and modal analysis, and the vibration analyses are performed under different operating conditions of AMDT. The wavelet packet transform (WPT) is used to perform detailed analysis of the vibration signal in the time domain and frequency domain to obtain the energy characteristic value of each frequency band, and it includes the frequency spectrum and waveform data under normal and fault conditions. After obtaining the energy characteristic thresholds of different frequency bands under different conditions, the operating status can be detected by comparing test data with the thresholds. The operation condition including mechanical failures induced by magnetostrictive actions can be accurately determined by the energy characteristic value, such as loose nuts and stress, etc.