Dry-Type Transformers Fault Diagnosis Method Based on Signal Wavelet Transform and DBN

Traction power supply system is a key infrastructure of urban rail transit, and dry-type transformer is an important equipment in traction power supply system, whose operation status is crucial to the safe and reliable operation of the whole power supply system. For the fault diagnosis of dry-type transformers, a fault diagnosis algorithm based on signal decomposition is proposed, and noise reduction is performed by reconstruction and hybrid feature extraction. First, the vibration signal is collected by vibration sensors, and the signal decomposition and reconstruction method based on wavelet decomposition is proposed for the noise reduction problem of the original signal. This method solves the problem that the traditional wavelet noise reduction method will lose feature information. Secondly, four relatively stable features are selected in the time domain and frequency domain respectively to form a new two-dimensional feature matrix, which enriches the fault feature information of the dry-type transformer. Finally, the extracted signals are put into the DBN deep confidence network for fault diagnosis.