Research on feature extraction and analysis of voiceprint signals in different operating states of transformers

During the operation of transformers, due to electrical, magnetic, mechanical, and other stresses, the iron core, winding, control mechanism, etc. will vibrate and generate mechanical waves. The vibration sound signal generated through the propagation of insulating media and cavities contains a large amount of equipment status information. When the core and winding of a transformer malfunction or operate abnormally, the vibration voiceprint will change. This article studies the development method of a transformer voiceprint signal acquisition device, and uses the device to record voiceprint signals of transformers under different faults and abnormal operating states. The voiceprint characteristics of transformers under normal operation, winding deformation, partial discharge, loose iron core, multi-point grounding of iron core, DC bias and other abnormal operating states are analyzed, which can provide feature values for subsequent artificial intelligence classification algorithm training.