Abstract:

Sound and vibration are the direct carriers of mechanical information.The promotion of vibro-acoustic detection technology provides an effective means for the state evaluation of the main equipment of the power grid and the diagnosis of mechanical defects.The development of artificial intelligence technology provides unprecedented opportunities for intelligent diagnosis of power grid main equipment based on voiceprint detection.This paper analyzes the vibro-acoustic signal of the main transformer and clean the initial voiceprint data of the transformer through multi-event detection technology,which applies artificial intelligence and data-driven algorithms to extract and mine the voiceprint characteristics of the equipment in different operating conditions,as well as extracting the corresponding relations between the high active value and the voiceprint of the transformer and explores effective ways to characterize its operating status.

Key words: data driven, multi event detection, deep learning, voiceprint recognition