Research On Transformer On-line Monitoring System Based On Vibration Signal Analysis

As one of the most important equipment in the power system,the operating condition of the transformer is directly related to the safe and stable operation of the entire power system.During the operation of the transformer for a long time,there will be various faults caused by the abnormal parts of the transformer,which will cause great economic losses.In order to prevent various hidden troubles of the transformer and improve the reliability of the power system,this paper researches on-line transformer monitoring system based on vibration signal analysis.This paper designs and implements an on-line transformer vibration monitoring system based on vibration signal analysis.The system is divided into a front-end acquisition system and a background analysis system.The front-end acquisition system is designed based on NI’s Compact RIO controller,and the supporting data acquisition card and sensor are selected.Lab VIEW and related software modules are used to jointly develop the data acquisition,data storage,and host computer communication functions.The design is divided into two parts: the communication and data processing subsystem and the web real-time monitoring and health evaluation subsystem.The communication and data processing subsystem is based on the C # language and implements functions such as receiving data,data processing,and database storage.The web real-time monitoring and health The evaluation subsystem is designed and developed based on the HTML5 standard,and realizes four major functional modules: online monitoring,vibration data analysis,threshold alarm,and system information.The combination of front-end acquisition system and background analysis system realizes real-time monitoring and analysis of transformer vibration signals.In order to solve the problem of transformer operating condition analysis under the premise of no fault samples,this paper proposes a fundamental frequency(100Hz)amplitude prediction method based on a recurrent neural network(RNN).This method predicts the fundamental frequency amplitude of the transformer under normal conditions based on historical operating voltage,load current,and oil temperature three operating conditions,and then judges the operating conditions of the transformer by comparing the predicted results with the measured values.The simulation experiments are compared with the existing formula method and GRNN-based prediction method.The experimental results show that the fundamental frequency amplitude prediction method based on RNN has lower prediction error and higher accuracy,which is an important reference for online monitoring of transformers based on vibration signal analysis value.