基于数字孪生的变压器故障诊断方法研究  

Research on Transformer Fault Diagnosis Method Based on Digital Twin

作　　者：江伦 王大江 孙文磊[1] 包胜辉 刘涵 常赛科 Jiang Lun;Wang Dajiang;Sun Wenlei;Bao Shenghuil;Liu Han;Chang Saike(School of Intelligent Manufacturing Modern Industry,Xinjiang University,Urumqi 830046,China;TBEA Co.,Ltd.,Tianjin 300000,China)

机构地区：[1]新疆大学智能制造现代产业学院,新疆乌鲁木齐830046 [2]特变电工有限公司,天津300000

出　　处：《系统仿真学报》2025年第3期775-790,共16页Journal of System Simulation

基　　金：工业互联网标识解析全要素集成平台项目(TC210A02E)。

摘　　要：针对已有的变压器故障诊断智能算法并不能快速高效的识别变压器故障,导致故障误检及不能被及时检测的问题,提出了一种利用改进麻雀优化算法优化XGBoost的双层故障诊断模型结合数字孪生技术的变压器故障诊断方法。采用先进传感器采集变压器的油气数据和温度数据,利用5G模块将实时数据传到数字孪生系统,设置设备告警阈值实时监控温度数据;使用改进麻雀优化算法优化XGBoost的双层故障诊断模型对油气数据进行实时故障识别处理,结合数字孪生技术对故障进行识别与预警。实验结果表明:该方法提高了故障识别与预警的效率和稳定性,且相较于现有的变压器故障诊断方法具有显著优势。Aiming at the inability of existing intelligent algorithms for transformer fault diagnosis to quickly and efficiently identify transformer faults,resulting in fault misdetection and untimely detection,this paper proposes a transformer fault diagnosis method using the improved sparrow optimization algorithm to optimize the two-layer fault diagnostic model of XGBoost combined with the digital twin technology.The method adopts advanced sensors to collect oil and gas data and temperature data of the transformer,uses 5G module to transmit the real-time data to the digital twin system.The system monitors the temperature data in real-time by setting the equipment alarm threshold;optimizes the twolayer fault diagnostic model of XGBoost using the improved sparrow optimization algorithm to process real-time fault identification of the oil and gas data,and finaly identifies and warns of faults by combining with the digital twin technology Experimental results indicate that this method significantly enhances the efficiency and stability of fault identification and early warning,demonstrating substantial advantages compared to existing transformer fault diagnosis methods.

关 键 词：变压器 数字孪生 故障诊断 实时监控 预警 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置] TM41[自动化与计算机技术—控制科学与工程]