魏敏,王刘旺
WEI Min,WANG Liuwang

• 1:国网浙江省电力有限公司嘉兴供电公司
• 2:国网浙江省电力有限公司电力科学研究院

摘要(Abstract)：

针对实际运行环境下变压器有载分接开关运行工况识别效果不佳的问题，提出一种基于MSSST（多重同步压缩S变换）和RLCNN（强化轻量级卷积神经网络）的工况识别方法。在该方法中，将MSSST理论引入电力设备状态监测领域，用于对有载分接开关振动信号进行分析处理，从而有效刻画信号的二维时频特征。此外，在MobileNetv2轻量级卷积神经网络中融合Adaboost自适应提升机制，提出一种新颖的RLCNN模型，以振动信号二维时频图作为样本对所构建的RLCNN模型进行训练，用于判定有载分接开关运行工况。实验结果表明，所提方法可实现有载分接开关不同运行工况的准确判定，与其他识别方法相比，该方法识别准确性更高、稳定性更好，具有实际工程应用价值。
The operating condition identification of on-load tap changer under actual service environment yields no desired effect. To solve this problem,the paper proposes an operating condition recognition method based on MSSST(multi-synchronous squeezing S transform)and RLCNN(reinforced lightweight convolution neural network) is proposed. In this method,the multi-synchronous squeezing S transform is firstly introduced into the field of power equipment condition monitoring and applied to analyze the vibration signal of on-load tap changer so that the twodimensional time frequency characteristic of signal can be effectively depicted. In additional,the MobileNetv2 lightweight convolution neural network is fused with the Adaboost adaptive lifting mechanism,and a novel RLCNN model is proposed. Then the two-dimensional time frequency maps of vibration signal are regarded as the samples to train the model,which is used to judge the operating condition of on-load tap changer. Experimental results show that the method can accurately judge the different operating conditions of on-load tap changer. Compared with other identification methods,this method has a higher precision rate and better stability as well as practical engineering application value.

关键词(KeyWords)： 有载分接开关;多重同步压缩S变换;Adaboost自适应提升机制;强化轻量级卷积神经网络;工况识别
on-load tap changer;MSSST;Adaboost adaptive lifting mechanism;RLCNN;operating condition identification

Abstract:

Keywords:

基金项目(Foundation): 国网浙江省电力有限公司科技项目（5211DS210007）

作者(Author): 魏敏,王刘旺
WEI Min,WANG Liuwang

参考文献(References)：

• [1]周翔，王丰华，傅坚，等.基于混沌理论和K-means聚类的有载分接开关机械状态监测[J].中国电机工程学报，2015,35(6):1541-1548.
• [2]徐亮，孙瑜，王广，等.变压器有载分接开关CMA9型电动机构故障的分析[J].浙江电力，2012,31(3):66-68.
• [3]李予全，寇晓适，郑含博，等.有载分接开关油箱渗漏导致变压器油色谱数据异常诊断[J].中国电力，2017,50(3):133-136.
• [4] SEO J,MA H,SAHA T K.A joint vibration and arcing measurement system for online condition monitoring of onload tap changer of the power transformer[J].IEEE Transactions on Power Delivery,2017,32(2):1031-1038.
• [5]韩帅，高飞，王博闻，等.基于Mel频谱滤波和CNN的有载分接开关可听声辨识方法[J].电网技术，2021,45(9):3609-3617.
• [6]高良玉，郭文元，李秀涛.有载分接开关温度的在线监测报警装置的研制[J].高电压技术，2001,27(5):24-26.
• [7]杨森，陈莎莎，李光茂，等.基于变分模态分解与特征选择的变压器有载分接开关机械故障诊断[J].南方电网技术，2019,13(1):39-47.
• [8]张知先，陈伟根，汤思蕊，等.基于互补集总经验模态分解和局部异常因子的有载分接开关状态特征提取及异常状态诊断[J].电工技术学报，2019,34(21):4508-4518.
• [9] SUN G D,HU Y,WU B,et al.Rolling bearing fault diagnosis method based on multisynchrosqueezing S transform and faster dictionary learning[J]. Shock and Vibration,2021,2021:1-13.
• [10]曾全昊，王丰华，郑一鸣，等.基于卷积神经网络的变压器有载分接开关故障识别[J].电力系统自动化，2020,44(11):144-151.
• [11]卓德兵，曹晖.基于小波时频图与轻量级卷积神经网络的螺栓连接损伤识别[J].工程力学，2021,38(9):228-238.
• [12] SANDLER M,HOWARD A,ZHU M L,et al.MobileNetV2:inverted residuals and linear bottlenecks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City,UT,USA:IEEE,2018:4510-4520.
• [13] BARANIUK R G,FLANDRIN P,JANSSEN A J E M,et al.Measuring time-frequency information content using the Renyi entropies[J].IEEE Transactions on Information Theory,2001,47(4):1391-1409.
• [14] WANG A L,HE X,GHAMISI P,et al.LiDAR data classification using morphological profiles and convolutional neural networks[J].IEEE Geoscience and Remote Sensing Letters,2018,15(5):774-778.
• [15] FREUND Y,SCHAPIRE R E.A decision-theoretic generalization of on-line learning and an application to boosting[J]. Journal of Computer and System Sciences,1997,55(1):119-139.
• [16]钱国超，彭庆军，程志万，等.基于参数自适应VMD和SA-ELM的有载分接开关故障诊断[J].电力工程技术，2020,39(1):157-164.
• [17]周建平，李聪，万书亭，等.基于优化型SVM的高压断路器故障诊断方法研究[J].浙江电力，2019,38(3):14-19.
• [18]王福忠，石秀立.AGA-BP神经网络的变压器分接开关机械故障诊断[J].现代制造工程，2017(4):11-16.
• [19]车一鸣，王冬梅，王国兴，等.基于IMDE和ORF模型的断路器工况识别[J].组合机床与自动化加工技术，2019(12):85-90.