摘要
有载分接开关(on-loadtapchanger,OLTC)作为变压器的核心组件易发生机械故障,为实现其机械状态的不停电检测,文章提出一种基于Mel频谱滤波与卷积神经网络(convolutional neural network,CNN)结合的OLTC机械故障可听声辨识方法。首先搭建110kV OLTC故障模拟平台,物理模拟传动机构卡涩和内部组件松动故障,并进行可听声信号采集;其次在变电站现场声源分析的基础上,采用基于相似矩阵的盲源分离法将OLTC动作可听声音号与变压器本体运行噪声进行分离,提高信噪比;再次,根据OLTC可听声信号的能量分布特性,采用Mel频谱滤波法对原始信号进行降维,有效提升了处理效率;最后引入CNN通过超参数调整和网络结构优化设计构建可听声辨识模型,实现OLTC机械故障的识别。研究结果表明:该方法对OLTC传动机构卡涩和内部组件松动故障具有较好的识别成功率和运算效率,为OLTC机械状态现场不停电监测与故障诊断提供了有效参考。
On-load tap changer(OLTC)as one of the core components of the transformer is prone to mechanical failure.In order to realize the on-line detection in the mechanical state of the on-load tap changer,an audible sound identification method combining the Mel spectrum filter with the CNN is proposed.Firstly,a 110 kV OLTC fault simulation platform is built to physically simulate transmission mechanism jamming and internal component loosening faults,and the sound signals are collected.Secondly,on the basis of on-site sound source analysis of the substation,the blind source separation method based on similarity matrix is used to separate the audible sound signal of the OLTC action from the noise of the transformer operation.Then,according to the energy distribution characteristics of the OLTC audible signal,the Mel spectrum filtering method is used to reduce the dimension of the original signal,which effectively improves the processing speed.Finally,the CNN is introduced through hyper-parameter adjustment and network structure optimization design,and the OLTC audible identification model is constructed to realize the identification of the mechanical fault in the OLTC.The research results show that this method has a good recognition success rate and calculation efficiency for OLTC transmission mechanism jamming and internal component loosening failures,and it provides an effective reference for OLTC’s power failure diagnosis.
作者
韩帅
高飞
王博闻
刘云鹏
王康
吴达
张晨晨
HAN Shuai;GAO Fei;WANG Bowen;LIU Yunpeng;WANG Kang;WU Da;ZHANG Chenchen(China Electric Power Research Institute,Haidian District,Beijing 100192,China;Hebei Provincial Key Laboratory of Power Transmission Equipment Security Defense(North China Electric Power University),Baoding 071003,Hebei Province,China;State Grid Fujian Electric Power Research Institute,Fuzhou 350007,Fujian Province,China;State Grid Anhui Electric Power Research Institute,Hefei 230022,Anhui Province,China)
出处
《电网技术》
EI
CSCD
北大核心
2021年第9期3609-3617,共9页
Power System Technology
基金
国家电网有限公司科技项目(5200-201955095A-0-0-00)。
