摘要
变压器声纹信号包含大量反映内部机械状态的有效信息。为实现变压器内部机械状态不停电检测,提出一种基于特征筛选和改进深度森林的变压器机械状态声纹识别方法。首先,利用自适应噪声完备集合经验模态分解(CEEMDAN)声纹信号得到本征模态函数(IMF),并通过频谱分析和皮尔逊相关系数对IMF分量进行筛选,得到包含故障信息的IMF分量。其次,利用各IMF分量在频段上的分布情况进行高、低频段划分,依据高、低频段IMF分量的差异性,将高频段IMF分量的时频能量和低频段IMF分量的幅值特性作为特征指标,构成特征向量,输入改进后的深度森林模型,得到10种机械松动状态的声纹识别结果。最后,通过现场试验验证了该方法的有效性。研究结果表明:所提方法对10种机械松动状态的平均识别准确率达99.2%。与传统变压器声纹特征相比,所提声纹特征区分度更高;与传统识别模型相比,所提改进深度森林识别模型复杂度更低,训练速度更快,识别准确率更高。
Transformer voiceprint signal contains a lot of effective information reflecting the internal mechanical state.In order to realize uninterrupted detection of internal mechanical state of transformer,a voiceprint recognition method of transformer mechanical state based on feature screening and improved deep forest is proposed.Firstly,the intrinsic mode function(IFM) is obtained by decomposing the voiceprint signal with the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN),and the IMF component containing the fault information is obtained by filtering the IMF components through spectrum analysis and Pearson correlation coefficient.Secondly,the distribution of each IMF component in the frequency band is used to divide the high and low frequency bands.According to the difference of the IMF components in the high and low frequency bands,the time-frequency energy of the IMF component in the high frequency band and the amplitude characteristic of the IMF component in the low frequency band are used as characteristic indicators to form a feature vector,which is input into the improved deep forest model,and the voiceprint recognition results of 10 mechanical loose states are obtained.Finally,the effectiveness of the method is verified by field experiments.The research results show that the average recognition accuracy of the proposed method is 99.2%for 10 mechanical loose states.Compared with the traditional transformer voiceprint feature,the proposed voiceprint feature has higher discrimination;Compared with the traditional recognition model,the proposed improved deep forest recognition model has lower complexity,faster training speed and higher recognition accuracy.
作者
李楠
马宏忠
张玉良
段大卫
崔佳嘉
何萍
LI Nan;MA Hongzhong;ZHANG Yuliang;DUAN Dawei;CUI Jiajia;HE Ping(College of Energy and Electrical Engineering,Hohai University,Nanjing 211100,China;State Grid Nanjing Power Supply Company,Nanjing 210008,China)
出处
《电机与控制应用》
2022年第9期57-65,74,共10页
Electric machines & control application
基金
国家自然科学基金项目(51577050)
国网江苏省电力公司科技项目(J2021053)。
关键词
电力变压器
特征筛选
深度森林
模态分解
声纹识别
power transformer
feature screening
deep forest
mode decomposition
voiceprint recognition