摘要
配电变压器油箱表面的振动信号富含绕组和铁芯的各类状态信息,是绕组和铁芯工作状况的最直接体现。采用希尔伯特黄(HHT)带通滤波提取配电变压器振动信号主成分,获得表征绕组状态的100Hz分量及表征铁芯状态的150~1000Hz分量;利用负载电流拟合法提取绕组振动信号的特征量,通过测到的已知振动信号估计指定负载下的绕组100Hz振动幅值,构成绕组振动的特征向量;利用具有良好泛化能力及鲁棒性的双谱分析结合奇异值分解的方法表征铁芯振动的特征。提取实验室试验测得的绕组松动、绕组变形、铁芯松动、铁芯两点接地以及铁芯接地不良等故障振动信号的特征向量,用基于信息融合的支持向量机(SVM)实现绕组和铁芯状态的识别,结果验证了本文方法的有效性和准确性。
The vibration signal on the surface of distribution transformer contains rich state information of the winding and iron core, which directly reflects the working conditions of the winding and iron core. The Hilbert-Huang Transform (HHT) band-pass filtering is adopted to extract the main components of the distribution transformer vibration signal. The 100 Hz and 150 ~ 1 000 Hz signals are obtained, which represent the working conditions of the winding and iron core, respectively. The load current curve fitting method is used to extract the features of the vibration signal, the 100Hz vibration amplitude of the winding under specified load is estimated from the measured vibration signal, and the feature vectors of winding vibration are constructed. The bi-spectrum analysis combined with Singular Value Decomposition (SVD) that has good generalization capability and robustness is adopted to represent the vibration features of the iron core. Large number of feature vectors were extracted from various fault vibration signals tested in the laboratory, such as winding looseness, winding deformation, core looseness, core two-point grounding, poor core grounding and so on. The Support Vector Machine (SVM) based on information fusion was used to achieve the condition identification of the winding and core. The results verify the effectiveness and accuracy of the proposed algorithm.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2016年第6期1299-1308,共10页
Chinese Journal of Scientific Instrument
基金
福建省自然科学基金(2016J01218)项目资助
关键词
配电变压器
振动信号
HHT带通滤波
双谱分析与奇异值分解
SVM分类
故障识别
distribution transformer
vibration signal
Hilbert-Huang transform (HHT) band-pass filtering
bi-spectrum analysis and singular value decomposition (SVD)
support vector machine (SVM) classification
fault identification