The Fault Identification Method for Distribution Transformer based on Support Vector Machine Classification of Vibration Signal Characteristics
Wei Xiaoying1, Song Shijiang2, Guo Moufa1, Lu Guoyi3
1. College of Electrical Engineering and Automation, Fuzhou University, Fuzhou 350000; 2. Shaowu Electric Power Supply Company, State Grid Fujian Electric Power Co., Ltd, Shaowu, Fujian 354000; 3. Fuzhou Metro Co., Ltd, Fuzhou 350000
Abstract：The tank surface of distribution transformer contains a wealth of vibration signals from the iron core, which can directly reflect the working conditions of the core. Extracting the principal component of the core from vibration signals via Hilbert-Huang Transform (HHT) band-pass filter, and then the vibration signal is decomposed in time-frequency domain via the second band-pass filter of HHT, calculating the energy and center frequency of each sub-band reconstructed signal, which constitute the 2-D feature vector of the vibration signal. The vibration signal of the core in 4 typical conditions including normal states, two-point grounding, looseness and poor grounding are measured through no-load experiment, SVM classification is applied to these 2-D feature vectors. The result shows that the feature vector can represent each state of the core accurately and effectively.
魏晓莹, 宋仕江, 郭谋发, 卢国仪. 基于振动信号二维特征向量的配变铁心故障诊断[J]. 电气技术, 2016, 17(1): 16-15.
Wei Xiaoying, Song Shijiang, Guo Moufa, Lu Guoyi. The Fault Identification Method for Distribution Transformer based on Support Vector Machine Classification of Vibration Signal Characteristics. Electrical Engineering, 2016, 17(1): 16-15.
2016, Vol. 17 
