Intrinsic Mode selection of transformer vibration signal based on correlation coefficient

In order to extract effectively the characteristics of every condition of transformer from vibration signal, a sensitive Intrinsic Mode Function (IMF) selection method which based on Ensemble Empirical Mode Decomposition (EEMD) is proposed. First, the transformer vibration signal is decomposed by using EEMD, and the sensitive components of obtained IMFs are extracted by using correlation coefficient. Then, the feature vector is constructed with the IMF energy entropy, which is then used as a criterion for the transformer winding state identification. The experimental results verify the validity of the method for fault diagnosis of transformer winding.