On-line state monitoring of distribution transformer windings with the vibration method depends on extraction of effective features from the vibration signals. In order to monitor and diagnose the state of the transformer windings more effectively

the authors built an on-line monitoring platform of vibration signals for the short-circuit impulse test and loading test on a power transformer

analyzed the vibration signals of the transformer winding via the ensemble empirical mode decomposition（EEMD）

and obtained the energy entropy of the vibration signals.Consequently

a state monitoring and mechanical fault diagnosis method based on EEMD energy entropy was proposed for distribution transformer windings. Experimental results show that EEMD can effectively extract the features of the vibration signals of distribution transformer windings

obtain the energy distribution within each frequency band of the vibration signal

and help to diagnose transformer winding faults accurately.