Feature extraction of the transformer core loosening based on variational mode decomposition

Abstract

The core is the key component in the transformer. The core loosening is one of common faults of the transformer, which will cause the noise and vibration obviously increasing then lead to facility damages. The feature extraction method for the transformer core loosening based on variational mode decomposition is proposed in this paper to analyze the core loosening vibration signal which has non-stationary and nonlinear characteristics. The variational mode decomposition is a new and adaptive time-frequency analysis method. This method is able to separate a multi-component signal into many single-component signals. The vibration signal in the loose conditions is obtained by changing the clamping pressure of the core in the experiment. The variational mode decomposition method is used to decompose the core vibration signal. The Hilbert transform is applied to each variational intrinsic mode functions and then the Hilbert spectrum of the core vibration signal is obtained. The time-frequency features of the core loosening vibration signal are extracted, which lays a solid foundation for diagnosing the fault of the transformer.