Typical Mechanical Fault Diagnosis of On-load Tap-changers Based on VMD-Hilbert Transform and 1D-CNN-SVM

The healthy operating state of the on-load tap changer (OLTC) plays a significant role in the safe and stable operation of the power system. Nevertheless, the nonlinear vibration signals of the on-load tap changer pose substantial challenges to fault diagnosis. The primary obstacle lies in the arduous task of extracting the characteristic information of diverse faults. Hence, to surmount this difficulty, this article presents an OLTC mechanical fault diagnosis approach that integrates a VMD-Hilbert transform feature extraction model with a 1D-CNN-SVM fault discrimination model. Firstly, the vibration signals are processed by means of the Variational Mode Decomposition (VMD) technique, and then K Intrinsic Mode Functions (IMF) are obtained. From these, the optimal IMF component that contains the richest information of the original signal is selected. Subsequently, the Hilbert transform operation is performed on this optimal IMF component, so as to obtain its Hilbert spectrum and marginal spectrum. Finally, the Hilbert spectral entropy, the marginal spectral energy entropy and the time-domain characteristic indicators are calculated to construct the feature vector, which is then input into the 1D-CNN-SVM model to fulfill the task of fault identification. After multiple tests, the accuracy rate of this model in fault classification is above 96%. In conclusion, the mechanical fault diagnosis method for OLTC proposed in this paper, which is based on the VMD-Hilbert transform and 1D-CNN-SVM, demonstrates a relatively accurate fault recognition rate and has application value.