Fault Diagnosis of On-Load Tap-Changer Based on the Parameter-adaptive VMD and SA-ELM

Abstract

Mechanical vibration signal can reflect the running state of on-load tap-changer. In order to realize effective mechanical fault diagnosis for on-load tap-changer, a fault diagnosis method based on the parameter-adapted Variational Mode Decomposition (VMD) and Extreme Learning Machine optimized by Simulated Anneal (SA-ELM) is proposed. Firstly, the signal is decomposed by VMD method, and the number of modals is selected based on energy criterion. A group of modal components with narrow band and great discrimination is obtained. Then the energy features of each modal component are calculated, which form the feature vector group, and the modal features of different fault states are clearly distinguished. Finally, the feature vector group is inputted to the extreme learning machine (ELM) optimized by simulated annealing algorithm to realize the recognition and fault diagnosis of the vibration signals. An experiment is carried out on the simulation experiment platform and the collected signals are processed. Compared with the method based on VMD and ELM, the fault diagnosis method proposed can effectively improve the diagnostic accuracy of mechanical fault of on-load tap-changer.