A Mechanical Fault Diagnosis Approach for Transformer Winding Core Based on VMD and IGOA-KELM

A transformer winding iron core mechanical fault diagnosis approach in view of improved grasshopper optimization algorithm (IGOA) optimized kernel extreme learning machine (KELM) is proposed to address the issue of challenge in precisely distinguishing transformer winding iron core mechanical faults. Firstly, using the transformer vibration signal as the monitoring measurement, the VMD is used to preprocess and feature separate the primary vibration signal. Secondly, the Pearson correlation coefficient is served as an indicator to calculate the correlation of each order of modal components and select the eigenvalues of modal components. In addition, the multi-scale fuzzy entropy of the selected modal components is calculated. Then, using IGOA to get the most out of the key kernel arguments of KELM, the IGOA-KELM diagnostic blueprint is constructed to achieve a high diagnostic accuracy model. Finally, the calculated MFE values of different states are constructed as feature datasets and input into IGOA-KELM for diagnostic recognition. The experimental data of 10kV transformer proves the excellent diagnostic effect of IGOA-KELM model, with a diagnostic accuracy rate of 99.29%.