Multiscale multivariate fuzzy entropy‐based technique to distinguish transformer magnetising from fault currents

As vectors' similarity is defined on the basis of the hard boundary of Heaviside function in multiscale multivariate sample entropy, the two families of statistics show high sensitivity to the similarity tolerance. Multiscale multivariate fuzzy entropy (MMFE), which applied fuzzy membership function, overcomes the problem. Hence, the visualisation inrush current discrimination algorithm based on MMFE is proposed. The proposed algorithm divides the MMFE‐ τ plane into two parts: active and braking zones. The ratio of the current entropy area to the given braking zone area is designated as the entropy area ratio. The magnetising inrush and fault current of the transformer are identified on the basis of the numerical value comparison between the entropy area ratio and the fixed value of 1. Through the ATP/EMTP platform and dynamic testing system, the transformer currents under different operating conditions are analysed. Results demonstrate that the proposed algorithm can effectively distinguish the magnetising inrush from the fault current of the transformer and is superior to other methods.