Hybrid algorithm for traction transformer differential protection based on intrinsic mode function energy entropy and correlation dimension

Abstract

This study addresses fault identification in differential protection of a V/x-type traction transformer used in a highspeed railway. To quickly and accurately identify an internal short circuit in a traction transformer, a hybrid algorithm is developed that combines intrinsic mode function (IMF) energy entropy with the correlation dimension from chaos theory. IMF energy entropy and correlation dimension are sufficiently fast and sensitive to reflect the dynamic changes in the differential-current signal from the traction transformer using different metrics; thus, this hybrid method can effectively identify an internal short circuit and magnetising inrush. Real-time simulations and actual measurements of faults illustrate the validity of the proposed method.

Nonlinear dynamic modeling and analysis for aviation axial piston pump in center spring return mechanism

“…For a periodic motion, there are a few of isolated points in the Poincaré map, For a chaotic behavior, the return points on the Poincaré map will form the geometrical fractal structure. 25–27 In conclusion, the behavior of the oil thickness of slipper/swash plate pair is in the periodic and steady motion.…”

Differential protection in power transformers using the statistical second central moment

“…It is well‐known inrush current has a high content of harmonics, especially the second harmonic, and this characteristic is used to block differential protection at the time the inrush current appears [3]. Other methods proposed to identify between inrush currents from fault currents are based on parameters of the power transformer [4], two‐terminal network [5], pattern recognition using principal component analysis (PCA) [6], waveform correlation analysis [7] morphology mathematic (MM) [8], gradient vector angle variation [9], wavelet transform analysis [10], empirical Fourier transform [11], multi‐region adaptive [12], neural network evaluation [13], fuzzy logic evaluation [14] and high‐order statistic [15]. However, in some cases, these methods operate incorrectly because of several factors as changes in the transformer parameters, current transformer (CT) saturation, modifications in the power system topology, and frequency system variations.…”

Using mathematical morphology to discriminate between internal fault and inrush current of transformers

“…Therefore, it was difficult for these methods to be implemented in practical applications. In addition, Li et al [16] and Tian et al [17] proposed a hybrid algorithm which combined the intrinsic mode function energy entropy with the correlation dimension theory for the V/x‐type traction transformer. A novel technique based on the moving windows was developed from the full‐cycle Fourier algorithm and least error square method [18, 19].…”