Determination of transformer excitation inrush current and short-circuit current based on wavelet theory

      The main protection of transformer microcomputer protection is differential protection. The key issue is how to distinguish between excitation inrush current and short-circuit current to prevent malfunction caused by excitation inrush current. Many methods have emerged around the identification of excitation inrush current, such as the discontinuity angle principle, the second harmonic braking principle, the voltage braking principle, the magnetic flux characteristics principle, the equivalent current principle and the waveform symmetry principle, etc. Each method has its own advantages and disadvantages, and the most widely used ones are the discontinuity angle principle and the second harmonic braking principle. Although there are many methods to identify excitation inrush current, they all have shortcomings. This article describes the application of wavelet theory in identification of excitation inrush current.

Wavelet theory analysis

Although wavelet analysis is an advanced mathematical theory, like Beaulieu transform, it is only a signal processing tool. Its application in excitation inrush current identification is only as a tool, and it still uses excitation. Basic characteristics of surges.

Figure 1 Saturation waveform of asymmetric inrush current and its wavelet transform

The modulus local maximum of wavelet transform, that is, the local maximum of |W ₂ ^j f(x)|, is of great significance for detecting the edges and singularities of signals. According to the local maximum value of wavelet transform and its cross-scale propagation, the location of the mutation point can be accurately determined. This paper uses this characteristic of wavelet analysis to extract the discontinuity angle characteristics of the transformer excitation inrush current.

Wavelet transform characteristics of excitation inrush current

The transformer protection processing of microcomputer is generally the secondary side current transformed by CT. Linear CT does not have saturation problem, but CT in actual operation may suffer from saturation due to the large excitation current and DC offset. CT saturation is one of the important problems that plagues the differential protection principle of transformers. The numerical values ​​of asymmetrical surges are much larger than those of symmetrical surges. An important reason for saturation is the influence of the non-periodic component of the primary current. This article mainly considers the saturation problem of asymmetric inrush current.

This article gives the saturation waveforms and wavelet transforms of asymmetric inrush current and fault current respectively, as shown in Figures 1 and 2.

Figure 2 Machine wavelet transform of the differential waveform of the primary current when the transformer is air-dropped with a fault

The modulus maximum value of the wavelet transform in Figure 1 and Figure 2 corresponds to the singular boundary of the current waveform and is transmitted along the scale. In order to remove the influence of high-frequency noise, the modulus maximum value of the wavelet transform at the third scale is selected as the extracted Characteristics. Comparing the wavelet transform of the third scale of asymmetric inrush current and short-circuit current, namely Figure 1(d) and Figure 2(d), it can be seen that the modulus maxima of the wavelet transform of fault current appear at equal intervals, one positive and one negative. , and the maximum values ​​of two adjacent modes of the wavelet transform of the inrush current have the same sign. These two maximum values ​​correspond to the discontinuity angles of the inrush waveform. This phenomenon is caused by the discontinuous angle of the inrush current. This feature can be used to identify symmetrical surges.

Figure 1 shows the simulation results of CT saturation. Figure 1(b) shows the unilateral inrush after saturation. Since the CT is severely saturated, the discontinuity angle of the inrush waveform no longer exists. Figure 1(c) is a waveform obtained by differencing the saturated waveform. Obviously, the discontinuity angle is restored after the difference, but the difference also amplifies the high-frequency component. Figure 1 (d), (e), and (f) are the results of wavelet transform 1~3 scale of the differential CT saturated inrush waveform. It is obvious from this that the wavelet transform of the third scale not only has the inrush characteristics of adjacent mode maxima with the same sign analyzed previously, but also has a very good effect in filtering out high-frequency components. This shows that the criteria obtained previously are not affected by CT saturation.

Figure 2 shows the difference of the fault current and its wavelet transform. The most serious situation (air-dropped transformer with fault) is considered here. At this time, the distortion of the waveform is very serious. However, after differential fault current is still a quasi-sinusoidal waveform without discontinuity angle, the modulus maximum value of its third scale wavelet transform is still adjacent with different signs. This feature can be used to identify asymmetric inrush and fault currents.

Conclusion
This paper proposes a new method based on wavelet theory to distinguish excitation inrush current and fault waveforms. The wavelet theory is used for feature extraction, and the discontinuity angle characteristics of the excitation inrush current are extracted through the modulus local maximum of the wavelet transform, so that the excitation inrush current and internal short circuit can be qualitatively identified. This principle is highly robust to CT saturation, and both theoretical analysis and simulation tests show that this principle is not affected by CT saturation.