Power Transformer Microprocessor Protection Study Based On Wavelet Transform Theory

Power transformers are important equipments in power system, which have a great influence on the secure operation of power system. But now the rate of transformer protection correctly act is lower, at the same time, the second harmonic restraint principle has its own fault when distinguishing between the inrush current and internal current. So it is necessary to find a new algorithm to improve performance of transformer protection. It depends on the extent to which people realize the mechanism and feature during magnetic inrush and faults. So the transient simulation is pretty necessary.The transformer model is established by simulation soft saber, which takes into account the limit hysteresis loop and eddy current lost. We can simulate the inrush current and kinds of fault current by it. The influence of remnant flux and closing angle to inrush current is discussed in this paper. Of all kinds of transformer faults, the turn-to-turn fault and turn-to-earth fault are the most frequent. Based on the available achievements of other scholars, we build a model with which turn-to-turn and turn-to-earth fault can be simulated wherever it takes place, and at the same time, all the other kinds of faults can be simulated with it.The wavelet analysis is introduced to correctly recognize the inrush current for it has the properties such as the multi-resolution analysis and localization characteristic in both the time domain and the frequency domain. This method can extract the transient feature of current. The wavelet analysis theory and its application in power system are introduced briefly in the thesis. After that, using the wavelet transform maximum modulus of the primary and the secondary current to set up a algorithm. It can correctly detect the internal fault and distinguish the external and internal fault. Specially, the inrush currents after both the external faults are removed and no-load closing, internal fault during the closing without load and fault during the closing with load can be distinguished. The threshold is presented on the basis of lots of simulations, and we find the threshold and sampling frequency which satisfy the algorithm. We can come to a conclusion that the saturation of the CT can't influence the algorithm through theory analysis.