The Research Of The Transformer Inrush Identification Based On Wavelet Analysis

With the fast development of artificial intelligence and network technology since late 1990's, there appears the digit-controlled intelligence protection system. Based on this system, the micro-computer relay technology has come to present a new trend where network and intelligence constitute the main feature and protection, control and measurement as well as data communications are integrated as a whole. One of the most important equipments in the power system is transformer(an energy transfer element), the normal operation of which bears greatly upon the whole power system, especially costly large-size transformer. In the case of malfunctions, economic damage incurred might be unimaginable.The main protection scheme of power equipments against internal faults has been differential protection, the application of which is comparatively simple in the generator and transmission line. Nevertheless, there still exist many difficulties. On the one hand, differential protection is expected toreflect the unbalanced current caused by various factors; On the other hand, it should also reflect internal faults. Therefore, some difficulties in transformer protection still fail to final satisfactory solutions so far. Without changing the traditional conception, principle and measurement, these obstacles would be rather hard to break through.In view of these, the research on transformer inrush identification is presented in this paper. The results of wavelet transform caused by either inrush or internal faults will be available to display distinctions between the two.In fact, not many functions can be used as wavelet function, and wavelet function is not the only one, too. The choosing to wavelet functions matters a lot. With improper functions, the distinguishing of different transform results would pose as a difficult task. Based on many tests, coif5 wavelet function is to be chosen in this paper. Because it shows the outstanding property to detect signal singularity and seek the result of their wavelet transform.Finally, PSCAD/EMTDC the software is adopted to simulate the inrush and internal faults, and then transform them with the help of coif5 wavelet function. The test reveals that coif5 wavelet function possesses the ability to detect the signal and find the singularity of transformer inrush, thus identifying correctly the inrush and internal faults and providing a reliable blocking criterion for transformer protection. In the meanwhile, in view ofthe complexity of inrush identifying, this criterion also requires testing in practice.