Transient Fault Simulation Of Transformer And Fault Identification Based On Support Vector Machine

Power transformer is essential and important element of power system. It plays an important role in the transmission of power supply. The reliability of power transformer directly determines whether the power system works safely or not. Especially for extra high voltage (EHV) transformer is expensive, great economic benefits would be lost as a result of the fault of EHV transformer. Differential protection is the master protection for transformers all along. There are many problems need to solve. Long-term run experiences make it clear that differential protection can distinguish between internal fault and external fault of transformer. So the key of transformer protection is to distinguish between magnetic inrush and internal fault currents.Decades lately, many domestic and foreign scholars put forward new theory and methods to differentiate magnetic inrush and internal fault currents. No matter the traditional methods or new intelligent methods are based on effective data for research. The transformers are so expensive that it is impossible to do fault experiments. It is also very expensive to do dynamic simulative experiments, and the experiments can't simulate many conditions of shutoff. So computer simulation is a feasible way to get the data of research. EMTDC (Electromagnetic Transients including DC) is a transient simulation software widely used in power system at present. This software can set up source-transformer-transformer current (CT) mode according to appointed differential protection transformer. So it can simulate integrally the transient of transformer. In this paper the magnetic inrush and internal fault currents of transformer are entirely simulated by EMTDC. And analyses are given to the CT's secondary differential currents. The characteristics about 2^nd harmonic and gap angle are got.Traditional statistical theory aims at the asymptotic theory when sample size is tend to infinity. However, in many practical cases, samples are limited.