A Research Of Power Line Fault Phase Selection Based On The Morphological Spectrum

Along with the development of economic, the demand for electricity power is growing and the requirement of power supply reliability is continuously improved. The accurate, fast fault phase selection is the basis for troubleshooting when the power line fault. It is meaningful for power supply reliability and safety. There will be an enormous impact on the system if all the power lines are removed when some faults occur. It will cause unpredictable losses. So in the event of the power line faults, the system must continue to supply power a certain period of time, which requires only remove of fault phase and the normal phase is with no tripping. Due to the need of phase selection tripping, phase selection is an important link to deal with the fault.In the current, widely used fault phase selector can select the fault phase in the single-phase to ground fault. In the multi-phase fault, it can judge the multi-phase fault and immediately jump three-phase. However, in order to get accurate measurement, it requires not only find the fault phase in the single-phase fault, but it requires find the fault phase in the multi-phase fault too. However traditional protection devices are too simple that it can only fit to some simple fault condition. All kind of fault phase selector alone can not fully meet the power line fault selection. It is necessary to approach the selection with further study. In this paper, the study of power line fault election in the new approach is necessary: not only can find the fault phase in a single-phase fault, but also to find the fault phase in the multi-phase fault.In this issue, we set up the simulation model basis of the Beijing-Tianjin inter-city rail electrical structure. We establish the actual fundamental power line model with relevant frequency existing in the PSCAD / EMTDC. When some fault occur, acquisition three-phase current, and then transform the phase current to mode current. We obtain the mode current, and then we use matlab.m to achieve the extraction of spectrum, and then extract the different type of fault patterns spectrum. We give a comparative analysis of spectrum in different fault distance and different resistance. The distance the fault occur and transition resistance give some effects of the spectrum's shape and size, and impact on the fault phase election. As the form of spectrum have very good ability to reflect the wave characteristics. Different types of fault mode current have different spectrum. It can be seen that using the spectrum to select fault phase could be achieved. We put spectrum as a form of artificial neural network input. The use of artificial neural network, which with a strong pattern recognition, non-linear mapping, contact imagination, as well as the ability of robustness, can judge the type of line faults.In this paper, we find a new method of using the extraction of the fault mode current spectrum to judge the fault phase. It is proved by simulation. This method is very good for the fault phase current selection to solute present components' problems. It is a feasible method.