Research On Methods Of Fault Line Selection And Location For Resonant Grounding System

Resonance grounding system is widely used in medium voltage distribution network in our country, which is similar to the petersen coil grounded system that we usually say. The single-phase grounding fault is the main form of distribution network fault. It will expand into two or more point earthing fault if the distribution network runs long time with single-phase grounding fault. Two or more point earthing fault can cause overvoltage to the whole system, then damage equipment and do harm to system security. Finding the fault line and deal with it quickly and accurately is an important way to ensure network security and improve the quality of electric energy. So, we must pay high attention to fault line selection and location.In recent years, much research has focused on the fault line selection and location from scholars of the world and remarkable gains have been produced. However, missing, wrong selection and misclassification often happen in reality, because distribution network structure is too complex and operating environment is changeable. There are a lot of technical problems in distribution network line selection and location. In order to further improve the accuracy and reliability of the fault line selection and location, further studies are needed. The main contents of this paper are as follows:(1) The following conclusions are demonstrated by analyzing transient process of single phase grounding fault in resonance grounding system. In the early stage of fault, the mutation directions of the zero sequence current are at opposite poles from faulty feeder and healthy feeder. The zero sequence of faulty feeder equals grounding capacity current from all healthy feeders. Arc suppression coil start-up adjustment needs at least 20 seconds. The fault feature of zero sequence transient current is not affected by arc suppression coil. Fault characteristics of transient zero sequence current is not affected by the phase inversionprocess in T/4 window after fault.(2) On the basis of the above conclusions, a new method of fault line selection based on the Hough transform to total amount of zero sequence current is put forward. Extract the line by the Hough transform to describe the overall direction of zero sequence current in the initial stage of change. And then raise a new algorithm based on overall direction of zero sequence current in the initial stage of change in order to apply it to fault line selection. The method is a good place for the fault with difference initial phase angle, random noise, high resistance to the ground, arc grounding and nonlinear load. At the same time, the method has some characteristics such as higher robustness, higher precision and lower sampling frequency.(3) According to the problem that the initial phase angle of fault affect line selection result, put forward a method of fault line selection based on extendable fusion with the peak and valley detection and the morphological spectrum analysis. The morphological spectrum analysis is sensitive for small initial angle fault and the peak and valley detection is reliable for large initial angle fault. This method effectively combines these advantages, and realizes extension fusion under the initial phase angle as the weight factor. Therefore, the influence from initial phase angle to fault line selection result is effectively overcomed.(4) In reality, there are a lot of arc grounding fault and the kind of fault line selection is very difficult. In reality, there is a lot of arc grounding fault and the kind of fault line selection is very difficult. According to arc grounding fault, this paper presents a continuous line selection method based on the the principal component analysis(PCA) and the evidence theory fusion. Because the PCA can reduce the data dimension, interference from other signals is reduced. In addition, this method makes use of continuous signal rather than a period of signal and the comprehensive basic reliability by fusing some groups of basic reliability is the basis for line selection. The above reasons improve the reliability and accuracy of the line selection. The method has good adaptability for permanent arc grounding fault.In the above three line selection method, the first one is a single line selection method that is applicable to all kinds of fault conditions. The latter two are intelligent fusion line selection methods. One method is based on the initial phase angle of fault and another method is according to the arc grounding fault. Three methods of line selection have universality and generality. Moreover, three methods of line selection have obvious advantages in both accuracy and reliability.(5) The following conclusions are demonstrated by analyzing transient traveling wave theory of straight line ground fault and simulating transient traveling wave of direct distribution overhead line fault. The premise for fault location with traveling wave is selecting traveling wave under different modes according to fault type. The key is identification the properties of wave head. However, distribution network has the characteristics with short feeder, complex structured and more wave impedance discontinuity points. In addition, traveling wave includes fault initial traveling wave and reflected wave or refracted wave from wave impedance discontinuity points. All kinds of traveling wave are superimposed together to brought great difficulties for identification of wave front.(6)Put forward a fault location method of k-NN based on waveform similarity. Firstly, find k nearest neighbor of test sample. Next, give the weight for k nearest neighbor according to their influence to the test sample by the similarity algorithm. Finally, get the fault distance through the regression algorithm. Because the method does not depend on identification of wave head, it has good reliability and accuracy for fault location of straight line distribution and radial distribution network. But for the large radial distribution network with a lot of branch node, if a failure occurs after multiple branch nodes, the accuracy of the method will be decreased or even failure. At this time, other methods must be combined.