A Traveling Wave Fault Location Method For Distribution Network Based On Graph Theroy

With the development of society and technology, it is increasingly important for the power grid to operate safely and reliably, especially for the medium voltage distribution network, which directly affects the quality of daily life. According to statistics, over 80% of the power outages are caused by ground fault. Fault location in distribution network plays an important role in ensuring the safe and stable operation of power grid by quickly detecting and determining the position of the fault, reducing maintenance work and speeding up the recovery of power supply. Traveling wave fault location has done good effects in transmission line. However, distribution networks are of tree structure with many branches, traditional methods of single-ended, double-ended fault localization cannot be used. This paper focuses on the current error and poor reliability problems in distribution network line fault location, and develops the related research based on double-end traveling wave method and combining with the characteristics of power distribution network topology, to aim at improving the accuracy and reliability of the power distribution line fault location.Detection accuracy is a significant factor which affects accuracy rate of travelling wave fault positioning, however, detection accuracy of current detection algorithm for travelling wave header is relatively low. The dissertation proposes a traveling wave extraction method based on local mean decomposition method and differential entropy(LMD-DE). In this method, local mean value of aerial mode component of fault voltage travelling wave is decomposed in the first place, then according to the achieved differential entropy of intrinsic scale component, the first discontinuity point of travelling wave signal can be detected, in this way, the arrive time of the first travelling wave can be judged. By using PSCAD and MATLAB simulations, the new method are compared with conventional ways such as wavelet transformation, HHT, traditional LMD, etc., which validate better accuracy and reliability in terms of wave header extraction.Considering the characteristics of distribution network topology, the paper constructs a distribution network model by PSCAD, which simplifies distribution network model and equivalents it to topology of graph theory, then, draw the minimum spanning tree, after the use of EBP pruning algorithm, a new method based on minimum spanning tree of graph theory and pruning algorithm is proposed. The new method can effectively exclude spurious point and find the exact location. At last, simulations are conducted under different fault conditions by using PSCAD and MATLAB co-simulation analysis. Results show that the proposed method can localize fault and the positioning accuracy has been significantly improved.