Research On Short-Circuit Faults Location In Mid-Voltage Distribution Network

The construction of the Strong&Smart Grid has been in full swing all over China. Distribution network is an important component of the Grid. The improvement of automatic level in distribution systems is of great significance to guarantee quality of power supply and realize the Strong and Smart Grid. The fault location in distribution network is a key technology for the power distribution automation system. Therefore, according to the current framework of power distribution automation, short-circuit faults of medium-voltage lines in urban diatribution network is selected as research subjects. It is studied in this dissertation from the following three aspects: identifying fault, locating fault section and getting accurate fault distance.(1) Aiming at identifying the fault types in distribution network,a new hybrid fuzzy H net(FHN) is proposed. Using Clarke transformation, the zero-sequence voltage amplitude is extracted from voltage signal of the fault. Voltage sag amplitude is also extracted by S-transformation. Then the zero-sequence voltage amplitude and voltage sag amplitude are fuzzified by fuzzy reasoning. Finally the recognition is performed by FHN. The experimental results show that the proposed method can recognize the type of faults exactly. And it is not affected by the factors such as transition resistance and fault position.(2) Considering the limitations of the artificial intelligence algorithms adopted past, such as large initial population, too many iterations and premature convergence, a novel algorithm is presented to realize fault section location, which combines EDA(Estimation of Distribution Algorithm) with bacterial foraging. To improve the EDA, which has the local optimization capability and the slow convergence in early stage, the dissertation applies the bacterial foraging behavior to the EDA. The experimental results show that the algorithm can exactly locate the fault section of multiple source ring-open distribution network. It has good accuracy, rapidity and fault tolerance.(3) To get accurate fault distance, it is presented that a fault distance interval based on D-type traveling wave theory. Firstly, it is obtained that the fault traveling waves on the both ends of fault section is obtained. The fault section is located by the improved EDA. Then the instantaneous energy curve from fault raveling wave is expressed by combining ensemble empirical mode decomposition(EEMD) with Hilbert Transform. The arrival instant of the first fault wavefront is confirmed through calculating the peak value of the instantaneous energy curve. Considering time synchronization problem of D-type traveling wave theory, the dissertation applies the nonsynchronized time in the formula of distance measuring. The results of simulations proved that this fault location algorithm not only can find the arrival time of the first fault wave front accurately, but also has a higher precision and a good application prospect in the practical engineering.