Fault line selection method based on local global consistency learning algorithm

For inaccurate line selection problem of existing fault line selection methods used in power system connected to earth through an arc suppression coil or fault selection for phase voltage zero-crossing point, a kind of small current line selection method was proposed which was based on local global consistency learning algorithm. In the method, primary grounding fault signals are processed by Fourier transform firstly, then characteristic value of each fault signal is processed by local global consistency learning algorithm, and fault characteristic signals are judged through cyclical transmitting labels at last, so as to select fault line. The Matlab simulation and laboratory testing results verify that the method has higher reliability and accuracy of fault line selection.