This paper used complementary ensemble empirical mode decomposition (CEEMD) to extractintrinsic mode function (IMF) and remaining component of zero sequence current. CEEMD not only overcame the modal aliasing,but also could effectively neutralize the added white noise. So, it made the decomposition more completeness. On the basis of it, the frequency band energy which correspondedwith IMF was calculated, while using correlation analysis to get the correlation coefficientbetween IMF and zero sequence current. Then, correlation coefficient was used as weight coefficients for givingit to the frequency band energy. Furthermore, the weighted frequency band energy was accumulated as the energy of each line. Eventually, the energy of each line have been normalized as energy proportion to form a line selection criterion. At last, a distribution network simulation model with hybrid transmission lines was set up by Matlab/Simulink. The results showed thatthis method could select the fault line regardless of fault close initial angles, fault location, transition resistanceand etc. To some extent, this methodimprove the precision of afault line-selection.
2016, Vol. 17 
