Fault selection technology of flexible DC distribution network plays a vital role in the development of DC distribution network. In this paper, to address the problems of less effective information provided after faults in flexible DC distribution networks and poor line selection accuracy, a support vector machine SVM (Support Vector Machine) optimized based on the Complementary Ensemble Empirical Mode Decomposition (CEEMD) algorithm, Bubble Entropy, and grey wolf algorithm is proposed as a fault line selection method. Vector Machine) for fault selection. Firstly, CEEMD is used to decompose the fault current signal to obtain each IMF component, and then the Bubble Entropy is used to reconstruct the feature vectors, which are finally fed into the SVM flexible DC distribution network fault selection model optimized by the Gray Wolf algorithm. The results show that the proposed line selection method has higher accuracy compared to other methods and has advantages in terms of insignificant fault information and different transition resistances.