Aiming at the traditional pole selection method, which has the problems of weak noise and high resistance resistance as well as complex threshold rectification, a flexible DC distribution line fault pole selection method based on Res-BiLSTM network is proposed. Firstly, the original fault signal is subjected to complete adaptive noise modal decomposition (CEEMDAN) , and then the reconstructed signal is obtained by using the correlation coefficient and Shannon entropy for reconstruction; secondly, the Res-BiLSTM network model is constructed for the pole selection, and in order to improve the network accuracy and the convergence speed, the channel attention module is introduced into the ResNeSt network, and the convolution BiLSTM network is used with the improved ResNeStM network for the pole selection. network and the improved ResNeSt network to extract the reconstructed signal features simultaneously, and use the Attention Feature Fusion Module to fuse the extracted features and classify the fused features; finally, the model is constructed and the proposed method is verified using PSCAD/EMTDC. The simulation results show that the proposed pole selection method is highly accurate, anti-interference, and independent of fault distance.