DC distribution network is the development trend of power distribution system in the future, in order to achieve the targeted improvement of DC power quality problems and promote the in-depth development of DC power distribution technology, it is necessary to propose an identification method suitable for DC power quality disturbance characteristics. In this paper, the formation mechanism and disturbance characteristics of four types of power quality problems in DC distribution network are deeply analyzed, and five characteristic indicators are proposed for the characteristics of various problems, which are used as the characteristic elements to identify DC power quality problems. The intra-class aggregation and inter-class separation of the proposed feature set were proved by k-means cluster analysis. Finally, the PSO-SVM classifier is used to accurately identify the DC power quality problem. Simulation examples verify the accuracy and effectiveness of the proposed method.