There are situations of complex and variable transient characteristics and rejection and misoperation of relay protection during faults in new DC distribution systems. To avoid the impact of incorrect relay protection actions on fault diagnosis, this paper presents a fault diagnosis method for a DC distribution network based on Bayesian network information fusion. First, the traditional relay protection Bayesian network model is improved, while considering the DC distribution network fault current limiting strategy, and three Bayesian network models of protection, circuit breaker and current-limiting strategy information are constructed respectively to make preliminary evaluation of the fault probability of each component in the fault area. Secondly, D-S evidence theory is used to fuse the corresponding failure probability information of each component to realize the identification of the faulty components. Then, the Bayesian network model corresponding to the faulty elements is applied to identify the protection devices and circuit breakers for misoperation or rejection to achieve fault diagnosis. Finally, the correctness and reliability of the proposed method are verified through an analysis of arithmetic cases.