Abstract: To improve both fault diagnosis accuracy and fault tolerance of protective relaying for distribution network under the lost of protection action information, a distribution network fault diagnosis method based on improved directional-weighted fuzzy Petri net was proposed. Firstly, considering the impact of near-backup protection on the diagnosis of suspected fault components during the occurance of the fault, the modeling for each fault spreading direction of suspected fault components was conducted. Secondly, on the basis of expert experience the random number was utilized to perform the random assignment of the input weight of the model. Finally, when protection action information was lost, the confidence value of not-acted suspected fault components was tracked and changed, and the adaptability and fault tolerance of the fault diagnosis method were analyzed. Simulation results show that the proposed distribution network fault diagnosis model based on improved directionality weighted fuzzy Petri net possesses higher fault diagnosis accuracy under incomplete information or rejection of protection and circuit breakers.