In order to ensure the reliable operation of the power system, when a fault occurs, it is necessary to determine the location and cause of the fault in the shortest possible time and troubleshoot it as soon as possible in order to return the service transmission to normal. In this paper, the power system and communication system are coupled with each other to construct a power communication network, and association rule mining is utilized to find valuable information for fault diagnosis from historical alarm data. With the help of a new parallel computing architecture, Spark, a weighted association rule parallel mining platform built on Spark, was constructed under the power communication network fault diagnosis technology. By mining the alarm information through parallel computing, we can quickly find the association relationships between the alarm information and utilize the mined alarm association rules. In this way, the fault diagnosis of the power communication network is completed. Finally, experiments are carried out in the built IEEE39 node simulation environment, and the simulation results show that the proposed method has a power transient fault response speed of 0.3μs , and is able to quickly and accurately complete the fault diagnosis of the power communication network.