The invention relates to the field of power communication network equipment fault prediction, in particular to a frequency domain feature-based power communication network equipment fault prediction method. According to the method, only based on the time sequence data of the equipment, the time sequence data of the equipment is converted into the frequency domain data through Fourier transform. After that, the frequency domain data is used for training the BP neural network model. Finally, the equipment state level is predicted, and the maintenance can be conducted in time when the equipment is in the potential fault state. The method only needs to obtain the time sequence data of the equipment, reduces the data size, and shortens the training duration of the neural network.