This paper use the Self-Organizing Feature Map neural network and BP neural network to diagnosis the network fault, use the cluster in training sample, add the result to the sample and put some weights, it can increase the qualities. Parallel computing is used in the neural network compute. The computer network fault diagnosis is used as a example to analysis. the work has been emulated and analyzed by computers, and a computer network fault diagnosis system has been excogitated. The whole works have been summed up and the future researching directions have been discussed at the end of this paper.