Fault Prognostics of Electronic Equipment: Temporal Point Process Based on Recurrent Neural Networks

The electronic device plays an important role in the avionics system. Most of these electronic devices work in the instable environment since the aircraft normally flies in at extremely high altitude. Therefore, such severe environmental condition significantly enhances the probability of device failures. Predicting the probability of the device failure is of great importance to reduce the loss of the crew and the equipment on the aircraft. In this paper, a method is proposed to predict the failure time of electronic devices, which uses recurrent neural network to model the strength function of point process. The experimental results on real datasets show that our method outperforms traditional point process methods.