Study of structural reliability evaluation method based on deep BP neural network

The back propagation (BP) neural network is commonly used to detect the fault of some equipment in the actual project. When the fault data is sufficient, BP neural network has achieved good results. But for practical engineering, it commonly meets with the rare failure event. The proposed detection model based on deep BP neural network in this paper is not only applicable to the case of sufficient fault data, but also applies to the rare fault data. In this structure reliability analysis, samples generated by Monte Carlo Simulation (MCS) are usually safe samples and the failure samples are very rare. The case is similar to the rare event in engineering. In this paper, a detection model based on deep BP neural network will be proposed to deal with the rare failure samples generated by MCS and calculate the failure probability. Finally, the efficiency and accuracy of the proposed method are demonstrated by a numerical example and an engineering application.