A Deep Network Based Fault Diagnosis Method for Stabilization Loop of Inertial Platform

The inertial platform serves as the fundamental component of system navigation, directly affects system performance and safety. Therefore, it is imperative to investigate accurate, fast and reliable fault diagnosis methods. In this paper, a hybrid deep neural network based fault diagnosis method for the inertial platform stable loop is proposed, which consists of the recurrent neural network (RNN), the gated recurrent unit (GRU), and the cross-attention mechanism. RNN is responsible for extracting local high-level features of time series, GRU is used to compensate for global high-level features, and cross-attention can effectively fuse the two features. Ultimately, the efficacy of this method is validated through application to the stabilization loop of the inertial platform. Ultimately, the efficacy of this method is validated through application to the stabilization loop of the inertial platform.