A Hybrid Network Based Disturbance Estimation Method for Stabilization Loop of Inertial Platform

The disturbance in the stabilization loop of the inertial platform seriously affects the system accuracy and performance. Conventional methods face challenges in identifying external disturbances. In this paper, a hybrid neural network based disturbance estimation method for inertial platform stabilization loop is proposed. The hybrid neural network model comprises a recurrent neural network (RNN) responsible for extracting local high-level disturbance features from time series, and gated recurrent units (GRU) used to compensate for global high-level disturbance features. Subsequently, these high-level features are fused and input into a fully connected layer for disturbance estimation. The effectiveness of the proposed method is verified through the stabilization loop data from the inertial platform.