Fault detection method of tightly coupled GNSS/INS integration assisted by LSTM
【摘要】：Aiming at the problem that the traditional fault detection method of tightly coupled GNSS/INS integration is not efficient for small-amplitude faults and gradual faults, a dual-threshold fault detection method based on long-short term memory neural network is proposed. This method establishes GNSS pseudo-range and pseudo-range rate prediction models through a long-short term memory neural network, sets the lower detection threshold based on the accuracy of the prediction model and the residual distribution characteristics, and assists the residual chi-square detection method for fault detection and isolation. The actual measurement data is used for simulation verification from the perspectives of fault detection performance and positioning accuracy. The results show that the proposed method retains the excellent detection performance of the residual detection method for large-amplitude faults, and has better detection sensitivity for small-amplitude faults and gradual faults, which can improve the positioning accuracy and fault detection performance of the system effectively.