Adaptive fault-tolerant in-motion alignment method for SINS/GPS/PNS integrated navigation system based on AFSTUKF

Abstract Accuracy and reliability are critical for integrated navigation systems. This paper presents an adaptive fault-tolerant in-motion alignment approach aimed at enhancing the reliability of the SINS/GPS/PNS integrated navigation system. Firstly, the error of polarization angle, influenced by environmental factors, is formulated for the purpose of real-time error estimation. Secondly, an improved adaptive federated strong tracking unscented Kalman filter (AFSTUKF) is proposed to proficiently address model uncertainties and to guarantee dependable performance in the presence of noise fluctuations or anomalies. Furthermore, the AFSTUKF integrates an adaptive feedforward information-sharing algorithm (ISA) and a fault detection and isolation (FDI) algorithm. The ISA is designed to dynamically modify information-sharing parameters in order to mitigate the risk of global information contamination resulting from inaccuracies in filtering or sensor malfunctions, thereby constituting a significant improvement over traditional federated filtering methodologies. The FDI algorithm employs Bayesian estimation techniques to recalibrate thresholds in accordance with system uncertainty, effectively addressing the shortcomings of conventional chi-square tests and enhancing overall robustness. Experimental results from a moving vehicle validate the high accuracy and reliability of the proposed method.