Enhanced MCUSUM Control Chart-Based CKF for Fault Detection and Identification in Fault-Tolerant Integrated Navigation System

Fault-tolerant integrated navigation is essential for ensuring the accurate, reliable, and effective operation of high-precision carriers. Effective fault detection and identification techniques enhance the quality and accuracy of integrated navigation systems. This paper proposes an enhanced multivariate cumulative sum (MCUSUM) control chart to improve fault detection sensitivity and identification reliability. Additionally, it integrates the Huber function with M-estimation theory and the cubature Kalman filter (CKF) for high-precision fault-tolerant integrated navigation. We discuss the advantages and disadvantages of the enhanced MCUSUM control chart in comparison to the multivariate exponentially weight moving average control chart in fault detection, highlighting its ability to efficiently smooth and abrupt to faults. Furthermore, fault identification through generalized canonical correlation analysis (GCCA) facilitates problem-specific analysis, ensuring that faults are justifiable and traceable. Consequently, a more suitable weight matrix can be constructed based on traceability results for constructing Huber function to suppress faults, thus ensuring high accuracy and reliability of the fault-tolerant integrated navigation system. Finally, both simulation and physical roadster experiments are conducted to validate the effectiveness of the proposed method. The results demonstrate that faults can be simultaneously identified and analyzed with high sensitivity, leading to greater accuracy compared to other methods.