SINS/GNSS/DVL/CNS Integrated Navigation Fault-Tolerant Design Based on SVMFFD

• Published in: 2023 9th Annual International Conference on Network and Information Systems for Computers (ICNISC) pp. 510 - 514
• Main Authors: Wang, Teng, Liu, Hongliang, Wang, Jian, Li, Enbao, Xiong, Hailiang, Yang, Pengpeng
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.10.2023
• Subjects: Fault detection; Fault tolerance; Fault tolerant systems; information distribution algorithm; integrated navigation system; Navigation; one-class SVM; Simulation; Stability analysis; Support vector machines
• DOI: 10.1109/ICNISC60562.2023.00078

To improve the fault tolerance of the integrated navigation system, a federal integrated navigation fault detection and isolation algorithm (SVMFFD) based on the One-Class Support Vector Machine (SVM) is proposed. The fault detection mechanism is incorporated into the sub-filter, which allows real-time monitoring of the operational status of each component and enables timely fault detection and isolation. One-class SVM algorithm is used in the fault detection module due to its excellent classification capabilities for scenarios with singular sample data, nonlinearity, and high dimensionality. Furthermore, an information distribution algorithm based on fault information is proposed to promptly adjust the state error covariance of each subsystem. The study delves into the isolated reconstruction mechanism, which holds great significance for system reconstruction in the event of a fault. Experimental results highlight the SVMFFD algorithm's capability to promptly and effectively detect subsystem faults, ensuring navigation stability and accuracy. The algorithm significantly enhances the positioning accuracy of the integrated navigation system, underscoring its superiority.