Fault detection is crucial to isolate positioning risks for safety-critical applications using Global Navigation Satellite Systems. Conventional Kalman filter-based fault detection methods mainly focus on satellite measurement faults and presume that the test statistics follow the chi-square distribution. These methods ignore the adverse effect of undetected faults occurring previously, and the detection performance would be restrained under a mis-matched distribution assumption. To solve these issues, an enhanced fault detection method is proposed in this paper, which combines the Maximum Correntropy Criterion (MCC) and Local Outlier Factor (LOF). The MCC is introduced to derive a robust extended Kalman filter to deal with the undetected faults. Simultaneously, a specific Kernel Bandwidth (KB) for each measurement is calculated by the innovation and innovation covariance matrix to handle the inherent restriction of a fixed KB. Moreover, the LOF is used to reconstruct the test statistics, and the threshold is calculated by an offline model. Simulations are conducted to evaluate the proposed method under different fault scenarios. The results illustrate that the adaptive robust estimation reduces the negative influence of undetected faults, which makes the filter innovation follow the actual fault amplitudes. The proposed algorithm effectively improves the fault detection rate and positioning accuracy.
Global Navigation Satellite System (GNSS) plays a crucial role in autonomous train positioning. However, GNSS measurements containing faults will cause a decreased positioning accuracy in the filter estimation. The traditional Kalman filter-based fault detection method assumes that the measurement noise variance is invariable, which results in an inaccuracy test statistic in a dynamic time-vary environment. To solve the problem, this paper proposed an improved two-stage fault detection method, including global and local tests. The variational Bayesian is introduced into the global test to construct the new test statistic with Mahalanobis distance for detecting the fault. Then, a subset-based local test is carried out to identify the fault measurement. The effectiveness of the proposed method was demonstrated under step and ramp faults scenarios. The results proved that the measurement noise variance directly influenced the ability of fault detection, and the proposed method provides a more reliable and accurate position.
Accurate and reliable train positioning is essential for railway operation control systems. However, conventional extended Kalman filter-based solutions are vulnerable to measurement faults, which can significantly degrade positioning performance. To address this issue, this paper proposes an adaptive maximum correntropy robust filter (AMCRF) for a GNSS/INS-based train positioning system. The loss function of the extended Kalman filter is reformulated from the minimum mean square error criterion to a maximumcorrentropy criterion, thereby improving the algorithm’s robustness against measurement faults. In AMCRF, considering the limitation of using a fixed kernel bandwidth, a lion swarm optimization strategy is introduced to adaptively tune the kernel bandwidth for each visible satellite, enabling the filter to adapt to time-varying measurement quality and fault magnitudes. By embedding the adaptive mechanism into an extended Kalman filtering framework, the proposed method achieves enhanced fault tolerance. The effectiveness of the proposed AMCRF is validated using experimental data collected along the Qinghai–Tibet Railway. Step and ramp faults of different magnitudes are injected into pseudorange measurements to evaluate fault tolerance. Experimental results demonstrate that the proposed method effectively suppresses the influence of faulty measurements and maintains positioning accuracy close to that under fault-free conditions.