Danish-based Fault-Tolerant Extended Kalman Filter Train Localization Method: Case Study and Performance Test

GNSS plays a promising role in the “trainborne centric” of the Next Generation Train Operation Control System (NGTC), the train position is obtained by fusing satellite and other multi-source localization information. Before the train control system putting into operation, it is compulsory to conduct comprehensive both functions and performances test in lab. Since GNSS signal is affected by uncertain factors during the train operation, there will be abnormal values and gross errors in satellite observation, which gives rise to the divergence of filtering algorithm or inaccurate localization results. The conventional FDE (Fault Detection and Exclude) algorithm excludes the faulty satellites and then provides the position solutions, but when the number of visible satellites is insufficient, the localization accuracy will also decline. This paper focuses on the research of a GNSS/INS fault-tolerant tightly-coupled localization algorithm based on Extended Kalman Filter, a Danish fault-tolerant algorithm is proposed, which reduces the weight of anomaly observation and reconstructs the observation noise covariance matrix of EKF to ensure the accuracy of localization results. The Beijing-Shenyang railway line filed data is used as input for the test scenario, and the localization method is verified in the proposed test architecture to evaluate the localization accuracy. The results demonstrate that the proposed Danish-based fault-tolerant Extended Kalman Filter localization method can improve the degraded localization performance. Compared with EKF innovation-Based FDE algorithm method, the localization accuracy obtained by the proposed method is further improved.