An Improved Seamless Train Attitude Determination Method Based on the Double-loop Quaternion Enhancement

The accumulation of inertial equipment errors in the GNSS/INS integrated system can degrade the accuracy of the train's position, velocity, and attitude (PVA) determination in the GNSS-difficult scenarios. To address this issue, a method for improving train attitude accuracy by employing inertial quaternion modelling is proposed. This method involves designing the quaternion estimation model assisted by the train acceleration in the inner-loop’s inertial derivation model (inner-loop estimation). When the INS operates in combined mode with the odometer, the inner-loop KF measurement model causes rotation errors due to the accelerometer’s bias. These errors can be transferred into the odometer, making it difficult to maintain the system's accuracy. To deal with this problem, the external-loop quaternion model is introduced. Utilizing error-quaternion as the coupling medium will ensure that the inner-loop’s quaternion and inertial-bias errors can be estimated and corrected in the external-loop, which involves integrating the inner-loop estimation model with the odometer/INS to generate the double-loop quaternion enhancement method. The experiment results on the Qinghai-Tibet Railway demonstrate that the proposed method has an obvious improvement in the PVA determination compared with odometer/INS method, and it can track the PVA references better and improve capability to maintain the PVA determination accuracy in GNSS-difficult scenarios.