130 / 2023-06-30 21:22:53

Research on Prediction and Restoration of Abnormal North Seeking Signal of Maglev Gyro Based on ARIMA

maglev gyro total station (GAT),data prediction and restoration,north-seeking,arima

The gyroscope can realize independent north seeking and orientation measurement by sensing the earth's rotation angular momentum without external GNSS and other auxiliary information. As an important product of gyroscope north finding technology in the field of surveying and mapping, the magnetic levitation gyroscope directional instrument plays an important role in large-scale underground engineering construction such as high-speed railway tunnels, urban subways, and mining. The magnetic levitation gyroscope orientation instrument uses magnetic levitation support technology to replace traditional suspension belt support technology, fundamentally reducing the effective contact between the sensitive part of the gyroscope and the outside world, and reducing the impact of external interference torque. Considering that high-speed gyroscopes are very sensitive to changes in the environment, they are inevitably subject to disturbances caused by wind vibrations during the north seeking process. These disturbances seriously affect the quality of the north seeking data collected by the torque converter rotor system directly connected to the gyroscopes. Therefore, it is necessary to carry out targeted processing of abnormal data in the rotor signal. To reduce the impact of external environmental disturbances on the rotor signal, noise reduction is mainly achieved by establishing a filtering model to process the rotor signal. However, current filtering models only filter the rotor signal as a whole, without considering that external interference is often instantaneous and does not run through the entire rotor signal acquisition process. Therefore, the current approach has limited noise reduction effect on rotor signals affected by instantaneous interference torque. This article proposes a Bidirectional Autoregressive Integrated Moving Average Model (BARIMA) to address the difficulty of effectively improving data quality of rotor signals under the influence of instantaneous interference torque through traditional filtering methods. Firstly, the non-stationary rotor signal sequence is differential processed and transformed into a stationary sequence. Then, the dependent variable is regressed only to its lag value and the present value and lag value of the random Error term in both forward and reverse forms. The corresponding forward and reverse regression models are established respectively and the forward and reverse regression models are weighted according to the length of the learning samples. Finally, the forward and reverse regression models are used to predict and complete the data at the location of abnormal data in the rotor signal, achieving effective noise reduction of the rotor signal affected by instantaneous interference torque. Compared with traditional filtering methods such as wavelet transform and empirical mode decomposition, the bidirectional autoregressive moving average model proposed in this paper performs better in suppressing instantaneous torque disturbances.