Geomagnetic Sensor Noise Reduction for Improving Calibration Compensation Accuracy Based on Improved HHT Algorithm

In navigation work, the geomagnetic sensor signal is susceptible to interference, so the calibration process is necessary to compensate sensor model error. However, we found the noise error in the geomagnetic sensor leads to a considerable calculation error in the calibration coefficient matrix, which is a big obstacle in obtaining the accurate signal. To improve the calibration effect, this paper analysis the characteristic of geomagnetic sensor's noise error, and then proposes the improved Hilbert-Huang transform (IHHT) algorithm to preprocessing these noise before calibration. IHHT creatively proposes an attention Hilbert spectrum, which is robust to strong noise recognition and localization. The adaptive median filtering is employed on strong noise regions. Additionally, a similarity criterion for Intrinsic Mode Function (IMF) processing is presented to save computational resources and time. Confirmed by simulation, IHHT algorithm depresses noise to be close to 0 mGuass, and declines sample variance to 13.963 mGuass(2) , which shows effective noise reduction capability. The experiment result shows IHHT algorithm considerably enhances calibration performance, as the calibration error drops by more than 50% compared with the results using state-of-the-art methods.