Compensate Three-Axis Magnetometers Based on Adaptive Cancellation Under Time-Varying Noise

Three-axis magnetometers (TAMs) are widely employed in a variety of applications,including geological surveys, navigation, and military fields. The interference introduced by the TAM itself and the mounted carrier significantly constrain the effectiveness of these measurements and the ability for magnetic anomaly detection (MAD), especially when the TAM is under shaking movements. Implementing compensation strategies by exploiting the geomagnetic field can reduce the interference. However, the geomagnetic field often encompasses time-varying magnetic noise with the progression of industrialization, which frequently leads to the failure of the TAM compensation. To address this critical challenge, we propose a compensation method that accounts for time-varying magnetic noise. The method first employs an uncompensated TAM to adaptively learn the time-varying magnetic noise, then applies the learned parameters directly within the TAM compensation model. Finally, we significantly reduce the impact of time-varying noise during the TAM compensation. After compensation, the TAM scalar output noise reduced a lot and the ability for MAD improved a lot. The experimental results demonstrate that for a MAD test during shaking movement, the peak-to-peak noise was 135.4 nT without any compensation, which was reduced to 36.9 nT by using the nonlinear least square-based calibration method, 31.2 nT by using the particle swarm optimization-based calibration method, and 5.6 nT by using the calibration method proposed in this study. Notably, the magnetic field of the target was 11.7 nT when the it passed by. In this case, only the compensation method proposed in this paper was able to effectively distinguish the target magnetic anomaly signal.