Magnetic Anomaly Detection Based on a Multi-Parameter-Constrained Mirror Dual-Branch Biased Monostable Stochastic...

Area of Science:

• Geophysics
• Signal Processing
• Nonlinear Dynamics

Background:

• Magnetic anomaly detection faces challenges from environmental noise and limited prior information, hindering the detection of non-periodic signals at low signal-to-noise ratios (SNRs).
• Existing methods struggle with effective noise suppression and feature preservation in complex noise conditions.

Purpose of the Study:

• To develop an advanced detection method for magnetic anomalies, particularly under low SNR and strong noise conditions.
• To enhance the robustness and accuracy of magnetic anomaly detection systems.

Main Methods:

• A multi-parameter-constrained mirror dual-branch biased monostable stochastic resonance (SR) system was proposed, incorporating nonlinear odd-order bias terms for improved potential well control.
• A composite metric combining correlation coefficient and wavelet-domain structural similarity was used for parameter optimization.
• Multi-fidelity robust Bayesian optimization was employed to determine optimal parameters for diverse magnetic anomaly signals.

Main Results:

• The proposed SR system demonstrated effective recovery of magnetic anomaly features even in the presence of simulated and measured strong noise.
• Detection probability exceeded 80% at -19 dB SNR, showcasing significant performance improvement.
• The method outperformed traditional techniques like empirical mode decomposition in noise suppression and feature preservation.

Conclusions:

• The developed multi-parameter-constrained mirror dual-branch biased monostable SR system offers a robust solution for magnetic anomaly detection in challenging low-SNR environments.
• This approach significantly enhances detection capabilities and preserves crucial signal features compared to existing methods.