Reference Signal-Guided Adaptive FitzHugh-Nagumo Stochastic Resonance Method for Weak Magnetic Anomaly Signal Restoration
Guangjing Du, Yao Wang, Ling Pei, Lei Chen, Guifu Gao, Yuhan Zhou, Yumei Wen, Ping Li · IEEE Transactions on Instrumentation and Measurement · 2025
Due to the severe attenuation of magnetic anomaly signal (MAS) over distance and inevitably dominant environmental noise, the measured MAS typically suffers from a low signal-to-noise ratio (SNR), which degrades the performance of target tracking, localization, and recognition applications. When the noise and signal are heavily overlapped in the time or frequency domain, conventional denoising methods often distort the MAS by separating them. To address this issue, a reference signal-guided adaptive FitzHugh-Nagumo stochastic resonance (RS-AFHN-SR) method is proposed to restore weak MAS under low SNR conditions. Instead of decomposing the noisy signal and subtracting corresponding noise components, the weak MAS is restored by transferring part of the noise energy to the signal based on stochastic resonance (SR) theory. On the one hand, the reference signal is estimated using the proposed correlation detection (CD) method based on the magnetic dipole model, which guides the adaptive optimization of FHN model parameters and enhances the accuracy of parameter estimation. On the other hand, residual noise in the SR-enhanced signal is further suppressed by integrating it with the reference signal, resulting in the more precise signal shape. Accordingly, the proposed denoising method provides the superior signal restoration performance under low SNR conditions by combining the advantages of SR and CD. Specifically, the RS-AFHN-SR method provides an SNR improvement of 27.8 dB for the noisy MAS with an initial SNR of -19.6 dB. Furthermore, experimental results show that the proposed approach effectively reduces false alarms and localization errors compared to previous denoising methods.