A rapid method that uses super-resolution magnetic array data is proposed to localize an unknown number of magnets in a magnetic array. A magnetic data super-resolution (SR) neural network was developed to improve the resolution of a magnetic sensor array. The approximate 3D positions of multiple targets were then obtained based on the normalized source strength (NSS) and magnetic gradient tensor (MGT) inversion. Finally, refined inversion of the position and magnetic moment was performed using a trust region reflective algorithm (TRR). The effectiveness of the proposed method was examined using experimental field data collected from a magnetic sensor array. The experimental results showed that all the targets were successfully captured in multiple trials with three to five targets with an average positioning error of less than 3 mm and an average time of less than 300 ms.

A Siamese neural network architecture is introduced to enhance the stability and generalization of deep neural networks for magnetic anomaly detection. Grounded in signal disparity contrastive learning, this study addresses the statistical disparity in signals from various regions and times. Within the proposed architecture, two identical one-dimensional convolutional neural networks with shared parameters are used as feature extractors for obtaining the embedding of paired input signals in the target space. Decision networks are then formulated to measure the discrepancies between these embeddings, shedding light on the differences between the original signals. A base signal family is crafted for detection using multiple noisy signals that are spatially and temporally aligned with the evaluated signal. The difference between the measured signal and those in the base family is computed. A voting mechanism subsequently determines if the assessed signal is a magnetic anomaly. Numerous semi-realistic datasets are employed for network training. The results indicate that the proposed network surpasses several existing networks in robustness with regard to detection area, time, and signal parameter variations and also has excellent detection capability and temperature in the face of measured magnetic anomaly signals. Notably, with changes in test parameters, the network only requires the background noise signal as the base, maintaining high detection performance without retraining.

Magnetic anomaly detection (MAD) plays a critical role in both geophysical exploration and environmental monitoring. Despite advancements, traditional MAD methods often struggle with low signal-to-noise ratios (SNR) and complex background noise, limiting their effectiveness. This study introduces a novel magnetic anomaly signal Restormer (MASR) approach, significantly outperforming existing signal restoration techniques in the MAD field regarding accuracy and reliability. Leveraging the unconventional computing power of magnetic tunnel junctions (MTJs), we propose an integrated measurement-computation system designed to implement the MASR method efficiently. Our system demonstrates superior detection performance and marks a departure from conventional von Neumann architectures, offering enhanced computational efficiency and energy savings. By integrating nanodevice-enabled artificial intelligence (AI) directly with signal measurement, our approach minimizes reliance on external processing units and data transmission, ensuring rapid, real-time MAD processing. This system is particularly suited to address the challenges of low SNR and harsh working conditions prevalent in geoscience data analysis. Experimental results obtained from real-world oceanic deployments validate the superiority of the MASR method and the integrated system in achieving high-fidelity magnetic anomaly signal restoration under challenging conditions. This research not only advances the field of MAD but also paves the way for future developments in energy-efficient, compact, and high-performance real-time remote sensing systems, laying the groundwork for broader applications of nanotechnology-enabled AI in autonomous and efficient remote sensing systems.

To achieve high-precision vector measurement values in a geomagnetic field, it is necessary to develop methods for overcoming alternating magnetic interference (AMI), which is generated by electrical equipment. This paper proposes the adaptive alternating magnetic interference suppression (AAIS) algorithm. In this algorithm, first, only a triaxial fluxgate sensor measures the magnetic field data. The time–frequency diagram of the total magnetic field is obtained quickly through short-time Fourier transform and wavelet transform. Additionally, the time and frequency of AMI appearance are analyzed. Then, the triaxial adaptive notch filter suppresses the three-component related magnetic interference. Herein, simulation and actual experiments are performed to verify the effectiveness of AAIS. The results indicate that the algorithm can quickly detect the frequencies of AMI from the total magnetic field and adaptively fit their amplitude and phase on the vector magnetic field. Finally, AAIS can suppress the interference effectively. The AAIS algorithm realizes error compensation for the vector measurement values by the total magnetic field, which effectively improves the vector measurement accuracy of the geomagnetic field. We highlight that the AAIS algorithm is effective for AMIs of different frequencies, numbers, and intensities without reference sensors. Our work has practical implications in airborne, vehicle-mounted, and shipborne geomagnetic vector detection.

Existing magnetic anomaly detection (MAD) methods are widely categorized into target-, noise-, and machine learning-based methods. This article first analyzes the commonalities and characteristics of these methods, unifying them into noise- and target-based frameworks. Focusing on the MAD problem under static sensing systems, and considering that the noise-based methods have better stability in real-world detection but suffer from poor performance at low signal-to-noise ratios (SNRs), this article proposes a novel MAD method based on deep support vector data description (Deep SVDD). The proposed method characterizes long-term magnetic background noise patterns in the region of interest. A deep neural network encoder is employed to extract time–frequency features of the signals, yielding a compact low-dimensional latent representation. The latent space is constrained by a prior distribution derived from a pretrained model, and the probability of noise signal features is maximized in the form of maximum likelihood estimation. To effectively avoid overfitting caused by hypersphere collapse, the Kullback–Leibler (KL) divergence is incorporated into the loss function. Statistical tests and visualizations confirm the method’s effectiveness and alignment with theoretical foundations, while comparative experiments demonstrate that the proposed method achieves significant performance improvements, especially at low SNRs over existing noise-based methods. Furthermore, to prevent the performance collapse in simulation-to-reality transfer that occurs in deep learning (DL) methods driven by semi-realistic data due to inaccurate prior information, this article also proposes a novel semi-supervised learning MAD method driven by sparse prior information about the magnetic anomaly. Experiments demonstrate that the proposed method exhibits superior stability, particularly showing enhanced robustness against $1/f^{\alpha }$ noise compared to supervised learning methods. Moreover, the controlled prior information integration mode enables the proposed method to achieve effective tradeoffs between sensitivity and stability in practical deployments. This confirms the method’s reduced dependence on simulation-derived prior information, making it particularly suitable for complex real-world detection scenarios. Finally, the method’s practical performance has been validated using both terrestrial and marine field data.

Multiple magnetic targets localisation (MMTL) based on magnetic vector sensors has been widely used in UXO detection, medical application and marine targets monitoring. However, locating an unknown number of non-cooperative targets and targets with highly overlapping horizontal positions under sparse measurements is challenging. This study proposes a multiple magnetic targets localisation method that combines a 3-D inversion neural network with a local optimisation algorithm. The method divides the inversion space into a fixed grid and employs a 3D U-net network with a super-resolution module to reconstruct the 3D magnetic moment distribution in space. Then, the reconstructed distribution is processed using clustering method for preliminary positioning. Finally, the positions and magnetic moments of multiple targets are refined by the trust region reflective (TRR) algorithm. Compared with other methods, this method reduces the average positioning error of multiple targets, the magnetic moment estimation error, and the running time. Terrestrial experiments further demonstrate its effectiveness for unknown number of multiple magnetic targets and targets with overlapping horizontal positions.