The integrity of geomagnetic data is a critical factor in understanding the evolutionary process of Earth's magnetic field, as it provides useful information for near-surface exploration, unexploded explosive ordnance detection, and so on. Aimed to reconstruct undersampled geomagnetic data, this paper presents a geomagnetic data reconstruction approach based on machine learning techniques. The traditional linear interpolation approaches are prone to time inefficiency and high labor cost, while the proposed approach has a significant improvement. In this paper, three classic machine learning models, support vector machine, random forests, and gradient boosting were built. Besides, a deep learning algorithm, recurrent neural network, was explored to further improve the training performance. The proposed learning models were used to specify a continuous regression hyperplane from a training data. The specified regression hyperplane is a mapping of the relation between the mock-up missing data and the surrounding intact data. Afterward, the trained models, essentially the hyperplanes, were used to reconstruct the missing geomagnetic traces for validation, and they can be used for reconstructing further collected new field data. Finally, numerical experiments were derived. The results showed that the performance of our methods was more competitive in comparison with the traditional linear method, as the reconstruction accuracy was increased by approximately 10%~20%.

The accuracy of measuring a magnetic field using a proton magnetometer depends on the accuracy of frequency measurement of the free induction decay (FID) signal generated by the proton magnetometer. Increasing the accuracy of frequency measurement could increase the accuracy of measuring magnetic fields. In this paper, we propose a multichannel frequency measurement to improve the accuracy after analyzing the error sources of existing frequency measurements based on the FID signal. In addition, we describe the basic principle of the proposed method and the formula for the calculation of the measured FID signal frequency and error in detail. Moreover, we implement a prototype device using the multichannel frequency measurement and compare the test results before and after using the proposed method. Finally, the indoor and outdoor experimental data in the case of using the prototype device are compared with those in the case of a very high-precision overhauser magnetometer. The comparisons indicate that the proposed method improves the accuracy of the FID signal frequency measurement, and therefore, improves the accuracy of magnetic field measurement.

Magnetic anomaly detection (MAD) is one of the most effective methods for engineering and environmental geophysical exploration, and plays an important role in scientific research and human technological innovation. Although numerous systematic and valuable articles have been published on the topic of MAD, they only briefly summarized relevant research from an algorithm perspective. Moreover, the practical applications in different scenarios are neglected and a comprehensive review on MAD is still lacking. In this case, it is particularly necessary to implement a comprehensive survey on the relevant publications of MAD, which could help readers to understand the current state-of-the-art techniques conveniently and quickly develop an effective solution to MAD problems in practice. Specifically, the developments of the MAD, mainly consisting of detection models and detection methods, are combed in this paper. In addition, the different application scenarios of the MAD have been argued. More importantly, the existing issues and future trends are further discussed. We believe that the work done in this survey could supply convenience and inspiration for the researchers who want to devote their efforts to MAD.

Composite materials have been widely used in the aerospace industry and are critical for safe operations. However, the delamination, caused by cyclic loads and corrosive service environment, poses a serious threat to the structural integrity of composite laminates. The acoustic emission technique has been adopted to assess the structural integrity by characterizing damage location, type, and size. This paper proposes and compares data-driven prognostic methods to quantify the delamination area efficiently and accurately. To address the problem of insufficient inspection data, the prediction model adopts the path length across the delamination area as the target value. The delamination area can then be estimated with the predicted path length based on formulated geometric relationships. This solution will augment the model training data sets and consequently avoid the overfitting problem during the training process. Experimental results on composite coupons demonstrate that the proposed ensemble learning based model outperforms other state-of-the-art methods in terms of prediction accuracy and efficiency.

Surface defects directly affect the mechanical properties of industrial strip steel products. To evaluate the integrity of the strip steel surface, a channel-wise global Transformer based dual-branch network for strip steel surface defect detection, dubbed CGTD-Net, is proposed in this study. First, the strip steel surface images are pre-processed using saturation adjustment and random flipping strategies to remove unnecessary background information and improve network generalization. Second, the Swin Transformer is employed at the end of the backbone network and the negative impacts of a single channel are then mitigated by using the multi-channel feature pyramid networks via Transformer, which improves the extraction ability of the global semantic information for tiny or narrow defects. Third, an edge detection branch network is constructed with a spatial-channel attention module (SCGA) to further enhance the feature extraction on both spatial and channel information. Finally, the CGTD-Net is compared with eleven state-of-the-art methods on the NEU-DET dataset, and ablation experiments are also implemented. The comparison results, conducted on a single 3090Ti GPU, reveal that the CGTD-Net achieves a mean intersection over union (mIoU) of 75.16% at 178 frames per second (FPS), outperforming other methods. The ablation experiment demonstrates that the CGTD-Net improves the mIoU by 7.83% and the F-score by 6.3% compared to the baseline.