Xiaotian Qiu

Aeromagnetic compensation is essential for enhancing the accuracy of aeromagnetic measurements by estimating and eliminating magnetic interference parameters. Traditional methods typically estimate invariant interference parameters, necessitating regular calibration flights for adjustments. In our previous work, we introduced an evolutionary aeromagnetic compensation method that updates magnetic interference parameters to adapt to variant interferences. While this evolutionary approach is effective, it assumes geomagnetic invariance, which is often not the case during regular low-altitude flights where the geomagnetic field is dynamic and complex. To enhance the applicability of the evolutionary compensation method, this paper presents a modified approach that accounts for geomagnetic variations. We first developed a geomagnetic field model with estimable parameters, which we then integrated with a model of magnetic interference parameters. Building on this integrative parameter model, we refined the evolutionary aeromagnetic compensation method to enable simultaneous estimation of both variable interference and geomagnetic parameters synchronously. The efficacy of our proposed method was validated through simulations and real-flight experiments. Results indicate that, in scenarios with dynamic interference and geomagnetic fluctuations, our method reduces magnetic interference from 0.941 nT to 0.195 nT, which is 21% better than the traditional ridge regression method (TraRR) and 16% better than the conventional evolutionary algorithm (TraEvo) from the literature.

Magnetic interference of the aircraft platform is one of the most primary restrictions for aeromagnetic missions, like magnetic anomaly detection (MAD) and geomagnetic navigation. In recent years, two main paradigms have emerged for aeromagnetic compensation: model-driven methods based on the linear Tolles-Lawson model and data-driven methods utilizing neural networks. However, model-driven methods struggle with nonlinear interference compensation, while data-driven methods exhibit a high dependence on the quality and quantity of available data. To address the limitations of existing paradigms, this paper introduces a novel approach: an interference model-guided neural network for aeromagnetic compensation. This method leverages the strengths of both model-driven and data-driven techniques. The synergy is achieved through the development of a network initialization method based on T-L estimation and the incorporation of an interference model-embedded loss function. With the proposed method, aeromagnetic compensation can consider both linear and nonlinear interferences, and eliminate them even with limited and noisy data without significant accuracy degradation. The proposed method’s efficacy was validated through real flight tests. Results demonstrate that the proposed method outperforms both the model-driven and data-driven methods, with standard deviations reaching 0.0550nT, compared to 0.0812nT and 0.1306nT, respectively.

Aeromagnetic compensation is important in aeromagnetic detection. Traditionally, a calibration flight would be implemented to estimate the interference model. However, the magnetic interference is not constant during the mission flight, which makes the interference model constructed in the calibration flight inapplicable in the mission. To solve that problem, this paper proposes an evolutionary aeromagnetic compensation method based on Woodbury equation. With this method, the magnetic interference model parameters can be evolved during the mission flight on basis of the previous interference model and the updated acquired data. To evaluate the performance of the evolutionary method, both simulation and the experiment were conducted. The results indicate that the proposed method can effectively reduce the influence of changing magnetic interference. In the experiment, the standard deviation (STD) of measured data before compensation in the last mission flight is 1.2070nT, the traditional compensation can reduce the STD to 0.0863nT, while our evolutionary aeromagnetic compensation method can reduce STD to 0.0450nT. The results show that increasing changes of magnetic interference will make evolutionary aeromagnetic compensation vital for isolating signals created by geologic features from signals created by the aircraft.

Magnetic interferential compensation plays a vital role in geomagnetic vector measurement applications. Traditional compensation accounts for only the permanent interferences, induced field interferences, and eddy-current interferences. However, nonlinear magnetic interferences are found, which also have a great impact on measurement, and it cannot be fully characterized by a linear compensation model. This paper proposes a new compensation method based on a back propagation neural network, which can reduce the influence of the linear model on compensation accuracy due to its good nonlinear mapping capabilities. The high-quality network training requires representative datasets, yet it is a common problem in the engineering field. To provide adequate data, this paper adopts a 3D Helmholtz coil to restore the magnetic signal of a geomagnetic vector measurement system. A 3D Helmholtz coil is more flexible and practical than the geomagnetic vector measurement system itself when generating abundant data under different postures and applications. Simulations and experiments are both conducted to prove the superiority of the proposed method. According to the experiment, the proposed method can reduce the root mean square errors of north, east, and vertical components and the total intensity from 73.25, 68.54, 70.45, and 101.77 nT to 23.35, 23.58, 27.42, and 29.72 nT, respectively, compared with the traditional method.

Owing to the ferromagnetism interferential field, the component compensation of magnetic field distortion is vital for geomagnetic field vector measurement system. However, magnetic interferences such as eddy-current field have not been considered in traditional component compensation methods due to the reference component magnetic field and its variation with time are difficult to obtain. In this paper, a new compensation method using a 3D Helmholtz coil is proposed, in which the eddy-current field related with the reference component magnetic field and its variation with time are integrated into compensation model. In addition, it is not necessary to need rotation equipment to construct the equations of error model, and the 3D Helmholtz coil is used to generate any required compensation data to estimate the compensation model error parameters, which makes compensation procedure more simple and convenient. In order to verify effectiveness of the proposed method, the experiment is conducted, and the results demonstrate that the proposed method contributes to the accuracy improvement of geomagnetic vector measurement system. After compensation, the RMS errors of North, East, Vertical components and total intensity are reduced from 3648.3 nT, 2986.2 nT, 3496.2 nT, and 3894.1 nT to 38.72 nT, 29.58 nT, 35.42 nT and 42.82 nT, respectively.

Recognition of surface targets has a vital influence on the development of military and civilian applications such as maritime rescue patrols, illegal-vessel screening, and maritime operation monitoring. However, owing to the interference of visual similarity and environmental variations and the lack of high-quality datasets, accurate recognition of surface targets has always been a challenging task. In this paper, we introduce a multi-attention residual model based on deep learning methods, in which channel and spatial attention modules are applied for feature fusion. In addition, we use transfer learning to improve the feature expression capabilities of the model under conditions of limited data. A function based on metric learning is adopted to increase the distance between different classes. Finally, a dataset with eight types of surface targets is established. Comparative experiments on our self-built dataset show that the proposed method focuses more on discriminative regions, avoiding problems like gradient disappearance, and achieves better classification results than B-CNN, RA-CNN, MAMC, and MA-CNN, DFL-CNN.