The remaining useful life (RUL) prediction is of significance to the health management of bearings. Recently, deep learning has been widely investigated for bearing RUL prediction due to its great success in sequence learning. However, the improvement of the prediction accuracy of existing deep learning algorithms heavily relies on feature engineering such as handcrafted feature generation and time–frequency transformation, which increase the complexity and difficulty of the actual deployment. In this paper, a novel spatial attention-based convolutional transformer (SAConvFormer) is proposed to establish an accurate bearing RUL prediction model based on raw vibration data without prior knowledge or feature engineering. In this algorithm, firstly, a convolutional neural network enhanced by a spatial attention mechanism is proposed to squeeze the feature maps and extract the local and global features from raw bearing vibration data effectively. Then, the extracted senior features are fed into a transformer network to further explore the sequential patterns relevant to the bearing RUL. An experimental study using the XJTU-SY rolling bearings dataset revealed the merits of the proposed deep learning algorithm in terms of root-mean-square-error (RMSE) and mean-absolute-error (MAE) in comparison with other state-of-the-art algorithms.

Evaluating the impact of geomagnetic disturbances on power grid infrastructure is critical to mitigate the risk posed by geomagnetically induced currents (GICs). In this paper, the geoelectric field and induced voltage distribution in North China were estimated from the SinoProbe magnetotelluric (MT) impedance data together with the geomagnetic observatory data of six INTERMAGNET stations recorded during the significant geomagnetic storm of 17th March 2015. The measured impedances from 119 SinoProbe MT sites were convolved with geomagnetic observatory data to account for the Earth's complex three‐dimensional electrical resistivity structure. The resultant geoelectric field was then used to model the induced voltage distribution across the regional power transmission network in North China. Due to the large inter‐site distances of the SinoProbe MT program, the derived geoelectric field is mostly homogeneous, except in the Ordos Basin that displays a polarization of the geoelectric field, and with higher magnitudes in the orogenic belts. The estimated geoelectric fields in Taihang‐Lvliang, Yanshan, and Luxi orogenic belts of North China were large (>1 V/km) during the storm, due to high‐resistivity lithosphere resulting in large voltage gradients in the Earth. However, in relation to locations of major power transmission lines, only the central part of North China experienced induced voltages exceeding 100 V.

Fine structures in the lower crust sounded by seismic reflection can be formed in various geological processes and their electrical resistivity properties can provide significant insights into the geodynamic processes of the lithosphere. The magnetotelluric (MT) method is popular for imaging the electrical resistivity structure of the lithosphere; however, it lacks high resolution to resolve the small heterogeneities in deep crust. To recover the fine electrical resistivity structures, we propose a strategy based on their equivalent resistivity that defines the relationship between the fine structures and the inverted bulk resistivities. Thus, we first formulated the equivalent resistivity for a 3‐D laminated structure with >95% accuracy. In this strategy, the bulk resistivity inverted from MT data is used to retrieve the fine resistivity structures, for which the corresponding equivalent resistivities should match with the inverted resistivities. The strategy was tested to demonstrate its feasibility for analyzing the electrical property and the cause of lamellae in the lower crust of the Basin and Range Province. From the bulk resistivities of 1‐D anisotropic inversion, we retrieved the resistivity of lamellae that was in the range of 0.15–0.3 and coincident with the resistivity of silicate melts, implying that the lamellae are probably magma chambers with >50% melt fraction, which is also supported by seismic and geological evidences. Our case study demonstrated that this strategy based on equivalence can be successfully used to recover the fine resistivity structures from MT data.