Deep Fault Research Articles

Rolling bearing intelligent fault diagnosis towards variable speed and imbalanced samples using multiscale dynamic supervised contrast learning

Deep learning-based fault diagnosis methods have already attained remarkable achievements in this field. However, rolling bearing frequently operates under variable speed conditions, and the number of healthy samples collected is often significantly larger than that of failure samples. In this paper, a multiscale dynamic supervised contrast learning (MDSupCon) framework is proposed. First, a multiscale adaptive feature extraction network is designed as the backbone, which utilizes multiple convolutional kernels to enhance feature extraction capabilities under variable speed conditions, and the branch attention mechanism is incorporated to adaptively adjust the weights of various scale branches. Second, the joint channel-space attention mechanism is constructed to enhance the importance of critical features while reducing redundant information, thereby improving fault identification accuracy and interpretability. Third, the dynamic supervised contrast loss function is designed to assign dynamic compensation factors to samples of various categories according to the training results, which reduces the impact of easily classified samples and enhances the contribution of hard-to-classify samples in imbalanced scenarios. Additionally, a dynamic cross-entropy loss is designed to train the backbone and the classifiers. The MDSupCon has achieved superior results of 89.49% and 92.15% on two bearing datasets with an imbalance ratio of 20:1 and variable speeds.

An unsupervised transfer learning approach for rolling bearing fault diagnosis based on dual pseudo-label screening

Deep transfer learning is an effective method for unsupervised fault diagnosis of rolling bearings. In some works, the pseudo-label of target domain prediction is used to improve the ability of target domain prediction in transfer learning. However, its validity depends on the quality of pseudo-label generated by the network itself, which is easy to cause the misclassification of the samples. Aiming to this, a dual sample screening (DSS) method based on the information of predicted label changes is proposed in the article, and it is applied to the fault diagnosis of rolling bearings with variable working conditions. DSS combines pre-screening and real-time screening and uses the continuous output of prediction label change information in the training process to improve the network training. It owes to eliminating part of the target domain samples with prediction errors in the stage of network training with pseudo-label. The proposed method improves the stability of the pseudo-label involved in the training and alleviates the negative effects caused by the pseudo-label. The experimental results on Paderborn University dataset show that, compare with the deep transfer learning fault diagnosis method based on pseudo-label cross-entropy, the average diagnostic accuracy of the six transfer tasks using DSS is increased by 5.97%, which effectively improves the fault diagnosis accuracy of rolling bearings.

A robust high-resolution potential-field edge detection filter with its application in the Indo-China Peninsula

Potential-field edge detection techniques are effective tools for identification of large-scale tectonic boundaries and geologic target bodies. Although a series of edge detectors have been developed, most of them suffer from a trade-off between edge resolution and noise immunity. To this end, we integrate multiple edge detectors such as normalized Harris filter (NHF) and ILTHG filter to propose a novel edge detector called ILGNH. The ILGNH edge detector takes advantage of both the enhanced noise immunity of NHF and the high resolution of ILTHG. Model experiments without/with noises both show that it is an alternative robust high-resolution edge detector. Afterwards, we apply this edge detection method to the Indo-China Peninsula and its adjacent areas to identify multi-phase tectonic boundaries. Combining the edge detection, residual crustal gravity anomalies (RCGA), and active faults, we identify four major NS transverse suture zones associated with microplates. These suture/boundary zones, together with deep faults, enclose 14 microplates with different interior feature of RCGA, while most of the boundaries are located in the gradient zones or the dislocation/distortion zones of the RCGA. Notably, inside the West Burma microplate, the ILGNH edge detection shows a distinct near NS-oriented arc-shaped deep boundary zone, which indicates that the underlying slab beneath the Indo-Burma Ranges was blocked during its eastward advance and shifted downward to form a high-angle deep subduction. Further, multi-scale edge detection at different depths confirms that the microplate boundary under the western Indo-China Peninsula vertically extends deeper than the eastern part. In addition, the seismicity below 50 km is concentrated to the west of the identified arc-shaped deep boundary zone and near the Naga-Kalaymyo-Arakan-Andaman Suture Zone.

Geochemistry and origins of hydrogen-containing natural gases in deep Songliao Basin, China: Insights from continental scientific drilling

The different reservoirs in deep Songliao Basin have non-homogeneous lithologies and include multiple layers with a high content of hydrogen gas. The gas composition and stable isotope characteristics vary significantly, but the origin analysis of different gas types has previously been weak. Based on the geochemical parameters of gas samples from different depths and the analysis of geological settings, this research covers the diverse origins of natural gas in different strata. The gas components are mainly methane with a small amount of C2+, and non-hydrocarbon gases, including nitrogen (N2), hydrogen (H2), carbon dioxide (CO2), and helium (He). At greater depth, the carbon isotope of methane becomes heavier, and the hydrogen isotope points to a lacustrine sedimentary environment. With increasing depth, the origins of N2 and CO2 change gradually from a mixture of organic and inorganic to inorganic. The origins of hydrogen gas are complex and include organic sources, water radiolysis, water-rock (Fe2+-containing minerals) reactions, and mantle-derived. The shales of Denglouku and Shahezi Formations, as source rocks, provide the premise for generation and occurrence of organic gas. Furthermore, the deep faults and fluid activities in Basement Formation control the generation and migration of mantle-derived gas. The discovery of a high content of H2 in study area not only reveals the organic and inorganic association of natural-gas generation, but also provides a scientific basis for the exploration of deep hydrogen-rich gas.

Model of Formation of Gold-Bearing Placers in Platform Areas (East of the Siberian Platform)

A model of the formation of gold-bearing placers for platform areas is proposed for the first time. Placers on the platforms are formed mainly due to the Precambrian ore sources spatially confined to the basement outcrops, and, locally, Mesozoic, formed when tectonic-magmatic activation occurs in the zones of paleorifts and deep faults. Placers formed due to the Precambrian sources are characterized by the presence of fine and thin gold with a size of 0.1‒0.25 mm and, as a rule, are not of commercial interest. They bar alluvial and coastal-marine allochtonous placers and are considered as complex, with associated mining of titanium, zirconium and other useful components. Placers formed due to the sources of the Mesozoic stage of ore formation belong to the class of small and medium reserves, the size of gold in them ranges from dust-like to 0.2‒0.25 mm or more and they are usually mined by prospectors' team. Placers on the platforms do not form sheet deposits, since they are formed mainly due to sources of non-placer-generating formations. The presence of such placers indicates the proximity of the ore occurrence, which the type and location can be determined based on the study of the mineralogical and geochemical features of the placer gold itself.

Deep Learning-Based Positioning Error Fault Diagnosis of Wire Bonding Equipment and an Empirical Study for IC Packaging

Little research has been done for artificial intelligence applications of semiconductor backend. This study aims to develop a deep learning based fault diagnosis framework as prognostics and health management (PHM) solutions with a comprehensive analytics process. A linear encoder sensor is employed to measure position, while data preparation is conducted to convert position information into a signal. Then, signal processing is used to extract the features, in which high pass filter is applied to normalize the signal and emphasize the features. High-dimensional features including time domain statistical features, time-frequency domain features obtained by continuous wavelet transform (CWT), and frequency domain features converted by fast Fourier transform (FFT) are extracted to prepare the dataset. An ensemble neural network model that integrates deep neural network (DNN) and convolutional neural network (CNN) is employed to recognize the pattern and diagnose the positioning errors. The accuracy and false alarm rate are considered to support maintenance decisions. An empirical study was conducted in a world leading IC assembly and testing company for validation. The results have shown that the proposed approach can effectively detect inappropriate installation of the bond head in wire bonding equipment for predictive maintenance and cost reduction.

Chemical Process Fault Diagnosis Method Based on Deep Learning

To address the problem that the traditional fault diagnosis method for chemical processes under big data relies too much on expert experience and fault features are difficult to distinguish, a deep learning-based fault diagnosis method is proposed, which combines convolutional neural network (CNN), long and short-term memory (LSTM) and attention mechanism (AM). In this method, the spatial sequence features of the input signal are extracted by the CNN adaptively, while the LSTM extracts the time-series features of the signal. Finally, the model performance is enhanced by introducing the attention mechanism and using the SoftMax layer as a classifier for fault diagnosis, so that the model can notice the important features of the faults with the interference of noise. Simulation validation of the method in this paper is performed using the TE chemical process data set, and it is demonstrated that the method can be used for chemical process fault diagnosis studies. Finally, compared with other fault diagnosis methods, the method is more accurate and has certain superiority.

Hydrothermal activity and metallogenic environment revealed by seismic wave tomography in Yunnan Province, China

S-wave velocity structures are revealed using double-difference tomography, which shows large-scale low-velocity anomalies around the Xiaojiang and Lijiang−Xiaojinhe fault zones at 10−25 km depth. In contrast, an obvioushigh-speed anomaly exists in the center of Emeishan area, which is consistent with the EmeishanLarge Igneousprovince. Combined with the geological and geophysical data, the low-velocity anomaly might relate to the lower crustal flowof central Tibetan Plateau. The high-velocity anomaly associated with mantle plume in Emeishan would obstruct the crustal flow and limit it around the large deep faults. Meanwhile, the results also show that many mineral resources are mainly concentrated on the major faults and low-velocity zones, indicating that the lower crustal flow and the large deep faults provide conditions for mineral fluids and minerals deposit.

Faults and gas chimneys jointly dominate the gas hydrate accumulation in the Shenhu Area, northern South China Sea

Gas hydrates possess significant potential as an energy resource and exert a notable influence on global climate change. The Shenhu Area is one of the globally recognized focal points for gas hydrate research, and additional investigation is required to fully comprehend its gas migration mechanism. By utilizing the most recent core-log-seismic data and gas geochemical data, a comprehensive analysis was conducted to determine the influence of gas migration pathways on gas hydrate accumulation in the study area. This study investigated the various types of gas migration pathways, employing integrated geological models that incorporate faults and gas chimneys to understand their respective contributions to the accumulation of gas hydrates. Based on these findings and drilling constraints, a three-gas combined production model was subsequently proposed. Thermogenic gas, secondary microbial gas, and in situ microbial gas are all potential sources of the gas responsible for hydrate formation. Thermogenic gas plays a significant role in the gas hydrate system, as evidenced by distinct features of late-mature thermogenic gas observed in gas samples extracted from hydrates in Well W18. In the study area, the primary conduits for gas migration encompass deep faults, branch faults, and gas chimneys. Among these, deep faults act as the most crucial pathways of thermogenic gas migration. The integration of geological models that incorporating deep faults and gas chimneys has profoundly impacted the accumulation of gas hydrates in the Shenhu Area, consequently influencing the distribution of shallow gas and gas hydrate. Furthermore, the proposed three-gas combined production model, which involves the simultaneous extraction of deep gas reservoirs, shallow gas reservoirs, and gas hydrates, holds significant implications for exploring and developing deep-water natural gas resources. However, its successful implementation necessitates interdisciplinary collaboration among scientists.

Effects of hydrothermal alteration on shear localization and weakening in the mantle lithosphere

Hydrothermal alteration occurs in upper-mantle fault zones and significantly modifies the rheological properties of ultramafic rocks. To understand how hydrothermal alteration affects shear localization and the strength of the fault zones, we conducted simple-shear deformation experiments on water-saturated harzburgite (70 wt% olivine +30 wt% orthopyroxene) gouges sandwiched between three types of shear pistons (tungsten, yttria-stabilized zirconia, and corundum) using two deformation-DIA apparatuses at confining pressures of 1–4 GPa, temperatures of 500–580 °C, average shear strain rates of 6.1 × 10−6 s−1 to 1.0 × 10−4 s−1, and water contents of 4–30 wt%. The harzburgitic gouges with water contents of 10 and 30 wt% exhibited steady-state sliding or strain-hardening behavior, and the shear strength of the latter sample was less than one-third of that of the former sample, possibly reflecting the difference in pore fluid pressure. Olivine/orthopyroxene grains exhibit distributed microfracturing and grain rotation towards P foliation, indicating the operation of cataclastic flow. Shear displacement is also accommodated by the development of serpentine- or talc-bearing multiple shear zones including B, R1, and/or Y shears, where intense cataclasis-related grain size reduction and dissolution/precipitation occur. For experimental runs using tungsten and zirconia shear pistons, the serpentine or talc blades that wrap finer-grained and dissolved olivine/orthopyroxene grains form an interconnected network, indicative of frictional sliding or dislocation glide on their basal planes as a dominant deformation mechanism. Raman spectra of the serpentine minerals indicate that they represent antigorite with low crystallinity. These findings suggest that when deep lithospheric fault zones undergo hydrothermal alteration, deformation is controlled by competition between cataclastic flow in olivine/orthopyroxene matrix and slip or glide of antigorite or talc in semi-brittle fault zones, the predominance of which depends on the amount of the hydrous phyllosilicates and the magnitude of pore fluid pressure.

Inert Gas—A New Geochemical Technology for Natural Gas Hydrate Exploration in Midlatitude Permafrost

The development of geochemical exploration technologies unaffected by marsh microorganisms is necessary to improve the prediction of wells and explore for natural gas hydrates in mid-latitude permafrost areas. Here, we examine the potential of inert gas as a new tool for the investigation of gas hydrates in permafrost areas. The study area, 150 km2 in size, is situated in the alpine wetland landscape of the Qilian Mountains. The sampling density and depth were 2 points/km2 and 60 cm, respectively. In total, 300 soil and headspace gas samples were collected. The chromatographic backflush technique was used to analyze the inert gases helium (He) and neon (Ne) in the headspace gas samples. A comprehensive interpretation was conducted based on geological and geochemical survey results. We propose a geogas migration mechanism of inert gases in the soil above the natural gas hydrate deposits. The inert gas anomalies near the ground surface of the Muli coalfield in the Qilian Mountains, derived from the deep hydrate deposits and fault structure, are not affected by marsh microorganisms. Thus, inert gas anomalies are practical tools for natural gas hydrate exploration in permafrost areas.

Genesis of Hg-Sb deposit controlled by strike-slip structures in Qinling orogenic belt: A case study of the Qingtonggou ore deposit, Shaanxi, China

The Qingtonggou deposit is the largest Hg-Sb deposit in the Qinling orogenic belt, but the deposit genesis remains unclear. Fluid inclusion study indicates that the homogenization temperature range from 139.9 °C to 217 °C and the salinities vary from 0.7 to 2.1 wt% (NaCleq). The δ13C value in ore-stage calcite is from-7.6 ‰ to −7.1 ‰, and the δ18OSMOW value is between 13.6 ‰ and 15.3 ‰. The δ18OV-SMOW value in hydrothermal quartz is from 21.5 ‰ to 24.8 ‰, and the δD value is between −138 ‰ to −99 ‰. The δ34SV-CDT value range from 4 ‰ to 11.6 ‰, the 206Pb/204Pb values range from 18.0351 to 18.5080, the 207Pb/204Pb values range from 15.6390 to 15.6672, and the 208Pb/204Pb values range from 38.1586 to 38.4128. The hydrothermal calcite Sm-Nd isochron age for ore-stage calcite is 98.02 ± 0.42 Ma. The Hg-Sb mineralization is controlled by tectonic space formed by the left strike-slip fault, the ore-forming fluid has the characteristics of medium–low temperature, low density, and supergene mineralization, the ore-forming fluid comes from meteoric water and formation water, the ore-forming material comes from the stratum and basement of Yaolinghe group and Lower Cambrian, the Hg-Sb mineralization occurred at early Late Cretaceous. The complete metallogenic process is that the Yaolinghe Group and Lower Cambrian in this area are the source areas for Hg-Sb mineralization, the subsequent high background values Hg and Sb strata which received the denudation deposit of the source bed all contributed ore-forming materials, with the Qinling collision orogeny and extrusion in the Indosinian period and the continuous contraction of the Paleozoic basin, porous meteoric water and formation water extracted the ore-forming materials in source bed, formed ore-forming fluids, during the strike-slip process in intracontinental environment, deep and large faults provide a channel for the upward discharge of ore-forming fluids, with the decrease of fluid temperature, sulfide precipitation occurs in secondary fracture.

A catastrophic, buried fault-generating earthquake: The 1937 M7.0 Heze earthquake in the south-central North China Plain

The North China Plain (NCP) is a region with a high level of seismic hazards. Since 1830, eight historical M7.0–7.9 earthquakes have been recorded in the NPC, seven of which have occurred on the boundary of the subblocks, and only the 1937 Heze earthquake has occurred in the interior of the Ji-Lu Subblock. Why did this intraplate earthquake occur there? What are its seismogenic conditions and seismogenic mechanisms? The Liaocheng-Lankao Fault (LLF) is an important buried active fault in the south-central NCP, and the 1937 Heze earthquake occurred 10 km east of it. What role did the LLF play in this earthquake? Using shallow seismic reflection detection and drilling combined with Quaternary dating methods reveals that the southern segment of the LLF is a Holocene active fault, and the northern segment is a late Pleistocene active fault. The BCS drilling profile reveals four paleoseismic events in the southern LLF since the late Pleistocene, and the average vertical slip rate of the southern LLF since 21 ka is approximately 0.35 ± 0.04 mm/a. The analysis of the intensity lines, surface rupture distribution, and focal mechanism solution of the 1937 Heze M7.0 earthquake implies that the NNE-striking Xiaoliu-Xieyuanji Fault and the WNW-striking Dongming-Chengwu Fault are the seismogenic faults of the 1937 Heze M7.0 earthquake. The LLF, as the deep major fault in the region, controlled the accumulation of stress, stimulated the earthquake with its deep movement, and reduced the effect of the seismic energy westward, acting as the regional seismic controlling fault of the 1937 Heze M7.0 earthquake. Another branch fault near Yangji, approximately 50 km north of Heze, is a new stress concentration area and may be at risk of a large earthquake. In addition, the northern segment of the LLF may also be reactivated in the future and produce an M ≥ 7.0 earthquake.

The vesiculated layer in the anomalous diogenite Northwest Africa 12973

AbstractDiogenites, which are part of the howardite–eucrite–diogenite (HED) group, are considered to represent rocks from the lower crust and mantle of a differentiated planetary body, likely the asteroid 4 Vesta. The Northwest Africa 12973 (NWA 12973) meteorite was classified as an anomalous diogenite due to the occurrence of a vesiculated layer. This work reports on the petrographic and geochemical study of two fragments of this meteorite, aiming to better constrain the origin of the vesiculated layer. Whereas the interior of NWA 12973 (here called host) presents the typical characteristics of an olivine diogenite, that is, coarse‐grained pyroxene, olivine ribbons, chromite, and accessory phases, the vesiculated layer presents a fine‐grained pyroxene groundmass with local rounded relics of olivine and interstitial chromite and metal, and is characterized by abundant large vesicles. The contact between the vesiculated layer and the host is sharply defined. The composition of individual minerals does not show any significant differences between the host and the vesiculated layer, suggesting in situ melting. Geothermobarometry indicates a slightly higher crystallization temperature at lower pressure for the vesiculated layer, consistent with melting and crystallization under lower crustal conditions upon exhumation. The trigger for the local melting was likely a large impact event, which was responsible for adiabatic decompression in the central area or deep faulting and frictional melting.

Leveraging deep learning techniques for ship pipeline valve leak monitoring

The monitoring of ship pipeline valve leaks holds immense significance as it serves to enhance ship safety, mitigate energy and material losses, and protect the health and sustainability of the marine environment. In comparison to traditional expert-based fault diagnosis methods, this study presents a deep learning-based fault feature extraction approach for end-to-end valve leak fault diagnosis. To address the challenge of incomplete leak fault description by a single sensor, a Multi-Channel Multi-Scale Convolutional Neural Network (MCMS-CNN) model is established. Diverging from existing methods, the proposed MCMS-CNN model automatically extracts fault features from grayscale images obtained from two sensors and performs feature-level fusion for fault classification, thereby mitigating the impact of data redundancy and noise and enhancing fault recognition accuracy. The effectiveness of the proposed approach is verified through valve leak experiments, and comparisons are conducted with different models and information fusion methods. The results demonstrate that the proposed MCMS-CNN model exhibits advantages in terms of accuracy and robustness, confirming its practicality in ship pipeline valve leak fault diagnosis.

Fault diagnosis method for proton exchange membrane fuel cell system based on digital twin and unsupervised domain adaptive learning

In recent years, supervised deep learning-based methods have achieved significant results in fuel cell system fault diagnosis. However, most existing deep learning-based fault prediction methods suffer from missing fault labels and data limitations due to the difficulty in obtaining fault and degradation data in real fuel cell systems. To address the above challenges, this paper proposes a fault diagnosis method based on digital twin and unsupervised domain adaptive learning. The method has two key features: First, a maximum-relevance minimum-redundancy algorithm is used to select the input signals. Then a high-order fuel cell system model is developed to obtain a large amount of digital domain fault data at low cost by simulating fault injection. Second, domain-invariant features are extracted using a domain-adaptive adversarial learning approach to reduce the distribution differences between the digital and real domains. The method successfully diagnosed nine typical faults in the fuel cell air, hydrogen, and thermal subsystems without real data fault labels. Under dynamic load conditions, the diagnostic accuracy reached 92.5 %. In addition, the method achieves a diagnostic accuracy of over 90 % under domain adversarial training using only normal real data. The experimental results show that the proposed method can achieve fuel cell system fault diagnosis without fault labels and significantly reduce the dependence on fault data.

A Semi-Supervised Siamese Network for Complex Aircraft System Fault Detection with Limited Labeled Fault Samples

Aircraft onboard systems typically have limited labeled fault samples and large amounts of unlabeled data. To better utilize the information contained in limited labeled fault samples, a deep learning-based semi-supervised fault detection method is proposed, which leverages a small number of labeled fault samples to enhance its performance. A novel sample pairing strategy is introduced to improve algorithm performance by iteratively utilizing fault samples. A comprehensive loss function is employed to accurately reconstruct normal samples and effectively separate fault samples. The results of a case study using real data from a commercial aircraft fleet demonstrate the superiority of the proposed method over existing techniques, with improvements of approximately 16.7% in AP, 9.5% in AUC, and 19.2% in F1 score. Ablation studies confirm that performance can be further improved by incorporating additional labeled fault samples during training. Furthermore, the algorithm demonstrates good generalization ability.

Water–Rock Interactions, Genesis Mechanism, and Mineral Scaling of Geothermal Waters in Northwestern Sichuan, SW China

Geothermal resources are the vital renewable energy for resolving energy crisis and environmental deterioration. Understanding hydrogeochemical processes, genesis mechanisms and scaling trends is crucial for securing the sustainable utilization of geothermal resources. In this study, fourteen geothermal waters were collected for hydrochemical and δ2H–δ18O isotopic analyses in northwestern Sichuan, SW China to clarify hydrogeochemical processes, genesis mechanisms, and scaling trends. Geothermal waters were recharged via atmospheric precipitation. Three different types of geothermal waters were identified using a piper diagram. Class 1 geothermal water with HCO3–Na and HCO3–SO4–Na types formed in the contact zone with Yanshanian intrusions and heated by residual radioactive heat. The hydrochemical processes were sodium/potassium silicate dissolution and positive cation–exchange. Class 2 geothermal water with HCO3–Ca and HCO3–Ca–Mg type was carbonate–type and heated by geothermal gradient. The dissolution of carbonate minerals dominated the hydrochemical process. Class 3 geothermal water with the SO4–Ca–Mg type was determined within deep faults. The dissolution of carbonatite and gypsum minerals and the oxidation of sulfides played a vital role in the hydrochemical process. The reservoir temperatures of geothermal waters followed the orders of Class 1 (74.9–137.6 °C) > Class 3 (85.9–100 °C) > Class 2 (38.7–93.5 °C). Calcium carbonate scaling should be paid attention to in Class1 and Class 3 geothermal water, and calcium sulfate scaling merely occurs in Class 3 geothermal water. This study provides vital information for geothermal exploitation in western Sichuan and other similar areas.

Earthquake relocation using a 3D velocity model and implications on seismogenic faults in the Beijing-Tianjin-Hebei region

To enhance the understanding of the geometry and characteristics of seismogenic faults in the Beijing-Tianjin-Hebei region, we relocated 14 805 out of 16 063 earthquakes (113°E−120°E, 36°N–43°N) that occurred between January 2008 and December 2020 using the double-difference tomography method. Based on the spatial variation in seismicity after relocation, the Beijing–Tianjin–Hebei region can be divided into three seismic zones: Xingtai–Wen'an, Zhangbei–Ninghexi, and Tangshan. (1) The Xingtai–Wen'an Seismic Zone has a northeast-southwest strike. The depth profile of earthquakes perpendicular to the strike reveals three northeast-striking, southeast-dipping, high-angle deep faults (>10 ​km depth), including one below the shallow (<10 ​km depth) listric, northwest-dipping Xinghe fault in the Xingtai region. Two additional deep faults in the Wen'an region are suggested to be associated with the 2006 M 5.1 Wen'an Earthquake and the 1967 M 6.3 Dacheng earthquake; (2) The Zhangbei-Ninghexi Seismic Zone is oriented north-northwest. Multiple northeast-striking faults (10–20 ​km depth), inferred from the earthquake-intensive zones, exist beneath the shallow (<10 ​km depth) Xiandian Fault, Xiaotangshan Fault, Huailai-Zhuolu Basin North Fault, Yangyuan Basin Fault and Yanggao Basin North Fault; (3) In the Tangshan Seismic Zone, earthquakes are mainly concentrated near the northeast-striking Tangshan-Guye Fault, Lulong Fault, and northwest-striking Luanxian-Laoting Fault. An inferred north-south-oriented blind fault is present to the north of the Tangshan-Guye Fault. The 1976 M 7.8 Tangshan earthquake occurred at the junction of a shallow northwest-dipping fault and a deep southeast-dipping fault. This study emphasizes that earthquakes in the region are primarily associated with deep blind faults. Some deep blind faults have different geometries compared to shallow faults, suggesting a complex fault system in the region. Overall, this research provides valuable insights into the seismogenic faults in the Beijing–Tianjin–Hebei region. Further studies and monitoring of these faults are essential for earthquake mitigation efforts in this region.

IDSN: A one-stage interpretable and differentiable STFT domain adaptation network for traction motor of high-speed trains cross-machine diagnosis

A surge of transfer fault diagnosis techniques has been proposed to guarantee the safe operation of traction motor systems. However, existing efforts highly depend on the availability of fault data in source domain, which is rare in practice due to the regular maintenance. Fortunately, self-customized testbeds provide an opportunity to easily obtain fault data, assuming that the simulated data can be utilized to monitor the real-world traction motor systems via the cross-machine diagnosis method. Besides, current deep learning-based cross-machine fault diagnosis methods suffer from the poor physical interpretability and the troublesome hype-parameter selection. To tackle aforementioned issues, a one-stage Interpretable and Differentiable STFT cross-machine dual-driven adaptation Network (IDSN) is proposed. In IDSN, a new paradigm termed interpretable differentiable STFT layer is devised, where a derivable coefficient is introduced to adjust pivotal parameters of STFT such as window length by the gradient descent. Prominently, it is a plug-and-play module, which can be embedded into the arbitrary typical network without conflict. Besides, a novel adaptive trade-off coefficient is developed to tackle the weight matching of the domain discrepancy metric. Finally, to ensure the reliability and effectiveness of cross-machine diagnosis, a concise yet valid smoothed joint maximum mean discrepancy is proposed, which simultaneously promotes intra-class compactness and inter-class separability. The results of experiments confirm that the proposed IDSN outperforms the state of the art.