Search Results (1,490)

CT scanners are essential tools in modern medical imaging. Sudden failures of their X-ray tubes can lead to equipment downtime, affecting healthcare services and patient diagnosis. However, existing prediction methods based on a single model struggle to adapt to the multi-stage variation characteristics of tube lifespan and have limited modeling capabilities for temporal features. To address these issues, this paper proposes an intelligent prediction architecture for CT tubes’ remaining useful life based on a dual-branch neural network. This architecture consists of two specialized branches: a residual self-attention BiLSTM (RSA-BiLSTM) and a multi-layer dilation temporal convolutional network (D-TCN). The RSA-BiLSTM branch extracts multi-scale features and also enhances the long-term dependency modeling capability for temporal data. The D-TCN branch captures multi-scale temporal features through multi-layer dilated convolutions, effectively handling non-linear changes in the degradation phase. Furthermore, a dynamic phase detector is applied to integrate the prediction results from both branches. In terms of optimization strategy, a dynamically weighted triplet mixed loss function is designed to adjust the weight ratios of different prediction tasks, effectively solving the problems of sample imbalance and uneven prediction accuracy. Experimental results using leave-one-out cross-validation (LOOCV) on six different CT tube datasets show that the proposed method achieved significant advantages over five comparison models, with an average MSE of 2.92, MAE of 0.46, and R² of 0.77. The LOOCV strategy ensures robust evaluation by testing each tube dataset independently while training on the remaining five, providing reliable generalization assessment across different CT equipment. Ablation experiments further confirmed that the collaborative design of multiple components is significant for improving the accuracy of X-ray tubes remaining life prediction. Full article

Accurate weld defect detection is critical for ensuring structural safety and evaluating welding quality in industrial applications. Manual inspection methods have inherent limitations, including inefficiency and inadequate sensitivity to subtle defects. Existing detection models, primarily designed for natural images, struggle to adapt to the characteristic challenges of weld X-ray images, such as high noise, low contrast, and inter-defect similarity, particularly leading to missed detections and false positives for small defects. To address these challenges, a multi-dimensional feature fusion model (MADet), which is a multi-branch deep fusion network for weld defect detection, was proposed. The framework incorporates two key innovations: (1) A multi-scale feature fusion network integrated with lightweight attention residual modules to enhance the perception of fine-grained defect features by leveraging low-level texture information. (2) An anchor-based feature-selective detection head was used to improve the discrimination and localization accuracy for five typical defect categories. Extensive experiments on both public and proprietary weld defect datasets demonstrated that MADet achieved significant improvements over the state-of-the-art YOLO variants. Specifically, it surpassed the suboptimal model by 7.41% in mAP@0.5, indicating strong industrial applicability. Full article

Analyzing the two-phase flow behavior in bearing lubrication is crucial for understanding friction and wear mechanisms, optimizing lubrication design, and improving bearing operational efficiency and reliability. However, the complexity of oil–air two-phase flow in high-speed bearings poses significant research challenges. Currently, there is a lack of comparative studies employing different simulation strategies to address this issue, leaving a gap in evidence-based guidance for selecting appropriate simulation approaches in practical applications. This study begins with a comparative analysis between experimental and simulation results to validate the reliability of the adopted simulation approach. Subsequently, a comparative evaluation of different simulation methods is conducted to provide a scientific basis for relevant decision-making. Evaluated from three dimensions—adaptability to rotational speed conditions, research focuses (oil distribution and power loss), and computational economy—the findings reveal that FVM excels at medium-to-high speeds, accurately predicting continuous oil film distribution and power loss, while MPS, leveraging its meshless Lagrangian characteristics, demonstrates superior capability in describing physical phenomena under extreme conditions, albeit with higher computational costs. Economically, FVM, supported by mature software ecosystems and parallel computing optimization, is more suitable for industrial design applications, whereas MPS, being more reliant on high-performance hardware, is better suited for academic research and customized scenarios. The study further proposes that future research could adopt an FVM-MPS coupled approach to balance efficiency and precision, offering a new paradigm for multi-scale lubrication analysis in bearings. Full article

The multiscale interactive system composed of wind, leaves, and droplets serves as a critical dynamic unit in precision orchard spraying. Its coupling mechanisms fundamentally influence pesticide transport pathways, deposition patterns, and drift behavior within crop canopies, forming the foundational basis for achieving intelligent and site-specific spraying operations. This review systematically examines the synergistic dynamics across three hierarchical scales: Droplet–leaf surface wetting and adhesion at the microscale; leaf cluster motion responses at the mesoscale; and the modulation of airflow and spray plume diffusion by canopy architecture at the macroscale. Key variables affecting spray performance—such as wind speed and turbulence structure, leaf biomechanical properties, droplet size and electrostatic characteristics, and spatial canopy heterogeneity—are identified and analyzed. Furthermore, current advances in multiscale modeling approaches and their corresponding experimental validation techniques are critically evaluated, along with their practical boundaries of applicability. Results indicate that while substantial progress has been made at individual scales, significant bottlenecks remain in the integration of cross-scale models, real-time acquisition of critical parameters, and the establishment of high-fidelity experimental platforms. Future research should prioritize the development of unified coupling frameworks, the integration of physics-based and data-driven modeling strategies, and the deployment of multimodal sensing technologies for real-time intelligent spray decision-making. These efforts are expected to provide both theoretical foundations and technological support for advancing precision and intelligent orchard spraying systems. Full article

Focusing on the problem of insufficient detection accuracy caused by blurred target boundaries, variable scales, and severe noise interference in side-scan sonar images, this paper proposes a high-precision detection network named ESL-YOLO, which integrates edge perception and adaptive quality assessment. Firstly, an Edge Fusion Module (EFM) is designed, which integrates the Sobel operator into depthwise separable convolution. Through a dual-branch structure, it realizes effective fusion of edge features and spatial features, significantly enhancing the ability to recognize targets with blurred boundaries. Secondly, a Self-Calibrated Dual Attention (SCDA) Module is constructed. By means of feature cross-calibration and multi-scale channel attention fusion mechanisms, it achieves adaptive fusion of shallow details and deep-rooted semantic content, improving the detection accuracy for small-sized targets and targets with elaborate shapes. Finally, a Location Quality Estimator (LQE) is introduced, which quantifies localization quality using the statistical characteristics of bounding box distribution, effectively reducing false detections and missed detections. Experiments on the SIMD dataset show that the mAP@0.5 of ESL-YOLO reaches 84.65%. The precision and recall rate reach 87.67% and 75.63%, respectively. Generalization experiments on additional sonar datasets further validate the effectiveness of the proposed method across different data distributions and target types, providing an effective technical solution for side-scan sonar image target detection. Full article

Transmission towers are particularly vulnerable to extreme wind events, which can lead to structural damage or collapse, thereby compromising the stability of power transmission systems. Enhancing the wind-resistant capacity of these towers is therefore critical for improving the reliability and resilience of electrical infrastructure. This study utilizes finite element analysis (FEA) to evaluate the structural response of a 220 kV transmission tower subjected to fluctuating wind loads, effectively capturing the dynamic characteristics of wind-induced forces. A comprehensive dynamic analysis is conducted to account for uncertainties in wind loading and variations in wind direction. Through this approach, this study identifies the most critical wind angle and local structural weaknesses, as well as determines the threshold wind speed that precipitates structural collapse. To improve structural resilience, a concurrent multi-scale modeling strategy is adopted. This allows for localized analysis of vulnerable components while maintaining a holistic understanding of the tower’s global behavior. To mitigate failure risks, the traditional perforated plate reinforcement technique is implemented. The reinforcement’s effectiveness is evaluated based on its impact on load-bearing capacity, displacement control, and stress redistribution. Results reveal that the critical wind direction is 45°, with failure predominantly initiating from instability in the third section of the tower leg. Post-reinforcement analysis demonstrates a marked improvement in structural performance, evidenced by a significant reduction in top displacement and stress intensity in the critical leg section. Overall, these findings contribute to a deeper understanding of the wind-induced fragility of transmission towers and offer practical reinforcement strategies that can be applied to enhance their structural integrity under extreme wind conditions. Full article

Facial expression recognition (FER) is a critical research direction in artificial intelligence, which is widely used in intelligent interaction, medical diagnosis, security monitoring, and other domains. These applications highlight its considerable practical value and social significance. Face expression recognition models often need to run efficiently on mobile devices or edge devices, so the research on lightweight face expression recognition is particularly important. However, feature extraction and classification methods of lightweight convolutional neural network expression recognition algorithms mostly used at present are not specifically and fully optimized for the characteristics of facial expression images, yet fail to make full use of the feature information in face expression images. To address the lack of facial expression recognition models that are both lightweight and effectively optimized for expression-specific feature extraction, this study proposes a novel network design tailored to the characteristics of facial expressions. In this paper, we refer to the backbone architecture of MobileNet V2 network, and redesign LightExNet, a lightweight convolutional neural network based on the fusion of deep and shallow layers, attention mechanism, and joint loss function, according to the characteristics of the facial expression features. In the network architecture of LightExNet, firstly, deep and shallow features are fused in order to fully extract the shallow features in the original image, reduce the loss of information, alleviate the problem of gradient disappearance when the number of convolutional layers increases, and achieve the effect of multi-scale feature fusion. The MobileNet V2 architecture has also been streamlined to seamlessly integrate deep and shallow networks. Secondly, by combining the own characteristics of face expression features, a new channel and spatial attention mechanism is proposed to obtain the feature information of different expression regions as much as possible for encoding. Thus improve the accuracy of expression recognition effectively. Finally, the improved center loss function is superimposed to further improve the accuracy of face expression classification results, and corresponding measures are taken to significantly reduce the computational volume of the joint loss function. In this paper, LightExNet is tested on the three mainstream face expression datasets: Fer2013, CK+ and RAF-DB, respectively, and the experimental results show that LightExNet has 3.27 M Parameters and 298.27 M Flops, and the accuracy on the three datasets is 69.17%, 97.37%, and 85.97%, respectively. The comprehensive performance of LightExNet is better than the current mainstream lightweight expression recognition algorithms such as MobileNet V2, IE-DBN, Self-Cure Net, Improved MobileViT, MFN, Ada-CM, Parallel CNN(Convolutional Neural Network), etc. Experimental results confirm that LightExNet effectively improves recognition accuracy and computational efficiency while reducing energy consumption and enhancing deployment flexibility. These advantages underscore its strong potential for real-world applications in lightweight facial expression recognition. Full article

Precisely recognizing diseases in tomato leaves plays a crucial role in enhancing the health, productivity, and quality of tomato crops. However, disease identification methods that rely on single-mode information often face the problems of insufficient accuracy and weak generalization ability. Therefore, this paper proposes a tomato leaf disease recognition framework FCMNet based on multimodal fusion, which combines tomato leaf disease image and text description to enhance the ability to capture disease characteristics. In this paper, the Fourier-guided Attention Mechanism (FGAM) is designed, which systematically embeds the Fourier frequency-domain information into the spatial-channel attention structure for the first time, enhances the stability and noise resistance of feature expression through spectral transform, and realizes more accurate lesion location by means of multi-scale fusion of local and global features. In order to realize the deep semantic interaction between image and text modality, a Cross Vision–Language Alignment module (CVLA) is further proposed. This module generates visual representations compatible with Bert embeddings by utilizing block segmentation and feature mapping techniques. Additionally, it incorporates a probability-based weighting mechanism to achieve enhanced multimodal fusion, significantly strengthening the model’s comprehension of semantic relationships across different modalities. Furthermore, to enhance both training efficiency and parameter optimization capabilities of the model, we introduce a Multi-strategy Improved Coati Optimization Algorithm (MSCOA). This algorithm integrates Good Point Set initialization with a Golden Sine search strategy, thereby boosting global exploration, accelerating convergence, and effectively preventing entrapment in local optima. Consequently, it exhibits robust adaptability and stable performance within high-dimensional search spaces. The experimental results show that the FCMNet model has increased the accuracy and precision by 2.61% and 2.85%, respectively, compared with the baseline model on the self-built dataset of tomato leaf diseases, and the recall and F1 score have increased by 3.03% and 3.06%, respectively, which is significantly superior to the existing methods. This research provides a new solution for the identification of tomato leaf diseases and has broad potential for agricultural applications. Full article

The development of deep coalbed methane (buried depth > 2000 m) in the Yan’an block of Ordos Basin is limited by low permeability, the pore structure of the coal reservoir, and the gas–water occurrence relationship. It is urgent to clarify the key control mechanism of pore structure on gas migration. In this study, based on high-pressure mercury intrusion (pore size > 50 nm), low-temperature N₂/CO₂ adsorption (0.38–50 nm), low-field nuclear magnetic resonance technology, fractal theory and Pearson correlation coefficient analysis, quantitative characterization of multi-scale pore–fluid system was carried out. The results show that the multi-scale pore network in the study area jointly regulates the occurrence and migration process of deep coalbed methane in Yan’an through the ternary hierarchical gas control mechanism of ‘micropore adsorption dominant, mesopore diffusion connection and macroporous seepage bottleneck’. The fractal dimensions of micropores and seepage are between 2.17–2.29 and 2.46–2.58, respectively. The shape of micropores is relatively regular, the complexity of micropore structure is low, and the confined space is mainly slit-like or ink bottle-like. The pore-throat network structure is relatively homogeneous, the difference in pore throat size is reduced, and the seepage pore shape is simple. The bimodal structure of low-field nuclear magnetic resonance shows that the bound fluid is related to the development of micropores, and the fluid mobility mainly depends on the seepage pores. Pearson’s correlation coefficient showed that the specific surface area of micropores was strongly positively correlated with methane adsorption capacity, and the nanoscale pore-size dominated gas occurrence through van der Waals force physical adsorption. The specific surface area of mesopores is significantly positively correlated with the tortuosity. The roughness and branch structure of the inner surface of the channel lead to the extension of the migration path and the inhibition of methane diffusion efficiency. Seepage porosity is linearly correlated with gas permeability, and the scale of connected seepage pores dominates the seepage capacity of reservoirs. This study reveals the pore structure and ternary grading synergistic gas control mechanism of deep coal reservoirs in the Yan’an Block, which provides a theoretical basis for the development of deep coalbed methane. Full article

Forest mortality (FM) is influenced by several independent factors, including forest fragmentation (FF) at different spatial scales and multi-scales, site conditions, and stand characteristics. The aim of this study was to investigate the relationship and effect of FF at various spatial scales on the probability of Scots pine FM. The presented study also analyzed the relationship of the multi-scale fragmentation index effect on forest dieback. The relationship between multiple stressors emphasizes the distinct role of FF in influencing pine FM probability. Data on forest cover, deadwood volume of Scots pine forest, and environmental variables were obtained from the Forest Information System for Europe, the Polish National Forest Inventory, and existing databases, respectively. A generalized additive model approach was used to develop models. The results showed that, at small (50–600 m), large (800–3000 m), and multi spatial scales, the FF effect on Scots pine FM probabilities was statistically significant. There is a partial effect of multi-scale fragmentation on the probability of Scots pine FM, given a holistic view of the fragmentation effect that captures both small and large-scale effects. The study concludes that to calculate FF for a particular area, analyzing different scales and capturing multi-scale level fragmentation indices is crucial to studying the cumulative effect of fragmentation on the probability of Scots pine FM. Full article

Solving stiff partial differential equations with neural networks remains challenging due to the presence of multiple time scales and numerical instabilities that arise during training. This paper addresses these limitations by embedding the mathematical structure of implicit–explicit time integration schemes directly into neural network architectures. The proposed approach preserves the operator splitting decomposition that separates stiff linear terms from non-stiff nonlinear terms, inheriting the stability properties established for these numerical methods. We evaluate the methodology on Allen–Cahn equation dynamics, where interface evolution exhibits the multi-scale behavior characteristic of stiff systems. The structure-preserving architecture achieves improvements in solution accuracy and long-term stability compared to conventional physics-informed approaches, while maintaining proper energy dissipation throughout the evolution. Full article

As critical components of planetary gearboxes, gears directly affect mechanical system reliability when faults occur. Traditional feature extraction methods exhibit limitations in accurately identifying fault characteristics and achieving satisfactory diagnostic accuracy. This research is concerned with the gear of the planetary gearbox and proposes a new approach termed multi-scale wavelet packet energy entropy (MSWPEE) for extracting gear fault features. The signal is split into sub-signals at three different scale factors. Following decomposition and reconstruction using the wavelet packet algorithm, the wavelet packet energy entropy for each node is computed under different operating conditions. A feature vector is formed by combining the wavelet packet energy entropy at different scale factors. Furthermore, this study proposes a method combining multi-scale wavelet packet energy entropy with extreme learning machine (MSWPEE-ELM). The experimental findings validate the precision of this approach in extracting features and diagnosing faults in sun gears with varying degrees of tooth breakage severity. Full article

Nodular Cast Iron (NCI, also known as ductile iron) is widely used in important components such as crankshafts for automotive engines and internal combustion engines, as well as storage and transportation containers for spent fuel in nuclear power plants, due to its good comprehensive mechanical properties such as strength, toughness, and wear resistance. The effect of temperature on the fracture behavior of NCI was investigated using compact tensile (CT) specimens at different temperatures. The results showed that the conditional fracture toughness parameter (K_Q) of the NCI specimens firstly increased and then decreased with decreasing temperature. The crack tip opening displacement δ_m shows a significant ductile–brittle transition behavior with the decreasing of temperature. δ_m remains constant in the upper plateau region but sharply decreases in the ductile–brittle region (−60 °C to −100 °C) and stabilizes at a smaller value in the lower plateau region. Multiscale fractographic analysis indicated that the fracture mechanism changed from ductile fracture (above −60 °C) to ductile–brittle mixed (−60 °C to −100 °C) and then to completely brittle fracture (below −100 °C). As the temperature decreased, the fracture characteristics changed from ductile dimples to dimple and cleavage mixed and then to brittle cleavage. Full article

With the intensification of urbanization, resulting in the growing building stock, building operations have become the main contributors to greenhouse gas emissions. However, the relationship between urban form and carbon emissions remains unclear, which limits the sustainable development of cities. This study reviews the definition of carbon sources, data characteristics, and evaluation methods of carbon emissions. In addition, the impact of urban form on building carbon emissions at the macro, meso, and micro scales is reviewed, and low-carbon design strategies for urban form are discussed. Finally, the existing problems in this field are pointed out, and future research directions are proposed. Our review found that small and medium-sized compact cities tend to have less carbon emissions, while large cities and megacities with compact urban forms have more carbon emissions. The carbon reduction design of urban form at the meso scale is often achieved by improving the microclimate. Developing a research framework for the impact mechanism of building carbon emissions in a coordinated manner with multi-scale urban forms can effectively promote the development of low-carbon sustainable cities. This review can assist urban planners and energy policymakers in selecting appropriate methods to formulate and implement low-carbon city analysis and planning projects based on limited available resources. Full article

Superhydrophobic coatings possess distinct wettability characteristics and hold significant potential in metal corrosion protection and underwater drag reduction. However, their practical application is often hindered by poor durability arising from the fragility of their micro/nanostructured surface roughness. In this study, a durable superhydrophobic coating featuring a hierarchical, hydrangea-like micro/nanostructure was successfully fabricated on an aluminum alloy substrate via a simple one-step cold-spraying technique. The coating consisted of hydrangea-shaped SiO₂ nanoparticles modified with 1H,1H,2H,2H-perfluorodecyltrimethoxysilane (PFDT) to produce multiscale roughness, while epoxy resin (EP) served as the binding matrix to enhance mechanical integrity. The hydrangea-like SiO₂ nanostructures were characterized by solid cores and wrinkled, petal-like outgrowths. This unique morphology not only increased the surface roughness but also provided more active sites for air entrapment, thereby enhancing the coating’s overall performance. The h-SiO₂@PFDT-EP composite coating exhibited excellent superhydrophobicity, with a WCA of 170.1° ± 0.8° and a SA of 2.7° ± 0.5°. Durability was evaluated through sandpaper abrasion, tape peeling, acid and alkali immersion, artificial weathering, and salt spray tests. The results demonstrated that the coating retained stable superhydrophobic performance under various environmental stresses. Compared with bare 6061 aluminum and EP coatings, its corrosion current density was reduced by four and three orders of magnitude, respectively. Furthermore, the coating achieved a maximum drag-reduction rate of 31.01% within a velocity range of 1.31–7.86 m/s. The coating also displayed excellent self-cleaning properties. Owing to its outstanding durability, corrosion resistance, and drag-reducing capability, this one-step fabricated superhydrophobic coating showed great promise for applications in marine engineering and defense. Full article