Search Results (9,797)

Aircraft icing prediction is crucial for aerodynamic design and airworthiness assessment. Traditional physics-based models struggle with complex multi-physical processes, while existing AI methods (function-based characterization or direct image learning) face issues like multi-valued mapping, high data dependency, or lack of physical interpretability. This study proposes a deep learning framework based on point set displacement description, transforming the icing process into airfoil boundary point movements. PCA dimensionality reduction mitigates the curse of dimensionality while retaining physical meaning. A neural network is used to map environmental parameters to low-dimensional principal components. Comparative analysis shows the 64 × 64 network achieves optimal fitting; 2000 samples reproduce complex ice shapes, and 800 low samples characterize simple ones. Balancing efficiency, accuracy, and interpretability with reduced data dependency, this method provides a new approach for rapid engineering icing prediction. Full article

Historical masonry minarets are highly vulnerable to seismic actions due to their slender geometry, limited tensile capacity, and material heterogeneity. However, their response to near-fault ground motions characterized by velocity pulses remains insufficiently explored. This study investigates the seismic response of the historical Tavanlı Mosque Minaret (1894, Trabzon, Türkiye) subjected to pulse-like (PL) and non-pulse-like (NPL) near-fault ground motions. A three-dimensional finite element model (FEM) was developed in ANSYS Workbench and systematically calibrated using empirical formulations to represent the current dynamic condition of the structure. Seismic performance was evaluated through linear dynamic analyses in terms of displacement demands, principal stress distribution, and drift-ratio-based performance levels. The results indicate that model calibration significantly modifies the dynamic characteristics, increasing the fundamental frequency from 0.734 Hz to 1.126 Hz and reducing displacement demands by approximately 35–76% across the considered records. Despite this improvement, PL ground motions consistently generate more critical deformation demands than NPL motions, frequently exceeding Collapse Prevention (CP) limits even when Peak Ground Acceleration (PGA) values are relatively low. A key finding is that seismic demand cannot be reliably predicted by peak intensity measures or pulse-period ratios ($T_p/T_1$) alone; rather, velocity-related parameters and pulse coherence govern the structural response. These results demonstrate that integrating empirical model calibration with pulse-sensitive seismic analysis is essential for reliable seismic assessment and conservation planning of slender historical masonry structures located in near-fault regions. The study offers a systematic framework that integrates model calibration and pulse-sensitive seismic analysis for evaluating the drift-controlled response of slender historical masonry minarets in near-fault regions. Full article

With the development of cross-modal image fusion in multi-sensor systems, current fusion technologies have made significant progress in feature extraction, facilitating more effective image analysis. However, insufficient fusion information may degrade the correlation between the source and fused images, often resulting in the omission of critical features from the original modalities. Therefore, in order to preserve as much information as possible, especially for the complete extraction of effective feature information in source images, this paper proposes a new cross-modal image fusion method based on low-rank representation and convolutional sparse learning named LRCFuse. Firstly, the learned low-rank representation (LLRR) blocks are employed to perform dimensionality reduction on the source images while simultaneously extracting their low-rank and sparse feature components. Nevertheless, considering that the low-rank representation has insufficient modeling ability for different modal images, we introduce common feature preservation module (CFPM) blocks based on convolutional sparse coding. By leveraging the CFPM module, LRCFuse recovers common features from both source images to mitigate the loss caused by the imperfect assumptions of low-rank representation. Based on this, a multi-level optimization strategy incorporating pixel loss, shallow-level loss, mid-level loss, deep-level loss, and sobel loss is proposed to hierarchically learn and refine diverse image features. Quantitative and qualitative evaluations are conducted across various datasets, revealing that LRCFuse can effectively detect targets infrared salient targets, preserve additional details in visible images, and achieve better fusion results for subsequent downstream tasks. Full article

Strengthening key competencies in veterinary preclinical education, such as anatomical identification, spatial–visual reasoning, and anatomical–surgical understanding, is essential for effective preclinical learning. In this context, veterinary preclinical education is undergoing a transformation process in which traditional theoretical–practical approaches show limitations in responding to current educational demands, making it necessary to adopt innovative strategies based on active learning and simulation. This study presents a simulation-based educational approach designed to support competency development within preclinical veterinary education. Using a reproducible and low-cost workflow applied to a real canine cranial case of extra-genital transmissible venereal tumor (TVCT) with frontal bone invasion, used exclusively as a teaching scenario. Fourteen veterinary medicine students from the same institution participated in two instructional conditions: Group A received traditional theoretical instruction (including cadaveric specimens) without the use of 3D-printed models, while Group B participated in simulation-based training supported by virtual planning and a 3D-printed cranial model. Learning outcomes are assessed through structured observation and descriptive analysis. A Likert-type survey was also used to assess satisfaction and engagement among students who participated in the model-supported training, as well as to map competencies across cognitive, visual-spatial, and anatomical-surgical reasoning domains, with evaluation conducted by veterinarians with clinical and teaching experience. Descriptive observations indicated that students participating in the simulation-based training engaged in three-dimensional anatomical exploration of cranial anatomy and case-based anatomical-surgical discussion. In addition, survey responses from Group B indicated high levels of engagement and interest, as well as high perceived usefulness of the model-supported training experience. These findings suggest that simulation-based educational frameworks may offer a safe, transferable, and pedagogically valuable strategy for competency development within preclinical veterinary education. Full article

Deep learning-based flow field prediction for microclimate pollutant dispersion represents an emerging and promising methodology, where effectively integrating meteorological, spatial, and temporal information remains a critical challenge. To address this, we propose a novel non-intrusive reduced-order model (ROM) that synergizes a Dilated Convolutional Autoencoder (DCAE) with pre-trained large language models (LLMs). The DCAE, leveraging nonlinear mapping, was employed for extracting low-dimensional spatiotemporal flow field features. These features were then combined with textual prototypes via text embedding to enable few-shot inference using the LLM-based flow field prediction method. To optimize the utilization of pre-trained LLMs, we designed a specialized textual description template tailored for pollutant dispersion data, which enhances the contextual input of meteorological conditions to guide model predictions. Experimental validation through three-dimensional urban canyon simulations conclusively demonstrated the efficacy of the convolutional autoencoder and LLM-based framework in predicting pollutant dispersion flow fields. The proposed method exhibits remarkable transfer learning capabilities across varying street canyon geometries and meteorological conditions while significantly representing a 9.85× acceleration in prediction compared to Computational Fluid Dynamics (CFD). Full article

Optimizing nitrogen (N) fertilization is essential for sustaining durum wheat yield and grain quality while reducing the environmental impacts associated with N over-application. Hyperspectral sensing provides a rapid and non-destructive approach for monitoring crop N status. However, high-dimensional data, phenology-dependent spectral responses, and spatial autocorrelation in field measurements limit robust nitrogen classification and interpretation. This study evaluated hyperspectral-based nitrogen status classification in durum wheat under Mediterranean field conditions and identified key spectral regions using explainable artificial intelligence. A field experiment was conducted in Southern Italy using ten N fertilization rates (0–180 kg N ha⁻¹). Canopy reflectance was acquired at the booting and heading stages from georeferenced sampling locations. Three nitrogen stratification strategies (binary Low–High, Extreme, and three-level) were evaluated using Random Forest, SVM-RBF, and XGBoost classifiers. Model performance was assessed using spatially independent Leave-One-Plot-Out cross-validation at both the sample and plot levels, with plot-level predictions derived through majority voting. Classification robustness was strongly influenced by the stratification strategy and phenological stage. The binary Low–High stratification achieved the highest sample-level accuracy, with a maximum of 0.78 at booting (SVM-RBF) and 0.75 at heading (SVM-RBF), whereas the Extreme stratification produced intermediate performance, with maximum accuracies of 0.73 at booting (SVM-RBF) and 0.63 at heading (XGBoost). Plot-level aggregation improved performance, reaching up to 0.90 at booting and 1.00 at heading. SHAP analysis highlighted red, red-edge, and near-infrared wavelengths as the dominant contributors, with increased reliance on longer wavelengths at the heading. Overall, explainable machine learning provides a robust framework for hyperspectral nitrogen monitoring in durum wheat. Full article

Increasing the proportion of renewable energy sources, such as photovoltaic power, in the grid can reduce fossil fuel consumption and build a low-carbon power system. However, the inherent instability of the photovoltaic power output makes it difficult to predict, thus increasing the cost of grid operation. Therefore, to improve the accuracy of power prediction and promote the development of the grid, a four-stage short-term photovoltaic power prediction framework, namely, CDT, is proposed, which includes decomposition, classification, reconstruction and forecasting. The initial power data are decomposed using complete ensemble empirical mode decomposition with adaptive noise. Next, an improved data classification and reconstruction method based on dynamic time warping is developed to process the data, which reduces the dimensionality of the data while preserving trend information. Finally, the reconstructed components are predicted using the improved TCN model. The results of the empirical study show that the proposed CDT has higher precision and scalability in processing and predicting the trend of photovoltaic power generation, compared to the other benchmark models. Full article

This paper investigates the finite-horizon survival probability for a system of correlated arithmetic Brownian motions with heterogeneous drifts and volatilities, focusing on the event in which one component remains strictly below all others. Using a whitening transformation of the covariance structure, we reduce the problem to the survival of a standard Brownian motion in a simplicial cone, characterized by its spherical cross-section. While explicit solutions are available in low dimensions, we address the computationally challenging tetrahedral angular case. We derive a semi-analytic formula for the survival probability via an eigenfunction expansion of the Dirichlet Laplace–Beltrami operator on this curved domain. For efficient implementation, we construct a diffeomorphism from the spherical tetrahedron to a fixed Euclidean tetrahedron, enabling the computation of angular eigenpairs through a stable finite-element scheme. For higher-dimensional regimes, we also introduce a covariance-based difficulty index and geometric bounds based on an inscribed spherical cap to assess spectral convergence and estimate long-time decay rates. Numerical experiments show that this offline–online approach achieves high accuracy and substantial speedups relative to Monte Carlo benchmarks. Full article

Large-scale backward-curved centrifugal fans without volutes are extensively employed in enclosed air-cooled electric motors owing to their exceptional heat dissipation performance. This category of fans features substantial blade dimensions and a multitude of optimization parameters, which introduce challenges such as diminished predictive accuracy in high-dimensional optimization spaces. To address these issues, this paper proposes a blade optimization design methodology based on a GA-Kriging surrogate model. Sobol’s global sensitivity analysis is first employed to reduce model dimensionality. Subsequently, a high-fidelity aerodynamic performance prediction model is constructed through the integration of a Genetic Algorithm (GA) and a Kriging model. A constrained optimization is then conducted with volumetric flow rate and static pressure as the design objectives, and shaft power along with geometric point coordinates as the constraints. Experimental test results demonstrate that the fan optimized via the surrogate model, while maintaining low prediction error, achieves a 14% increase in volumetric flow rate and a 20% improvement in static pressure. This outcome indicates a significant enhancement in the overall aerodynamic performance. Full article

The borehole-to-surface electromagnetic (BSEM) method is widely employed in oil and gas exploration and downhole monitoring. However, the strength of the ground observation signals of the BSEM method is affected by the metal steel casing in the well. To investigate the response characteristics of the BSEM method under metal casing conditions, this study performed three-dimensional BSEM forward modeling based on a cylindrical grid. The finite volume method was adopted to discretize and solve the governing equations of the electromagnetic field, and the cylindrical grid was partitioned in accordance with the axisymmetric geometric features of the wellbore-casing system, thereby achieving high-precision adaptation to the well structure. To explore the impact of metal casing in an alternating electromagnetic field, four typical models were established: a linear source, a long metal wire, a metal casing, and a casing with a cement sheath. The characteristics of ground signals under low-frequency alternating emission conditions were systematically studied. By comparing the simulation results with the 1D analytical solution, this method was verified to have high numerical accuracy, which can accurately reflect the responses of a metal casing and multiple media interfaces to the alternating electromagnetic field. Based on comparative analysis, the differences in underground electromagnetic field distributions among different source models and their applicable ranges were clarified, and the applicable scenarios and effective detection depths of different models in actual monitoring were explored. This research provides numerical simulation cases to investigate the role of metal casings in BSEM observations, and also lays a theoretical foundation for the interpretation of downhole electromagnetic data, which is of positive significance for improving the effect of applying BSEM technology in oil and gas exploration. Full article

As a unique cultural heritage of Chinese ethnic minorities, Dong architecture provides rich historical and cultural information. Rapid and accurate extraction of ethnic building information from remote sensing images in complex terrain and high-density settlement environments is highly important for the protection of architectural heritage and the management of rural space. Huanggang Dong Village in Liping County, Guizhou Province, China, is taken as a case study. This paper develops a multifeature fusion machine learning framework for the automatic recognition of Dong ethnic architecture based on centimeter-level visible images captured by UAV. First, the vegetation index, HSI color features and texture features based on the gray level co-occurrence matrix are extracted from the UAV visible light orthophoto image. Through the random forest feature importance ranking and correlation test, six key features, namely, the VDVI, HSI-S, HSI-I, mean, variance and contrast, are selected to construct a multifeature space. This step constitutes the feature construction stage of the proposed methodology and provides the basis for subsequent classification. Second, on the basis of a support vector machine (SVM) and random forest (RF), classification models are constructed. The effects of different feature combinations and different algorithms on classification accuracy are systematically compared, and the results are evaluated in terms of overall accuracy (OA), the kappa coefficient, user accuracy (UA) and producer accuracy (PA). This second part highlights the classification phase of the methodology, which tests the feature space using different algorithms and evaluates the performance of the models. The experimental data fully show that under the condition of a single feature, the SVM model dominated by texture features performs best, with an OA of 85.33% and a kappa of 0.799; under the condition of multifeature fusion, the RF algorithm has a stronger ability to integrate multisource features. The accuracy of building category recognition based on the total feature and dimensionality reduction feature space is particularly prominent. The total feature and overall accuracy reach 89.00%, and the kappa coefficient is 0.850. The UA and PA reached 89.66% and 94.55%, respectively. Through in-depth comparative analysis, the vegetation index–color–texture multifeature fusion and machine learning classification framework based on UAV visible light images can achieve high-precision extraction of Dong architecture without relying on high-cost sensors. It can effectively alleviate the confusion between water bodies and shadows and between dark roofs and vegetation and effectively separate traditional Dong architecture from roads, vegetation and other elements. It provides a low-cost and feasible way for digital archiving, dynamic monitoring and protection management of the traditional village architectural heritage of ethnic minorities. Full article

With the deepening implementation of the dual-carbon strategy, the penetration rates of distributed power sources and flexible loads in new distribution grids continue to rise, posing significant challenges to system security and stability due to output fluctuations and randomness. To enhance voltage quality and achieve low-carbon economic operation in distribution grids, this paper proposes a multi-objective optimization model for Distributed Energy Storage System allocation. The model integrates power quality, economic benefits, and net carbon emissions. To efficiently solve this high-dimensional nonlinear problem, an improved Multi-Objective Gray Wolf Optimization algorithm is proposed. It employs a chaotic map to initialize the population, enhancing global distribution uniformity. A nonlinear convergence factor is introduced to dynamically balance global exploration and local exploitation. A dynamic grouping collaboration strategy is designed, combining Lévy flight and the elite crossover strategy to enhance search capability and convergence accuracy. Simulations on an IEEE 33-node system show that the improved MOGWO-optimized energy storage scheme reduces average voltage deviation by 37.0%, total operating costs by 7.0%, and net carbon emissions by 4.1%, compared to a no-storage scenario. Compared to the standard MOGWO algorithm, the proposed method achieves further optimization across all objectives, validating its effectiveness and superiority in realizing coordinated energy storage planning that balances safety, economy, and low-carbon goals. Full article

The development of cement composites that simultaneously achieve high compressive strength and low density remains a fundamental scientific challenge, particularly because optimizing weight reduction often compromises mechanical performance under sustained high-pressure conditions. In modern construction—especially high-rise buildings, large-span structures, and underground projects—there is an urgent applied need for lightweight materials that can reduce structural self-weight, enhance seismic resilience, simplify foundation design, and improve construction efficiency without sacrificing load-bearing capacity or long-term durability. To address this dual problem, this study investigates high-pressure-resistant lightweight cement composites incorporating hollow glass microspheres (HGMSs) of three different particle sizes as functional fillers, modified with isobutyl triethoxy silane (IBTES) to strengthen interfacial bonding. Ten formulations with varying HGMS types and dosages (5%, 10%, and 15% by volume) were systematically evaluated through creep tests, uniaxial compression experiments, X-ray diffraction (XRD), and thermogravimetric analysis (TGA). The scientific results demonstrate marked qualitative and quantitative improvements: the optimal formulation (25 μm HGMS at 5% dosage) exhibited a 22.01% reduction in creep deformation and a 67.85% increase in compressive strength compared to plain cement, while bulk density was reduced by 8.8–19.0%. Enhanced hydration was confirmed by a 23.6% reduction in residual Ca(OH)₂ content and a 31.2% increase in chemically bound water, indicating more complete formation of calcium silicate hydrate (C–S–H) gel. Energy evolution analysis revealed a prolonged elastic energy accumulation stage (increasing from 56% to 95% of total compression duration), signifying a transition toward quasi-ductile failure behavior. From an applied perspective, these quantitative enhancements translate directly into practical construction benefits: the 8.8–19.0% density reduction enables lighter structural components, easing transportation and installation; the 67.85% higher compressive strength ensures reliable performance in high-pressure environments; and the 22.01% lower creep deformation guarantees long-term dimensional stability. Collectively, these findings confirm that the HGMS-IBTES-modified composite offers a scalable, high-performance solution for advanced construction applications where both weight reduction and superior pressure resistance are critical. Full article

Gel polymer electrolytes are key components in next-generation energy storage systems, particularly supercapacitors, due to their high ionic conductivity, mechanical robustness, and operational safety. Ionomer-based gels derived from perfluorosulfonic acid (PFSA) are particularly promising, as their nanophase-segregated morphology enables the formation of three-dimensional ionic clusters capable of absorbing and retaining aqueous electrolytes. In this study, the commercial PFSA D50-U (Thasar S.r.l.) membrane was investigated for the first time as a gel-state ionomer electrolyte and separator in symmetric supercapacitors using coconut shell-derived activated carbon (YP-80F Kuraray Co., Ltd.). The effects of cation type on gel swelling, ionic conductivity, and electrochemical performance were investigated using Na₂SO₄ and Li₂SO₄ aqueous electrolytes. The results showed that PFSA D50-U formed stable gel structures, resulting in low internal resistance, high specific capacitance, and excellent long-term cycling stability. These findings demonstrate that PFSA D50-U is a novel proton-exchange gel membrane with strong potential for high-performance symmetric supercapacitors and other gel-based energy storage devices. Full article

Detecting small insulator defects in unmanned aerial vehicle (UAV) imagery remains challenging due to low resolution, complex backgrounds and scale variation, which degrade the performance of existing detectors. This study aims to develop a highly efficient and accurate model for real-time insulator defect inspection on resource-constrained UAV platforms. This paper proposes POLD-YOLO, a novel lightweight object detector based on YOLO11. The key innovations include: (1) A backbone enhanced by a PoolingFormer module and Channel-wise Gated Linear Units (CGLUs) to boost feature extraction efficiency; (2) An Omni-Dimensional Adaptive Downsampling (OD-ADown) module for multi-scale feature extraction with low complexity; (3) A Lightweight Shared Convolutional Detection Head (LSCD-Head) to minimize the number of parameters; (4) A Focaler-MPDIoU loss function to improve bounding box regression. Extensive experiments conducted on a self-built UAV insulator defect dataset show that POLD-YOLO achieves a state-of-the-art mAP@0.5 of 92.4%, outperforming YOLOv5n, YOLOv8n, YOLOv10n, and YOLO11n by 3.6%, 1.6%, 1.4%, and 1.6%, respectively. Notably, it attains this superior accuracy with only 1.55 million parameters and 3.8 GFLOPs. POLD-YOLO establishes a new Pareto front for accuracy-efficiency for onboard defect detection. It demonstrates great potential for automated power line inspection and can be extended to other real-time aerial vision tasks. Full article