Search Results (17,075)

This paper proposes an accelerated Riemannian conjugate gradient method based on the Barzilai-Borwein (BB) technique, termed ABBSRCG, for unconstrained optimization on Riemannian manifolds. Building upon classical Riemannian conjugate gradient frameworks, the method enhances step-size selection through a Wolfe-condition-informed strategy and incorporates a dynamic mechanism that adaptively adjusts the computed step length. The resulting algorithm achieves both high efficiency and numerical stability. Compared to conventional approaches such as the Fletcher-Reeves (FR)- type Riemannian conjugate gradient method, the Dai-Yuan (DY)- type Riemannian conjugate gradient method, ABBSRCG maintains the sufficient descent property regardless of whether a line search is used or not. Under mild assumptions, we establish the global convergence of ABBSRCG for u-strongly geodesically convex functions on Riemannian manifolds. Experiments on sphere and oblique manifolds show that ABBSRCG requires fewer iterations and achieves higher computational efficiency than existing Riemannian conjugate gradient methods, confirming its efficiency and reliability for large-scale Riemannian optimization problems. Full article

Amid the global transition to carbon neutrality, tidal current energy has become a strategic sustainable energy resource due to its high predictability, power density, and environmental compatibility. Horizontal-axis turbines show great potential for marine energy harvesting, yet the large-scale commercialization of tidal turbines is severely hindered by complex wake dynamics and the lack of reliable, efficient prediction tools for out-of-distribution (OOD) operating conditions. Traditional high-fidelity CFD methods are computationally prohibitive for engineering optimization, while conventional data-driven surrogate models suffer from poor extrapolation performance, extrapolation collapse near training parameter boundaries, and the absence of uncertainty quantification. To address these bottlenecks, this study focuses on the OOD extrapolation of wake flow prediction across tip speed ratio (TSR) distributions for a single horizontal-axis tidal turbine. A CFD-generated spatiotemporal benchmark dataset is constructed for comparative OOD evaluation across various TSR conditions with 9504 total samples. A novel physics-constrained Fourier neural operator framework named TSR-FNO is proposed to improve OOD generalization. The model integrates TSR–Lipschitz regularization to suppress extrapolation collapse and Monte Carlo Dropout to provide reliable uncertainty estimation. Extensive experiments demonstrate that the proposed method effectively reduces prediction error in unseen TSR regimes, mitigates performance degradation in far-field extrapolation, and produces well-calibrated uncertainty estimates consistent with actual prediction confidence. This work provides a data-driven surrogate modeling strategy for fast and reliable wake prediction on a common CFD-generated benchmark, supporting the efficient design, array layout optimization, and engineering deployment of tidal current energy systems. Full article

Accurate vegetable disease detection in complex cultivation environments remains challenging because early lesions are often small, low-contrast, and easily confounded by cluttered backgrounds. To address this issue, we propose VDD-Net, a feature-enhanced detection network based on YOLOv10 for robust vegetable disease detection in protected agriculture. The proposed framework integrates three modules: a receptive field enhancement (RFE) module to improve local perception of small lesions, an adaptive channel fusion (ACF) module to strengthen multi-scale feature aggregation and suppress background interference, and a global context attention (GCA) module to capture long-range dependencies and improve contextual discrimination. Experiments on a custom vegetable disease dataset showed that VDD-Net achieved an mAP@0.5 of 95.2% with only 7.78 M parameters. To further evaluate robustness, zero-shot cross-domain testing was conducted on the PlantDoc dataset, where VDD-Net achieved an mAP@0.5 of 76.5%, outperforming the baseline and showing improved generalization to natural scenes. In addition, after TensorRT optimization and FP16 quantization, the model maintained real-time inference on edge platforms, reaching 89.3 FPS on Jetson AGX Orin and 24.2 FPS on Jetson Nano. These results indicate that VDD-Net provides a practical balance among detection accuracy, cross-domain robustness, and deployment efficiency for intelligent disease monitoring in modern agriculture. Full article

Automated Storage and Retrieval Systems (AS/RSs) are pivotal hubs in modern intelligent logistics, yet their operational efficiency is often constrained by the complex coupling of storage allocation, equipment scheduling, and path planning. This study proposes a systematic optimization framework to address these three critical control challenges. First, a multi-objective mathematical model for storage location allocation is established, considering efficiency, stability, and correlation. To solve this high-dimensional discrete problem, a Tabu Variable Neighborhood Search (TVNS) algorithm is proposed, integrating short-term memory mechanisms with multi-structure exploration to prevent premature convergence. Second, regarding stacker crane and forklift collaborative scheduling, a Pheromone-guided Artificial Hummingbird Algorithm (PT-AHA) is introduced. By incorporating pheromone feedback into foraging behavior, the algorithm significantly enhances global search capability to minimize total task completion time. Third, stacker crane path planning is modeled as a constrained Traveling Salesman Problem (TSP) and solved using a hybrid Simulated Annealing-Whale Optimization Algorithm (SA-WOA). Quantitative simulation results demonstrate that the TVNS algorithm improves storage allocation fitness by 1.1% over standard Genetic Algorithms, while the PT-AHA reduces task completion time (Makespan) by 21.9% for small-scale batches and consistently outperforms ACO by up to 3.6% in large-scale operations. Validation through an Intelligent Warehouse Management System (WMS) confirms that the integrated framework maintains high industrial resilience by triggering fault alarms and initiating recovery within 3.2 s during simulated equipment failures, providing a robust solution for enterprise-level deployments. Full article

Accurate 3D documentation of large and complex structures is essential for long-term stability assessment, structural monitoring, and conservation planning, particularly for heritage sites exposed to environmental and anthropogenic threats. This study develops an integrated workflow combining Terrestrial Laser Scanning (TLS), Global Navigation Satellite System (GNSS), and Total Station geodetic control for large-scale, high-precision documentation. The approach was implemented at the Saqqara archaeological zone, a UNESCO World Heritage Site facing significant deterioration risks, to document four major pyramids: Djoser, Unas, Teti, and Userkaf. More than 2.1 billion georeferenced points were acquired from 16 scan positions with sub-centimeter registration errors and overall geometric accuracy better than ±1 cm. From these datasets, detailed mesh models, orthoimages, Digital Elevation Models (DEMs), contour maps, and 2D plans were derived. These enabled quantitative analyses of height loss and volumetric change, indicating severe structural degradation in Unas (~53%), Teti (~66%), and Userkaf (~63%), as well as localized deformations such as 4.2 cm displacement at Teti’s south flank. The degradation results from environmental factors and anthropogenic influences. Beyond this case study, the workflow proves that integrated TLS documentation can be applied to large and complex structures, supporting deformation monitoring, stability assessment, and digital twin development. Full article

(1) Background: Curve data compression plays a critical role in efficient storage, transmission, and multi-scale visualization of vector spatial data, especially for complex geographic boundaries. Achieving high compression efficiency while preserving geometric fidelity remains a challenging task. (2) Methods: This study proposes a vector curve compression framework based on a convolutional autoencoder. Curve data are segmented and resampled to unify network input, after which coordinate-difference sequences are encoded into low-dimensional latent vectors through convolutional layers and reconstructed via a symmetric decoder. (3) Results: Experiments conducted on a global island boundary dataset demonstrate that the proposed method achieves effective data reduction with stable reconstruction accuracy. Specifically, compared with the classical Douglas–Peucker (DP) algorithm, Fourier series (FS) methods, and fully connected autoencoders (FCAs), the 1D CAE exhibits superior and more robust reconstruction performance, especially under high compression ratios. It achieves the lowest positional deviation (PD = 42.41) and the highest spatial fidelity (IoU = 0.9991, with a relative area error of only 0.0067%), while maintaining high computational efficiency (57.32 s). Sensitivity analyses reveal that a convolution kernel size of 1 × 7 and a segment length of 25 km yield the optimal trade-off between representational capacity and model stability. (4) Conclusions: The proposed method enables efficient vector curve compression and reliable coastline reconstruction, and is particularly suitable for small- and medium-scale cartographic applications up to a map scale of 1:250 K. Full article

Identifying small-scale burn scars is critical for global carbon accounting, yet remains computationally challenging due to spectral complexity and ground truth scarcity in heterogeneous landscapes. Conventional deep learning models often fail to generalize in such environments, lacking both domain-specific priors and representative training distributions required for precise segmentation. Here, we show that optimizing the fine-tuning of the Prithvi Earth Foundation Model (EFM) via Multidimensional Latin Hypercube Sampling (LHS) establishes a robust framework for this task. Our comparative analysis reveals that the domain-adapted Prithvi model achieves a Mean Intersection over Union (mIoU) of 0.91, outperforming standard Vision Transformers (ViT) by 31.9% and significantly surpassing reconstruction-based architectures, such as Scale-MAE. We demonstrate that LHS is superior to Simple Random Sampling (SRS) for optimizing foundation models, as it ensures statistical fidelity with a Kolmogorov–Smirnov (KS) statistic below 0.1 and effectively captures the tail distributions of fire weather indices. Furthermore, our framework exhibited exceptional data efficiency, retaining 94.5% of its peak accuracy with only 100 training samples. These findings provide a scalable solution for monitoring small-scale disasters in data-constrained regions and validate the synergy between rigorous sampling strategies and EFMs. Full article

Objective: Clinical remission is a major therapeutic goal in systemic lupus erythematosus (SLE) because of its association with improved long-term outcomes. However, its relationship with patient-reported burden, quality of life, and disease perception remains incompletely understood. This study aimed to evaluate patient-reported outcomes (PROs) in patients with SLE in clinical remission, identify factors associated with impaired health-related quality of life (HRQoL), and assess physician–patient discordance in disease activity perception. Methods: A total of 106 adult patients with SLE in clinical remission according to the definition proposed by Zen et al. were enrolled at a single rheumatology center. Patients were classified into complete remission, clinical remission off corticosteroids, or clinical remission on corticosteroids. Demographic, clinical, and treatment-related data were collected, including organ damage (SLICC-SDI) and disease activity (SLEDAI-2K). Patients completed PRO measures including SF-36, Global Health (GH), pain VAS, STAI-Y1 and STAI-Y2, Zung Depression Scale, Insomnia Severity Index, and HAQ. Disease activity was assessed by both the patient (PGA) and the physician (PhGA); a PGA–PhGA difference >25 mm was considered clinically relevant discordance. Results: Among patients in clinical remission, mild anxiety was observed in 17.1% according to STAI-Y1 and in 27.9% according to STAI-Y2, mild-to-moderate depressive symptoms in 47.1%, and mild insomnia in 25.5%. Of the 106 patients, 24 (22.6%) were in complete remission, 27 (25.5%) in clinical remission off corticosteroids, and 55 (51.9%) in clinical remission on corticosteroids. Patients in clinical remission on corticosteroids showed worse patient-reported outcomes than those in complete remission or clinical remission off corticosteroids. In multivariable analyses, poorer physical HRQoL was independently associated with functional disability, pain intensity, and depressive symptoms, whereas poorer mental HRQoL was independently associated with trait and state anxiety. Clinically relevant physician–patient discordance was observed in 22.6% of the cohort and was almost exclusively driven by higher patient than physician scores. Pain intensity emerged as the most robust independent correlate of discordance. Conclusions: A substantial patient-reported burden may persist in patients with SLE despite clinical remission. Pain, psychological distress, insomnia, and functional disability contribute to impaired HRQoL, while physician–patient discordance appears to reflect a broader mismatch between inflammatory disease control and the patient’s lived experience of illness. These findings support a more comprehensive and patient-centered approach to remission assessment in SLE. Full article

Hyperspectral image (HSI) classification requires the extraction of discriminative features from high-dimensional spatial–spectral data. While the Mamba architecture has shown promise in long-sequence modeling with linear complexity, its application to HSI remains constrained by two major hurdles: the unidirectional causal scanning which fails to capture non-causal global dependencies, and the serialization-induced loss of two-dimensional spatial topology and local textures. To overcome these limitations, we propose HAMamba, a novel Hybrid Attention State Space Model. HAMamba facilitates deep representation learning through two core components: a Multi-Scale Dynamic Fusion (MSDF) module and a Hybrid Attention Mamba Encoder (HAME). Specifically, the MSDF module augments spatial perception through parallelized feature extraction and dynamically weighted integration. The HAME synergizes a Bidirectional Sequence Scan Mamba (BSSM) to establish global semantic context and a Spatial–Spectral Gated Attention (SSGA) module to refine local structural details. Comprehensive experiments on four public benchmark datasets demonstrate that the proposed HAMamba significantly outperforms state-of-the-art approaches, achieving a superior balance between classification accuracy and computational efficiency. Full article

In response to the significant challenges posed by ice accumulation and contamination from various fluids in complex operating conditions for metallic materials, this study utilises picosecond laser precision machining to develop a ‘slippery surface’ featuring a micro-channel network structure. The core innovation of this study lies in the use of laser-machined micrometre-scale array textures to overcome the limitations of traditional isolated pores. These globally interconnected micro-channels serve as highly efficient reservoirs and dynamic transport channels for lubricants, significantly enhancing the interfacial capillary locking force of the lubricant. Experimental results demonstrate that this unique network geometry endows the surface with exceptional fluid replenishment and self-healing properties, enabling it to exhibit outstanding broad-spectrum hydrophobicity towards various fluids—including water, crude oil and ethanol (surface tension range: 17.9–72.0 mN m⁻¹)—with sliding angles consistently below 12°, whilst effectively slowing the dehydration and solidification processes of biological fluids. At a low temperature of −15 °C, the surface achieved an ice formation delay of up to 286 s, with an ice adhesion strength of only 33.9 kPa, ensuring that accumulated ice could be spontaneously detached under minimal external force. Furthermore, the micro-channel network structure serves as a key protective mechanism against mechanical wear, maintaining robust slippery properties even after three hours of high-pressure water jet scouring (Weber number of 300). This reliable interface, achieved through structural management, provides an efficient and scalable platform for addressing the all-weather anti-icing and antifouling requirements of outdoor infrastructure. Full article

Agricultural robots and unmanned farmland management require real-time and precise parsing of crop leaves at the edge to support variable application of pesticides, seedling condition monitoring, and phenotypic analysis. However, the field environment features drastic changes in light, leaf occlusion, and interference from background weeds, which can cause semantic fragmentation and boundary artifacts in lightweight models. This paper presents ES2-LeafSeg, a lightweight framework for leaf semantic segmentation tailored for edge deployment. The method employs EfficientNetV2 as the backbone encoder and introduces the State Space Semantic Enhancement Module (S2FEM) on skip connection features, modeling long-range dependencies and suppressing local texture noise through SSM pooling in row and column directions. Meanwhile, a cross-scale decoder (CSD) and a global context transformation (GCT) are designed to achieve multi-scale semantic fusion and boundary refinement. On the three-class segmentation task of the SoyCotton dataset, ES2-LeafSeg achieved mIoU of 0.817, mDice of 0.869, Fβw of 0.925, and MAE of 0.011, outperforming multiple classic and recent baselines while maintaining 23.67 M parameters and 49.62 FPS. Ablation experiments further verified the complementary contributions of S2FEM and GCT to regional consistency and boundary quality. Full article

Interpretability in corporate financial risk profiling must support not only predictive performance but also governance-oriented decision-making. This study proposes a three-class financial risk assessment workflow for B2B settings and introduces Risk-Weighted Uncertainty-Conditioned Feature Importance (RW-UCFI) as a post-explanation prioritization framework. RW-UCFI is not a new attribution method; rather, it reorganizes existing explanation outputs according to class sensitivity, predictive uncertainty, and asymmetric risk relevance. The empirical analysis uses a single cross-sectional dataset of 954 Indonesia Stock Exchange-listed firms with organizationally provided Low Risk, Medium Risk, and High Risk labels. A stacked ensemble model is used as the explanatory substrate, followed by calibration analysis, uncertainty analysis, and governance-oriented explainability aggregation. On the held-out validation set, the model achieved an accuracy of 0.7487 and a macro ROC-AUC of 0.8630. Repeated stratified validation indicated moderately stable aggregate performance, although class-level reliability remained uneven, with High Risk recall emerging as the weakest and most variable component. The original model showed the most favorable probability reliability among the evaluated variants, whereas temperature scaling and one-vs-rest isotonic regression did not improve calibration. Uncertainty analysis further showed that the most uncertain cases concentrated substantially more misclassifications and High Risk misses; the top 30% most uncertain cases captured 52.1% of all errors and 43.8% of High Risk misses. RW-UCFI produced a materially different feature-priority structure from standard global SHAP ranking, suggesting that explanation outputs may become more decision-relevant for governance-oriented review when contextualized by uncertainty and asymmetric risk conditions in the present setting. Full article

Against the backdrop of the global trends toward lightweighting, multi-functionalization, and greening of materials, polypropylene (PP) has been extensively applied owing to its advantages of low density and low cost. However, its inferior foaming performance fails to meet high-end application requirements, which is primarily attributed to its low melt strength and restricted crystallization behavior. In this paper, the five-dimensional selection mechanism and classification of components for PP micro/nanocomposites fabricated via supercritical foaming are systematically summarized. The regulatory effects of micro/nano additives on the crystallization, rheological properties, and foaming behavior of PP are quantitatively analyzed. The parameter optimization windows of three foaming processes, namely batch foaming, extrusion foaming, and injection foaming, are integrated (e.g., a foaming temperature of 150–170 °C and a saturation pressure of 8–20 MPa). Additionally, the application progress of PP micro/nanocomposite foams in fields such as automotive lightweighting (with a weight reduction rate of 64.29%) and building thermal insulation (with a thermal conductivity as low as 29 mW/(m·K)) is outlined. The core novel insight of this work lies in clarifying the unified mechanism of crystal refinement induced by reinforcing agents with different geometric morphologies, which is dominated by the synergy between heterogeneous nucleation and steric hindrance. This finding provides theoretical and technical guidelines for the industrial-scale preparation of high-performance PP foams. Full article

Skeleton-based action recognition is a critical technology for intelligent sports analysis. Although the human skeletal structure exhibits inherent bilateral symmetry, sensor noise on resource-constrained edge devices frequently induces geometric distortion and topological asymmetry. Consequently, achieving a balance between high accuracy and real-time performance remains a significant challenge. To this end, we propose EMS-GCN, an Efficient Multi-scale Shift Graph Convolutional Network that integrates geometric priors. Specifically, we design a Gaussian kernel-driven topology refinement module to mitigate structural noise inherent in sensor data. By leveraging geometric symmetry and Gaussian distances among nodes, this module dynamically constrains graph topology learning, thereby effectively rectifying the structural asymmetry and ambiguity induced by noise. Furthermore, we construct a Multi-scale Shift Linear Attention (MSLA) module to replace computationally intensive temporal convolutions. Leveraging temporal shift invariance, this module captures multi-scale contexts via parameter-free shift operations. Furthermore, we introduce a linear temporal attention mechanism to model global temporal dependencies with linear complexity, effectively resolving the information asymmetry inherent in long-range interactions. Finally, EMS-GCN incorporates a dual-branch attention structure to adaptively calibrate feature responses. Extensive experiments demonstrate that our model maintains high recognition accuracy with only 0.56M parameters, representing a reduction of over 60% compared to mainstream baselines. These results validate the efficacy of leveraging geometric and temporal symmetries to enhance real-time sports analysis. Full article

Stock price prediction is vital for quantitative investment but challenging due to multi-source data complexity, including endogenous, exogenous, and noise components. Standard deep learning models rely on end-to-end modeling of raw market data, failing to disentangle these distinct drivers and hindering prediction accuracy. To address this, we propose MMCAD-Net, a novel model based on time series decomposition. It first decomposes the original stock series into an exogenous cyclical component, endogenous temporal component and residual component, thereby disentangling the mixed temporal patterns. Subsequently, deep feature extraction and information refinement are applied to each component: multi-scale convolutions capture diverse patterns in the cyclical component; multi-level convolutional networks refine local and global features in the temporal component; and an attention mechanism sifts for potentially informative signals within the residuals. Finally, a multi-source feature aggregation mechanism fuses all enhanced information. Experiments on real-world stock market datasets demonstrate that MMCAD-Net surpasses mainstream models in both prediction accuracy and efficiency. Ablation studies further confirm the necessity and effectiveness of each core module. Full article