Search Results (10,303)

Against the global carbon neutrality target, predictive maintenance (PdM) of automotive engines represents a core technical strategy to advance the sustainable development of the automotive industry. Conventional single-modal diagnostic approaches for engine abnormal sound defects suffer from low accuracy and weak anti-interference capability. Existing multi-modal fusion methods fail to deeply mine the physical coupling between cross-modal features and often entail excessive model complexity, hindering deployment on resource-constrained on-board edge devices. To resolve these limitations, this study proposes a Physical Prior-Embedded Cross-Modal Attention (PPE-CMA) mechanism for lightweight multi-modal fusion diagnosis of engine abnormal sound defects. First, wavelet packet decomposition (WPD) and mel-frequency cepstral coefficients (MFCC) are integrated to extract time-frequency features from engine audio signals, while a channel-pruned ResNet18 is employed to extract spatial features from engine thermal imaging and vibration visualization images. Second, the PPE-CMA module is designed to adaptively assign attention weights to audio and image features by exploiting the physical coupling between engine fault acoustic and visual characteristics, enabling efficient cross-modal feature fusion with redundant information suppression. A rigorous theoretical derivation is provided to link cosine similarity with the physical correlation of engine fault acoustic-visual features, justifying the attention weight constraint (β = 1 − α) from the perspective of fault feature physical coupling. Third, an improved lightweight XGBoost classifier is constructed for fault classification, and a hybrid data augmentation strategy customized for engine multi-modal data is proposed to address the small-sample challenge in industrial applications. Ablation experiments on ResNet18 pruning ratios verify the optimal trade-off between diagnostic performance and computational efficiency, while feature distribution analysis validates the authenticity and effectiveness of the hybrid augmentation strategy. Experimental results on a self-constructed multi-modal dataset show that the proposed method achieves 98.7% diagnostic accuracy and a 98.2% F1-score, retaining 96.5% accuracy under 90 dB high-level environmental noise, with an end-to-end inference speed of 0.8 ms per sample (including preprocessing, feature extraction, and classification). Cross-engine and cross-domain validation on a 2.0T diesel engine small-sample dataset and the open-source SEMFault-2024 dataset yield average accuracies of 94.8% and 95.2%, respectively, demonstrating strong generalization. This method effectively enhances the accuracy and robustness of engine abnormal sound defect diagnosis, offering a lightweight technical solution for on-board real-time fault diagnosis and in-plant online quality inspection. By reducing engine fault-induced energy loss and spare parts waste, it further promotes energy conservation and emission reduction in the automotive industry. Quantified experimental data on fuel efficiency improvement and carbon emission reduction are provided to substantiate the ecological benefits of the proposed framework. Full article

Cancer immunotherapy has transformed the clinical management of several malignancies; however, its efficacy remains limited in tumors with low mutational burden and restricted availability of classical mutation-derived neoantigens. In this context, increasing evidence indicates that the tumor immunopeptidome extends far beyond canonical protein-coding regions, incorporating peptides derived from non-coding transcripts through non-canonical translation mechanisms. Long non-coding RNAs (lncRNAs), traditionally regarded as transcriptional or post-transcriptional regulators, have recently emerged as an unexpected source of small open reading frame-encoded peptides (lncPEPs). A subset of these peptides is processed and presented by major histocompatibility complex class I molecules, generating tumor-specific neoantigens capable of eliciting CD8⁺ T cell responses. Owing to the high tissue and context specificity of lncRNA expression, lncRNA-derived neoantigens offer unique advantages over mutation-based targets, including increased tumor selectivity and potential recurrence across patient subsets. In this review, we synthesize current knowledge on the biogenesis, detection, and immunogenic potential of lncRNA-derived peptides, highlighting experimental and computational strategies for their identification within the cancer immunopeptidome. We discuss the challenges associated with their validation and clinical translation, as well as their relevance for the development of vaccines and adoptive T cell–based therapies. Finally, we illustrate these concepts using epithelial ovarian cancer as a representative model of low-mutational-burden tumors, where lncRNA-derived neoantigens may help overcome current limitations of immunotherapy and enable patient stratification for personalized treatment approaches. Full article

The integration of Geographic Information Systems (GISs) with hydrologic science has evolved over seven decades from manual catchment delineation and output visualization to AI-native spatial water intelligence, reshaping how the water cycle is observed, modeled, and managed. This review explores that evolution, from the progressively tightening coupling between GIS software and hydrologic models to an AI-assisted future in which the line between these two fields blurs and eventually dissolves completely. The evolution of GISs in hydrology is traced through four eras, stratified as: (1) the formalization of governing equations and digital terrain representations (1950–1985); (2) the initial GIS–model coupling era and the rise in watershed simulation (1985–2000); (3) open source and the start of the open data deluge (2000–2015); and (4) machine learning and cloud-native computing (2015–present). A four-level vision for the role of artificial intelligence in the next generation of spatial hydrology is then articulated, from AI-assisted GIS operation to spatially aware AI water intelligence that reasons directly over geospatial data without requiring a traditional GIS or simulation software as an intermediary. Broader limitations and challenges are also discussed. Full article

Transboundary river basins face water stress exacerbated by data scarcity and political instability, and most allocation models require ideal conditions that ordinarily do not exist. This study operationalizes Water Diplomacy Theory (WDT) for data-scarce, conflict-prone basins through quantifiable allocation rules—a critical gap as 310 transboundary basins worldwide face similar challenges. We address: (1) How can a four-stage allocation framework reduce basin-wide water stress under varying Institutional Capacity (IC), Data Transparency (DT), and Stakeholder Inclusion (SI)? (2) What treaty provisions achieve bindingness under upstream-downstream power asymmetries? (3) How does this framework advance beyond existing models in equity, efficiency, and adaptive capacity? We synthesize Water Diplomacy Theory with Hydro-political Security Complex Theory to construct a novel four-stage framework: initial allocation with ecological floors, conditional reallocation triggers, interannual water banking, and satellite-verified compliance. Drawing on 14 treaty precedents and 30-year hydrological data for the Salween River, we embed these rules in an open-source water banking model. Results demonstrate that increasing IC from low to high reduces basin-wide water stress by 34% (±7%, 95% IC) under drought conditions. Stakeholder Inclusion decreases allocation conflicts by 52%. Water banking outperforms priority rules by 23% across climate scenarios. Cooperation becomes self-enforcing when IC exceeds 0.55. The novelty and contribution to existing literature our study provides are: (1) first operationalization of hybrid commons-to-treaty transition with 85.7% empirically grounded clauses; (2) evidence that binding cooperative treaty design is achievable in weak-state contexts through institutional design; and (3) a portable template for data-scarce conflict-affected basins. Full article

In challenging environments, there often exist problems of false alarms and missed detections in real-time kinematic (RTK) ambiguity resolution, which significantly reduce the reliability and availability of position information. To address these issues, a machine-learning method is proposed to conduct a correctness check on RTK ambiguity fixing, aiming to reduce the occurrences of false alarms and missed detections. The inter-system differential RTK model is adopted. Compared with the traditional RTK model, this model can provide an effective feature, namely the differential inter-system biases (DISB), to improve the accuracy of machine-learning classification. This is because when the RTK ambiguity is correctly fixed, the DISB usually appears as a stable constant. In addition to DISB, features that are strongly related to ambiguity fixing, such as the ratio value, DOP value, and residuals, are also comprehensively utilized. This method is verified by an open-source, large-scale, and diverse GNSS/SINS dataset—SmartPNT-POS. The experimental results show that, compared with the traditional method of relying solely on the empirical ratio value for ambiguity fixing verification, the missed detection probability of this method is reduced by 2%, the false-alarm probability is decreased by 29%, and the positioning accuracy is improved by approximately 7%. Moreover, compared with other features, the DISB feature provides the highest contribution rate in the machine-learning classification model. Full article

Building height is a key indicator of vertical urbanization and urban morphological complexity, yet accurately mapping building height at fine spatial resolution and large spatial scales remains challenging. This study proposes an attention-based deep learning framework (ABHNet) for building height estimation at a 10 m spatial resolution by integrating multi-source remote sensing data and socioeconomic information. The model jointly exploits Sentinel-1 synthetic aperture radar data, Sentinel-2 multispectral imagery, and point of interest (POI) data. The proposed framework is evaluated in Shanghai, a megacity with dense and vertically complex urban structures, using Baidu Maps-derived building height data as reference information. The results demonstrate that the proposed method achieves accurate building height estimation, with a root mean squared error (RMSE) of 3.81 m and a mean absolute error (MAE) of 0.96 m for 2023, and an RMSE of 3.30 m and an MAE of 0.78 m for 2019, indicating robust performance across different time periods. Also, this model is applied in two other cities (Changzhou and Guiyang) and the results indicate good performance. In addition, the expandability of the framework is examined by incorporating higher-resolution ZY-3 imagery, for which the spatial resolution was increased to 2.5 m, highlighting the potential extension of the model to heterogeneous data sources. Overall, this study demonstrates the effectiveness of attention-based deep learning and multimodal data fusion for large-scale and fine-resolution building height estimation using open-source data. Full article

Urban wetlands are acoustic hotspots where vegetation structure, hydrological dynamics, and anthropogenic noise interact, yet multi-season assessments of how vegetation influences avian soundscapes are limited. This study explored bird soundscape dynamics across forest, open forest grassland, and meadow habitats in Nanjing Xinjizhou National Wetland Park, eastern China, using passive acoustic monitoring during spring and autumn 2023. Twelve sampling points (four per vegetation type) were established, and six acoustic indices were calculated, including the Acoustic Complexity Index (ACI), Acoustic Diversity Index (ADI), Acoustic Evenness Index (AEI), Bioacoustic Index (BIO), Normalized Difference Soundscape Index (NDSI), and Acoustic Entropy Index (H). were calculated from 48-h recordings each season. Random forest models and redundancy analysis assessed the relationships between acoustic indices, fine-scale vegetation parameters (e.g., crown width, tree height, species richness), and anthropogenic factors (e.g., distance to roads/trails, surface hardness). Vegetation structure, particularly crown width, was the primary driver of avian acoustic diversity, with broad-crowned forests consistently exhibiting the highest acoustic complexity. In spring, anthropogenic factors such as trail and road proximity dominated soundscape variation, suppressing biological sounds. In autumn, with reduced human presence, vegetation structure emerged as the dominant factor, while bioacoustic activity remained elevated despite reduced peaks in acoustic complexity. Proximity to roads increased low-frequency (1–2 kHz) noise and suppressed mid-frequency (4–8 kHz) bird vocalizations, but trees with crown widths ≥4 m maintained higher acoustic diversity even near disturbance sources. This study demonstrates that vegetation structure mediates both resource availability and sound propagation, buffering the effects of anthropogenic disturbance in frequency-specific ways. Multi-season sampling is crucial for understanding the dynamic interplay between vegetation phenology and human activity that shapes urban wetland soundscapes. Full article

Machining feature recognition is a key enabling technology in intelligent manufacturing that extracts manufacturing semantics from the boundary representation (B-Rep) of 3D CAD models to bridge design and process planning. Recent advances in deep learning have accelerated data-driven feature recognition methods. Among these, graph neural networks (GNNs) have gained significant attention due to their natural compatibility with the non-Euclidean, hierarchical topological structure of B-Rep data, enabling efficient and lossless encoding of geometric and topological attributes. However, existing GNN-based methods primarily leverage the topological structure and geometric attributes of B-Rep models, often neglecting the inherent geometric relationships present in the B-Rep data structure. To address this gap, we propose a dual-graph fusion network (Topo-Geom DualGNN) that integrates a topological attribute adjacency graph and a geometric relationship graph. Our approach employs a GatedGCN-based graph encoder and an FiLM-based cross-stream fusion mechanism to jointly encode topological and geometric information from the B-Rep model. Evaluations on open-source synthetic datasets, including MFInstSeg and MFRCAD, demonstrate that the proposed method achieves competitive comprehensive recognition performance and exhibits promising capability in recognizing machining features in complex parts. Full article

This paper presents an open-source ROS 2 navigation and sensor platform for quadruped robots, demonstrated on Boston Dynamics Spot in a laboratory environment. The platform integrates SLAM Toolbox for mapping and localisation, Navigation2 with MPPI and Smac Hybrid-A* for global path planning, and a frontier-based autonomous exploration module with practical handling of unreachable frontiers. The paper validates and verifies current, open-source algorithms deployed on off-the-shelf hardware. A greedy wavefront-based frontier selection method is presented that prioritizes Time-to-Closest-Viable-Frontier (TCVF) by terminating the search as soon as a feasible frontier is identified. On a real robot dataset replayed across five goal scenarios, the method reduces median selection latency from $94.31$ ms to $51.08$ ms (95th percentile: $109.54$ ms to $56.99$ ms), corresponding to a $1.85$-times improvement in compute time compared to a standard implementation. The system also employs Zenoh middleware and Foxglove for remote monitoring and control, enabling flexible, high-bandwidth operation. The platform, including configuration files and launch scripts, is released openly to support future research and deployment on quadruped robots. Full article

The Korteweg–de Vries (KdV) equation is a foundational model in geophysical fluid dynamics (GFD), governing the propagation of long internal and surface gravity waves in stratified and shallow ocean environments where the interplay between nonlinear steepening and frequency-dependent dispersion gives rise to solitons. Although the analytical tractability of the KdV equation through inverse scattering is well established, systematic numerical exploration of multi-soliton interactions remains valuable for benchmarking solvers, probing conservation properties under varied oceanic initial conditions, and building intuition for more complex ocean wave phenomena. This article presents sangkuriang, an open-source Python library that solves the KdV equation using Fourier pseudo-spectral spatial discretization and adaptive eighth-order Runge–Kutta time integration. The implementation leverages just-in-time (JIT) compilation to achieve research-grade computational efficiency on standard hardware, making it readily accessible for coastal and ocean engineering applications, including idealized modeling of internal solitary waves on continental shelves, rapid parameter studies for solitary wave propagation in stratified basins, and pedagogical investigations of nonlinear dispersive wave dynamics. The solver is validated through four progressively complex idealized scenarios motivated by oceanic wave dynamics: isolated soliton propagation, symmetric interactions, overtaking collisions, and three-body interactions. High-fidelity conservation of mass, momentum, and energy is demonstrated, with relative errors remaining below $\mathcal{O}\left({10}^{-4}\right)$ across all test cases. Measured soliton velocities align with theoretical predictions within 5%, confirming the capture of the amplitude-dependent dispersion characteristic of oceanic solitary waves. Complementary diagnostics, including spectral entropy and recurrence quantification analysis (RQA), verify that the numerical solutions preserve the regular phase-space structure characteristic of integrable Hamiltonian systems. These results establish sangkuriang as a robust, lightweight platform for reproducible numerical investigation of idealized nonlinear dispersive wave dynamics relevant to coastal and ocean engineering applications. Full article

Background: Tumor cells can reprogram their metabolism, constituting a hallmark of cancer that plays a crucial role in tumor progression. As tumor cells exhibit an increased demand for nutrients, e.g., amino acids, they rely on extracellular sources and show deregulation of transport proteins. Among these, SNAT1 (SLC38A1) is described as the loader for glutamine that is responsible for the main influx of this amino acid. The aim of this study was to assess the molecular function of SNAT1 in melanoma regarding its role in amino acid transport and regulation of cellular metabolism. Methods: siPool-mediated downregulation of SNAT1 expression in melanoma cell lines was used to investigate the molecular function of this protein. Glutamine transport was assessed by measuring the intracellular and extracellular concentrations of glutamine. Regulation of downstream effectors was evaluated with qRT-PCR and Western Blot. Metabolism was investigated by performing Seahorse flux analysis. Mitochondrial staining was examined via flow cytometry. Protein interaction was assessed with Co-IP, and in silico modeling of protein interaction was performed with AlphaFold3. Results: In this study, we uncovered the new finding that SNAT1 is not primarily implicated in glutamine influx into melanoma cells but in signaling in response to extracellular glutamine. We identified P62 and cMYC as downstream effectors of SNAT1. By activating the P62/cMYC-axis and target genes of cMYC, SNAT1 modulates the metabolism of melanoma cells depending on the glutamine level. SNAT1 and P62 are interaction partners. Conclusions: This finding newly suggests that SNAT1 may function as a sensor or receptor (“transceptor”) for glutamine rather than being a direct and primary glutamine transporter, and could open up new therapeutic options targeting melanoma cells. Full article

It is difficult to capture the realities of the illicit antiquities market because of the lack of accessible, unsiloed data from underground trade networks. Despite existing literature on social network analyses and machine-learning experiments with antiquities data, there is a gap in simple open-source methodologies accessible to the non-academic public. By using a provenance-based analysis, we present a case study of the Italian antiquities trafficking networks that more fully captures their complexity. This study culls provenance data from repatriated antiquities gathered in the Museum of Looted Antiquities’ dataset to create a network visualization for analysis. Using open-source provenance and repatriation data from 1950 to July 2025, we built a dataset of 233 repatriation events with 15.858 objects to produce a network that reveals central actors, roles, and locations while staying within ethical privacy limits. This study captures large portions of the trafficking network by using accessible data and produces a reproducible, ethically framed workflow. Full article

Plantain (Musa paradisiaca) peel is an agro-industrial waste product with remarkable functional potential, attributed to its composition of bioactive compounds with antioxidant and antimicrobial properties. Given this scenario, this scoping review aimed to map and synthesize the scientific evidence regarding the nutritional composition and potential functionalities of plantain peel. A scoping review approach was used, and data were reported using the PRISMA-ScR checklist. The studies evaluating the use of plantain peel were included without restrictions on language or publication date. The following databases were searched: Embase, MEDLINE (via PubMed), Scopus, and Web of Science. Additional searches were conducted through Google Scholar. The protocol has been registered prospectively on the Open Science Framework. This review’s findings included 53 studies. All of them presented methodological limitations that hindered further analysis and the generation of robust evidence. This analysis detailed the chemical composition of the peel, showing that it varies with ripeness stage and processing and is an excellent source of fiber and minerals. Several technological applications are explored, including the use of peel in the production of functional foods, the development of nanoparticles with antimicrobial activity, and its use as a substrate for the biosynthesis of industrial enzymes and citric acid. This review also addresses the possible health benefits that have already been studied in animal and in vitro models. Plantain peel is a promising agro-industrial by-product with high fiber, starch, and bioactive compound content and functional properties. Despite advances, challenges in sensory acceptance and process standardization limit industrial application. A key research gap remains in the systematic evaluation of antinutrient reduction (e.g., oxalates, phytates) and pesticide residue levels during the processing of plantain peel, a mandatory step before its widespread application in the food industry (e.g., flours and food additives). Further research on optimization and bioactive mechanisms is essential to enable its large-scale use and strengthen its role in the circular bioeconomy and human health. Full article

Background: Large language models (LLMs) are being used by surgeons for education and reference yet concerns about hallucinations and reliability limit safe adoption. Retrieval-augmented generation (RAG) can offer a potential solution by grounding responses in a high-quality external database (e.g., medical textbooks) to enhance accuracy. However, performance tradeoffs across different RAG configurations—many of which exponentially increase computational cost—remain poorly characterized. Methods: In total, 120 lightweight, open-source RAG configurations were evaluated across 40 plastic surgery-focused question-answering tasks (20 single-hop, 20 multi-hop), spanning multiple subspecialties (4800 total evaluations). Configurations varied by base LLM (Phi-3-mini-128k-instruct vs. BioMistral-7B), embedding model, database size, chunk size, and query hop type. Performance was assessed using semantic similarity (Ragas) to physician-validated reference answers. Performance was analyzed using linear mixed-effects regression with query as a random effect and fixed and interaction effects selected via likelihood testing and AIC. Results: High performance was achievable using lightweight, open-source models. While BioMistral-7B had high mean sematic similarity under specific configurations (mean semantic similarity up to 0.786), Phi-3-mini-128k-instruct demonstrated more consistent performance across query complexity. Larger database sizes significantly improved semantic similarity, with the largest gain at intermediate sizes (e.g., size 5: +0.043, p = 0.001). Embedding choice had a strong effect, with bge-large-en-v1.5 improving performance (p = 0.0016) and Bio_ClinicalBERT markedly reducing it (p < 0.001). Multi-hop queries substantially reduced performance (p < 0.001), though this effect was attenuated for Phi-3-mini-128k-instruct via a strong model × hop-type interaction (p < 0.001). Conclusions: RAG systems for plastic surgery do not require large proprietary models, as performance depends on configuration choices and interaction effects rather than isolated components. With advancements, predictive modeling may enable resource-efficient, safe deployment of clinical RAG systems. Full article

Higher education is undergoing a constant paradigm shift, transforming itself into a system of innovation for society. This study has explored and determined the relationship between educational innovation and research in university contexts in order to distinguish, compare, and establish dynamics of interaction. The contributions of scientific articles published in WoS-indexed journals between 2019 and 2025 in a total of 108 sources were analyzed using the PRISMA method and an analysis inspired by grounded theory with open coding and axial coding (mixed method). As a result, both functions have been conceptually differentiated while establishing these points of contact, productive interactions, and their relationship with university institutional management. It is concluded that higher education is facing a paradigm shift, transforming itself from a center of knowledge and professional training to the hub of innovation systems. The main contribution of this study is its exposition of how this profound change is taking place and the conditions of research–innovation interaction in the university setting. Full article