Search Results (469)

Tropical cyclones are among the most destructive meteorological systems on Earth. Accurate track forecasting of tropical cyclones remains a core challenge in atmospheric science, and it is of great significance for disaster prevention and mitigation. This study targets the critical limitations of existing tropical cyclone track forecasting models: the insufficient ability to extract non-linear spatiotemporal features from 3D atmospheric circulation fields and the long-standing bottlenecks in multi-source heterogeneous meteorological data fusion. To address these issues, we propose a Dual-Stream Gated Fusion Network (DGF-Net), a high-precision track forecasting method tailored to the Northwest Pacific basin. The proposed framework takes the Best Track dataset and ERA5 Reanalysis Dataset as primary inputs: a Bidirectional Gated Recurrent Unit (Bi-GRU) is adopted to capture the temporal evolution characteristics of 2D tropical cyclone trajectory sequences, and a SpatioTemporal Convolutional Gated Recurrent Unit (STConvGRU) is used to extract complex non-linear features from 3D atmospheric environmental fields. Then, a multimodal fusion module integrating gating and attention mechanism is constructed to achieve deep fusion of cross-dimensional features, which effectively mines the intrinsic physical correlations between tropical cyclone track evolution and environmental driving factors. Comparative experiments based on historical observational datasets of the Northwest Pacific show that DGF-Net achieves superior forecasting performance, with the 6 h, 12 h, and 24 h Great Circle Distance (GCD) errors of 35.62 km, 43.53 km, and 135.49 km, respectively. The results significantly outperform mainstream baseline models, which validates the effectiveness of DGF-Net in feature extraction and multimodal fusion and provides solid technical support for tropical cyclone disaster prevention and operational decision-making. Full article

Smart city decision-making increasingly relies on heterogeneous urban data sources. Dense traffic sensor streams provide continuous quantitative measurements, while citizen-generated textual reports offer event-driven contextual information. However, integrating these modalities remains challenging due to temporal misalignment, textual sparsity, and semantic noise. This paper investigates multi-modal learning for traffic congestion severity prediction through an experimental integration of open traffic sensor data (METR-LA: Los Angeles, USA) and citizen-generated textual reports (NYC 311: New York City, USA). Congestion severity is formulated as a four-class classification task derived from traffic speed measurements. We propose an end-to-end framework that combines: (i) sensor time-series encoding using a GRU-based temporal encoder, (ii) textual representation learning using a BERT-based encoder, (iii) a symmetric time-window alignment strategy (±Δ) to associate irregular reports with sensor time steps, and (iv) multiple fusion architectures, including early fusion, late fusion, and a cross-attention module for cross-modal interaction modeling. Experiments on publicly available datasets show that multi-modal early fusion achieves the best overall performance (Accuracy = 0.8283, Macro-F1 = 0.8231) compared to uni-modal baselines. In the studied cross-city setting with sparse and weakly aligned textual signals, the proposed cross-attention fusion does not outperform the strong sensor-only baseline, suggesting that the sensor modality dominates when cross-modal signal strength is limited. These results highlight both the potential and the practical constraints of multi-modal fusion in heterogeneous smart-city environments, emphasizing the importance of alignment design, modality relevance, and transparent experimental validation. Full article

This paper investigates ship–pier collision risk caused by yaw deviation in inland bridge waterways. The proposed framework is conceived for fixed auxiliary thruster installation in bridge areas, rather than retrofitting shipboard propulsion systems. A proactive intervention scheme is developed based on state estimation and short-horizon prediction. A Kalman filter is used for state fusion and short-horizon motion prediction. Yaw events are detected via a threshold rule with consecutive-decision logic. An extended state observer (ESO) is adopted to estimate lumped disturbances and model uncertainties. A fuzzy self-tuning PID law is then applied to generate thruster commands for closed-loop corrective control. Numerical simulations suggest that, relative to rudder-only recovery, thruster-assisted intervention yields improved restoration behavior, reduced lateral deviation accumulation, and increased minimum clearance to bridge piers under the tested conditions. Additional tests with cross-current disturbances indicate that the risk-triggered scheme with ESO-based compensation can maintain stable recovery and a higher safety margin. The proposed approach provides an engineering-oriented pathway to extend bridge-area risk management from warning-level assessment to executable control intervention. Full article

In modern distribution networks (DNs), extreme weather events and cascading faults pose severe challenges to operational safety. However, existing defense mechanisms struggle with a core question: How to maintain high-fidelity situational awareness and make precise active decisions when physical parameters drift and historical fault data is scarce? To address this, this paper proposes a situational deduction and active defense framework based on a service-oriented digital twin. First, regarding the modeling fidelity gap, a data–physics fusion mechanism is constructed. By integrating Kirchhoff’s laws with data-driven error correction, it dynamically calibrates time-varying parameters to resolve mapping distortion. Second, regarding the data scarcity bottleneck, a predictive perception method is introduced. Utilizing the digital twin as a generative engine, it augments rare fault samples to enable super-real-time deduction of future trends. Third, regarding the decision-making passivity, a service-driven simulation model is established. It transforms abstract indicators (safety, economy, resilience) into executable constraints, shifting the paradigm from ‘passive response’ to ‘active defense.’ Case studies on a modified IEEE 123-node system demonstrate that the proposed method significantly enhances resilience and decision accuracy under complex conditions. Full article

Retrieval-Augmented Generation (RAG) has been observed to encounter challenges in heterogeneous query scenarios characterised by varying evidence requirements and reasoning depths. In order to address this limitation, the present paper puts forward a proposal for an Adaptive Multi-Source RAG framework (AMSRAG) that integrates query complexity awareness with confidence-aware fusion. The framework performs query complexity classification with a pretrained language model, calibrates the classification confidence to guide the dynamic scheduling of retrieval paths and the adjustment of fusion weights, and enables a controllable balance between answer quality and retrieval efficiency through hierarchical path selection and cross-source weighting. The experiments conducted on multiple open-domain question-answering datasets demonstrate that the query complexity classifier achieves an accuracy of 85.9% and a Macro-F1 score of 85.4%. These outcomes indicate the potential for the classifier to generate a reliable decision signal, which can subsequently be utilised to guide the process of adaptive retrieval and fusion. The proposed framework demonstrates a marked improvement in terms of both answer accuracy and retrieval relevance when compared to the fixed-pipeline RAG. In scenarios involving high-confidence queries, the system has been shown to effectively avoid redundant retrieval, thereby reducing the average number of retrievals. In instances of low-confidence complex queries, the system has been shown to enhance evidence coverage and completeness of answers through multi-source retrieval and confidence-weighted fusion. This study proposes a novel methodology for enhancing the adaptability and resource efficiency of RAG systems in response to heterogeneous query conditions. Full article

Unmanned aerial vehicles (UAVs) are increasingly employed in precision agronomy to support high-resolution monitoring and management of crops; however, the extent to which UAV-derived data can be translated into reliable, scalable, and decision-ready applications remains inconsistent. This review addresses this gap by critically synthesising the recent literature with a specific focus on the end-to-end data pipeline, from acquisition planning and pre-processing to data fusion, analytics readiness, and operational decision support. A systematic analysis of peer-reviewed studies published over the last five years was conducted to evaluate core agronomic applications, including crop health monitoring, precision irrigation, soil and field variability assessment, spraying, and yield prediction, with particular attention to indicators used, validation strategies, and reported agronomic outcomes. The findings indicate that monitoring and diagnostic applications are the most mature and consistently validated, whereas interventional uses and absolute yield prediction remain strongly context-dependent and constrained by operational, methodological, and regulatory factors. Across applications, pipeline robustness, uncertainty management, and reproducibility emerge as more critical determinants of agronomic value than sensor resolution alone. The review further identifies key barriers to scaling, including technical limitations, skills requirements, data integration challenges, and regulatory constraints, and outlines an innovation roadmap distinguishing currently deployable solutions from emerging developments over the next three to five years. Overall, this work provides a decision-oriented framework to support more transparent, validated, and sustainable integration of UAV technologies into modern agricultural systems. Full article

Although existing Chinese word segmentation systems have achieved substantial progress on standard benchmarks, prediction disagreements among heterogeneous models remain prevalent when processing texts containing complex ambiguities and out-of-vocabulary words, and traditional static ensemble methods such as majority voting often fail to make reliable decisions in low-consensus scenarios. To address this issue, this paper proposes SegFusion, a stacked heterogeneous ensemble framework for Chinese word segmentation based on word lattice re-scoring. The framework first constructs a candidate word lattice to consolidate diverse outputs from heterogeneous segmenters into a unified lattice representation, and then incorporates unsupervised statistical features, including mutual information and branching entropy, as external discriminative evidence to perform dynamic arbitration at the word level, followed by global decoding to obtain the optimal segmentation path. Experimental results on multiple standard datasets demonstrate that SegFusion consistently outperforms individual models and mainstream ensemble baselines in terms of overall segmentation performance and out-of-vocabulary (OOV) recall. In particular, on the MSR dataset with severe ambiguity, SegFusion achieves improvements of 3.71% in F1 score and 4.10% in OOV recall. Further fine-grained analysis shows that the introduction of unsupervised statistical features effectively mitigates model consistency bias in low-support scenarios. These results indicate that integrating language statistical priors independent of training data into the ensemble arbitration stage is an effective way to enhance the robustness and consistency of Chinese word segmentation systems. Full article

Clients need to visit project sites periodically during construction to visualize progress and identify deviations from expectations. However, physical site visits are time-consuming, costly, and potentially unsafe, especially for remote and overseas projects. More fundamentally, existing remote-site-visit solutions focus primarily on automatic recognition and visualization, while insufficiently addressing the scientific challenge of how heterogeneous, dynamic site data can be fused and operationalized to support timely, collaborative decision making. This research proposes a framework for clients’ remote site visits. It develops an RASE system that enables multi-source data fusion and real-time collaborative decision support by integrating UAVs, 360° cameras, BIM, and VR/AR technologies. RASE allows clients to synchronize real-world visual data with BIM models within predefined scenes, annotate issues directly on BIM components, and seamlessly switch among heterogeneous image-capture sources to maintain situational awareness in highly dynamic construction environments. The proposed framework emphasizes an operational data-fusion mechanism and an interaction paradigm that reduces the cognitive and coordination burdens of remote decision making. A case study shows that RASE reduces site-visit time by 78.0%, though initial equipment costs increase total expenses by 44.1%. Sensitivity analyses indicate that projects with greater remoteness or higher visit frequency significantly improve both time and cost effectiveness. The core contribution of RASE lies in enabling a scalable, operational data-fusion mechanism that supports collaboration for remote site visits, with the associated issues for the corresponding BIM components. Automatic image and voice recognition functionality may be incorporated with RASE to improve the efficiency of system control, textual input, and BIM association in the future. Full article

Background/Objectives: Lung cancer (LC) is the leading cause of cancer-related mortality worldwide, making early and accurate diagnosis crucial for improving patient outcomes. Although chest computed tomography (CT) enables detailed assessment of lung abnormalities, manual interpretation is time-consuming, requires expert expertise, and is prone to diagnostic variability. To address these challenges, this study proposes DE-SAMNet, a hybrid deep learning framework for automated multi-class LC classification from CT scans. Methods: The model integrates two pre-trained convolutional neural networks—DenseNet121 and EfficientNetB0—operating in parallel to extract complementary multi-scale features. A Spatial Attention Module (SAM) is applied to each feature stream to emphasize clinically important regions. Final classification is performed through a compact fusion mechanism involving global average pooling, batch normalization, and a fully connected layer. DE-SAMNet was evaluated on two datasets: a public dataset (IQ-OTH/NCCD) with benign, malignant, and normal cases, and a private clinical dataset including benign, malignant, cystic, and healthy cases. Results: On the public dataset, the model achieved a 99.00% F1-score, 98.41% recall, 99.64% precision, and 99.54% accuracy. On the private dataset, it obtained 95.96% accuracy, 95.99% precision, 96.04% F1-score, and 96.21% recall, outperforming existing approaches. To enhance reliability, explainable AI (XAI) techniques such as Grad-CAM were used to visualize the model’s decision rationale. The resulting heatmaps effectively highlight lesion-specific regions, offering transparency and supporting clinical interpretability. Conclusions: This explainability strengthens trust in automated predictions and demonstrates the clinical potential of the proposed system. Overall, DE-SAMNet delivers a highly accurate and interpretable solution for early LC detection. Full article

Reliable estimation of rice yield is essential for food security planning, climate-resilient agriculture, and informed policy decisions. This review synthesizes recent research on the integration of remote sensing and crop simulation models for rice yield estimation. The analysis shows that optical and Synthetic Aperture Radar (SAR) data are the most commonly used remote sensing sources, with SAR proving especially valuable in monsoon-affected regions due to its ability to provide consistent observations under cloud cover. Among crop simulation models, DSSAT, APSIM, ORYZA, and WOFOST are most frequently applied, either independently or in combination with satellite-derived information. Across the reviewed studies, integrated approaches, particularly those using data assimilation and hybrid modeling, consistently achieve higher accuracy and better spatial representation of yield compared to standalone remote sensing or crop model methods. Despite these advances, limitations related to data availability, model calibration, scale mismatches, and climate-induced uncertainty remain significant. Based on the reviewed evidence, future efforts should focus on developing practical hybrid frameworks, improving multi-sensor data fusion, and designing scalable systems suited to data-limited regions. Overall, integrating remote sensing with crop simulation models offers a robust pathway for improving rice yield forecasting and supporting climate-adaptive agricultural management. Full article

Accurate and real-time monitoring of root soil water content (RSWC) is key in optimizing field irrigation decisions and enhancing crop water productivity. However, relying only on the vegetation index as the input to the inversion model may result in lower inversion accuracy due to the canopy spectral saturation effect. To break through the limitation of a single data source, this study constructed an integrated network model (ATT-LSTM) incorporating the attention mechanism based on the long and short-term memory network (LSTM) to enhance the inversion performance by integrating heterogeneous data from multiple sources. The experiment used canopy spectral data based on UAV remote sensing and weather station monitoring data as input features. A control group was set up for cross-validation to realize the accurate inversion of RSWC in kiwifruit plants. The results show that the coefficient of determination (R²) of the ATT-LSTM model on the test set reaches 0.868. This study confirms that the multi-source data fusion framework effectively overcomes vegetation index saturation, improves rhizosphere moisture monitoring accuracy, supports precision irrigation decisions in kiwifruit orchards, and provides a reference for smart agriculture water management optimization. Full article

Multi-criteria decision-making (MCDM) involves evaluating alternatives under uncertain, vague, and conflicting criteria. While fuzzy set theories, such as intuitionistic, pythagorean, fermatean, and q-rung orthopair fuzzy sets have advanced uncertainty modeling, they remain limited to capturing extreme judgments where membership reaches a value of one alongside significant non-membership. The recently introduced q-fractional fuzzy set (q-FrFS) addresses these shortcomings via a flexible constraint, making it suitable for extreme contexts. However, existing q-FrFS methodologies lack robust aggregation mechanisms capable of balancing trade-offs and modulating compensation during information fusion. To overcome this, this study proposes a novel class of Frank-based aggregation operators tailored specifically to q-FrFS environments. Leveraging the parameterized structure of Frank t-norms and t-conorms, we develop two operators: q-FrFFWA (Frank weighted averaging) and q-FrFFWG (Frank weighted geometric) alongside their essential algebraic properties. These operators enhance the representation and fusion of complex and uncertain data. Furthermore, we present a comprehensive MCDM framework utilizing the proposed operators and demonstrate its applicability by selecting optimal vehicle routing software for last-mile delivery. Sensitivity and comparative analyses affirm the stability and credibility of the proposed methodology. This research contributes to the evolving landscape of fuzzy decision-making by integrating the expressive power of q-FrFS with the adaptive flexibility of Frank aggregation, offering a potent tool for modeling and analyzing multidimensional uncertainties in complex decision environments. Full article

A safe physical human–robot interaction (pHRI) in rehabilitation requires reliable perception and low-latency decision making under heterogeneous and unreliable sensor inputs. This paper presents a multimodal sensor-fusion-based safety framework that integrates physical state estimation, semantic information fusion, and an edge-deployed large language model (LLM) for real-time pHRI safety control. A dynamics-based virtual sensing method is introduced to estimate internal joint torques from external force–torque measurements, achieving a normalized mean absolute error of $18.5\%$ in real-world experiments. An asynchronous semantic state pool with a time-to-live mechanism is designed to fuse visual, force, posture, and human semantic cues while maintaining robustness to sensor delays and dropouts. Based on structured multimodal tokens, an instruction-tuned edge LLM outputs discrete safety decisions that are further mapped to continuous compliant control parameters. The framework is trained using a hybrid dataset consisting of limited real-world samples and LLM-augmented synthetic data, and evaluated on unseen real and mixed-condition scenarios. Experimental results show reliable detection of safety-critical events with a low emergency misdetection rate, while maintaining an end-to-end decision latency of approximately 223 ms on edge hardware. Real-world experiments on a rehabilitation robot demonstrate effective responses to impacts, user instability, and visual occlusions, indicating the practical applicability of the proposed approach for real-time pHRI safety monitoring. Full article

Small airway dysfunction (SAD) is an early functional abnormality associated with multiple chronic airway diseases. However, clinical assessment often relies on spirometry-based indices, which require forced maneuvers and are sensitive to subject effort, thereby increasing patient burden and complicating quality control. In contrast, Impulse Oscillometry (IOS) requires only tidal breathing, imposing minimal subject burden while providing respiratory impedance indices informative for SAD identification. This study proposes a dual-domain complementary deep learning framework based on IOS for SAD identification, leveraging within-breath impedance dynamics. Specifically, raw IOS time-series signals are transformed into time–frequency respiratory impedance maps (TFRIM) capturing impedance over frequency and within-breath time. A two-stream architecture is then used to jointly learn complementary features from TFRIM and the original time-series signals. To mitigate inter-subject baseline variability, we further introduce a demographics-driven adaptive feature modulation module for subject-specific calibration. The model jointly predicts multiple small-airway indices, with decision-level fusion applied during inference. Experimental validation on 2510 subjects using five-fold cross-validation demonstrates that the proposed framework achieves an accuracy of 81.39%, outperforming representative baselines. These results suggest the potential utility of combining within-breath IOS dynamics with subject-specific calibration for SAD identification, warranting further external validation before screening deployment. Full article

Precise monitoring of sow behaviour is essential for enhancing animal welfare and production efficiency in precision husbandry. This study proposes an improved RT-DETR model to address real-time detection challenges in complex farming environments. By integrating innovative multi-scale feature fusion and lightweight attention mechanisms, the model achieves high-precision detection of four key postures (standing, sitting, sternal recumbency, and lateral recumbency). Experimental results show that the model attains an mAP@0.5 of 96.6% and a processing speed of 56 FPS, significantly outperforming existing methods. Furthermore, a Unity3D-based digital twin system was constructed to enable real-time bidirectional mapping, achieving a low latency of 320 ms. This system proposes a potential technical framework for intelligent pig farm management, providing a reliable tool for automated welfare assessment and operational decision support. Full article