Search Results (1,753)

This study presents an AI-assisted long-distance fiber Bragg grating (FBG)-based sensing approach for simultaneous temperature and vibration measurement using a single bare FBG sensor. To address the strong coupling between temperature- and vibration-induced effects in the wavelength time series, a signal processing framework based on adaptive variational mode decomposition (AVMD) is developed. With power-spectral-density-guided parameter selection, the mixed wavelength signal is separated into a low-frequency temperature-related component and a high-frequency vibration-related component, enabling stable temperature–vibration decoupling within a single-sensor architecture. Experiments conducted with a 10 km fiber link between the sensor and the interrogator demonstrate that the proposed method can stably track the dominant vibration frequency under various temperature and vibration conditions, while the reconstructed low-frequency component remains consistent with the thermal evolution trend even in the presence of vibration. Random vibration tests and low-frequency vibration resolution analysis further confirm the stability and practicality of the proposed approach under long-distance fiber transmission conditions. In addition, an AI-assisted condition-monitoring scheme is demonstrated using a one-dimensional convolutional autoencoder trained solely with normal wavelength time-series data. Rather than relying on raw reconstruction error alone, the diagnostic layer derives a latent transition score from encoder bottleneck features through temporal pooling, L2 normalization, cosine-distance evaluation, smoothing, and baseline removal. Deviations from steady operating conditions can thereby be preliminarily indicated, highlighting the potential for integrating physics-driven signal processing with data-driven artificial intelligence in long-distance fiber sensing systems. Full article

The growing environmental crisis and rapid urbanization have made the shift to renewable energy systems even more important for smart city development. In today’s cities, such renewable energy sources as solar photovoltaics, wind energy, hybrid systems, and battery energy storage are no longer just separate assets. They are now important parts of smart grids, intelligent buildings, and urban infrastructure that work together. However, putting these systems in cities on a large scale makes it harder to monitor, control, integrate, scale, and work with them in real time. In this setting, the Internet of Things (IoT) and edge computing are technologies that make it possible to turn traditional renewable energy systems into smart, responsive, and self-sufficient urban energy systems. IoT-based monitoring and control systems let city operators, utilities, and policymakers gather real-time data, improve grid stability, optimize energy flows, and better integrate distributed renewable energy sources into smart city ecosystems. Edge computing makes these features even better by allowing for low-latency processing, more localized decision-making, and less reliance on centralized cloud infrastructures. This paper offers a thorough and methodical examination of contemporary IoT- and edge-enabled technologies used to monitor, control, and integrate renewable energy systems; specifically highlighting their significance in smart city and smart grid applications. The review combines the most recent research on hardware platforms, communication protocols, data processing architectures, and edge–cloud coordination mechanisms used in solar, wind, and hybrid energy systems. Additionally, this review synthesizes architectural design principles extracted from analyzed studies to guide the development of scalable, resilient, and cost-efficient renewable energy monitoring systems. This study offers a structured foundation for the design of scalable, resilient, and cost-effective renewable energy management systems that align with the sustainability, efficiency, and intelligence goals of future smart cities by analyzing cutting-edge solutions and pinpointing significant technological trends, challenges, and research deficiencies. This review also highlights its contribution vis-à-vis previous surveys by stressing the inter-domain comparison across solar, wind, and hybrid systems. It focuses, in particular, on edge–cloud coordination and architecture-level trade-offs pertinent to smart grid and smart city deployments. Full article

Single-frame remote sensing image super-resolution can improve the spatial resolution of imaging without requiring a change in hardware, making it an important research direction in remote sensing. In recent years, owing to their excellent generative capability and training stability, diffusion models have shown great potential in the field of super-resolution remote sensing imaging. Given the rapid development of research in this field, it is necessary to conduct a comprehensive review of existing diffusion model-based super-resolution algorithms for remote sensing to help researchers accurately grasp the technical context and development trends. This paper surveys the application of diffusion models to single-frame SR remote sensing imagery. First, we elaborate on the intrinsic characteristics of remote sensing images from three perspectives—the imaging physics process, semantic features, and data scale and task objectives—thereby laying a tailored foundation for algorithm design. Subsequently, we categorize existing algorithms based on three core technical approaches: model architecture optimization, training paradigm innovation, and prior-knowledge fusion. For a comprehensive and objective evaluation of current algorithms, we compile mainstream datasets and multi-level evaluation metrics applicable to super-resolution remote sensing image tasks. Finally, combining the current technical bottlenecks, we propose four future development directions, aiming to provide theoretical support and technical guidance for the development of diffusion models in the field of super-resolution remote sensing imaging. Full article

Aspect-Based Sentiment Analysis explores sentiment trends related to specific opinion aspects and holds significant commercial potential for monitoring brand reputation, understanding customer satisfaction, and personalizing recommendations. However, traditional methods rely exclusively on textual input and often struggle when the target aspect is not mentioned in the sentence. Multimodal Aspect-Based Sentiment Analysis addresses this limitation by incorporating both textual and visual modalities to enable more comprehensive sentiment understanding. Despite advancements in deep learning and transformer-based architectures, existing models often suffer from suboptimal modality fusion and weak aspect grounding, limiting their classification accuracy. To overcome these challenges, we propose an Attention-Driven Feature Fusion (ADFF) approach based on a three-stage hierarchical attention mechanism. First, it only fuses text and image embeddings. Second, it incorporates aspect-level features. Third, a multi-head attention layer further enhances cross-modal dependencies. The resulting representation is passed to a Long Short-Term Memory (LSTM) classifier for sentiment polarity prediction. We evaluate our model on three benchmark datasets, namely Twitter-2015, Twitter-2017, and MASAD. The experimental results demonstrate that the proposed model substantially outperforms state-of-the-art multimodal and unimodal baselines, improves both accuracy and F1-score, achieving 82.55% accuracy and 81.05% F1-score on Twitter-2015, 77.07% accuracy and 77.15% F1-score on Twitter-2017, and up to 99.67% accuracy and F1-score in the Plant domain of MASAD, where we observe consistent improvements across all seven domains. These results highlight the effectiveness and scalability of the hierarchical attention-based fusion strategy for real-world aspect-based sentiment analysis tasks. Full article

Accurate carbon price forecasting is critical for trading decisions, risk management and policy formulation in carbon markets. However, mainstream decomposition-ensemble models suffer from two key drawbacks: point-wise modeling fails to capture long-term temporal dependencies, while independent modeling of decomposed trend and seasonal components leads to serious information loss. To address these limitations, this paper proposes a novel Dual-Path Interactive Attention Network (DPIANet) for carbon price time series forecasting, whose dual-parallel architecture consists of a Dual Interaction Attention (DIA) Block and a Decomposition–Subsequence Interaction Attention (DSIA) Block. First, DPIANet employs a patch-wise partitioning strategy to extract local temporal semantic information inaccessible to traditional point-wise segmentation. The DIA Block jointly captures temporal dependencies between different patches within the same sequence and inter-feature dependencies within the same time step. In parallel, the DSIA Block extracts interactive features between decomposed trend and seasonal subsequences, fusing these features with original subsequences to enhance representation and mitigate decomposition-induced information loss. A dual-layer feature selection method (PMI and XGBoost-SHAP) is adopted to identify key driving factors. Experiments on four representative Chinese regional carbon trading markets covering 2014-2020 show that DPIANet achieves superior prediction performance over state-of-the-art models in terms of MSE and MAE, with competitive robustness across different market characteristics, providing practical decision support for carbon market stakeholders. Full article

Predicting high-resolution Normalized Difference Vegetation Index (NDVI) in mountainous ecosystems is challenging due to topographic complexity and climate heterogeneity. Existing methods often struggle to balance fine-grained spatial patterns with multi-scale meteorological drivers. This paper introduces the Hierarchical Multi-Modal Fusion Network (HMMFN), which employs a conditioned reconstruction strategy to decouple spatial learning from environmental forcing. The architecture utilizes a dual-stream encoder to process NDVI imagery and meteorological data in parallel. A Transformer module captures long-term temporal dependencies, while a multi-level fusion decoder integrates climate semantics with local vegetation details. The model is optimized using a hybrid loss function that combines Mean Squared Error and Structural Similarity Index Measure to ensure both numerical precision and spatial fidelity. Evaluated in the Liupan Mountains, HMMFN consistently outperforms baseline models across multiple lead times. For prediction horizons ranging from one to five months, the model maintains high accuracy with $R^2$ values between 0.9123 (1-month horizon) and 0.8195 (5-month horizon), achieving a 10.1% and 3.6% reduction in RMSE compared to the optimal baseline model, respectively. These results demonstrate that HMMFN effectively preserves fine-scale spatial structures while maintaining accurate temporal trends across various time steps. Full article

The transformative role of Digital Twins (DTs) in the Architecture, Engineering, and Construction (AEC) sector lies in their capacity to generate dynamic, data-driven representations of physical assets that support design, construction, and lifecycle management. To achieve their full potential, DTs must integrate accurate geometric models with continuously updated information reflecting real-world conditions. This information is inherently multidisciplinary and heterogeneous, encompassing structural, environmental, operational, and monitoring data characterized by different spatial and temporal scales. Integrating these diverse datasets into a unified DT environment presents significant challenges related to data heterogeneity, interoperability, varying resolutions, data quality, and uncertainty. This paper presents a PRISMA-based systematic literature review of data fusion techniques applied to DTs within the AEC sector, with particular emphasis on geotechnical and underground infrastructure. A Scopus search conducted on 31 March 2026 retrieved 10,124 records. After sequential screening, 1916 geotechnical-related records were retained for quantitative characterization, 719 records were assessed for eligibility, 454 reports were retained for manual assessment, and 82 studies were finally included in the detailed qualitative review. Existing approaches are classified according to their integration paradigms, methodological foundations, and application domains. Particular attention is given to applications in Geotechnical Engineering, where DTs must integrate sparse, indirect, and highly uncertain subsurface data. Geological conditions are characterized by strong spatial variability, limited observability, material heterogeneity, and epistemic uncertainty, which introduce additional complexities for data fusion compared to surface infrastructure systems. By synthesizing current developments and identifying methodological trends and research gaps, this review provides a structured framework to support the selection and adaptation of data fusion strategies for geotechnical DTs and other complex AEC applications operating under high uncertainty. Full article

Low-latency content delivery is essential in the Internet of Vehicles (IoV) to support autonomous driving, cooperative perception, and infotainment services. However, rapidly changing vehicular mobility and demand patterns limit the effectiveness of existing content prediction and caching strategies, which often capture either short-term temporal trends or long-range dependencies, but not both. This paper proposes a hybrid deep learning architecture that integrates Long Short-Term Memory (LSTM) networks with Transformer encoders to jointly model fine-grained temporal dynamics and global correlations in content requests. The resulting popularity predictions are incorporated into a reinforcement learning (RL)-based caching policy, enabling proactive and adaptive cache placement at roadside units (RSUs) within an end-to-end optimization framework. Simulation results across representative IoV scenarios show that the proposed approach consistently improves cache hit ratio, retrieval latency, and prediction accuracy compared with LSTM-only, Transformer-only, Least Frequently Used (LFU), and Least Recently Used (LRU) baselines. Ablation studies further demonstrate the complementary strengths of the hybrid components, highlighting improved convergence behavior and robustness under varying demand distributions. Full article

This review focuses on research outputs of water purification, wastewater treatment, metallic remediation, and sorption-based experimental studies. It aims to identify the leading nations contributing to these areas and identify the journals that have published the highest number of papers from 2010 to 2025, and centers on yearly publication trends. A thorough quantitative analysis was carried out to examine key characteristics of adsorbents derived from various materials, as well as symmetry and asymmetry of wastewater treatment for the removal of metallic pollutants. Key adsorption mechanisms—including ion exchange, surface complexation, electrostatic attraction, and pore filling—are discussed alongside the structural roles of symmetric (ordered) and asymmetric (heterogeneous) adsorbent architectures. Data was collected from the Scopus database, focusing on specific keywords like “metal,” “water,” “removal,” “adsorption,” “purification,” “drinking water,” “nano adsorbent,” etc. Among approximately 29,598 publications encompassing research papers, reviews, short communications, conference papers, and book chapters, China emerged as the leading publisher with 11,957 papers, trailed by India (4324 papers), the USA (1825 papers), Iran (1739 papers), Saudi Arabia (1484 papers), Egypt (1318 papers), and Republic of Korea (1194 papers). The bibliometric mapping of conventional adsorbents and nanomaterials used in sorption-based technologies was analyzed using VOSviewer, revealing major research clusters, research hotspots, networks, and evolutionary patterns in wastewater treatment and sorption-based water purification. This study indicates that several journals from Elsevier Ltd. and Springer Nature are leading the field with a large number of publications per year. The analysis reveals a consistent upward trend in the number of research publications in recent years. In sum, the bibliometric data provided highlights the growing relevance of these areas among academicians and acts as a catalyst for further research, motivating researchers to investigate new adsorbents or modifications that could improve adsorption performance while maintaining economic viability and efficiency. Full article

Dynamic gravity measurement, crucial for engineering application, suffers accuracy degradation due to errors induced by carrier maneuvers. To address the limitations of conventional swarm optimization algorithms in precision, stability, and generalizability, this study proposes a Swarm Cooperation Evolution Strategy (SCES) that efficiently integrates multiple algorithms. The proposed SCES is extensively evaluated on the CEC2022 benchmark suite in comparison with several cooperative fusion-related algorithms and representative single optimization algorithms. The experimental results demonstrate that SCES achieves an overall effectiveness score of 0.034 and an optimal accessibility rate exceeding 95%. Compared to the best-performing fusion-based algorithm, these metrics represent improvements of 54.67% and 31.11%, respectively. Moreover, relative to the best-performing single optimization algorithm, the improvements amount to 37.73% and 32.69%, respectively. These findings robustly validate the superior performance of the proposed algorithm. Moreover, an in-depth investigation based on SCES into dynamic error compensation methodologies is conducted. Firstly, a polynomial compensation model is established through error mechanism analysis, with parameters identified via SCES. Secondly, a data-driven compensation model employing a multi-layer long short-term memory (LSTM) network optimized via neural architecture search (NAS) guided by SCES is proposed, circumventing the performance limitations inherent in manually designed networks. Furthermore, an innovative two-stage hybrid strategy is introduced. Systematic trend errors are compensated using the polynomial model, followed by the NAS-LSTM model addressing complex residual nonlinear errors, effectively combining mechanism-based and data-driven approaches. Validation on three lines exhibiting varying maneuverability shows all methods significantly improve accuracy. The hybrid strategy delivers optimal performance, achieving 0.58 mGal internal coincidence accuracy on stable lines and up to 91.58% improvement in external coincidence accuracy under high maneuverability. This research provides an effective high-precision dynamic gravity measurement and compensation solution, advancing engineering applications. Full article

With the large-scale integration of offshore wind farms (OWFs), harmonic issues caused by the interaction between high-frequency switching of converters and complex network impedances pose severe challenges to power quality. Traditional harmonic monitoring heavily relies on post-event fixed-threshold alarm mechanisms, which struggle to achieve early warning during the low-distortion sub-health operation stage and lack the capability for multi-dimensional tracing of harmonic degradation sources. To address these limitations, this paper proposes a deep warning network for grid harmonics combining multi-step regression and multi-dimensional tracing within a unified multi-task learning (MTL) architecture. First, a deep shared feature encoder, integrating a bi-directional long short-term memory (Bi-LSTM) network with a multi-head self-attention (MHSA) mechanism, is utilized to extract high-order temporal coupling features between meteorological evolution and multi-node electrical states. Subsequently, the main task branch executes a k-step-ahead multivariate time-series regression to accurately predict the evolution trend of total harmonic distortion (THD) at both the point of common coupling (PCC) and the turbine terminal. Simultaneously, the auxiliary task branch performs multi-label micro-state classification based on relative degradation thresholds, achieving fine-grained multi-dimensional tracing covering spatial nodes, electrical attributes, and their joint micro-states. Experimental results on real-world OWF operational data demonstrate that through the joint optimization of regression and tracing tasks, the proposed MultiDimKStepMTL model significantly improves time-series prediction accuracy, achieving a $10.3\%$ relative improvement over single-task baselines, while substantially reducing computational overhead. This research successfully advances grid harmonic monitoring from passive response to proactive micro-state early warning, providing a solid, highly interpretable data-driven foundation for active filter control of offshore wind clusters. Full article

Depression is a critical global public health challenge, and the demand for accurate automated detection methods has generated considerable research interest in Transformer-based models. Despite their substantial promise, a comprehensive investigation into their architectural efficacy, intrinsic mechanisms, and barriers to practical implementation remains lacking. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines, this systematic review was conducted across six databases (IEEE Xplore, Elsevier, Springer, MDPI, PubMed, and arXiv). The final search was performed in October 2025, covering English-language empirical studies published between 2020 and 2025 that employed Transformer-based architectures for depression detection. Risk of bias and methodological quality were independently appraised by two authors using a six-dimension structured rubric, with disagreements resolved by a third author. Findings were narratively synthesized given substantial cross-study heterogeneity. This systematic review analyzed 46 studies and provided the first comprehensive, mechanism-level, architecturally stratified comparison of encoder-only, decoder-only, hybrid, and multimodal fusion paradigms, examining self-attention dynamics and transfer learning strategies. Since 2019, these frameworks have evolved from text-centric approaches to advanced multimodal systems. Encoder-only models show consistently strong results in high-throughput text-based screening, decoder-only models demonstrate stronger few-shot learning capabilities, hybrid architectures show the highest observed median performance in clinical interview settings across the reviewed studies, and multimodal fusion systems offer complementary advantages when heterogeneous signal integration is critical. These trends are task-contextualized and should not be interpreted as unconditional rankings, given heterogeneity in evaluation metrics and tasks across studies. Nonetheless, four principal challenges hinder clinical translation: overreliance on self-reported data, cross-linguistic bias, absence of uncertainty quantification, and substantial computational overhead. Future efforts should shift from incremental benchmark improvements toward clinical utility through standardized psychiatric validation, uncertainty-aware architectures, fairness-enforced training across diverse populations, and the integration of Transformer-based models with wearable and mobile health data to improve detection stability and reduce translational risk. This systematic review was registered on the Open Science Framework (OSF; DOI: 10.17605/OSF.IO/SYF9N). This research was funded by the Faculty of Information Science and Technology and by Universiti Kebangsaan Malaysia under Grant TAP-K014364. Full article

Background: Neglected diseases caused by protozoan parasites remain a major public health burden, particularly in low- and middle-income countries. Among the chemical motifs explored in antiparasitic drug discovery, guanidine-containing compounds have attracted considerable attention due to their strong cationic character, high capacity for hydrogen bonding, and versatility in interacting with biological targets. Methodology: This review summarizes advances reported in the last decade regarding guanidine derivatives with activity against pathogens associated with Chagas disease, human African trypanosomiasis, Leishmaniasis, tuberculosis, toxoplasmosis, dengue and schistosomiasis. Results: Evidence gathered from synthetic, natural, and drug-repurposing studies indicates that the guanidine, guanidine-containing and guanidine-related compounds contribute to modulating biological activity by changing electrostatic interactions, hydrogen-bonding networks, and physicochemical properties, with enzymes, nucleic acids, and membrane-associated targets essential for parasite survival. Across the analyzed studies, several emerging structure–activity relationship trends were identified, including the contribution of polycationic or dicationic architectures, the influence of halogenated or lipophilic substituents, and the dependence of biological activity on the complete molecular framework, including heterocyclic systems, macrocycles, peptide conjugates, hybrid scaffolds, and repurposed drugs. In addition to direct antiparasitic effects, certain guanidine-containing and guanidine-related compounds demonstrate immunomodulatory or host-protective properties, expanding the therapeutic relevance of this class. Despite promising in vitro results, protonation trapping, efflux pump susceptibility, and pharmacokinetic limitations such as poor oral absorption, high polarity, plasma protein binding and limited membrane permeability remain significant challenges for clinical translation. Nonetheless, the integration of medicinal chemistry, computational modeling, and biological screening continues to accelerate the identification of optimized scaffolds. Conclusions: Overall, guanidine-based compounds constitute a promising scaffold for the development of new therapeutic strategies targeting neglected parasitic diseases, and further structural optimization may enable the emergence of candidates with improved efficacy, selectivity, and drug-like properties. Full article

Accurately estimating the state of health (SOH) of lithium-ion batteries remains challenging for battery management systems. Traditional data-driven methods, such as long short-term memory (LSTM), lack physical interpretability and often fail to generalize across varying operating conditions. To address this, a physics-informed SP-LSTM framework is proposed that integrates the single particle model (SPM) with a bidirectional LSTM network. A hybrid optimization strategy combining particle swarm optimization and the limited-memory Broyden–Fletcher–Goldfarb–Shanno with bounds (L-BFGS-B) is first used to identify key SPM parameters, which are then combined with macro external features (charging time, discharge energy, IC peak) to form a seven-dimensional fusion vector. A dual-stream Bi-LSTM architecture separately models fast-varying macro trends and slow-varying physical parameters, achieving robust SOH mapping. Validated on the NASA PCoE dataset, the proposed SP-LSTM achieves a root mean square error (RMSE) of 0.0136 and a mean absolute error (MAE) of 0.0089 on an independent test set (B0018), outperforming the baseline LSTM by 38.2% in RMSE. Noise robustness tests (0–3% voltage noise) and Sobol global sensitivity analysis further confirm its stability and interpretability. By embedding electrochemical priors into the data-driven pipeline, this work provides a practical physics-data collaborative framework for accurate and trustworthy battery SOH estimation. Full article

Early-exit cascades over a frozen convolutional backbone enable adaptive inference but suffer from three sources of train–inference mismatch: branches train on samples they will never see at inference; their per-class precision thresholds are calibrated on the wrong distribution; the standard cross-entropy target on backbone argmax labels discards the backbone’s uncertainty signal. We close all three gaps with CalexNet (cascade-aligned early exits), a training-recipe-only modification: branches train under continuously weighted importance sampling that matches the cascade-survivor distribution; per-class precision thresholds are calibrated on the actual cascade-survivor subset of the validation set; the classification head is trained against the backbone’s full softmax via a temperature-scaled KL objective. Combined with an augmented prototype-pooling branch head, CalexNet is evaluated on ResNet18 and ResNet50 backbones across CIFAR-100 (20-supe-class coarse, the harder primary setting) and CINIC-10 (10-class, the easier cross-validation counterpart). On the accuracy–FLOPs Pareto frontier, CalexNet matches or exceeds three published baselines (PTEEnet, ZTW, BoostNet) and a within-paper “no-alignment, no-KD” reference. The largest gains appear in the practically relevant 30–70% FLOPs-reduction regime and show consistent trends across $n=3$ training seeds. CalexNet requires no inference-time architectural change and is a drop-in for any frozen-backbone early-exit cascade. Full article