Wind farms are often located at high latitudes, which entails a risk of icing for wind turbine blades. Traditional anti-icing methods rely primarily on manual observation, the use special materials, or external sensors/tools, but these limited by human experience, additional costs, and understanding mechanical mechanism. Model-based approaches heavily prior knowledge subject to misinterpretation. Data-driven can deliver promising solutions require large datasets training, might face...

Accurate and reliable optical remote sensing image-based small-ship detection is crucial for maritime surveillance systems, but existing methods often struggle with balancing performance computational complexity. In this article, we propose a novel lightweight framework called HSI-ShipDetectionNet that based on high-order spatial interactions (HSIs) suitable deployment resource-limited platforms, such as satellites unmanned aerial vehicles. includes prediction branch specifically tiny ships...

Wind energy is of great importance for future development. In order to fully exploit wind energy, farms are often located at high latitudes, a practice that accompanied by risk icing. Traditional blade icing detection methods usually based on manual inspection or external sensors/tools, but these techniques limited human expertise and additional costs. Model-based highly dependent prior domain knowledge prone misinterpretation. Data-driven approaches can offer promising solutions require...

Wind farms are usually located in high-latitude areas, which bring a high risk of icing. Traditional methods anti-blade-icing limited by extra costs and potential damages to the original mechanical structure. Model-based heavily dependent on mathematical models blade icing, prone produce erroneous estimation. As data-driven better able achieve competitive performances for icing estimation, this article proposes temporal attention-based convolutional neural network (TACNN). This novel model...

Wind farms are typically located at high latitudes, resulting in a risk of blade icing. Data-driven approaches offer promising solutions for icing detection, but they rely on considerable amount data. Data exchange between multiple wind would improve the performance detection models, due to spatio-temporal dependencies capable reflecting different meteorological conditions. The traditional centralized approach faces many challenges, including requirement storage and computational capacity...

Wind energy is a fast-growing renewable but faces blade icing. Data-driven methods provide talented solutions for icing detection, considerable amount of Internet Things data needs to be collected central server, which may lead the leakage sensitive business data. To address this limitation, article proposes <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BLADE</i> , Blockchain-empowered imbalanced federated learning (FL) model detection....

Semantic segmentation is of great importance and a challenge in computer vision. One its main problems how to efficiently obtain rich information (geometric structure) identify useful features from higher dimensions. A light field camera, due special microlens array structure, can completely record the angular-spatial scenes, which attractive has potential improve performance semantic task. Inspired by this, we propose an end-to-end network that process macro-pixel image robustly extract...

Data from multimodalities bring complementary information for deep learning-based medical image classification models. However, data fusion methods simply concatenating features or images barely consider the correlations complementarities among different modalities and easily suffer exponential growth in dimensions computational complexity when modality increases. Consequently, this article proposes a subspace network with tensor decomposition (TD) to heighten multimodal classification. We...

Terahertz (THz) spectroscopy is now achieving increasing attention in security inspection owning to its non-destructiveness and deep penetrability of most packaging materials, such as leather, wood wrapper. However, two major obstacles remain spectral classification liquid contraband: the complex components some contraband overlapping effect similar types contraband. In this paper, we establish THz datasets propose a real-time multi-class multi-concentration framework based on convolutional...

Blade icing detection becomes increasingly significant as it can avoid revenue loss and power degradation. Conventional methods are usually limited by additional costs, model-driven heavily depend on prior domain knowledge. Data-driven methods, especially deep learning approaches without needing the time-consuming handcraft feature engineering, offer a promising solution for blade detection. However, monitoring signals normally have complex diverse features wind turbine operates in...

Prediction of icing on wind turbine blades is crucial, particularly in high-latitude areas where ice accumulation a frequent occurrence. Traditional centralized data-driven approaches for predicting blade have demonstrated promising performance, but they require large amount storage and computational resources may also raise concerns about data privacy. Federated Learning (FL) presents potential solution to address these issues. These challenges include redundant features the collected data,...

Unidirectionally propagated electromagnetic waves are rare in nature but heavily sought after due to their potential applications backscatter-free optical information processing setups. It was theoretically shown that the distinct bulk band topologies of a gyrotropic metal and an isotropic can enable topologically protected unidirectional surface plasmon polaritons (SPPs) at interface. Here, we experimentally identify such interfacial modes terahertz frequencies. Launching SPPs via tailored...

With the extensive application of artificial intelligence, ship detection from optical satellite remote sensing images using deep learning technology can significantly improve accuracy. However, existing methods usually have complex models and huge computations, which makes them difficult to deploy on resource-constrained devices such as satellites. To solve this problem, paper proposes an enhanced lightweight model called ShipDetectionNet replace standard convolution with improved units....

Road cracks on highways and main roads are among the most prominent defects. Given inherent inaccuracy, time-consuming nature, labor intensiveness of manual road crack detection, there's a compelling need for automated solutions. The irregular shape cracks, along with complex background conditions encompassing varying lighting, tree shadows, dark stains, poses significant challenge computer vision-based approaches. Most exhibit edge patterns, which pivotal features accurate detection. In...

We have reported the realization of a plasmonic random fiber laser based on localized surface resonance gold nanoparticles (NPs) in liquid core optical fiber. The material contains dispersive solution NPs and dye pyrromethene 597 toluene. It was experimentally proved that fluorescence quenching is restrained fiber, which considered one main sources loss traditional system. Meanwhile, lasing can be more easily obtained system with overlap between photoluminescence spectrum molecules.

Visual object tracking is of great importance in the field computer vision. One main challenges difficulty identifying moving targets from nearby similar distractors with a single-view image scene. To overcome this challenge, article, we acquire multiview images scenes by using light-field camera. The are able to capture 4-D structure instead 2-D plane objects but more difficult process. Therefore, propose novel representation for images, i.e., macro-epipolar (macro-EPI), which highlights...