In this research work, a new approach to compensate for both the response In this paper, we highlight the crucial role of substation digitalization in identifying the origins of harmonic disturbances within iron and steel plants connected to the point of common coupling (PCC). Our method capitalizes on the benefits of a digital substation infrastructure by utilizing merging units that operate at a 28.8 kHz sampling rate, in line with the IEC 61850-9-2 standard. This setup allows for the real-time measurement of voltage and current from all feeders, streaming these high-resolution data directly to substation servers. This digital approach not only facilitates precise and timely data acquisition but also enhances the overall reliability and efficiency of the monitoring process. By digitalizing the substation, we enable a more dynamic response to power quality issues and improve the capacity for preventive maintenance. Integrating our method into existing digital substations supports a robust framework for managing and mitigating the effects of current harmonics and interharmonics, showcasing the practical benefits of advanced digital infrastructure in critical utility operations. We propose attention based long short-term memory as a novel method for determining the harmonic responsibility of each plant without the need for instantaneous impedance measurements. This method relies on the correlation of time-synchronized voltage and current measurements taken from all feeders at the PCC. The approach is first validated using synthetic data generated in a PSCAD/EMTDC simulation environment, as well as real-world field data. Further verification is conducted through measurements of voltage and current waveforms using mobile Power Quality (PQ) measurement systems at one of the PCCs in the electricity transmission system. Our results demonstrate that the harmonic responsibility of each IS plant can be accurately determined for each harmonic order. Unlike existing methods, our proposed approach eliminates the need for online system impedance measurements or in-plant measurements.

Remote sensing visible-light small target detection is a challenging and common problem. Due to the persistent objective factors such as complex backgrounds, it is still difficult to balance the detection accuracy and the number of parameters for densely packed and partially occluded small targets. Existing small target detection methods have neglected the issue of synchronous growth in the number of parameters while improving accuracy. To address these issues, we propose the Higher Precision Small Object Detection Transformer (HPS-DETR), a model optimized for remote sensing imagery with improved precision and real-time capability with fewer parameters. HPS-DETR introduces four key innovations. First, the Faster Convolutional Gated Selection Unit (Faster-CGSU) enhances multi-scale feature extraction by integrating gating mechanism with partial convolutions, improving the aggregation of local and global features while reducing redundant parameters. Second, the Attention-based Intra-scale Feature Interaction using Cascade Group Attention (CGA-AIFI) module focuses on critical feature regions while suppressing irrelevant background noise to improve attention diversity. Third, the Cross-Scale Dynamic Feature Fusion Module (CDFF) enriches small object detection through dynamic scale fusion and precise feature encoding, mitigating feature loss during upsampling and downsampling. Lastly, the Inner-MPDIoU loss function enhances bounding-box regression by prioritizing hard samples, improving accuracy and convergence. Experimental results in SIMD, DOTA-v1.0, and NWPU VHR-10 datasets demonstrate mAP improvements of 2.4%, 1.7%, and 3.1%, respectively, with a 22.11% reduction in parameter count compared to baseline. The mAPs also improves by 2.5%, 2.2%, and 3.3%, respectively. In VisDrone2019 and VEDAI datasets, mAP increases by 2.1% and 3.8%, demonstrating superior performance in small-object detection and computational efficiency.

The circulatory system ensures continuous blood flow to all body tissues. During gestation, progressive changes in the female body and immune system lead to homeodynamic alterations, characterized by increased venous return and cardiac output, enhancing blood flow to meet tissue metabolic demands. This paper presents the Homeodynamic Maternal-Placental Molecular Communication Model (HMP-MC) in the cardiovascular network to accurately represent circulatory components and interactions. The model also analyzes the dynamics of exosomes carrying Zika virus particles throughout pregnancy, integrating new parameters from gestational physiological changes. We quantify the risk of pregnancy over different stages accounting the impact of infection on the placenta viral. This interdisciplinary approach explores cardiovascular interactions during pregnancy analysis over time with the quantification of pregnancy risk.

Social networks, with their intricate web of connections, serve as a lens for studying human interaction patterns and community structures. The SNAP Social Circle dataset for Facebook provides a platform for such analysis, representing users as nodes and their interactions as edges. By employing modularity analysis, hierarchical clustering, and k-clique detection, this study investigates the complex social dynamics within the Facebook network. The results shed light on tightlyknit communities, influential nodes, and structural hierarchies, contributing to strategies for enhancing online engagement and community development.

The urgent need to meet Environmental, Social, and Governance net-zero commitments and the financial risks posed by rising energy costs, are placing increasing pressure on Mobile Network Operators (MNOs) to optimise energy use. Given the significant energy consumption from Base Stations (BSs) and the fact that Fifth Generation (5G) deployments are not meeting expectations for reduced energy consumption when compared to Fourth Generation (4G), MNOs are shifting from a load-centric optimisation approach to a service-centric paradigm that balances spectral efficiency and energy efficiency. This paper introduces the Energy Sustainability Framework, which provides a standardised index to evaluate the spectral and energy efficiencies of BSs, summarised across four classes: A, B, C and D. This framework supports network auditing, benchmarking, monitoring and energy efficiency optimisation while ensuring the desired Quality of Service (QoS). By leveraging Machine Learning (ML) and Explainable Artificial Intelligence (XAI) techniques applied to live data integrated with domain expertise, the framework goes beyond deterministic computation to identify the root-cause factors influencing the BSs' energy class, thereby enabling targeted optimisation. The framework's model-agnostic nature allows it to be applied at various aggregation levels, such as location (site level), BS (cell level), or equipment type (hardware level). Notably, the implemented ML and XAI model achieved an F1-Score of 0.78 for 5G scenarios and 0.83 for 4G scenarios in predicting the Energy Sustainability Class, relying on a reduced set of indicators encompassing the most influential factors. An additional analysis, based on domain expertise, provided evidence of potential causal relationships between the class and observed variables involving channel quality, Multiple Input Multiple Output (MIMO) utilisation, network topology and energy-saving functionalities. For example, the results show that using higher order modulation coding schemes reduces the probability of a BS falling into class D by 50%. These findings provide practical insights for developing more sustainable network operations.

This review explores the burgeoning field of virtual and augmented reality (VR/AR) applications for mental health, focusing on the integration of conversational artificial intelligence (AI) with animated photorealistic avatars created from photographs. This approach facilitates the development of personalized, accessible, and engaging therapeutic experiences across multiple platforms, including mobile phones and VR headsets. By combining the immersive qualities of VR/AR with the interactive capabilities of AI, we present a novel solution to addressing mental health needs, while also expanding to elderly care applications through computer vision technology. Special consideration is given to the potential for adoption in the Japanese market, which is characterized by a rapidly aging population and a growing interest in technological solutions for healthcare in a smart city context. This review draws upon existing research, clinical trials, and the development of the hAI! Friend MR application, highlighting its innovative cross-platform approach and discussing future directions and potential challenges.

In many contexts, particularly in Electromagnetic Compatibility (EMC), a common-mode Current Probe (CP) is intended to be a non-invasive transducer. We introduce its applications generally, but focus on transient waveform measurements. The complex form of our 10 Hz to 1 GHz CP Transfer Impedance (Zt) is measured using Vector Network Analysers. Measured data are compared to manufacturer calibration data and a choice made on the best form of the complex Zt for transient analysis. We then inject a time waveform into a test system and use the Zt to reconstruct the signal, using the inverse Fast Fourier Transform algorithm. The reconstructed current waveform is compared to a reference observation for validation. For frequencies between DC and 240 MHz, a good correlation can be achieved. We raise relevant discussions on filtering, calibration system resonance and de-embedding.  

Large language models (LLMs) have revolutionized the field of artificial intelligence, achieving unprecedented performance in tasks such as text generation, translation, and reasoning. Despite their capabilities, the enormous size and computational demands of these models limit their accessibility and deployment in resource-constrained settings. Knowledge distillation has emerged as a promising approach to address these challenges by transferring knowledge from a large, complex teacher model to a smaller, more efficient student model. This process retains much of the teacher's performance while significantly reducing the computational and memory requirements. This survey provides a comprehensive overview of knowledge distillation techniques tailored for LLMs. We discuss foundational approaches, such as logit matching and feature alignment, alongside advanced methods that leverage intermediate layer supervision, task-specific adaptations, and multimodal extensions. Applications of knowledge distillation in real-world scenarios are explored, emphasizing its role in enabling efficient deployment of LLMs on edge devices and in low-latency environments. Key challenges are identified, including the preservation of emergent behaviors, domain-specific generalization, and the scalability of distillation techniques for increasingly larger models. The survey also highlights ethical and environmental considerations, such as bias transfer and the carbon footprint of model compression. Finally, we outline future research directions, focusing on adaptive frameworks, integration with other compression techniques, and the development of standardized benchmarks to evaluate distilled models.

Reconstructing the human connectome at the neuron level presents a formidable challenge, requiring approximately 400 trillion hours of manual work by trained neuroscientists. However, with advancements in machine learning and deep learning, it has become feasible to automate this process. This paper explores various automated methods for reconstructing the 3D geometry of neural tissue through electron microscopy (EM) image segmentation. We focus on enhancing existing segmentation pipelines by improving both their efficiency and accuracy. Our contributions include the development of a modular segmentation pipeline, DeepSeg, written in Python, allowing for flexible experimentation at different stages of the computational pipeline. Furthermore, we investigate multiple neural network architectures to improve segmentation accuracy and robustness, mitigating error propagation across stages. By leveraging learned alignment strategies and evaluating the performance of different approaches, this work offers a promising path towards the ultimate goal of generating neuron-level connectivity graphs, which can significantly advance our understanding of neural structures and brain function.

Robust autonomous navigation in dynamic and cluttered environments is a persistent challenge for mobile robotics. Conventional methods, reliant on static maps and classical algorithms, often falter when faced with frequently reconfigured obstacle layouts. Deep Reinforcement Learning (DRL) enables adaptive, map-free navigation, yet the complexity of high-resolution LiDAR data and the presence of subtle, moving obstacles complicate training and scalability. To address these issues, we present the Multi-Head Memory Contextualising TD3 (mhmcTD3) architecture, a novel DRL framework that unites multiple specialised network heads-LiDAR fusion, LiDAR Feature, Robot States, and LiDAR Memory-within a unified actor-critic backbone. This integrated design emphasises critical LiDAR states through tailored preprocessing (inversion and exponential scaling), leverages CNNs for refined feature extraction, and employs an LSTM-based LiDAR Memory head for temporal awareness. Additionally, the incorporation of the SiLU activation function and the CoRE optimiser enhances learning stability and convergence efficiency. Extensive evaluations in both simulation (using GAZEBO with ROS2) and real-world tests on a Turtlebot3 waffle_pi platform demonstrate that mhmcTD3 delivers state-of-the-art performance under diverse LiDAR resolutions. It excels at detecting and avoiding small, dynamic obstacles while maintaining adaptability in dense, fast-changing scenarios. Ablation studies further confirm the indispensable synergy among the multiple heads. Taken together, these contributions establish mhmcTD3 as a robust and generalisable DRL-based navigation solution, bridging the gap from controlled simulations to complex real-world environments.

This study aimed to explore the use of Foundation Models (FM) for skeletal muscle segmentation from magnetic resonance (MR) images, which is a critical step in extracting morphological and functional biomarkers in neuromuscular disorders (NMDs). The research addresses challenges such as fat infiltration and ambiguous muscle boundaries by comparing the performance of traditional convolutional neural networks (CNNs) and emerging FMs like the Segment Anything Model (SAM) and its medical adaptation, MedSAM. A dataset of MR thigh images from 76 NMD patients, splitted in Early, Moderate, and Severe according to the degree of fat infiltration, was annotated for 12 muscle groups across 152 volumes. Fine-tuned SAM and MedSAM configurations (encoder-decoder and decoderonly) were evaluated alongside 2D and 3D nnU-Net CNN models. Performance was assessed using Dice Similarity Coefficient (DSC), Average Symmetric Surface Distance (ASSD), and 95% Hausdorff Distance (HD95). Additionally, uncertainty quantification (UQ) metrics, including Expected Calibration Error (ECE) and Negative Log-Likelihood (NLL), were employed to assess model reliability. SAM's finetuned encoder-decoder configuration achieved segmentation accuracy comparable to state-of-the-art 3D nnU-Net models (DSC: 0.925, 0.883, and 0.857 for Early, Moderate, and Severe cases) and significantly outperformed 2D nnU-Net model. SAM achieved superior calibration, particularly in the Severe group (ECE: 0.032 vs. 0.071 of nnU-Net 3D). Deep ensemble methods further improved segmentation reliability and UQ assessment. MedSAM did not surpass the SAM performance. In conclusion, SAM has demonstrated accurate and better calibrated segmentation, demonstrating its potential for medical imaging applications in challenging NMD datasets.

The field of artificial intelligence (AI) and machine learning (ML) has seen remarkable advances with the advent of neural networks. This paper presents a comprehensive comparison between Convolutional Neural Networks (CNNs) and traditional Backpropagation Neural Networks (BPNNs). By exploring the underlying mechanisms, performance metrics, and applications of both types of neural networks, we aim to highlight their respective strengths and limitations. The study provides a detailed examination of the architecture, training processes, and suitability for various tasks, supported by recent empirical research and theoretical analysis.

The fabrication and assembly of High Temperature Superconducting (HTS) magnets can be significantly streamlined by (1) direct deposition of HTS onto modular components and (2) laser-ablated grooves to bound and guide the electric currents over the superconducting surfaces. “Coils” bounded by consecutive grooves can be individually powered, or connected in series, or in parallel. Applications include plasma-confinement devices such as stellarators and magnets for particle accelerators. On the way to HTS magnets generating more complicated three-dimensional fields, this paper validates the technique for two cylindrically symmetric magnets made of standard conductors. The optimized grooving pattern is considered as an inverse problem that is resolved using a least squares approach with Tikhonov regularization. This approach was first applied to design a magnet that replicates the magnetic field configuration of a microwave source of the gyrotron type. Gyrotrons require a particular profile of magnetic field, which our aluminium prototype successfully reproduced with 10−1 precision. The second one, in copper, is a small-size, reduced-field magnet for Magnetic Resonance Imaging (MRI). This application requires highly uniform longitudinal fields. The achieved precision (about 50 ppm) is exceptional for an MRI magnet just getting out of the factory, and can easily meet hospital standards after minimal shimming or other error field correction. Future work and other potential applications of wide, patterned conductors or superconductors are also discussed.

This paper presents an advanced hierarchical detection and real-time text recognition system tailored for UAE traffic signs. Our system is designed to support critical applications in autonomous vehicle navigation, assistance for visually impaired individuals, and automated traffic sign maintenance. Emphasizing sensor-based applications our approach uniquely incorporates text recognition within the traffic sign analysis pipeline, addressing the complex challenge of bilingual (Arabic-English) signs-a capability not previously explored in the literature.The system employs a three-stage pipeline, combining object detection, symbol detection, and advanced text recognition, all optimized for high-speed, realworld conditions. Trained on DoTaS, a custom dataset with 1,500 UAE traffic sign images, the system achieves a mean Average Precision (mAP) of 0.9124 using YOLOv8. For English text recognition, PARSeq achieves 89.3% word accuracy, setting a new benchmark for real-time, high-accuracy sign interpretation. Operating at around 250 ms per inference (around 4 FPS), our framework enhances safety, accessibility, and efficiency in urban environments.

The complexity and variation inherent in the tasks processed through language models increasingly demand that such systems not only retain semantic coherence but also adapt responsively to diverse contextual shifts across prompts. In response to this challenge, Neural Modulation for Dynamic Semantic Convergence (NMDSC) introduces an innovative neural architecture enhancement designed to enable on-the-fly adjustments to the language model's interpretative pathways. NMDSC accomplishes this through embedding modulation gates at strategic points within the model's intermediate layers, which recalibrate neural pathways dynamically, ensuring responses remain aligned with the semantic demands of changing prompts. Such modulation allows the model to maintain interpretative continuity and thematic relevance, particularly when handling complex or contextually complex inputs. Experimental implementation of NMDSC within an open-source language model showcased substantial advancements in response adaptability and coherence, with quantifiable improvements observed across various performance metrics and qualitative dimensions. Detailed assessments revealed that the NMDSC-enabled model exhibited enhanced semantic alignment and reduced error rates when responding to prompts with complex thematic requirements. The findings emphasize NMDSC's capacity to improve language model functionality, offering a scalable framework that enriches context-sensitive language generation, ultimately fostering more refined, consistent, and reliable interpretative outcomes across varied application scenarios. This approach to modular neural modulation thus positions NMDSC as a forward-looking solution to the adaptive challenges faced in current-generation language modeling tasks.

There is a lot of pressures from organizations to continually redesign the business processes and the technology infrastructure to support the organization, but to also help organizations manage competing stakeholder needs, changing objectives and behaviour, and so on. A key part of this is making sure that all their activities in the different components of the business are interconnected and that they are moving in the same direction. Enterprise Architecture (EA) is an effective way of dealing with the challenge of designing business activities and supporting IT. This paper will summarize the organization of EA and look more closely at some major components of EA: governance architecture, business architecture, information architecture and technical architecture and consider the overlap and constructive interaction involved. Enterprise Architecture can be best understood through business process redesign. So, it is useful to briefly review the leading frameworks for business process redesign, including COBIT from ISACA, LEAN from the Toyota Group, TOGAF from the Object-Managemen t Group, and the Capgemini Integrated Architecture Framework (IAF). All these frameworks allow us to more successful transformation, to return to the central idea of this paper.

High-resolution underwater imagery is essential for advancing marine exploration, offering critical data to support resource analysis and discovery. However, underwater images frequently face issues like color loss and diminished contrast. To overcome these challenges, we propose CCMCS-Net, a network for enhancing underwater images that operates across multiple color spaces and is structured in two stages: color correction and contrast augmentation. A Color Correction Subnet (CC-Net) and a Multi Color-space Stretching Subnet (MCS-Net) are the two primary parts of the network. Initially, the Static Correction Module (SCM) leverages green channel data to adjust the red and blue channels, correcting color distortion in degraded images. At the same time, the Dynamic Correction Module (DCM) obtains weight mappings through end-to-end training. It combines SCM to achieve adaptive dynamic compensation of the image, which generalizes the deviation during correction to a certain extent. Next, the RGB, Lab, and HSI color distributions are individually adjusted by applying histogram stretching to the corrected image. The adjusted channels are then concatenated and fed into a Dynamic Fusion Module (DFM) for feature fusion across color spaces. This further enhances image visibility and contrast. The DCM and DFM are trained on our modified U-shaped network architecture. The network is able to adaptively select and emphasize the most useful features for feature aggregation so that the aggregated features better promote high-quality image reconstruction. Extensive experiments demonstrate that CCMCSNet delivers outstanding results in both visual assessments and objective metrics for real and synthetic underwater images.

As humanity advances toward permanent lunar exploration and habitation, the establishment of reliable, autonomous power systems becomes a cornerstone for the success of lunar missions. Lunar microgrids, which are responsible for supplying power to habitats and critical infrastructure on the Moon, must operate with a high degree of autonomy due to the unique challenges of the lunar environment. One critical aspect of lunar power system operation is frequency regulation. Unlike Earth’s power grids, which benefit from a wide range of dynamic grid support mechanisms, lunar microgrids must manage frequency regulation autonomously due to the absence of external support and the extreme conditions on the Moon. This paper explores the importance of autonomous frequency regulation in lunar microgrids, discussing the challenges of maintaining grid stability, ensuring energy efficiency, and mitigating power fluctuations in a microgrid system. Additionally, the role of autonomous control systems, energy storage technologies, and advanced predictive algorithms in achieving effective frequency regulation on the Moon is analyzed. By examining the operational and technical requirements for lunar energy systems, this paper highlights how autonomous frequency regulation is crucial for ensuring reliable and continuous power supply for long-term lunar habitation and exploration. Through a combination of AI-driven optimization techniques, real-time monitoring, and predictive maintenance, lunar microgrids can achieve the resilience and stability required to meet the energy demands of lunar habitats, rovers, and other essential systems.