Search Results (18,464)

Facial expression recognition (FER) for assistive and telepresence robotics remains challenging under resource-constrained conditions because landmark normalization is often unstable, many datasets have limited variability, and full facial landmark sets introduce redundancy. This paper proposes a lightweight, privacy-preserving FER framework for assistive healthcare robotics based on geometric facial landmarks rather than raw RGB images. The objective is to improve recognition robustness and deployment suitability on low-power edge devices through two complementary contributions: a revised nose-centered landmark normalization method and an optimized Facial Feature Mapping, FFM-L03. The proposed normalization replaces the expression-sensitive upper-lip reference with a geometrically stable nose-center anchor, while FFM-L03 combines FACS-guided anatomical priors with ANOVA F-score, LASSO, PCA, and t-SNE/UMAP to retain 60 informative landmarks. In addition, a heterogeneous Freepik dataset was constructed to increase variability in lighting, background, resolution, and subject appearance. Experimental evaluation across 15 landmark groups, four datasets, and four classifiers shows that the proposed method consistently improves performance over prior landmark configurations, achieving gains of up to 22.4 percentage points over the Ciraolo baseline and 22.1 percentage points over the full-landmark baseline in accuracy, precision, recall, and F1-score, while maintaining lightweight operation. These results demonstrate that principled normalization and targeted landmark selection can substantially improve FER for real-time, privacy-aware assistive robotic systems. Full article

Driven by the global energy transition and carbon neutrality goals, hybrid power systems have become a core technical path for energy conservation and carbon reduction in the transportation and power sectors, and the performance of energy management strategies directly determines the system’s overall energy efficiency. Traditional energy management methods have inherent bottlenecks of high model dependence and poor adaptability, making it difficult to satisfy real-time decision-making requirements under complex operating conditions. Deep Reinforcement Learning (DRL) provides an innovative solution to this technical bottleneck, and has become a cutting-edge research direction in this field. However, existing reviews have not yet constructed a full-chain analysis framework covering its algorithms, applications, verification, challenges and prospects. Focusing on the engineering application of DRL in the real-time energy management of hybrid power systems, this paper systematically sorts out domestic and international research results up to the first quarter of 2026. The core quantitative findings of this review are as follows: (1) DRL-based strategies can achieve 93–99.5% of the Dynamic Programming (DP) theoretical global optimum in fuel economy, which is 5–25% higher than rule-based methods; (2) DRL strategies only have 3.1–4.8% performance degradation under unseen operating conditions, which is significantly better than the 10.3–14.7% degradation of the Equivalent Consumption Minimization Strategy (ECMS); (3) Actor–Critic (AC) algorithms (Twin Delayed Deep Deterministic Policy Gradient (TD3)/Soft Actor–Critic (SAC)) have become the mainstream in this field, with a 3–5 times higher sample efficiency than value function-based algorithms; and (4) offline DRL and transfer learning can reduce the training time of DRL strategies by more than 80% while maintaining equivalent optimization performance. This paper first analyzes the essential attributes and core technical challenges of hybrid power system energy management; second, classifies DRL algorithms from the perspective of control engineering and analyzes their technical characteristics; third, disassembles the application design logic of DRL around four major scenarios: land vehicles, water vessels, aerial vehicles and fixed microgrids; fourth, summarizes the mainstream verification platforms and evaluation systems; fifth, analyzes core bottlenecks and cutting-edge solutions; and finally, prospects the development trends of next-generation intelligent energy management systems combined with cross-fusion technologies. This paper aims to build a complete technical system map for this field and promote the engineering deployment and practical application of intelligent energy management technologies integrating data and knowledge. Full article

Forecasting hydropower plant power production is a great challenge in the context of maintaining power system stability, reliability and efficiency, especially in an age with variable renewable energy sources when demand for electricity is steadily rising. Accurate forecasting methods are a crucial enabler for the operational existence of power systems that rely on renewable sources. And while in the pursuit of increased accuracy of predictions, many recent research works rely on artificial intelligence and machine learning techniques, this study proposes and adopts a more conventional approach with standardized mathematical models to address the problem of hydropower production forecasting. The model predicts the runoff–power relationship. It starts with the normalization of different rain phenomena as a part of the statistical characterization of runoff events. The system transforms rain occurrence to runoff events via the USDA SCS CN model and then feature vectors are composed, which are used to generate kernel coefficients via interpolation. Contrary to models based on artificial intelligence, the proposed approach has several practical advantages requiring a minimal set of input parameters, which significantly reduces data preprocessing demands and allows for a straightforward integration into existing systems, thereby lowering the cost and the implementation and deployment time. Furthermore, the simplicity and universality of the model make it so that it can be adapted across a wide range of hydropower plants of varying scales and with diverse hydrological and meteorological conditions. The model’s performance and prediction accuracy are evaluated using empirical data records of time series over a five-year period for the meteorological parameters and production of an existing real-life hydropower plant in Bulgaria. The performance of the newly proposed model is assessed using widely accepted statistical error metrics, namely, Root Mean Square Error (RMSE), Mean Absolute Error (MAE), the Nash–Sutcliffe Efficiency (NSE) coefficient, and the Pearson correlation coefficient (R). These metrics provide a comprehensive assessment of the forecasts’ precision and effectiveness. The results show that the proposed model offers admissible accuracy with low computational effort. Thus, it can be successfully implemented in practice in a number of hydropower plant production forecasting applications. Full article

Deploying post-quantum cryptography on highly constrained devices remains challenging due to the large key sizes and substantial storage and memory-traffic demands of leading lattice-based schemes. Although constructions such as Kyber, Dilithium, and NTRU offer strong resistance against quantum adversaries, their multi-kilobyte public keys and intensive memory access patterns limit practical adoption in microcontrollers, smart cards, and low-power edge environments. This work proposes a hybrid key-encapsulation mechanism that integrates a compact, seed-generated Module-LWE structure with a quantum-secure hash-based authentication layer. The design employs a small public seed to instantiate lattice matrices on demand via a lightweight pseudorandom generator and incorporates a Merkle-tree commitment to represent compressed auxiliary error information. Additional design considerations—including sparsity-aware secret keys, SIMD-friendly polynomial operations, and cache-efficient decryption paths—are intended to reduce runtime memory usage and computational overhead. The security of the proposed construction is analysed under both Module-LWE and hash-based one-way assumptions, with further consideration of constant-time execution and cache-line alignment to mitigate side-channel risks. This hybrid approach outlines a design pathway toward post-quantum key-encapsulation mechanisms suitable for deployment on memory-limited and energy-constrained platforms. Full article

With the rapid and ongoing expansion of urban areas, the need for accurate, reliable, and regularly updated topographic maps has become increasingly critical for planning and sustainable development. While traditional aerial photogrammetry—whether analog or digital—has long been the standard for such tasks, it remains costly, time-consuming, and logistically demanding, particularly when large or inaccessible regions are involved. This study proposes an alternative approach based on very-high-resolution satellite imagery, focusing specifically on data acquired from Morocco’s Mohammed VI A and B satellites. The research evaluates the capacity of this satellite imagery to support large-scale topographic mapping, both in terms of geometric accuracy and the ability to identify essential urban features. To validate the results, we conducted a comparative analysis of satellite data with conventional photogrammetric imagery from analog cameras (RMK TOP) and digital sensors (ADS, DMC), using ground control points (GCPs) and differential GPS (DGPS) measurements for calibration and accuracy assessment. The outcomes demonstrate that planimetric accuracy from satellite imagery meets the required standards for mapping at 1:10,000 and 1:5000 scales. However, altimetric accuracy is closer to the upper permissible limits, especially in applications requiring finer detail. While major urban elements such as roads, buildings, and vegetation are well identified, smaller infrastructure components, such as power lines, remain challenging to detect. Despite these limitations, the study highlights the growing potential of satellite imagery as a cost-effective and operationally efficient alternative to traditional methods, particularly in rapidly evolving urban environments where frequent map updates are essential. Integration with GeoAI workflows is identified as a key direction for future research and is not part of the current methodology. Full article

Deploying highly generalizable deepfake detection systems on resource-constrained edge devices poses a significant technical challenge for conventional end-to-end large models that rely heavily on computational resources. Extracting multi-source physical prior features is a viable approach under limited computational power; however, in few-shot scenarios, the dimensional mismatch of heterogeneous features is prone to causing downstream classifiers to overfit. To mitigate this bottleneck, this paper proposes a “static feature extraction–central normalization alignment–independent downstream decision” decoupled detection system for few-shot cross-domain tasks on edge devices. The front end of the system constructs an 856-dimensional comprehensive feature reservoir, and a lightweight residual normalization adapter $g_{\varphi }$ is introduced as the central support module. This module explicitly compresses the intra-class variance of heterogeneous features, providing a smoothly aligned manifold base for downstream classifiers. Experimental results indicate that this decoupled architecture demonstrates consistent stability in few-shot ($K=10$) cross-domain evaluations. When encountering intra-family cross-domain shifts and cross-mechanism distribution shifts from diffusion models, the accuracy reaches 84.9% and 76.1%, respectively. Compared to representative end-to-end meta-learning baselines (e.g., MAML), the relative error rate is reduced by over 30%. Furthermore, after completing the asynchronous offline pre-processing (approximately 897 ms) at the front end, a single-image online classification query requires only 7.7 ms under a simulated single-core CPU constraint, satisfying the low-latency requirements for lightweight deployment on edge devices. Finally, combined with empirical observations, this paper discusses the performance boundaries of the architecture in cross-mechanism metric mismatch scenarios, providing a low-barrier, robust engineering defense scheme for resource-constrained environments. Full article

Power-system reliability is crucial for sustainable development, but large-scale, long-term monitoring remains challenging. Existing nighttime light (NTL)-based outage detection methods often rely on fixed thresholds or prior information, limiting cross-regional application. To address this, we develop an adaptive thresholding framework using daily NASA Black Marble data. Observations are grouped by view angle to mitigate radiometric instability, and a per-pixel dynamic baseline is constructed from high-radiance statistics, enabling robust anomaly detection without prior outage timing. From the detected anomalies, we formulate a population-weighted NTL power reliability index (NTPRI) to quantify regional electricity service reliability. Validation across six diverse outage events yields an F1 score of 0.807. National-scale analysis shows NTPRI correlates significantly with the World Bank’s System Average Interruption Duration Index (SAIDI). The derived Light Anomaly Rate (LAR) further supports pixel-level frequency analysis. Together, this framework provides a transferable remote-sensing tool for large-scale power-reliability assessment in data-scarce regions, supporting disaster impact evaluation and energy vulnerability analysis. Full article

The operation of medical cyclotrons for PET radiopharmaceutical production presents significant radiological and environmental challenges that require systematic risk assessment and evidence-based mitigation strategies. In this study, an integrated framework combining Failure Mode and Effects Analysis (FMEA) with a quantitative Defense Effectiveness Factor (DEF) approach to evaluate and reduce residual risk in a real urban cyclotron facility. High-criticality failure modes (Risk Priority Number $\ge 120$) affecting HVAC systems, stack exhaust, and power supply were identified and validated through a Delphi expert consensus process. These modes were addressed with multi-layered defense-in-depth strategies: redundant systems (occurrence reduction, 60–80% effectiveness), real-time monitoring (detection reduction, 40–50% effectiveness), and design robustness (severity reduction, 70–85% effectiveness). The combined DEF yielded a 96–97% risk reduction. One-way sensitivity analysis confirmed the robustness of these results, with residual annual effective dose to the representative person remaining between 50–88 $\mathsf{\mu }$Sv/year (well below the IAEA 1 mSv/year public dose constraint) even under pessimistic scenarios. Primary exposure pathways were inhalation and cloud gamma from ${}^{18}$F and ${}^{41}$Ar during the early-morning production window, while secondary pathways were negligible due to the short half-lives of the radionuclides. These findings demonstrate that the integration of FMEA with DEF-based defense-in-depth and Gaussian plume modeling provides a transparent, robust, and regulatory-compliant framework for managing radioactive atmospheric emissions in urban medical cyclotron facilities. Full article

Varistors are critical overvoltage protection components in modern power electronic systems. They effectively absorb and dissipate surge energy to ensure the safe and stable operation of electrical equipment. However, surface defects can lead to substandard performance or even trigger equipment failure, compromising overall system stability. Therefore, high-precision surface defect detection is essential for quality assurance. To address these challenges, we propose a lightweight model termed Defect-Aware and Edge-Reconstruction Enhanced YOLO (DAER-YOLO) for efficient varistor inspection. First, we construct a C3k2-based defect-aware enhancement module (C3k2-iEMA). This module tackles the difficulty of extracting features from small or morphologically complex defects. By integrating multi-scale feature extraction, an attention mechanism, and efficient nonlinear mapping, it strengthens the perception of defect details. Second, to enhance the reconstruction capability for edge damage and small-object defects, we introduce the Efficient Up-Convolution Block (EUCB). This block improves multi-level feature fusion and generates clearer enhanced feature maps. Based on these improvements, DAER-YOLO outperforms the YOLOv11n baseline on a custom varistor dataset, with mAP@50 and mAP@50:95 increasing by 1.6% and 2.3%, respectively. Experimental results demonstrate that the model effectively improves detection accuracy while exhibiting significant potential for real-time industrial applications. Full article

In the evolving landscape of next-generation wireless networks, ensuring seamless mobility and high-quality service delivery for millions of devices and end users in dynamic scenarios, where the speed of a wireless device keeps changing with time, is important. The mobility, seamless and continuous connectivity, and ultra-dense deployment of wireless networks pose a significant challenge. Seamless and successful transition of a wireless device from point A to point B in variable-speed scenarios is one of the major challenges in future networks. This paper presents a novel Deep Q-Network (DQN)-based reinforcement learning (RL) framework integrated with Software-Defined Networking (SDN) for intelligent mobility management in hybrid 5G cellular networks consisting of macro and small base stations. The proposed system architecture utilizes a SDN controller to receive real-time user measurement reports, including Reference Signal Received Power (RSRP), Signal-to-Interference Noise Ratio (SINR), and user velocity, thereby classifying user mobility into distinct subclasses and dynamically determining optimal handover parameters. Leveraging the DQN’s capability to learn adaptive strategies, the model enables seamless transitions between macro and small cells based on mobility profiles, thereby enhancing Quality of Service (QoS) metrics such as latency, throughput, and handover efficiency. Simulation results demonstrate consistent performance improvements over baseline and existing models in ultra-dense network environments, with handover success rates 10–15% higher across SINR and different speed scenarios, while maintaining a packet failure rate of 9% across different speed scenarios, allowing more users to transition during various environmental changes seamlessly. Our proposed model is compared with our previous work and Learning-based Intelligent Mobility Management (LIM2) models. Specifically, our previous work focused on adaptive handover management primarily for high-speed train scenarios using a learning-assisted approach tailored to fixed high-mobility scenarios, with a limitation to single mobility conditions. This work contributes to the field of merging SDN’s centralized control with the predictive power of RL, paving the way for more resilient and responsive mobile networks in high-mobility scenarios. The proposed approach incorporates subclass-based mobility action abstraction, joint optimization of TTT and hysteresis margin, and dynamic target cell selection using global network information available at the SDN controller. Full article

Photovoltaic (PV) power generation is inherently intermittent due to unpredictable irradiance variations, posing significant challenges for grid integration. While conventional power smoothing strategies mitigate short-term fluctuations, they do not explicitly enforce the tracking of a scheduled power trajectory. This paper proposes a dispatchable PV framework that integrates a hybrid convolutional neural network-long short-term memory (CNN-LSTM) model for precise day-ahead power forecasting with a real-time supercapacitor (SC) compensation strategy. The CNN-LSTM network captures complex spatiotemporal meteorological dependencies to generate a robust day-ahead reference trajectory. Concurrently, a supercapacitor energy storage system (SC-ESS) integrated at the DC-link level via a bidirectional buck–boost converter actively balances the instantaneous mismatch between this forecast trajectory and the actual PV generation. Unlike filter-based hybrid methods, the SC-ESS is employed as a direct forecast error actuator in a closed-loop control scheme. This strategy strictly enforces real-time forecast tracking while preserving maximum power point tracking (MPPT) and DC-link voltage stability. Simulations and laboratory experiments under rapidly varying irradiance confirm that the proposed method significantly reduces power deviations from the forecast reference and improves short-term power predictability without imposing excessive stress on the SC. This forecast-aware strategy effectively enhances the dispatchability of PV systems, providing a practical solution for grid-supportive operation. Full article

Electric vehicles are an important carrier for achieving energy savings and emission reductions in the transportation sector. As the decision-making core of the powertrain, the energy management strategy is responsible for power allocation and energy scheduling and directly determines vehicle economy, power-source lifetime, and overall performance. Model predictive control can handle multiple constraints and objectives within a prediction horizon and realize online closed-loop decision-making via receding-horizon optimization and has become an important research direction for energy management of electric vehicles. This paper presents the basic principles and typical modeling framework of model predictive control and reviews its research progress in hybrid electric vehicle energy management. The related studies are categorized and comparatively analyzed from three perspectives—prediction methods, solution strategies, and optimization objectives—and the characteristics of different approaches are summarized. The review shows that model predictive control has advantages in multi-objective trade-offs and adaptation to time-varying operating conditions. However, practical implementation still faces significant barriers, including prediction uncertainty and computational complexity. Finally, the challenges and future directions of model-predictive-control-based energy management strategies are discussed. Full article

Recent advancements in deep learning have shown considerable potential to enhance radar target detection, particularly in improving detection probability under complex environmental conditions. However, existing deep learning approaches largely operate in the real number domain, neglecting the complex-valued nature of radar data, and often inherit vision-oriented architectures that fail to address radar-specific challenges—such as sparse target echoes, the necessity for phase preservation, and constraints imposed by scanning radar systems. Meanwhile, conventional radar signal processing methods, including CA-CFAR, are limited by their dependence on idealized statistical models and often underperform in dynamic and cluttered electromagnetic environments.To overcome these issues, this paper proposes Radar Transformer for Detection (RaTDet), an end-to-end detection network that integrates complex-valued convolutional neural networks (CNNs) and Transformers. RaTDet fully leverages complex-valued data to preserve critical phase and amplitude information, enabling automated feature learning directly from raw radar signals. The model operates effectively with very few pulses, making it suitable for resource-constrained scenarios, and can serve as a pre-trained foundation model for various radar downstream tasks. Experimental results demonstrate that RaTDet achieves excellent detection performance, characterized by high detection probability (Pd) and low false alarm rate (Pfa), outperforming both traditional signal processing and conventional deep learning methods. This work bridges the gap between deep learning and radar signal processing, offering a flexible and powerful network for next-generation radar systems. Full article

As the global response to climate change and energy crises accelerates, the large-scale integration of heterogeneous distributed energy resources (DERs) is rapidly transforming traditional passive distribution networks into active distribution networks. However, the massive quantity and high stochasticity of these underlying devices trigger a severe “curse of dimensionality,” creating significant computational and communication bottlenecks for coordinated system dispatch. To overcome these challenges, the “clustering followed by equivalence” aggregation modeling paradigm has emerged as a critical technical pathway. This paper reviews the state-of-the-art clustering and aggregation methodologies for distribution networks with high DER penetration. The review begins by synthesizing multi-dimensional feature extraction techniques and cutting-edge clustering algorithms that establish the foundation for dimensionality reduction. It then delves into refined aggregation models tailored to heterogeneous resources, including dynamic data-driven equivalence for renewable generation, Minkowski sum-based boundary approximations for energy storage, and thermodynamic alongside Markov chain mapping methods for flexible loads. Building upon these models, the paper comprehensively discusses the practical applications of generalized aggregators, such as microgrids and virtual power plants, in feasible region error evaluation, coordinated network control, multi-agent market games, and privacy-preserving architectures. Finally, the review outlines future research trajectories, emphasizing hybrid data-model-driven architectures for real-time dispatch, distributionally robust optimization (DRO) for enhancing grid resilience and self-healing, and decentralized trading ecosystems to ensure equitable system-level surplus allocation. This review aims to provide a systematic theoretical reference for the coordinated management and aggregated trading of flexibility resources in novel power systems. Full article

Financial fraud has become a critical challenge for modern financial systems due to the rapid growth of digital transactions, online banking services, and electronic payment platforms. Traditional rule-based fraud detection systems are increasingly inadequate in addressing the evolving and adaptive strategies employed by fraudsters. Consequently, Machine Learning (ML) and Deep Learning (DL) techniques have emerged as powerful tools for detecting fraudulent activities in large-scale financial datasets. This paper presents a comprehensive survey of ML/DL approaches for financial fraud detection. The survey systematically reviews existing research across multiple methodological paradigms, including classical supervised learning, anomaly detection, graph-based methods, deep neural networks, multimodal architectures, and cost-sensitive learning frameworks. Particular emphasis is placed on emerging techniques such as graph neural networks, transformer-based architectures, and federated learning approaches designed to address privacy and scalability challenges. In addition to reviewing model architectures, this work analyzes key challenges inherent to fraud detection systems, including extreme class imbalance, concept drift, adversarial behavior, data privacy constraints, and real-time deployment requirements. Furthermore, the survey examines evaluation methodologies, highlighting the limitations of commonly used metrics and discussing more realistic evaluation strategies that incorporate operational costs and risk management considerations. This paper also provides a structured taxonomy of fraud detection methods, comparative analyses of commonly used datasets, and a synthesis of current research trends. Finally, open challenges and promising research directions are identified, including adaptive learning systems, interpretable Artificial Intelligence models, graph-based behavioral modeling, and privacy-preserving collaborative fraud detection frameworks. Full article