Search Results (4,318)

Fatigue crack propagation in friction stir welded joints significantly affects aircraft structural integrity. This study investigates the influence of welding speed, rotational speed, specimen thickness, loading frequency, and stress ratio on the fatigue crack growth rate. Four classical machine learning models with different structures—Deep Back-Propagation Network, Random Forest, Support Vector Regression, and K-Nearest Neighbors—were employed to predict fatigue crack growth behavior. The results show that all models achieve strong predictive performance. For FSWed joints, Deep BP and KNN exhibit comparable performance (R² > 0.98) on the training data, indicating similar learning capabilities with sufficient data coverage. Notably, KNN achieves the fastest training time (<0.3 s), while all models require less than 5 s of computation time. These results demonstrate that machine learning-based models provide an efficient and reliable alternative for rapid fatigue crack growth evaluation, supporting damage-tolerant design and structural integrity assessment in aircraft engineering. Full article

This study introduced the Digital Image Correlation (DIC) technique into the axial tensile and compression tests of ice materials. The surface strain distribution measured by DIC was compared with experimental phenomena to verify the accuracy of DIC measurement technology. Additionally, the strain data obtained from DIC were used to correct the stress–strain rate curves of ice materials under axial tension and compression, as measured by the universal testing machine. The study found that the constitutive relationship of a type of ice material under tension and compression can be fitted to a bi-linear model. After correction, the bi-linear two-stage moduli of the ice specimens frozen at −30 °C during tensile testing were approximately ${\overline{E}}_1$ = 687.50 MPa and ${\overline{E}}_2$ = 1.12 GPa; During compression, the bi-linear two-stage moduli are approximately ${\overline{E}}_1^{\prime }$ = 1.521 GPa and ${\overline{E}}_2^{\prime }$ = 7.734 GPa. The above research results are similar to those of previous studies and have a high degree of credibility. The mechanical properties of ice materials were found to be more stable at a freezing temperature of −30 °C compared to −10 °C. When microcracks form in ice materials under load, these cracks may refreeze internally, leading to viscoelastic behavior in the early stages of loading. Full article

Climate change affects surface water quality parameters, including river quality. This study analyses changes in climate parameters, specifically air temperature and solar radiation, and their impact on river water temperature. It also examines how changes in river water temperature and organic matter load affect oxygen saturation levels, a key indicator of river water quality. Using water quality data, the status as well as temporal and spatial trends of the analysed parameters were assessed for the period between 2007 and 2024 on the three largest Slovenian rivers: the Drava, Mura, and Sava. Relative importance analysis of temperature and biochemical oxygen demand (BOD) using the Random Forest machine learning method showed that water temperature in the analysed rivers has an impact ranging from 51% to 66% on predicting oxygen saturation. The selected approach to analysing watercourse quality parameters enables the assessment of the impact of these parameters on river water quality. Based on these results, it will be possible to implement appropriate measures promptly to achieve sustainable river management by establishing a strategy that, under climate change conditions, safeguards water quality and maintains ecosystem protection, ensuring long-term ecological and socio-economic benefits. Full article

Background: Orientation discrimination tasks provide a core measure of visual sensitivity and are widely used to study how perceptual performance varies with stimulus uncertainty and visual field location. Here, we examined how external noise, retinal eccentricity, and individual perceptual efficiency shape orientation discrimination thresholds. Methods: Forty-two adults (mean age = 32.35 years, SD = 7.23) completed a two-alternative forced-choice task judging the orientation (clockwise vs. counterclockwise) of briefly presented Gabor patches under varying levels of external noise (low, medium, high) and eccentricity (0°, 5°, 10°). Orientation offsets ranged from −8° to +8°. Thresholds were estimated using psychometric functions and analyzed via rm ANOVA, linear mixed-effects models, and supervised machine learning. Results: Accuracy declined with increasing noise (ω² = 0.48, p < 0.001) and improved with larger orientation offsets (ω² = 0.62, p < 0.001). Thresholds increased with both noise (ω² = 0.31, p = 0.002) and eccentricity (ω² = 0.27, p = 0.003). Signal-to-noise efficiency was the strongest predictor (β = −0.72, p < 0.001); age alone was nonsignificant, but its interaction with eccentricity showed selective peripheral declines. Mixed-effects models confirmed spatial effects (β = 0.058, p < 0.001) and residual between-subject variability (σ² = 0.14). Predictive models generalized well (R² = 0.54). Conclusions: Orientation discrimination is shaped by both stimulus-level difficulty and individual differences in perceptual efficiency, which account for variability in sensitivity across visual conditions. Age-related differences emerge primarily under spatial load and depend on interactions between observer traits and task demands. Full article

The growing demand for electricity in residential sectors and the global need to decarbonize power systems are accelerating the transformation toward smart and sustainable energy networks. Smart homes and smart grids, integrating renewable generation, energy storage, and intelligent control systems, represent a crucial step toward achieving energy efficiency and carbon neutrality. However, ensuring real-time optimization, interoperability, and sustainability across these distributed energy resources (DERs) remains a key challenge. This paper presents a comprehensive review of artificial intelligence (AI) applications for sustainable energy management and low-carbon technology integration in smart grids and smart homes. The review explores how AI-driven techniques include machine learning, deep learning, and bio-inspired optimization algorithms such as particle swarm optimization (PSO), whale optimization algorithm (WOA), and cuckoo optimization algorithm (COA) enhance forecasting, adaptive scheduling, and real-time energy optimization. These techniques have shown significant potential in improving demand-side management, dynamic load balancing, and renewable energy utilization efficiency. Moreover, AI-based home energy management systems (HEMSs) enable predictive control and seamless coordination between grid operations and distributed generation. This review also discusses current barriers, including data heterogeneity, computational overhead, and the lack of standardized integration frameworks. Future directions highlight the need for lightweight, scalable, and explainable AI models that support decentralized decision-making in cyber-physical energy systems. Overall, this paper emphasizes the transformative role of AI in enabling sustainable, flexible, and intelligent power management across smart residential and grid-level systems, supporting global energy transition goals and contributing to the realization of carbon-neutral communities. Full article

A new approach to end-to-end power delivery for increasingly sought-after hierarchical power delivery units (PDUs) is presented, improving the power efficiency of portable systems. The benefits of the technique are demonstrated through a PDU comprising multiple DC–DC converters, such as low-dropout regulators (LDOs), and the support of heterogeneous loads. A properly tailored Q-algorithm is combined with power gating to manage the power supplied by a multi-level PDU. The effectiveness of the proposed method is evaluated via a realistic PDU for different combinations of loads. The learning-based technique yields up to 13% higher total end-to-end power efficiency in the case of similar loads by utilizing four available LDOs compared to the case of a single LDO, which supports the same span of loads. Moreover, the proposed method improves power efficiency by up to 5% in the case of heterogeneous loads when compared to other autonomous state-of-the-art power management units. Full article

Atmospheric deposition of emerging contaminants, including toxic trace elements, remains a critical environmental and public health concern. Moss biomonitoring offers a sensitive and cost-effective tool for assessing airborne pollutants, yet traditional analyses rely on descriptive statistics and lack predictive and mechanistic insights. Here, we introduce Mosses ML, a machine-learning-enhanced framework that integrates moss biomonitoring with bulk and dry deposition measurements to improve detection, interpretation, and risk assessment of atmospheric contaminants. Using Hylocomium splendens transplants exposed for 90 days across industrial, urban, and rural sites in Upper Silesia (Poland), we combined trace element accumulation (Cd, Pb, Zn, Ni, Cr, Fe), relative accumulation factors (RAFs), PCA-derived gradients, and site-level metadata with Random Forest and Gradient Boosting models. ML algorithms achieved high predictive performance (R² up to 0.91), accurately estimating moss metal concentrations from deposition metrics and environmental variables. SHAP feature-importance analysis identified dry deposition load and co-occurring metal signals as the dominant predictors of contamination, confirming the primary role of particulate emissions in shaping moss chemistry. Compared with classical threshold-based classification, the ML approach improved high-risk site identification by 24–38%. Mosses ML combines biologically meaningful indicators with modern computational tools, strengthening the role of mosses as early-warning systems for atmospheric pollution. The framework is broadly applicable to bryophyte biomonitoring and supports regulatory decision-making for emerging contaminants. Full article

Machine learning inference is increasingly deployed on shared and cloud infrastructures, where both user inputs and model parameters are highly sensitive. Confidential computing promises to protect these assets using Trusted Execution Environments (TEEs), yet existing TEE-based inference systems remain fundamentally constrained: they rely almost exclusively on low-level, memory-unsafe languages to enforce confinement, sacrificing developer productivity, portability, and access to modern ML ecosystems. At the same time, mainstream high-level runtimes, such as Python, are widely considered incompatible with enclave execution due to their large memory footprints and unsafe model-loading mechanisms that permit arbitrary code execution. To bridge this gap, we present the first Python-based ML inference system that executes entirely inside Intel SGX enclaves while safely supporting untrusted third-party models. Our design enforces standardized, declarative model representations (ONNX), eliminating deserialization-time code execution and confining model behavior through interpreter-mediated execution. The entire inference pipeline (including model loading, execution, and I/O) remains enclave-resident, with cryptographic protection and integrity verification throughout. Our experimental results show that Python incurs modest overheads for small models (≈$17\%$) and outperforms a low-level baseline on larger workloads (97% vs. 265% overhead), demonstrating that enclave-resident high-level runtimes can achieve competitive performances. Overall, our findings indicate that Python-based TEE inference is practical and secure, enabling the deployment of untrusted models with strong confidentiality and integrity guarantees while maintaining developer productivity and ecosystem advantages. Full article

The residential energy sector contributes substantially to global energy-related emissions. Effective energy management requires an understanding occupant behavior through activity detection and habit identification. Recent advances in artificial intelligence (AI) and machine learning (ML) enable the automatic detection of user activities and prediction of energy needs based on historical consumption data. Non-intrusive load monitoring (NILM) facilitates device-level disaggregation without additional sensors, supporting demand forecasting and behavior-aware control in Home Energy Management Systems (HEMSs). This review synthesizes various AI and ML approaches for detecting user activities and energy habits in HEMSs from 2020 to 2025. The analyses revealed that deep learning (DL) models, with their ability to capture complex temporal and nonlinear patterns in multisensor data, achieve superior accuracy in activity detection and load forecasting, with occupancy detection reaching 95–99% accuracy. Hybrid systems combining neural networks and optimization algorithms demonstrate enhanced robustness, but challenges remain in limited cross-building generalization, insufficient interpretability of deep models, and the absence of dataset standardized. Future work should prioritize lightweight, explainable edge-ready models, federated learning, and integration with digital twins and control systems. It should also extend energy optimization toward occupant wellbeing and grid flexibility, using standardized protocols and open datasets for ensuring trustworthy and sustainability. Full article

Non-Intrusive Load Monitoring is an effective method for disaggregating the power consumption of individual appliances from the aggregate load data of a building. The advent of smart meters, Internet of Things devices, and artificial intelligence technologies has significantly advanced the capabilities of non-intrusive load monitoring. However, challenges such as varying sampling frequencies and measurement sensitivities remain. This paper introduces an innovative model incorporating an Adaptive Multi-Scale Attention Integration Module (AMSAIM) to address these issues. The model leverages deep learning and attention mechanisms to improve the accuracy and real-time performance of non-intrusive load monitoring. Validated on the standard UK-DALE dataset, the model consistently demonstrated superior performance. In seen scenarios, our model achieved average F1-scores approximating 0.94 and notably reduced Mean Absolute Error (MAE) values. For washing machines, it achieved an F1-score of 0.99 and MAE of 41.64, outperforming the next best method’s F1-score by 1 percentage point. In challenging unseen scenarios, the model showcased strong generalization, achieving an F1-score of 0.91 for washing machines and reducing MAE to 7.66. Furthermore, an ablation study rigorously confirmed the necessity of the AMSAIM module, showing that the synergistic integration of the efficient multi-scale attention (EMA) and the selective kernel (SK) adaptive receptive field unit is crucial for enhancing model robustness and generalization. Our results highlight the model’s potential for enhancing energy efficiency and providing actionable insights for energy management across various conditions. Full article

Background: To compare the biomechanical strength of three fixation techniques for the elbow anterior capsule and coronoid process using a synthetic ulna model. We hypothesize that a cortical suture button would be equivalent to the bone tunnel model but inferior to a screw-post construct. Methods: A biomechanical study was conducted using a composite ulna bone model to simulate coronoid process fixation with three techniques: traditional trans-osseous bone tunnel repair, suspensory fixation using a cortical button, and a screw-post construct using a 3.5 mm cortical screw. All constructs were assembled using high-strength suture. Each specimen underwent axial loading on an Instron machine until failure, defined as loss of fixation through the dorsal cortex. Peak ultimate strength was recorded. Statistical analysis was performed using one-way ANOVA and Tukey’s HSD test. Results: The suture button construct demonstrated the highest mean ultimate strength at 490.3 ± 125.2 N, significantly greater than both the bone tunnel (328.8 ± 86.4 N, p < 0.01) and screw-post constructs (273.4 ± 54.5 N, p < 0.001). While the bone tunnel construct exhibited a 20.3% higher strength than the screw-post construct, this difference was not statistically significant (p = 0.13). The screw-post construct showed the least variability in strength to failure but the lowest overall strength. The suture button demonstrated the greatest mechanical strength but also the most variability. Conclusions: Suspensory fixation using a titanium cortical suture button provides significantly greater mechanical strength compared to traditional bone tunnel and screw-post techniques in a synthetic ulna model. While variability was greatest with the suture button construct, its superior load-bearing capacity suggests potential advantages in stabilizing the elbow through anterior capsule and coronoid fracture repair. These findings support further clinical investigation of suture button fixation as a viable technique in complex elbow injuries. Full article

The integration of Software-Defined Networking (SDN), blockchain (BC), and machine learning (ML) has emerged as a promising approach to securing Internet of Things (IoT) and Industrial IoT (IIoT) networks. This paper conducted a comprehensive review of recent studies focusing on multi-layered security across device, control, network, and application layers. The analysis reveals that BC technology ensures decentralised trust, immutability, and secure access validation, while SDN enables programmability, load balancing, and real-time monitoring. In addition, ML/deep learning (DL) techniques, including federated and hybrid learning, strengthen anomaly detection, predictive security, and adaptive mitigation. Reported evaluations show similar gains in detection accuracy, latency, throughput, and energy efficiency, with effective defence against threats, though differing experimental contexts limit direct comparison. It also shows that the solutions’ effectiveness depends on ecosystem factors such as SDN controllers, BC platforms, cryptographic protocols, and ML frameworks. However, most studies rely on simulations or small-scale testbeds, leaving large-scale and heterogeneous deployments unverified. Significant challenges include scalability, computational and energy overhead, dataset dependency, limited adversarial resilience, and the explainability of ML-driven decisions. Based on the findings, future research should focus on lightweight consensus mechanisms for constrained devices, privacy-preserving ML/DL, and cross-layer adversarial-resilient frameworks. Advancing these directions will be important in achieving scalable, interoperable, and trustworthy SDN-IoT/IIoT security solutions. Full article

Modern power transmission systems are required to meet increasingly stringent demands, including a wide range of transmission ratios, compact dimensions, high precision, energy efficiency, reliability, and thermal stability under dynamic operating conditions. Among the solutions that satisfy these requirements, cycloidal reducers are particularly prominent, with their application continuously expanding in industrial robotics, computer numerical control (CNC) machines, and military and transportation systems, as well as in the satellite industry. However, as with all mechanical power transmissions, friction in the contact zones of load-carrying elements in cycloidal reducers leads to power losses and an increase in operating temperature, which in turn results in a range of adverse effects. These undesirable phenomena strongly depend on lubrication conditions, namely on the type and properties of the applied lubricant. Although manufacturers’ catalogs provide general recommendations for lubricant selection, they do not address the fundamental tribological mechanisms in the most heavily loaded contact pairs. At the same time, the available scientific literature reveals a significant lack of systematic and experimentally validated studies examining the influence of lubricant type on the energetic and thermal performance of cycloidal reducers. To address this identified research gap, this study presents an analytical and experimental investigation of the effects of different lubricant types—primarily greases and mineral oils—on the thermal stability and efficiency of cycloidal reducers. The results demonstrate that grease lubrication provides lower total power losses and a more stable thermal operating regime compared to oil lubrication, while oil film thickness analyses indicate that the most unfavorable lubrication conditions occur in the contact between the eccentric bearing rollers and the outer raceway. These findings provide valuable guidelines for engineers involved in cycloidal reducer design and lubricant selection under specific operating conditions, as well as deeper insight into the lubricant behavior mechanisms within critical contact zones. Full article

Background: Solid organ transplantation (SOT) is a life-saving treatment for patients with end-stage diseases and/or organ failure. However, access to healthy organs is often limited by challenges in organ preservation. Furthermore, upon transplantation, ischemia–reperfusion injury (IRI) can lead to increased organ rejection or graft failures. The work presented aims to address both challenges using an innovative nanomedicine platform for simultaneous drug and oxygen delivery. In recent studies, resveratrol (RSV), a natural antioxidant, anti-inflammatory, and reactive oxygen species (ROS) scavenging agent, has been reported to protect against IRI by inhibiting ferroptosis. Here, we report the design, development, and scalable manufacturing of the first-in-class dual-function perfluorocarbon-nanoemulsion (PFC-NE) perfusate for simultaneous oxygen and antioxidant delivery, equipped with a near-infrared fluorescence (NIRF) reporter, longitudinal, non-invasive NIRF imaging of perfusate flow through organs/tissues during machine perfusion. Methods: A Quality-by-Design (QbD)-guided optimization was used to formulate a triphasic PFC-NE with 30% w/v perfluorooctyl bromide (PFOB). Drug-free perfluorocarbon nanoemulsions (DF-NEs) and RSV-loaded nanoemulsions (RSV-NEs) were produced at 250–1000 mL scales using M110S, LM20, and M110P microfluidizers. Colloidal attributes, fluorescence stability, drug loading, and RSV release were evaluated using DLS, NIRF imaging, and HPLC, respectively. PFC-NE oxygen loading and release kinetics were evaluated during perfusion through the BMI OrganBank^® machine with the MEDOS HILITE^® oxygenator and by controlled flow of oxygen. The in vitro antioxidant activity of RSV-NE was measured using the oxygen radical scavenging antioxidant capacity (ORAC) assay. The cytotoxicity and ferroptosis inhibition of RSV-NE were evaluated in RAW 264.7 macrophages. Results: PFC-NE batches maintained a consistent droplet size (90–110 nm) and low polydispersity index (<0.3) across all scales, with high reproducibility and >80% PFOB loading. Both DF-NE and RSV-NE maintained colloidal and fluorescence stability under centrifugation, serum exposure at body temperature, filtration, 3-month storage, and oxygenation. Furthermore, RSV-NE showed high drug loading and sustained release (63.37 ± 2.48% at day 5) compared with the rapid release observed in free RSV solution. In perfusion studies, the oxygenation capacity of PFC-NE consistently exceeded that of University of Wisconsin (UW) solution and demonstrated stable, linear gas responsiveness across flow rates and FiO₂ (fraction of inspired oxygen) inputs. RSV-NE displayed strong antioxidant activity and concentration-dependent inhibition of free radicals. RSV-NE maintained higher cell viability and prevented RAS-selective lethal compound 3 (RSL3)-induced ferroptosis in murine macrophages (macrophage cell line RAW 264.7), compared to the free RSV solution. Morphological and functional protection against RSL3-induced ferroptosis was confirmed microscopically. Conclusions: This study establishes a robust and scalable PFC-NE platform integrating antioxidant and oxygen delivery, along with NIRF-based non-invasive live monitoring of organ perfusion during machine-supported preservation. These combined features position PFC-NE as a promising next-generation acellular perfusate for preventing IRI and improving graft viability during ex vivo machine perfusion. Full article

This study investigates the design of bio-hybrid sandwich composites by combining polyester/glass skins with cores reinforced by continuous Musa paradisiaca fibres. The aim is to quantify how fibre weight fraction and alkaline surface treatment control the Charpy impact performance of these systems. Sandwich laminates were manufactured with three fibre loadings in the core (20, 25 and 30 wt.%), using fibres in the as-received condition and after alkaline treatment in NaOH solution. Charpy impact specimens were machined from the laminates and tested according to ISO 179-1. Fibre morphology and fracture surfaces were examined by scanning electron microscopy, while Fourier-transform infrared spectroscopy was used to monitor changes in surface chemistry after alkaline treatment. The combined effect of fibre content and treatment on absorbed energy was assessed through a two-way analysis of variance. Increasing Musa paradisiaca fibre content up to 30 wt.% enhanced the impact energy of the sandwich composites, and alkaline treatment further improved performance by strengthening fibre–matrix adhesion and promoting fibre pull-out, crack deflection and bridging mechanisms. The best Charpy impact response was obtained for cores containing 30 wt.% NaOH-treated fibres, demonstrating that surface modification and optimised fibre loading are effective design parameters for toughening polyester/glass bio-hybrid sandwich composites. Full article