Search Results (41,723)

Convolution is spatially fixed and channel-specific, whereas involution is location-specific and channel-agnostic, capturing spatially varying patterns efficiently. Existing involutional networks, however, generate point-estimate kernels and expose no native measure of where the operator is uncertain, while prevailing uncertainty and calibration methods act on the network output rather than on the aggregation operator itself. We propose Evidential Involution–Attention (EvIA) networks, which recast involution as a distributional operator whose per-location neighborhood aggregation is a Dirichlet distribution. This yields, in a single forward pass and at the same parameter cost as involution, a closed-form epistemic-uncertainty (vacuity) map. The vacuity drives a parameter-free, precision-weighted gate that fuses the local involution branch with a global branch, instantiated as windowed self-attention (EvIA-W) or lightweight channel attention (EvIA-C). A lemma and three propositions establish that normalized involution is the infinite-evidence limit of the operator, that convex aggregation makes it non-expansive, that the Dirichlet strength is the precision of the aggregated feature, and that the gate is the minimum-variance unbiased fusion of the two branches. An evidential head trained with a differentiable calibration objective produces reliable confidences. Over five seeds with paired tests on brain magnetic resonance imaging, chest radiography, and dermatoscopy, the windowed variant EvIA-W is significantly stronger on brain MRI, attaining 0.803 ± 0.026 accuracy against 0.714 ± 0.019 (p = 0.006) by EvIA-C, and reducing the area under the risk–coverage curve from 0.187 to 0.092 (p = 0.001). Architecture-matched controls show that the discrimination gain on brain MRI comes from the global branch, while substituting evidential for standard involution leaves accuracy, calibration, and selective risk statistically unchanged, so the operator supplies its uncertainty machinery at no measurable cost. We further report that the spatial vacuity map does not localize input corruption, because the aggregation weights are invariant to the evidence scale and no objective term supervises it, an analysis that motivates the operator-level regularizers we define for future work. Full article

Laser-induced graphene (LIG) has great potential for multiple applications because of its large specific surface area, facile fabrication process, and tunable properties. However, abundant lattice defects severely degrade its conductivity. Herein, from an innovative perspective of precursor design, the poly-phenylamines (P-PAs) with improved solubility and strong light absorption were synthesized, which act as an intercalated polymer for graphene oxide (GO) nanosheets. On this basis, the composite precursors show remarkably enhanced photothermal conversion capability and a compact stacked structure. These bring a 60% reduction in I_D/I_G in LIG after laser irradiation. To explain the above phenomenon, an isolation effect induced by the compact stacking precursor is proposed based on experimental results. Furthermore, the cooperative effect between P-PAs and Fe²⁺ is introduced, and a fluffy LIG aerogel with the lowest I_D/I_G ratio of 0.17 is prepared, which is barely achievable in conventional LIGs. When the obtained graphene aerogel is compounded with PDMS, the as-prepared thermal-conductive composite gel reaches a thermal conductivity of 1.05 W·m⁻¹·K⁻¹ and an ultralow interfacial thermal resistance of 37.2 mm²·K·W⁻¹ under a low graphene loading of 3.3 wt%. This intercalation strategy in GO precursor supplies a new route for preparing high-quality LIGs and thermal-conductive gels, which show great application prospects in thermal management devices. Full article

Integrated transportation hubs are characterized by fast-varying and strongly coupled electricity, hydrogen-refueling, and thermal demands driven by traffic activities. To characterize their short-term dynamic interactions, this paper develops a compact system-level model of a core electricity–hydrogen–heat coupling subsystem comprising a PEM electrolyzer, a hydrogen storage tank, a fuel cell, and a thermal side. Power- and temperature-dependent off-design models are established for the PEM electrolyzer and fuel cell, while a lumped-parameter thermodynamic model with real-gas correction is developed for the hydrogen storage tank. The electrolyzer and fuel-cell models achieve calibration MAPEs of 0.39% and approximately 0.81%, respectively, against published reference data. Two typical disturbance scenarios are then investigated. Under a 30 kW electrical-load step, the grid-power deviation is reduced from a peak of approximately 29.4 kW to about 9.1 kW, while the hydrogen-refueling-demand disturbance produces only a minor thermal-side temperature variation. The results reveal distinct propagation magnitudes and time-scale characteristics across the electrical, hydrogen, and thermal domains. The proposed framework provides a compact and physically interpretable tool for short-term cross-domain dynamic analysis of integrated transportation hubs. Full article

This study analyzes changes in precipitation over the extended Brazilian region (10° N–40° S; 75° W–25° W) based on data from meteorological stations, reanalyses (ECMWF ERA5, NASA MERRA-2), the NOAA PREC reconstruction, and an ensemble of 32 CMIP6 models. According to weather station data and the ERA5 reanalysis for the period 1996–2025, a slight decrease in mean precipitation (P) is observed in the whole region (ΔP_{Weather station} = −44 mm/year, ΔP_ERA5 = −174 mm/year). A comparison of the periods 1996–2025 and 1940–1969 using CMIP6 ensemble also indicates a reduction in regional average precipitation (ΔP_Ave = −18 mm/year). The reanalyses show a decline in precipitation beginning approximately in the 1980s, accompanied by significant interannual and interdecadal variability associated with natural climate variability. Future projections under high radiative forcing scenarios (SSP3-7.0 and SSP5-8.5) indicate a continued decline in average precipitation in the study region throughout the 21st century. Low forcing scenarios (SSP1-2.6 and SSP2-4.5) suggest only weak changes in average precipitation over the period 2026–2099. However, the precipitation changes are spatially heterogeneous in the study region. According to the ERA5 reanalysis (1940–2025), precipitation has increased over the northern and southern parts of the Brazilian territory, while a decrease was observed in its central portion. All CMIP6 SSP ensemble projections considered (2026–2099) indicate a decrease in precipitation in the north and an increase in the south of the Brazilian region. Moreover, with increasing radiative forcing, the reduction in precipitation in the Amazon River basin and adjacent territories will intensify. Full article

To address the challenge of quantitatively identifying thermal-resistance contributions from multi-layer packaging structures within liquid-cooled insulated-gate bipolar transistor (IGBT) modules, this study establishes a three-dimensional conjugate heat-transfer model incorporating the chip, solder layer, direct-bonded copper (DBC) ceramic layer, substrate, elliptical pin-fin heat sink, and fluid domain. The IGBT and fast-recovery diode (FRD) power losses under typical motor controller operating conditions are modeled as volumetric heat sources applied to the chip region. Based on model validation, an internal thermal bottleneck evaluation method is proposed using inter-layer temperature-drop decomposition, introducing the thermal bottleneck number (BN) to quantify the temperature-drop contribution of each packaging layer along the target chip’s heat-dissipation path. Results show that in the baseline structure, the DBC ceramic layer is the critical internal thermal bottleneck, with a BN value of 11.05%. When the DBC thermal conductivity increases from 20 W/(m·K) to 80 W/(m·K), the maximum junction temperature of the IGBT decreases from 413.32 K to 396.79 K, and the BN drops from 11.05% to 3.69%. Conversely, when the DBC thickness increases from 0.20 mm to 0.50 mm, the maximum junction temperature rises from 405.45 K to 424.42 K. Further analysis reveals that after weakening the DBC thermal bottleneck, the relative temperature-drop contribution of the solder interface becomes increasingly apparent; notably, a central 20% low-conductivity defect in the solder layer raises the maximum junction temperature to 517.16 K, 103.84 K higher than under normal solder conditions. These findings provide valuable insights for identifying internal thermal bottlenecks and optimizing packaging structures in liquid-cooled IGBT modules. Full article

This paper presents the design, implementation, and evaluation of a fuzzy logic-based adaptive lighting strategy for Phaseolus vulgaris cultivation. The objective of this study was to integrate an image-derived plant growth indicator obtained through projected leaf-area estimation into an adaptive pulse-width modulation framework. The proposed system used two input variables: the error between the desired and measured Photosynthetic Photon Flux Density (PPFD) and the estimated projected leaf area. PPFD was indirectly estimated from illuminance measurements acquired using a BH1750 sensor (ROHM Semiconductor, Kyoto, Japan) Mostrar más líneas connected to an Arduino Uno microcontroller, while projected leaf area was obtained from RGB images captured with an Intel RealSense D435 camera and processed in MATLAB R2021a. The fuzzy inference system was implemented using the Mamdani approach and the Fuzzy Logic Designer application. The experimental results showed that the system regulated the PWM signal within a range of 17.00% to 87.85%, corresponding to a power range of 13.65 W to 34.58 W. The projected leaf-area indicator increased from 284.94 cm${}^2$ to 392.57 cm${}^2$ during the 21-day evaluation period, representing a relative increase of 37.77%. These findings provide preliminary evidence supporting the integration of image-derived projected leaf-area information, illuminance sensing, and fuzzy logic for adaptive lighting regulation in controlled-environment cultivation systems. Full article

Phosphonates in the ocean can serve as an alternative phosphorus (P) source for microorganisms when phosphate is scarce. Dinoflagellates cannot utilize phosphonates, but some associated bacteria can degrade these compounds and release phosphate. However, the community composition of these bacteria and their phosphonate degradation pathways remain poorly understood. In this study, the dinoflagellate Amphidinium carterae was cultured with 2-aminoethylphosphonic acid (2-AEP) as the exclusive P source. Metagenomic and genomic analyses were conducted to identify bacterial taxa and genes related to phosphonate utilization (phn genes). Specific strains were isolated from 2-AEP cultures to characterize phosphonate degradation gene clusters. Ten of the 20 most abundant genera were found to possess the genetic potential for phosphonate utilization, with Labrenzia, Phycocomes, and Pseudosulfitobacter as the dominant genera. The identified phn genes constituted multiple pathways, including the C-P lyase, PhnW-PhnX, and PhnW-PhnY-PhnA pathways, indicating that A. carterae-associated bacteria can utilize phosphonates through diverse mechanisms, with some strains even possessing more than one pathway. Five isolated bacterial strains were confirmed to be capable of degrading 2-AEP. These findings underscore the ecological significance of bacterial diversity and metabolic versatility in helping dinoflagellate hosts adapt to P scarcity, providing novel insights into bacteria-algae interactions and marine P cycling. Full article

Reliable estimation of humanoid robot body configuration from monocular RGB frames is useful for external monitoring in traffic, logistics, and service-robotics environments. However, direct transfer of human pose models is challenged by differences in body proportions, rigid surface appearance, joint morphology, and local texture, while robot-specific annotated data are often limited. This study presents a detection-guided framework for full-body 2D keypoint estimation from monocular video frames. A robot-specific detector first localises the target, and the detected box is expanded and normalised into a local region of interest (ROI) for keypoint recovery. Rather than treating the task as direct coordinate regression, the proposed integration uses ResNet18 feature extraction, convolutional block attention module (CBAM) refinement, heatmap-based keypoint representation, differentiable spatial to numerical transform (DSNT)-based continuous coordinate decoding, and skeleton-aware regularisation. A compact 13-keypoint annotation protocol is defined to describe the head, shoulders, elbows, hands, hips, knees, and feet. On the held-out test set, the proposed model reduces mean per-joint position error (MPJPE) from 27.98 px to 15.49 px relative to the ResNet18 direct-regression baseline and improves object keypoint similarity (OKS) from 0.80 to 0.96. Under the same 13-keypoint protocol, the proposed model also achieves a lower MPJPE than fine-tuned YOLOv8-Pose with detector-ROI input (18.85 px) and HRNet-W18 (30.82 px). Direct transfer of COCO-pretrained YOLOv8-Pose performs substantially worse, with an MPJPE of 201.50 px and an OKS of 0.28. These results support the effectiveness of robot-specific local spatial modelling and continuous coordinate recovery for monocular humanoid robot keypoint estimation under the evaluated limited-data setting. Full article

The recovery of bioactive compounds from industrial wastewaters is a key strategy for improving resource efficiency and reducing the environmental impact of high-load effluents. In this work, two protein-rich tannery wastewater streams with different origins and compositions, unhairing–liming wastewater and lime-fleshing wastewater, were comparatively evaluated as secondary raw materials for the production of amino acid-based protein hydrolysates with potential biostimulant activity. Both streams were characterised in terms of physicochemical composition, mineral content and amino acid profile, and subsequently subjected to a comparative sequential enzymatic hydrolysis approach using endo- and exo-proteolytic enzymes. Alcalase, followed by Pancreatin, was selected as the most effective enzymatic system for both streams, although different optimal enzyme loadings were required depending on the wastewater composition. For unhairing–liming wastewater, the optimal conditions were 1.5% Alcalase and 1.0% Pancreatin, yielding 70.76 ± 1.95% hydrolysate, 80.87 ± 1.98% protein recovery and 11.42 ± 1.03% free amino acids. For lime-fleshing wastewater, 0.7% Alcalase and 1.0% Pancreatin provided 98.42 ± 1.97% yield, 92.19 ± 1.92% protein recovery and 16.97 ± 0.98% free amino acids. The two hydrolysates showed differentiated amino acid profiles, reflecting the keratin- and collagen-derived nature of the original streams. Germination assays indicated a growth-stimulating effect of the hydrolysates, increasing seed growth by up to 20.7% for the unhairing–liming wastewater hydrolysate at 0.10% (w/v) and by up to 27.1% for the lime-fleshing wastewater hydrolysate at 0.07% (w/v). These results demonstrate that enzymatic hydrolysis can be an effective and environmentally friendly route for converting tannery wastewaters into value-added bio-based products for agricultural applications, while also highlighting the industrial relevance of adapting the enzymatic process to the origin, composition and protein accessibility of each wastewater stream. Full article

The Xianglin Sn polymetallic deposit, located on the northwestern margin of the Cuonadong dome in the eastern Tethyan Himalaya, represents an important Sn-W-Be polymetallic system recently recognized in southern Tibet. This study investigates the nature and evolution of the ore-forming fluids and the mechanisms of cassiterite precipitation. Fluid inclusion petrography, microthermometry, and laser Raman spectroscopy were applied to inclusions from the cassiterite–sulfide, proximal cassiterite–quartz vein, distal cassiterite–quartz vein, and late fluorite–quartz vein stages. The fluid inclusions mainly comprise liquid-rich, vapor-rich, H₂O-CO₂ three-phase, and minor daughter-mineral-bearing inclusions. The cassiterite–sulfide stage contains the most complex inclusion assemblage, with homogenization temperatures of 326–399 °C and salinities of 2.4–10.5 wt.% NaCl equiv. The proximal cassiterite–quartz vein stage yields homogenization temperatures of 280–361 °C and salinities of 2.4–9.6 wt.% NaCl equiv., whereas the distal cassiterite–quartz vein stage records lower temperatures of 195–295 °C. The fluorite–quartz vein stage is characterized by low-temperature and low-salinity fluids. Raman spectra identify CO₂, N₂, CH₄, H₂S, and CaCO₃ daughter minerals, indicating a volatile-rich, relatively reduced H₂O-CO₂-NaCl hydrothermal system. These features suggest that the principal ore-forming fluids were dominantly magmatic–hydrothermal fluids exsolved from highly fractionated Miocene leucogranites. During upward and outward migration from the main detachment zone to peripheral fractures, the fluids experienced boiling, cooling, mixing, and multistage hydrothermal overprinting. Cassiterite precipitation during the cassiterite–sulfide stage was mainly triggered by boiling or phase separation, whereas the proximal cassiterite–quartz veins were controlled by progressive cooling and local phase separation. In contrast, distal cassiterite precipitation was more strongly influenced by fluid mixing and cooling. These results indicate that the Xianglin deposit records a structurally controlled proximal-to-distal magmatic–hydrothermal Sn mineralizing system related to the Cuonadong dome. Full article

In this research, terbium-doped strontium tungsten oxide samples were synthesized using the solid-state synthesis method. It was found through X-ray powder diffraction (XRD) that the samples belong to the tetragonal crystal system and have the space group I4₁/a. The SEM-EDS images indicate that the Na, K, W, and Tb atoms are uniformly distributed throughout the samples. Photoluminescence (PL) measurements were applied for sample characterization. All of them exhibited green emission with the highest peak at 545 nm attributable to the ⁵D₄ → ⁷F₅ transition. The highest emission intensity was observed in the Sr_0.996Tb_0.004WO₄ sample, while higher doping levels resulted in a decrease in the PL intensities due to concentration quenching. Samples with additional alkali metal ions for charge compensation were also prepared, and their PL properties were measured; the enhancement of co-doping alkali metal (A⁺ = Li⁺, Na⁺, K⁺) on the green luminescence is demonstrated thoroughly. Doping with 0.5 mol% Na⁺ ions yielded a 9.14-fold increase in peak intensity of 545 nm, a 8.02-fold enhancement in operational lifespan, and a 88.82% improvement in quantum yield. These results indicate that Sr_0.991Tb_0.004Na_0.005WO₄ materials have potential application as green phosphors in LEDs. Full article

In this paper, the solar irradiance is investigated in a part of central Serbia, where a low-cost IoT sensor network was deployed at three locations: a mountain slope, an open field near a village, and an obstructed position in the city center. All three locations are in the same NASA POWER grid cell. After multi-stage quality control, 26,145 valid daytime records were compared to the satellite reference. The satellite assigns identical values to all three positions but the measured mean daytime irradiances are 267.7, 351.5 and 19.6 W/m², respectively. The rural station is in the best agreement with the reference (R² = 0.567); the mountain station suffers from a persistent shading bias (MBE = −138.9 W/m²); the signal at the urban station is attenuated by surrounding buildings and vegetation by a factor of ~12. Also, 25 regression models (gradient boosting, recurrent, convolutional, fully connected and graph-based) were trained on the 25-year monthly NASA POWER record for the same cell. XGBoost obtained R² = 0.998, the hybrid TCN-GNN R² = 0.968 and a plain ReLU network R² = 0.912, and a seasonal model with calendar features was used to reconstruct the satellite reference for two months of 2026 not yet available in the archive. The deployed hardware, network and energy behavior are documented quantitatively: per-site link delivery ratios of 96.8%, 83.5% and 58.2%, the photovoltaic harvesting record of the nodes, and the absence of any energy-aware transmission scheduling. Two of the trained models were compiled for the ESP32 nodes and measured on the deployment hardware: a depth-limited gradient-boosted corrector runs in 46.3 µs using 11.1 kB of flash, and an INT8 fully connected network in 138.1 µs using 3.0 kB, while the graph hybrid cannot be converted for microcontroller execution at all. This defines an edge–cloud partition in which local inference performs bias correction and fault detection while the cloud tier retains the heavy models and periodic retraining. To the best of the authors’ knowledge, this is the first multi-site ground-based IoT irradiance record for central Serbia with such different terrain types. Full article

This paper presents a comparative analysis of three magnetic-integrated LCL-derived filter configurations for suppressing high-frequency switching harmonics in avionics power-conversion systems: the integrated LLCL filter, the magnetic-integrated multi-trap LCL filter, and the magnetic-integrated trapped-LCL filter. The high-frequency harmonics considered in this study originate primarily from the pulse-width-modulated (PWM) switching process and can increase current distortion, electrical and magnetic losses, component stress, and the potential for electromagnetic interference in aircraft electrical systems. To enable a consistent comparison, the three filter configurations are evaluated within a unified analytical framework under common system-level operating conditions, while retaining their topology-specific passive and magnetic parameters according to their respective design requirements. The framework distinguishes aggregate total harmonic distortion (THD) from attenuation at targeted switching-frequency bands and incorporates the effects of resonant branches and magnetic coupling. The simulated current THD values of the LLCL, multi-trap LCL, and trapped-LCL filters are 0.82%, 0.84%, and 1.13%, respectively, while their estimated total filter losses are 24.98 W, 59.74 W, and 40.49 W. At the 1 kW operating point, these losses correspond to 2.50%, 5.97%, and 4.05% of rated power, respectively. The results show that the LLCL filter achieves the lowest aggregate THD and estimated loss under the investigated conditions, whereas the multi-trap and trapped-LCL configurations provide additional capability for targeted attenuation of selected switching-frequency components through their resonant structures. Hardware-in-the-loop (HIL) results further support the electrical filtering and dynamic behavior of the investigated concepts. The comparison demonstrates that minimum THD, minimum loss, targeted switching-harmonic attenuation, and magnetic integration are distinct design objectives; therefore, topology selection should be based on the specific requirements and constraints of the intended avionics application. Full article

This study reports the sonocatalytic degradation of malachite green (MG) in media using synthesized Co₃O₄ nanoparticles and a CTAB-modified zeolite supported Co₃O₄ hybrid nanocatalyst (M-ZLT@Co₃O₄). The catalysts were characterized by XRD, FT-IR, SEM-EDX, TEM, XRF, N₂ adsorption–desorption, BET, and BJH analyses, verifying crystalline Co₃O₄ formation, successful immobilization on the modified zeolite support, porous architecture, morphology and elemental distribution. Under optimized conditions of 1 g L⁻¹ catalyst dosage, 20 mg L⁻¹ MG concentration, natural pH, and 300 W L⁻¹ ultrasonic power, M-ZLT@Co₃O₄ achieved 99.31% ± 0.45% degradation within 120 min. Decreasing the MG concentration to 5 mg L⁻¹ increased the efficiency to 99.96% ± 0.64%, confirming high activity at low pollutant loading. The degradation followed pseudo-first-order kinetics, with k_app values of 0.0042–0.0595 min⁻¹ depending on catalyst dosage, 0.0046–0.0667 min⁻¹ for initial MG concentration, and 0.0117–0.0620 min⁻¹ under pH conditions. Water matrix inhibition followed distilled water (99.31%) > tap water (96.08%) > river water (88.84%) > lake water (86.90%), whereas higher ultrasonic power increased degradation from 91.98% at 200 W L⁻¹ to 99.90% at 500 W L⁻¹. Dye degradation followed MG (99.31%) > CR (95.78%) > BY2 (92.15%) > MV (90.24%) > AO7 (85.62%) > RB19 (82.49%). Comparative tests demonstrated performance for US/M-ZLT@Co₃O₄ (99.31%) compared with US/Co₃O₄ nanoparticles (78.48%), US/ZLT (72.86%), and US/MG alone (35.71%). The best activity was obtained at 0.8–2 g L⁻¹ catalyst dosage and 5–20 mg L⁻¹ MG concentration, while reusability tests showed a decrease from 99.31% to 78.82% after six cycles, indicating promising stability. Full article

Solar irradiance forecasting accuracy depends on the prediction horizon because the use of recent observations, meteorological variables, and seasonal patterns changes with lead time. However, weak persistence baselines, temporally unreliable validation, and inconsistent test samples can overstate the advantage of complex models. This paper presents a leakage-safe, horizon-specific solar irradiance forecasting framework combining boosted trees, validation-weighted convex forecast combinations, and seasonal reference models. Direct forecasts are evaluated at 10-minute and 1-, 3-, 6-, 12-, and 24-hour horizons. Models are tuned using expanding-window validation; preprocessing is fitted only to the training data; and every method is evaluated on a predefined canonical test support. XGBoost achieves the lowest root mean square error (RMSE), 10.78 W/m${}^2$, at 10 min, whereas CatBoost achieves the lowest RMSE, 16.08 W/m${}^2$, at 1 h. At 3, 12, and 24 h, the lowest RMSE is obtained by seasonally guided forecast combinations. At 6 h, the XGBoost and two-day seasonal-mean combination ranks first by RMSE, although its gain over the two-day seasonal mean results in a larger mean absolute error (MAE). Daily-block significance tests show an improvement over the reference models at 10 min, while improvements over seasonal references at longer horizons are not statistically significant after adjustment. The framework provides a reproducible benchmark for horizon-dependent solar irradiance forecasting. Full article