Search Results (15,018)

The increasing penetration of inverter-interfaced photovoltaic (PV) generation in active distribution networks (ADNs) intensifies fast voltage violations and makes real-time Volt-VAR control (VVC) challenging, especially when each inverter has only partial and noisy measurements and communication is limited. Existing local droop-type strategies lack coordination, while fully centralized optimization/learning is often impractical for online deployment. To address these gaps, an attention-enhanced multi-agent deep reinforcement learning (MADRL) framework is developed for inverter-based VVC under the centralized training and decentralized execution (CTDE) paradigm. First, the voltage regulation problem is formulated as a decentralized partially observable Markov decision process (Dec-POMDP) to explicitly account for system stochasticity and temporal variability under partial observability. To solve this complex game, an attention-enhanced MADRL architecture is employed, where an agent-level attention mechanism is integrated into the centralized critic. Unlike traditional methods that treat all neighbor information equally, the proposed mechanism enables each inverter agent to dynamically prioritize and selectively focus on the most influential states from other agents, effectively capturing complex intercorrelations while enhancing training stability and learning efficiency. Operating under the CTDE paradigm, the framework realizes coordinated reactive power support using only local measurements, ensuring high scalability and practical implementability in communication-constrained environments. Simulations on the IEEE 33-bus system with six PV inverters show that the proposed method reduces the average voltage deviation on the test set from 0.0117 p.u. (droop control) and 0.0112 p.u. (MADDPG) to 0.0074 p.u., while maintaining millisecond-level execution time comparable to other MADRL baselines. Scalability tests with up to 12 agents further demonstrate robust performance of the proposed method under higher PV penetration. Full article

Object detection is a fundamental task in computer vision. Bounding box regression (BBR) losses are critical to detector performance. However, evaluation measures that rely on the Intersection over Union (IoU) between the predicted and ground truth boxes are highly sensitive to positional deviations, which can hinder optimization. To alleviate this issue, we propose ATDIoU, a novel arctangent-differential loss for bounding-box regression. ATDIoU computes distance similarity between a predicted and a ground truth box by modeling the distances between their corresponding vertices as a two-dimensional arctangent differential distribution (ATD). This arctangent differential-based design mitigates bounding box drift and reduces sensitivity to localization errors. As a result, it guides the model to learn target positions more effectively. We evaluate ATDIoU by integrating it into YOLOv6 and conducting experiments on PASCAL VOC and VisDrone2019. The results demonstrate that ATDIoU yields improvements of 1.4% and 0.7% in mean average precision (mAP) relative to MPDIoU. Full article

Ultrafine hematite particles (<10 μm), commonly generated in beneficiation circuits, exhibit poor flocculation and slow settling, posing challenges for solid–liquid separation. This study investigates the influence of the anionic polyacrylamide (APAM) molecular weight on ultrafine hematite flocculation under controlled laboratory conditions, combining macroscopic experiments with molecular dynamics simulations (MDSs). Sedimentation tests show that the APAM molecular weight strongly affects settling kinetics, supernatant clarity, and floc structure, with the settling rate, flocculation-stage reaction time, supernatant turbidity, and underflow concentration exhibiting a non-monotonic trend and optimal performance at seven million. Under this condition, particles aggregate most efficiently, achieving a turbidity of 182 NTU, an underflow concentration of 51.5%, and the largest compact flocs, averaging 379.8 μm with a fractal dimension of 1.71. Higher molecular weights (≥9 million) induce chain coiling, reduce floc compactness, increase water retention, and impair settling. MDS indicates that polymer–surface interactions improve with an increasing polymerisation degree only up to an intermediate chain length; a polymerisation degree of 30 exhibits the most favourable extended–flexible conformation, maximal surface enrichment, strongest coordination between carboxyl groups and surface Fe atoms, lowest adsorption energy, and fastest adsorption kinetics. The functional-group distribution and hydrogen-bond analyses show that –NH₂ and –COO^– groups dominate interfacial interactions, with a polymerisation degree of 30 yielding the highest density of interfacial hydrogen bonds. By correlating macroscopic experiments with molecular-scale observations, this work provides mechanistic insight into how the APAM chain length governs ultrafine hematite flocculation, highlighting the role of polymer conformation and multipoint adsorption in controlling the settling performance. Full article

Effective electrochemical CO₂ reduction to liquid fuels requires that the local catalytic environment facilitates the desired reactivity, yet a microscopic understanding of this environment is difficult to achieve from experiment alone. In this work, a 3D reaction-diffusion model was developed to explore the effects of electrode surface area and local geometry on the performance of a heterogeneous catalyst that performs a two-step CO₂ reduction cascade reaction to CO and then CH₃OH under aqueous conditions. Kinetic parameters for the model were inspired by experimental results using a cobalt phthalocyanine (CoPc) catalyst. Three-dimensional architectures composed of arrays of square pillars with varying dimensions and either smooth or periodically modulated surfaces were tested, revealing the extent to which geometry modulates the performance of the cascade reactions. Although structural variations modulate local concentration gradients, we find that electrochemically active surface area predominantly governs the overall cascade reaction. Moreover, the results suggest that supersaturation of CO, with concentrations up to ten-fold higher than the equilibrium solubility limit, might be critical for more efficient conversion to CH₃OH. For any given geometry, the spatially averaged ratio of [CO] to [CO₂] is dictated by the electrochemically active surface area and determines the yield of CH₃OH. For a fixed surface area, geometries that spatially confine the electrolyte yield moderate local [CO] to [CO₂] ratios within small volumes. In contrast, less confining geometries result in a broader distribution of local ratios spread over larger volumes, with both configurations yielding the same spatially averaged [CO] to [CO₂] ratio. These insights provide valuable design principles—highlighting the critical importance of surface area and possibly CO supersaturation—for engineering advanced electrode architectures that leverage intermediate trapping and CO supersaturation to enhance overall performance in tandem CO₂ reduction systems. Full article

Forefoot offloading footwear is widely used in postoperative care, trauma management, and the prevention of diabetic foot ulceration, where redistribution of plantar load must be achieved without compromising gait stability. This study evaluated plantar pressure and centre of pressure characteristics of a new side-specific forefoot offloading footwear design in comparison with commonly used clinical and retail footwear. Twelve healthy adults completed treadmill walking trials at 4.0 km/h under five footwear conditions. Plantar pressure data were collected using an in-shoe pressure measurement system and analysed for peak pressure, average pressure, force–time impulse, centre of pressure velocity, and centre of pressure excursion index across seven anatomically defined plantar regions. Across all conditions, consistent left–right asymmetry in plantar loading was observed, although overall variability between footwear designs was modest. The experimental footwear demonstrated pressure and impulse distributions comparable to retail and universal offloading footwear, without increasing hallux loading. Centre of pressure measures were generally consistent between side variability, indicating controlled rollover and preserved gait stability. These findings suggest that side-specific sole geometry can support balanced forefoot load management without introducing instability in healthy walking and provide a foundation for future bilateral testing in clinical populations at risk. Full article

Existing hyperspectral denoising networks typically rely on large amounts of high-quality paired noisy–clean images for training, which are often unavailable. Moreover, the noise distribution in real hyperspectral images (HSIs) is complex and variable, making it challenging for existing networks to handle noise distributions not present in the training dataset, resulting in poor generalization. To address these issues, this paper proposes an unsupervised Hyperspectral image Denoising approach exploiting the spectral learning preference of neural networks with an adaptive early stopping strategy (termed HyDePre). Inspired by the Deep Image Prior, which reveals that neural networks tend to capture natural image structures before fitting noise, we observe that deep neural networks exhibit a similar learning preference in the spectral domain. Specifically, as training progresses, the network first fits smooth spectral feature curves and only later adapts to Gaussian noise and complex impulse noise. This observation provides an opportunity to use an early stopping strategy, allowing the network to fit only the clean spectral signals and thus achieve denoising. Our method does not require clean images for training, but instead optimizes network parameters to automatically learn prior spectral information from a single noisy image, modeling the intrinsic structure of the input data to uncover its underlying patterns.However, finding the optimal stopping point is challenging without access to clean images as sources of prior information. To tackle this challenge, we introduce an adaptive early stopping strategy based on the average spectral maximum variation of the reconstructed image, effectively preventing overfitting. The experimental results demonstrate that HyDePre outperforms existing methods in terms of both visual quality and quantitative metrics. Full article

The North China Craton (NCC) hosts numerous world-class Au deposits and these Au deposits can be classified into the Au-only and Ag-Au polymetallics, respectively. The former is mostly located in the eastern NCC, such as in the giant Jiaodong Province, and the latter is mostly distributed along the northern and southern margins of the NCC. Compared with Au-only deposits, the ore genesis of the Ag-Au deposits remains controversial. This paper focuses on the Niujuan Ag-Au deposit in the Fengning ore cluster of the northern margin of the NCC. Detailed deposit geology investigation, texture analysis, and analyses of the in situ trace element and sulfur isotope compositions of pyrite, coupled with H-O isotope compositions of quartz from different stages, were conducted to elucidate the ore-forming processes and metal sources. The results showed that the formation of the Niujuan deposit can be divided into four stages, including a pre-ore siliceous breccia stage (stage 1), syn-ore quartz-pyrite stage (stage 2), syn-ore polymetallic sulfide stage (stage 3), and post-ore fluorite-calcite stage (stage 4). Among these, stage 3 represents the major Ag-Au mineralization stage. Pyrite is well developed within stage 2 and stage 3, representing the intensive sulfidation of the wall rock. Microscopic analytical techniques including gamma-enhanced reflected light and scanning electron microscopy backscattered electron (BSE) reveal that pyrite samples from stage 2 and stage 3 have distinct textures. Pyrite (Py1) from stage 2 is homogeneous but with numerous pores. In contrast, pyrite (Py2) from stage 3 has overgrowth textures, and be divided into three sub-stages from core to rim (Py2a, Py2b, and Py2c) with different BSE brightness levels. LA-ICP-MS trace elements analyses results show that these different stages of pyrite show different composition such as Au, As, Ag, Co, and Ni. Py1 has low Au and Ag concentrations ranging from < 0.1 ppm to 0.02 ppm and < 0.1 ppm to 21.8 ppm, respectively. Py2a has low Au and Ag concentrations ranging from < 0.1 ppm to 0.4 ppm and 0.4 ppm to 118.4 ppm, respectively. Py2b is characterized by high As and low Au contents, with average values of 6670.8 ppm for As and 1.4 ppm for Au. Py2c shows relatively low Co and Ni concentrations ranging from 0.02 ppm to 255.2 ppm and < 0.1 ppm to 9.9 ppm, respectively. The sulfur isotope composition of Py1 and Py2 is relatively consistent, ranging from 3.8‰ to 6.7‰. The H and O isotope compositions of quartz from stage 1, stage 2, and stage 3 have insignificant variations, ranging from −119.5‰ to −101.3‰ for δD and −6.8‰ to −3.7‰ for δ¹⁸O_fluid, respectively. The results show that sulfur and, possibly, Au and Ag were mainly derived from magmatic hydrothermal fluids, and a significant amount of meteoric water was involved. Combined with the published mineralizing ages (~140 Ma), this paper suggests that the Niujuan Ag-Au deposit formed during the Early Cretaceous under an extensional setting in response to the eastward retreating subduction of the Paleo-Pacific oceanic plate. Evidence from deposit geology and geochemistry reveals that the mixture of magmatic and meteoric water, together with intensive sulfidation, is the key factor controlling Au and Ag deposition. Full article

Time series data find extensive applications in finance, healthcare, and industrial monitoring domains. However, analytical models targeting such data are subject to notable constraints imposed by the rigid independent and identically distributed (IID) assumption and the high cost of data annotation. Unsupervised Domain Adaptation (UDA) offers an effective remedy for these challenges, and Contrastive Learning (CL) has been widely integrated into UDA frameworks, owing to its robust feature representation and clustering capabilities. Nonetheless, existing CL-based UDA methods suffer from two key limitations: (1) fixed data augmentation strategies result in imbalanced intensity—excessive augmentation erodes sample semantics, while insufficient augmentation induces model overfitting; (2) distribution alignment strategies neglect hard samples which are the core carriers of domain shift, causing their domain adaptation signals to be overshadowed by a large number of normal samples and thus degrading alignment accuracy. To address these drawbacks, this paper proposes a time-series UDA algorithm, termed Adaptive Contrastive Learning Domain Adaptation (ACLDA), which incorporates two key components: (1) an adaptive feature enhancement module that integrates adaptive sample augmentation and CL, enabling the model to capture high-quality transferable features; (2) sample-level adaptive weights, introduced on the basis of class-level alignment via supervised CL, to emphasize the value of hard samples. Comparative experiments on multiple time-series datasets demonstrate that our ACLDA outperforms state-of-the-art domain adaptation methods in terms of average accuracy, verifying its superiority and providing a more robust solution for cross-domain time series analysis. Full article

To address the challenges of separating roots from soil and the high soil carryover during harvesting of Ophiopogon japonicus in heavy clay soils, a variable-gap tooth roller chain-rod-and-slat separation device was designed, integrating variable-gap tooth roller soil-crushing with vibrating chain-rod-and-slat conveying and separation functions. A coupled “soil–plant–equipment” model was established using the discrete element method. Conveying speed, vibration frequency, and amplitude were selected as key operational parameters. Interaction effects were analyzed, and dual-objective optimization was performed using response surface methodology. The contact number was used to characterize soil–plant particle adhesion, whereas D80 (the distance corresponding to 80% soil fallout) represented the spatial distribution of soil fallout. Optimization results indicate that, within the experimental parameter range, a combination yielding low contact number and low D80 is achievable. The simulations predicted a D80 of 563.25 mm and a contact number of approximately 6. Conversion of particle-mass data indicated the average soil mass adhering to plants is about 0.0096 kg. Field validation tests conducted at a conveying speed of 0.80 m/s, vibration frequency of 12.00 Hz, and amplitude of 15.00 mm yielded an average soil mass carried by separated plants of 0.012 kg. These results demonstrated that the constructed discrete element model and response surface optimization can be applied to parameter matching for Ophiopogon japonicus root–soil separation equipment, providing a reference for optimizing root–soil separation machinery in hilly and mountainous regions for Chinese medicinal herbs. Full article

This study takes 18CrNiMo7-6 wind power gear steel as the object. Following the first holding stage of isothermal normalizing, the 18CrNiMo7-6 wind power gear blanks were cooled to the isothermal temperature via air cooling (AC) and forced-air cooling (FA), respectively. The influence of cooling rate on the roughness of the mechanically polished surface of wind power gear blanks was comprehensively studied by means of white light interference, EBSD, TEM, DSC and other technical characterization methods. The results show that a difference in cooling rate leads to a variation in the morphology and distribution of Cr-rich carbides (mainly Cr₇C₃), which affects the roughness of the mechanically polished surface. During air cooling (slow cooling), atoms diffuse fully. Owing to the relatively low cooling rate in the inner ring of the blank, C and Cr segregate, and abundant Cr-rich carbides precipitated and accumulated at grain boundaries, forming coarse blocky structures. This resulted in uneven mechanically polished surfaces and bright spot defects. The average roughness of the inner and outer ring is 2.648 nm and 2.096 nm, respectively. Forced-air cooling (fast cooling) eliminates surface quality defects by inhibiting long-range atomic diffusion. Meanwhile, radial elemental segregation in the original cast blanks was inherited in subsequent processes, which affected the uniformity of carbide precipitation during cooling. In addition, the differences in cooling rates will also cause variations in the precipitation temperatures of carbides in steel, which in turn further affects the homogenization distribution of carbides in steel. This research provides a theoretical basis and an optimization method for the microstructural regulation and surface quality enhancement of wind power gear steel. Full article

Accurate monitoring of soybean seedlings in the field is a core component for implementing scientific management during the seedling stage and unlocking yield potential. Traditional manual survey methods are inefficient and highly subjective, making them inadequate for real-time assessment at the field scale. This study addresses challenges such as the small size of individual seedlings, dense inter-plant overlap, blurred boundaries, and complex interferences from soil residue and varying illumination by proposing a high-precision method for soybean seedling instance segmentation and georeferenced localization based on low-altitude (12 m) Unmanned Aerial Vehicle (UAV) imagery. By implementing targeted improvements in the YOLOv11n-seg model, we developed the YOLOv11-seg-SSC model, which integrates the SCSA (Shared Cross-Semantic Space and Progressive Channel Self-Attention) mechanism, the Context-Guided (CG) Block, and a lightweight Slim-Neck structure based on GSConv and VoV-GSCSP. While significantly reducing computational complexity (approximately 9.5 GFLOPs and 2.96 M parameters), the model improved the mean average precision for segmentation (mAP@0.5 Mask) from the baseline of 80.6% to 83.3%, maintained a stable detection mAP@0.5 (Box) at 95.9%, and achieved an overall segmentation precision of 85.1% and recall of 80.3%. This approach not only meets the requirements for near-real-time field processing but also outputs seedling spatial distribution results with true geographic coordinates through georeferenced mapping, thereby providing directly applicable data support for seedling count statistics, missing seedling diagnosis, population spatial pattern analysis, and variable-rate management. This study establishes a complete technical pipeline from precise UAV image segmentation to spatially informed seedling status decision support, offering a theoretical foundation for efficient and accurate monitoring of soybean seedlings in the context of smart agriculture. Full article

Multi-UAV target localization relies on cooperative fusion of local, perception-derived geometric measurements over an edge network. While distributed fusion improves scalability and robustness compared with a centralized architecture, the iterative message exchanges may leak sensitive information to external eavesdroppers or honest-but-curious peers. This paper proposes a privacy-preserving distributed fusion method for multi-UAV localization via free-noise masking. The key idea is a double-injection mechanism. Specifically, each UAV masks its transmitted iterate with a locally generated bounded noise vector, while injecting the same noise into its local update so that the perturbations cancel exactly in the network-average dynamics under doubly stochastic mixing. As a result, the proposed PPDO-FN scheme preserves the practical convergence and weighted least squares localization accuracy of non-private distributed gradient descent, without requiring heavy cryptography or a trusted server. We further introduce reconstruction-based privacy metrics under transcript attacks and quantify the privacy–accuracy tradeoff. Simulation results demonstrate (i) near-identical accuracy and consensus behavior to the non-private baseline, (ii) monotonic privacy improvement with increasing masking strength, and (iii) the necessity of double-injection canceling compared with a naive single-injection baseline. Finally, we provide an end-to-end case study to connect the image-level detection to the geometric localization and then to privacy-preserving distributed fusion, illustrating engineering viability for our proposed approach. Full article

With the development of science and technology, heat dissipation has become a bottleneck problem restricting the development of fields such as transportation, machinery, electronics, and aerospace. Aiming to resolve the bottleneck problem of low thermal conductivity in traditional commercial magnesium alloys, this paper designed alloy compositions to investigate the effects of the solid solubility of La and Ce, and the size, morphology, distribution, and volume fraction of the second phase in the microstructure of magnesium alloys during the heat dissipation performance of the Mg-RE binary system and the Mg-Mn-La(Ce) system. The research shows that through CAFE simulation calculations, regulation can be achieved via the following methods: increasing the average nucleation undercooling, which leads to larger grain sizes; reducing the nucleation density, which results in larger grain sizes; and increasing the standard deviation of the average nucleation undercooling, which reduces the area of small grains while increasing the area of large grains. The thermal conductivity of both as-cast and solid-solution Mg-La (Ce) binary alloys gradually decreases with the increase in the added elements. However, after solution treatment, the thermal conductivity of the Mg-La (Ce) binary alloys is higher than that of the as-cast alloys. The addition of the Ce element helps refine the as-cast microstructure of the Mg-0.5Mn alloy. With the increase in Ce addition, the volume fraction of the Mg12Ce phase also increases. The thermal conductivity of the as-cast Mg-0.5Mn-xCe alloy gradually increases with rising temperature. Meanwhile, at room temperature, the thermal conductivity of the as-cast Mg-0.5Mn alloy gradually decreases with the increase in Ce addition, and the rate of decline gradually slows down due to the precipitation of the Mg12Ce phase. Full article

This work systematically investigates the influence of Mn content (0.1, 0.3, and 0.5 wt.%) on the microstructure, mechanical properties, and high-temperature stability of asymmetric extruded Mg-4.0Al-0.8Sn-0.3Ca-xMn alloys. The results demonstrate that Mn addition effectively promotes the formation of multi-scale secondary phases. Increasing the Mn content refines the average grain size from ∼2.82 µm to ∼1.89 µm and significantly modulates the recrystallization behavior of the alloy. The ATX4103-05Mn alloy (0.5 wt.% Mn) exhibits an optimal strength–ductility synergy, achieving a yield strength of 281.8 MPa and an elongation of 19.1%. Quantitative analysis reveals that this enhancement is predominantly governed by dispersion strengthening (∆σ_p∼34.1 MPa), with supplementary contributions from grain boundary and dislocation strengthening. Furthermore, the ATX4103-05Mn alloy shows superior resistance to abnormal grain growth after thermal exposure at 400 °C for 10 h, which is attributed to effective Zener pinning by the uniform distribution of short rod-shaped Al8Mn5 phases along the grain boundaries. This study elucidates the multi-scale strengthening and thermal stabilization mechanisms enabled by Mn microalloying, providing a viable pathway for developing high-performance, thermally stable magnesium alloys. Full article

This paper presents a comprehensive analysis and simulation of a cost-effective optical Intensity-Modulation/Direct-Detection (IM/DD) Orthogonal Frequency Division Multiplexing (OFDM) system. Implemented via a MATLABR2024a and OptiSystem 23 co-simulation environment, the study evaluates a 4-QAM modulated link over a 120 km transmission distance, providing detailed investigations into signal spectral properties and constellation characteristics. To address the critical performance limitation posed by high Peak-to-Average Power Ratio (PAPR), a novel Hybrid Whale Optimization Algorithm with Selective Mapping (HWOA-SLM) is proposed. Simulation results demonstrate that the proposed scheme significantly outperforms conventional reduction techniques; specifically, at a Complementary Cumulative Distribution Function (CCDF) of 10⁻² and a fixed computational budget of 256 evaluations, the HWOA-SLM achieves a PAPR reduction gain of 3.9 dB relative to the original OFDM signal. Furthermore, in terms of algorithmic efficiency, it outperforms standard Genetic Algorithm (GA) and WOA-based SLM techniques by approximately 0.4 dB under identical computational budgets. Parametric analysis further confirms that increasing population size and iteration numbers consistently improves convergence, thereby minimizing non-linear distortions and enhancing signal integrity. Moreover, the technique exhibits superior Bit Error Rate (BER) performance, delivering Optical Signal-to-Noise Ratio (OSNR) gains of 0.63 dB, 1.31 dB, and 2.0 dB over standard WOA-SLM, GA-SLM, and conventional SLM, respectively. Conclusively, the HWOA-SLM offers a favorable trade-off between computational complexity and reduction efficiency, validating its potential for reliable, high-speed optical communication networks. Full article