Search Results (202)

Reliable identification of maize leaf diseases is critical for mitigating crop losses, particularly in regions where farmers have limited access to experts. Although vision transformers (ViTs) have recently demonstrated strong performance in image recognition, their weak inductive bias and limited modeling of local texture patterns make them non-ideal for fine-grained maize leaf disease classification. To address these limitations, we propose ConvDeiT-Tiny, a lightweight hybrid ViT that improves DeiT-Ti by placing depthwise convolutions in parallel with multi-head self-attention modules in the first three transformer blocks. The local and global features captured by the convolution and attention modules are concatenated along the embedding dimension and fused using a multilayer perceptron. This results in richer token representations without significantly increasing model size. Across three datasets, ConvDeiT-Tiny (6.9 M parameters) consistently outperformed DeiT-Ti, DeiT-Ti-Distilled, and DeiT-S (21.7 M parameters) when trained from scratch. With transfer learning, ConvDeiT-Tiny achieved an accuracy of 99.15%, 99.35%, and 98.60% on the CD&S, primary, and Kaggle datasets, respectively, surpassing many previous studies with far fewer parameters. For explainability, we present gradient-weighted transformer attribution visualizations showing the disease lesions driving model predictions. These results indicate that injecting local inductive bias in early transformer blocks is beneficial for accurate maize leaf disease classification. Full article

Urban–industrial environments can generate mixed geogenic and traffic-related metal signatures in paediatric scalp hair, yet interpretation is challenged by left-censoring and limited health-based guidance values for hair. We quantified barium (Ba), strontium (Sr) and vanadium (V) in archived scalp hair collected in 2001 from children (6–9 years, n = 120) and adolescents (13–16 years, n = 97) residing in Alcalá de Henares (central Spain). Samples were washed, digested and quantified by Inductively coupled plasma mass spectrometry (ICP–MS; laboratory processing in 2025); results below the limit of detection (LoD) were treated as left-censored using NADA2 (no substitution). In children, Ba and Sr were frequently quantifiable (medians 0.193 and 0.412 µg/g; 38.3% and 23.3% <LoD), whereas V was heavily censored (74.2% <LoD; median 0.003 µg/g). Adolescents showed higher Ba and Sr and broader upper tails (Ba median 0.287 µg/g, P95 2.061 µg/g; Sr median 1.105 µg/g, P95 4.995 µg/g), while V remained low (median 0.011 µg/g, P95 0.052 µg/g). Ba and Sr displayed strong spatial gradients across four residential zones in adolescents (censored-data Peto–Peto tests p < 1 × 10⁻⁸), but V did not (p = 0.162). Co-located residential topsoils were available only for V and showed limited between-zone contrast; soil–hair correspondence was weak overall but moderate in adolescent girls (Spearman ρ = 0.433). These findings provide a historical baseline and support a cautious tracer-oriented interpretation in which the observed Ba–Sr spatial patterning is consistent with heterogeneous contact with dust- and traffic-influenced surface materials, while V appears less discriminatory in low-contrast community settings. Full article

Document-level Event Extraction (DEE) requires identifying complex event records and arguments dispersed across unstructured texts. However, applying general Large Language Models (LLMs) to DEE is intrinsically hindered by their lack of inductive bias for rigid structural constraints, often leading to schema violations and suboptimal performance in complex structural prediction tasks. To address this, we propose the S tructure-Aware Lightweight DEE, termed SALE, which leverages the structural reasoning potential of Code-Based LLMs (Code-LLMs) as a favorable inductive preference. We leverage the natural isomorphism between event schemas and programming object definitions, formulating event extraction as a Python 3.9 class instantiation task to bridge the gap between semantic understanding and structural adherence. Specifically, SALE employs a novel two-stage training paradigm: First, a Structure-Aware Fine-tuning stage injects general structural knowledge via diverse code-style instruction tasks derived from broad Information Extraction (IE) datasets; second, an Event Extraction Alignment stage utilizes a reward-based alignment loss—optimized via policy gradient—to adapt this capability to document-level intricacies. The effectiveness of SALE stems from the synergy between its structure-aware prompting and the specialized alignment stage built on a code-oriented backbone. Extensive experiments on established news-domain benchmarks (RAMS and WikiEvents) demonstrate that our approach significantly outperforms representative supervised and general LLM baselines in cross-task zero-shot and few-shot transfer settings (e.g., surpassing supervised baselines by over 7% in F1 score). Furthermore, SALE maintains a highly efficient inference profile and parameter-efficient footprint, offering a practical and scalable solution for vertical domain applications. Full article

A high-throughput assay system is developed for measuring communication in microbial biofilms in a 96-well microtiter plate format. In this assay, bioluminescent microbial whole cell biosensor systems (MWCBs) for quorum-sensing molecules (QSMs) are embedded into biofilms, and their response to chemical cues relevant for bacterial communication is assessed. For measuring the response to stress, a sigma factor 54 (σ⁵⁴, RpoN)-dependent MWCB was developed. Biofilms generated in this platform were exposed to gradients of communication signals (QSMs such as N-acetyl-homoserine lactones (AHLs), 3,5- dimethylpyrazin-2-ol (DPO), or phytochemicals that can act as natural quorum-sensing inhibitors (QSIs) such as curcumin or 3,3′-diindolylmethane (DIM)), and the response pattern was monitored. Further, the regulatory role of stressors such as oxidants (H₂O₂) or antibiotics (ciprofloxacin, trimethoprim/sulfamethoxazole) on the communication response is assessed. QSMs induced the MWCBs at 1 h and 4 h in biofilms, but high concentrations inhibited them at 24 h. Curcumin and DIM at higher concentrations lead to inhibition of quorum sensing in biofilms after 4 h and 24 h, but this is not followed by biofilm disintegration. H₂O₂ above 0.002% efficiently inhibited the MWCB activities and led to biofilm disintegration. At lower concentrations of H₂O₂, we observed induction of MWCBs. The antibiotics inhibited the MWCB activity at concentrations above their minimal inhibitory concentration (MIC), but this did not necessarily lead to disintegration of the biofilm. Like low concentrations of H₂O₂, the antibiotics activated the MWCBs at concentrations close to their MIC, possibly as a result of H₂O₂ generated during their bactericidal action. Interestingly, the induction of communication in response to antibiotics can be quenched by iron chelators, suggesting involvement of H₂O₂ and free radicals generated by the Fenton reaction. We hypothesize that the observed response to these stressors reflects increased communication in the biofilm, possibly enhancing tolerance and increasing survival. Full article

Antimicrobial resistance (AMR) has become a major global public health challenge, and widely residual disinfectants in the environment are one of the key drivers of bacterial AMR development. This study aimed to investigate the inductive effects of three commonly used disinfectants—benzalkonium bromide (BAB), glutaraldehyde (GTA), and povidone-iodine (PVP-I)—on the resistance of Escherichia coli (E. coli), as well as the resultant bacterial phenotypic and genetic alterations. Three disinfectants frequently detected in clinical and environmental settings were selected as the research objects: first, their bactericidal efficacy against environmental bacteria was determined; subsequently, a concentration-increasing gradient approach was adopted to systematically explore the evolutionary patterns of E. coli resistance under the stress of sub-inhibitory concentrations (SICs). After induction, the bacterial resistance levels to disinfectants and various antibiotics, growth characteristics, and biofilm-forming ability were detected, and combined with whole-genome analysis to investigate genetic-level changes. The results showed that all three disinfectants could enhance E. coli resistance to themselves (12–48-fold) and antibiotics, and the induced antibiotic resistance exhibited favorable genetic stability. Among them, BAB induced the strongest resistance, with the most significant increase in resistance levels to multiple antibiotics (16–64-fold); GTA had the weakest inductive effect, only slightly enhancing bacterial resistance to a small number of antibiotics. Notably, all induced strains exhibited reduced growth rates yet markedly enhanced biofilm-forming capacity, alongside acquired genomic structural variations. Their gene functions displayed shared adaptive signatures in coping with environmental stress, while core pathogenicity-associated genes remained conserved. This study demonstrates that inducing E. coli using environmentally relevant low concentrations of disinfectant residues as initial induction doses drives the evolution of bacterial antimicrobial resistance (AMR), with distinct resistance induction risks among the three disinfectant types. These findings offer critical insights for standardizing disinfectant application, mitigating the transmission of bacterial AMR, and underscore the imperative of interdisciplinary collaboration to tackle the environmental risks posed by disinfectant residues. Full article

Rare event detection in large tabular data remains a computationally challenging problem due to class imbalance, heterogeneous feature distributions, and unstable thresholds. Traditional machine learning approaches based on individual models and fixed thresholds often exhibit limited robustness and reproducibility in such settings. This paper proposes a hybrid ensemble framework for rare event detection that integrates heterogeneous machine learning models through threshold-aware probabilistic aggregation. The framework combines gradient-boosted decision trees, regularized linear models, and neural networks, leveraging their complementary inductive biases. To ensure reproducibility and robust performance evaluation under severe class imbalance, a leaky-controlled evaluation protocol is employed, including rootwise summation, probability calibration, and validation-based threshold optimization. The proposed approach is evaluated on a large tabular dataset containing approximately 50,000 observations. Experimental results demonstrate improved rare event detection and robust generalization performance compared to individual baseline models. Explainability is achieved through Shapley Additive Explanations (SHAP)-based attribution analysis and clustering in the explanation space, enabling transparent analysis of ensemble decision-making behavior. The proposed framework represents a general-purpose computational solution for rare event detection and can be applied to a wide range of data-driven decision-making and anomaly detection problems. Full article

Dynamic wireless power transfer (DWPT) systems for in-motion electric vehicle (EV) charging often suffer from unstable power delivery due to spatial variations in magnetic coupling caused by vehicle misalignment. This study presents a stabilization-oriented DWPT design methodology that prioritizes minimizing spatial variations of mutual inductance rather than maximizing peak coupling under perfect alignment. A ferrite-backed double-D coil configuration is analyzed and refined using three-dimensional finite-element electromagnetic modeling integrated with circuit-level co-simulation to evaluate coupling behavior, magnetic field homogeneity, and power transfer efficiency under realistic dynamic misalignment conditions. The proposed design achieves a coupling coefficient of 0.50–0.55 under aligned conditions and exhibits smooth, predictable degradation for lateral offsets up to 40–50 mm. Quantitative analysis demonstrates a low spatial coupling gradient of approximately 0.001 mm⁻¹, indicating that abrupt coupling transitions are effectively suppressed during vehicle motion. The system attains a maximum power transfer efficiency of 84.37% at an 80 mm air gap, while maintaining stable performance under both lateral and vertical displacement. Comparative evaluation shows improved misalignment tolerance and coupling stability relative to conventional double-D configurations. The results demonstrate that electromagnetic field shaping focused on coupling smoothness is an effective and practical strategy for reliable dynamic wireless charging of electric vehicles. Full article

Understanding the pathways through which heavy metals accumulate in medicinal plants and enter herbal infusions is essential for linking environmental quality with human exposure. This study investigated multi-matrix metal transfer in Tilia spp. along an urban–forest gradient by quantifying twelve elements (Pb, Cd, Zn, Cu, Ni, Cr, Mn, Co, As, Hg, Al, and V) in soil, bark, leaves, flowers, and corresponding infusions using inductively coupled plasma mass spectrometry and by estimating daily intake for different age groups based on EFSA default body weights and two consumption scenarios (150 and 400 mL day⁻¹). The results revealed clear spatial patterns, with significantly higher metal loads in urban sites and a consistent transfer from environmental compartments to plant tissues and infusions. Mn, Al, Pb, and Cd exhibited the highest extractability, leading to elevated estimated daily intakes in young children, identified as the most vulnerable group due to their lower body mass. However, all exposure values remained below EFSA and JECFA toxicological reference limits, while As and Hg were undetectable in all infusions. These findings indicate that Tilia infusions contribute minimally to overall dietary metal exposure and confirm Tilia spp. as reliable bioindicators of soil- and airborne metal deposition, supporting the safe consumption of linden tea under realistic intake conditions. Full article

Efficient on-site connection of power cable conductors is critical for ensuring the safe operation of the power grid. Traditional thermite welding methods pose significant safety risks, including open flames and fumes. Meanwhile, induction heating, when applied to cable conductors, faces challenges of severe magnetic field dispersion, low heating efficiency, and a high risk of damaging adjacent insulation layers. This paper proposes a novel magnetic flux control system based on gradient permeability ceramics to address these issues. The core of this system is the synergistic utilization of a gradient permeability composite ceramic mold and a high-permeability shielding shell. A 2D axisymmetric multiphysics coupled model was established to compare the performance of the optimized system with a conventional case and single control components. Simulation results demonstrate that the optimized system increases the magnetic flux density at the weld seam to 3.7 times that of the conventional setup (0.263 T). Consequently, the weld seam of the 240 mm² copper conductor is rapidly heated to the melting point of copper (1083 °C) within 7.78 s. Due to the high heating rate, upon completion of the welding process, the temperatures of the inner shielding and insulation layers are only 48.8 °C and 24.3 °C, respectively, well below the materials’ safety thresholds. These findings suggest that the proposed magnetic flux control strategy achieves rapid and precise heating, offering a theoretical foundation for the development of high-performance on-site equipment for fabricating cable joints. Full article

Accurate estimation of the State-of-Health (SOH) is essential for the reliable operation of lithium-ion batteries in electric vehicles and energy storage systems. However, conventional data-driven models often lack interpretability and show limited robustness under non-linear aging conditions. In this study, a physics-informed Transformer model is proposed for long-term SOH estimation by incorporating physically interpretable, degradation-sensitive indicators into a self-attention framework. Incremental Capacity Analysis (ICA)-derived features and thermal-gradient indicators are used as auxiliary inputs to provide physics-consistent inductive bias, enabling the model to focus on degradation-relevant regions of the charging trajectory. The proposed approach is validated using four lithium-ion battery cells exhibiting diverse aging behaviors, including severe non-linear capacity fade. Experimental results demonstrate that the proposed model consistently outperforms an LSTM baseline, achieving an RMSE below 1.5% even for the most degraded cell. Furthermore, attention map analysis reveals that the model autonomously emphasizes voltage regions associated with electrochemical phase transitions, providing clear physical interpretability. These results indicate that the proposed physics-informed Transformer offers a robust and explainable solution for battery health monitoring under practical aging conditions. Full article

Motor position sensors are critical parts for traction motors control in electrified automotive powertrains. As motors are becoming more compact due to the advance of technology the packaging space for motor position sensors is becoming increasingly restricted. This study presents a two-layer (2L) printed circuit board (PCB) routing strategy for inductive motor position sensors with limited area. A prototype was fabricated and tested on a test bench using a comprehensive design of experiments that contains 625 combinations of X- and Y-offsets, tilt angle, and airgap at various levels (±0.5 mm in X/Y, ±0.5° tilt, 1.9–3.1 mm airgap). Across the tolerance box, the accuracy under all test cases remained within ±1 electrical degree. The accuracy analysis through Fourier series on a circle shows that the DC offset and magnitude mismatches of the 3 Rx signals are the dominant error contributors due to the routing modification. An Extreme Gradient Boosting (XGBoost) model was trained and validated with R² = 0.9951. A comparison with a Multiple Linear Regression baseline (R² = 0.0565) demonstrates that installation-induced accuracy degradation is inherently non-linear. The SHapley Additive exPlanations (SHAP) and interaction intensity analysis identified tilt and Y-offset as dominant error drivers, revealing a strong coupled influence (interaction intensity = 0.9581). The model revealed a mild Y-axis asymmetry introduced by routing modifications. This integrated workflow provides a general, quantitative framework for optimizing and analyzing inductive sensor layouts and establishing installation tolerances. Full article

Speed breeding represents a pivotal technology for enhancing crop breeding efficiency. This study systematically examined the regulation of LED light environments, planting density, and gibberellic acid (GA₃) on rice growth cycle progression in plant factories, establishing an integrated speed breeding protocol. The experimental design comprised three components: (1) coupling seedling age (9–25 days, variety-dependent) with LED environments and planting densities (25–100 plants/tray); (2) combining light intensity gradients (450 and 900 μmol·m⁻²·s⁻¹) with photoperiod control; (3) applying GA₃ gradients (0–120 ppm) to enhance immature seed germination. Results indicated that high planting densities (>50 plants/tray) prolonged the growth cycle and decreased yield, whereas 25 plants/tray optimally balanced growth cycle shortening and yield maximization. Under short-day induction, Nipponbare (Nip) and Wufeng B (WFB) reached heading at 39 and 58 days after sowing (DAS), respectively. Stage-specific light responses were observed: 450 μmol·m⁻²·s⁻¹ during the basic vegetative phase (BVP) promoted morphological development, whereas 900 μmol·m⁻²·s⁻¹ during the photoperiod-sensitive phase (PSP) accelerated tillering and panicle differentiation. GA₃ treatment (60 ppm) enhanced the germination rate of immature seeds by 31%. The optimized lightregimes comprised natural light + 900 μmol·m⁻²·s⁻¹ (NL–900) and 450 μmol·m⁻²·s⁻¹ + 900 μmol·m⁻²·s⁻¹ (450–900), combined with density control (25 plants/tray) and GA₃-mediated immature seed utilization, shortened the generation time to 54 days and 70 days for Nip and WFB, respectively. This integrated protocol establishes an efficient strategy for rice speed breeding in plant factories. Full article

During liquid drainage from intermediate vessels in various industrial processes such as continuous steel casting, aircraft fuel supply, and chemical separation, free-surface vortices commonly occur. The formation and evolution of these vortices not only entrain surface slag and gas, but also lead to deterioration of downstream product quality and abnormal equipment operation. The vortex evolution process exhibits notable three-dimensional unsteadiness, multi-scale turbulence, and dynamic gas–liquid interfacial changes, accompanied by strong coupling effects between temperature gradients and flow field structures. Traditional macroscopic numerical models show clear limitations in accurately capturing these complex physical mechanisms. To address these challenges, this study developed a mesoscopic numerical model for gas-liquid two-phase vortex flow based on the lattice Boltzmann method. The model systematically reveals the dynamic behavior during vortex evolution and the multi-field coupling mechanism with the temperature field while providing an in-depth analysis of how initial perturbation velocity regulates vortex intensity and stability. The results indicate that vortex evolution begins near the bottom drain outlet, with the tangential velocity distribution conforming to the theoretical Rankine vortex model. The vortex core velocity during the critical penetration stage is significantly higher than that during the initial depression stage. An increase in the initial perturbation velocity not only enhances vortex intensity and induces low-frequency oscillations of the vortex core but also markedly promotes the global convective heat transfer process. With regard to the temperature field, an increase in fluid temperature reduces the viscosity coefficient, thereby weakening viscous dissipation effects, which accelerates vortex development and prolongs drainage time. Meanwhile, the vortex structure—through the induction of Taylor vortices and a spiral pumping effect—drives shear mixing and radial thermal diffusion between fluid regions at different temperatures, leading to dynamic reconstruction and homogenization of the temperature field. The outcomes of this study not only provide a solid theoretical foundation for understanding the generation, evolution, and heat transfer mechanisms of vortices under industrial thermal conditions, but also offer clear engineering guidance for practical production-enabling optimized operational parameters to suppress vortices and enhance drainage efficiency. Full article

Groundwater is the main water source for irrigated agriculture, accounting for an increasing share of the domestic supply in the hyper-arid district of La Yarada Los Palos (Tacna, Peru); however, at the sector scale, concerns about arsenic, boron and salinity remain poorly quantified. Arsenic and boron were selected as target contaminants because of their naturally elevated concentrations associated with coastal and volcanic hydrogeological settings, and their well-documented implications for human health and irrigation suitability. This study reports a 12-month monitoring program (September 2024–August 2025) in three irrigated sectors, in which wells were sampled monthly and analyzed by inductively coupled plasma–mass spectrometry (ICP-MS) for total arsenic, boron, lithium and sodium, along with electrical conductivity, pH, temperature and total dissolved solids. The sector–month total arsenic means ranged from 0.0089 to 0.0143 mg L⁻¹, with 33 of 36 exceeding the 0.010 mg L⁻¹ drinking water benchmark recommended by the World Health Organization (WHO). Total boron ranged from 1.11 to 2.76 mg L⁻¹, meaning that all observations were above the 0.5 mg L⁻¹ irrigation guideline for agricultural use proposed by the United Nations Food and Agriculture Organization (FAO). A marked salinity gradient was observed from the inland Sector 1-BH (median Na ≈ 77 mg L⁻¹; EC ≈ 1.2 mS cm⁻¹) to the coastal Sector 3-LC (median Na ≈ 251 mg L⁻¹; EC ≈ 3.3 mS cm⁻¹), with Sector 2-FS showing intermediate salinity but the highest median boron and lithium levels. Spearman rank correlations indicate that sodium, electrical conductivity and total dissolved solids define the main salinity axis, whereas arsenic is only moderately associated with boron and lithium and is not a simple function of bulk salinity. Taken together, these results show that groundwater from the monitored wells is not safe for drinking without treatment and is subject to at least moderate boron-related irrigation restrictions. The sector-resolved dataset provides a quantitative baseline for La Yarada Los Palos and a foundation for future work integrating expanded monitoring, health-risk metrics and management scenarios for arsenic, boron and salinity in hyper-arid coastal aquifers. Full article

Our team has initiated work to determine residual stresses by means of monitoring magnetic properties, namely differential permeability, magnetoacoustic emission, and surface field components. Concerning surface field measurements, Hall, AMR, and TMR sensors have been used, with AMR and TMR sensors enabling 3D field determination. In this paper, we compare the surface magnetic field components with residual stresses in 2 mm thick AISI 4130 steel coupons. The steel samples were in a dog-bone structure with residual stresses induced by localized RF induction heating to create a temperature gradient, followed by quenching to transform the temperature gradient into a residual stress one. GMI and AMR sensors were used to determine the localized magnetic field component distribution on the surface of the steel coupons and at the same areas where the residual stresses were determined. The GMI sensor was able to monitor the field component perpendicular to the surface of the steel coupon, while the AMR sensor was able to monitor the three field components at the same points. The results illustrated that both sensors were able to monitor residual stresses, with the GMI sensor illustrating better sensitivity at a higher cost, while the AMR sensor had a lower sensitivity with a significantly lower cost as an integrated sensor. Full article