Search Results (298)

The timely identification of key rice growth stages supports water management, crop monitoring, and yield estimation. This study evaluated whether single-date unmanned aerial vehicle (UAV) RGB imagery can identify early tillering, booting, heading, and milk-ripe rice without a continuous image time series. UAV imagery and synchronous field observations were collected from 64 fields in the Zhanghe Irrigation District, China, during 2022–2023. Four visible-light indices and 12 gray-level co-occurrence matrix (GLCM) texture features were used with k-nearest neighbors (KNN), a support vector machine (SVM), random forest (RF), a gradient boosting decision tree (GBDT), and corresponding stacking ensembles. Stacking-RF achieved the highest test performance, with 91.7% accuracy and a macro-F1 of 0.917, exceeding KNN, the best individual model, by 6.8 percentage points and 0.067, respectively. Early tillering and milk-ripe were identified more accurately than booting and heading. Visible-light indices supplied most of the discriminative information, while texture features contributed complementary canopy-structure information. The results show that an RGB-only, single-date input can support a rapid within-site growth-stage assessment. However, the sample-level split, single study area, and absence of radiometric calibration limit inference to new fields, years, and regions. Full article

High-resolution atmospheric modeling over topographically complex coastal zones requires multi-source spatial data fusion to bridge the scale gap between satellite-assimilated global reanalysis and localized boundary layer dynamics. This study presents a spatial data fusion framework coupling convection-permitting Weather Research and Forecasting (WRF) simulations (3 km grid spacing) with machine learning calibration and GIS multi-criteria decision analysis across continuous computational grids (24,871 cells). Focusing on a high-resolution seasonal case study during the peak Caribbean Low-Level Jet (CLLJ) and Trade Winds regime along the Colombian Caribbean coast, predictive algorithms were benchmarked across day-grouped and Leave-One-Station-Out (LOSO) spatial transfer protocols against classical meteorological baselines. Support Vector Regression (SVR) with an RBF kernel achieved superior residual calibration ($R^2=0.9523$, $\mathrm{RMSE}=0.4601$ m/s) under day-grouped cross-validation, and an average Leave-One-Station-Out (LOSO) spatial transfer RMSE of $1.1030$ m/s under blind geographic transfer to unmonitored stations, corresponding to a >76% RMSE reduction relative to raw, uncalibrated WRF output in that same spatial-transfer setting. Chronological forward-forecasting tests further showed that this seasonally augmented kernel model does not extrapolate reliably to calendar months absent from training ($R^2=-0.09$), so a reduced 8-variable configuration is recommended for genuine forward forecasting into unseen seasons. Scalar calibrations were transferred to horizontal wind vectors preserving simulated flow direction, and vertically scaled to 80 m hub height using WRF prognostic shear. Calibrated fields were integrated into a Boolean GIS-MCDA model incorporating wind power density, slope limits, environmental/indigenous reserves, and infrastructure proximity. The framework delineated an $8.9\%$ high-feasibility macro-corridor (2214 cells; ≈19,926 ${\mathrm{km}}^2$) in northern La Guajira, while substantially reducing uncalibrated mesoscale overestimation in the sheltered wake of the Sierra Nevada de Santa Marta. The methodology provides a robust screening-level tool for coastal renewable resource assessment. Full article

Thyroid disease classification is an important healthcare task because thyroid dysfunction affects metabolism, hormonal balance, cardiovascular activity, and overall well-being. Accurate detection can support timely treatment; however, prediction remains challenging because of nonlinear relationships among biochemical markers, clinical-history variables, missing values, and imbalanced classes. This study proposes ThyroTrans-XGB, a hybrid framework combining Transformer-based feature representation with XGBoost >= 1.0.0 classification. Two publicly available thyroid disease datasets were examined in separate benchmark experiments. The Kaggle Thyroid Disease Dataset was used for binary classification, whereas the UCI Thyroid Disease Dataset was used for a separate multiclass benchmark evaluation. The model was separately trained and evaluated within each dataset-specific setting. The Transformer module learns attention-based representations of thyroid-related variables, while XGBoost performs final classification. Performance was assessed using accuracy, precision, recall, F1-score, ROC-AUC, PR-AUC, Matthews correlation coefficient, calibration analysis, statistical testing, and SHAP-based explainable artificial intelligence (XAI). ThyroTrans-XGB achieved an F1-score of 0.9828, ROC-AUC of 0.9997, and PR-AUC of 0.9948 on the Kaggle benchmark, and a macro-F1-score of 0.9891, ROC-AUC of 0.9999, and PR-AUC of 0.9928 on the UCI benchmark. Since both datasets were separately trained and evaluated and may share public-data provenance, the UCI experiment represents benchmark evaluation rather than external validation. Full article

Accurate maturity recognition of Pleurotus citrinopileatus fruiting bodies is important for growth monitoring and harvest assistance in controlled cultivation. However, multiple developmental stages often coexist within one image, and near-mature fruiting bodies share visual characteristics with both adjacent stages. We developed a two-stage RGB-based framework for instance-level maturity recognition in complex cultivation scenes. After fruiting-body localization, each detected instance was represented by paired local and context views and classified using a DINOv3 ViT-S/16-based ordinal model. The task-specific methodological contribution is a morphology-guided fusion module that incorporates automatically extracted geometric, color, intensity, and edge descriptors into the dual-view visual representation. Object-level matching was used to quantify how detection and classification errors affected the complete pipeline, and temperature scaling was applied for probability calibration. Among five candidate detectors, RF-DETR-Small achieved validation mAP50 and mAP50–95 values of 0.9747 and 0.8813, respectively. On the independent ROI test set, the final classifier achieved an accuracy of 0.8779, a macro F1 of 0.7950, a near-mature F1 of 0.5546, and a QWK of 0.9068. Temperature scaling reduced ECE from 0.1094 to 0.0564 without changing the predicted labels. On 370 original cultivation images containing 475 ground-truth instances, the unfiltered pipeline achieved an object-level macro F1 of 0.7072 and a near-mature F1 of 0.4672. Candidate-box filtering increased these scores to 0.7305 and 0.4885, respectively, but increased the number of missed instances from 14 to 23. These results demonstrate the feasibility of non-destructive instance-level maturity monitoring while identifying near-mature discrimination and candidate-box control as priorities for practical application. Full article

Credit-scoring systems must translate predictive risk into calibrated, uncertainty-aware, and governable decisions rather than ranking applicants alone. This study aims to develop and evaluate a governed credit-scoring architecture integrating predictive performance, probability quality, uncertainty, and governance within a unified decision-oriented framework. Existing studies typically evaluate these functions separately, leaving uncertainty and governance outside the core modeling pipeline. We present the Calibrated, Adaptive, Knowledge-aware Ensemble for Credit Scoring (CAKE-CS), a structure-aware architecture integrating fold-local dependency representation, heterogeneous experts, sparse routing, cross-fitted stacking, probability calibration, and cross-conformal decisions within a fold-disciplined pipeline. Across eleven public credit and default portfolios containing 70,518 observations, CAKE-CS achieves a macro-averaged Area Under Curve (AUC) of 0.869 versus 0.862 for the strongest gradient-boosting competitor, together with the lowest macro-averaged expected misclassification cost, Brier score, and log loss among nineteen competitors under the primary protocol. Bidirectional ablation identifies stacking as the clearest positive contributor to AUC, while routing and structure serve primarily architectural roles. Empirical cross-conformal coverage ranges from 0.882 to 0.920. The originality is architectural rather than algorithmic. CAKE-CS reconfigures credit scoring from a model-centric prediction problem into a governed decision-system problem by linking prediction, probability quality, uncertainty, action, monitoring, and audit evidence. Theoretically, this extends credit-scoring research by treating these dimensions as interconnected components of decision-system performance. Practically, the architecture provides a deployment-oriented template in which calibrated predictions and uncertainty support decision thresholds, human review, monitoring, and audit. Full article

Offshore wind power cables may experience drag and lift loading where submarine landslides or related density flows cross a cable corridor. To prioritise conditions for detailed analysis, a leakage-controlled ensemble-learning workflow was developed from multi-source submarine landslide–pipeline/cable impact data and evaluated by source-grouped cross-validation. Separately predicted peak drag and peak lift were converted to empirical percentile ranks and averaged to form a dataset-relative peak-load severity index. The resulting Low–Extreme categories achieved accuracy of 0.748, macro F1 of 0.690 and High/Extreme recall of 0.758; binary High/Extreme identification achieved accuracy of 0.903, F1 of 0.800 and ROC-AUC of 0.947. Missingness analyses gave four-category accuracy of 0.717–0.771 across full, low-missingness and complete-case settings, whereas weight and threshold changes showed that category boundaries remain calibration choices rather than universal limits. Scenario and feature-group results indicate combined associations with hydrodynamic forcing, cable geometry, rheology and exposure/cover conditions. The output is intended to rank candidate conditions for computational fluid dynamics, physical modelling or structural verification. It is not a design load, a code check or a full risk estimate, which would additionally require occurrence probability, cable vulnerability, consequence and project-specific validation. Full article

Physiological-signal-based emotion recognition has received attention in human–computer interaction. Electrocardiograms (ECGs) are readily acquired, and short windows contain repeated beat morphology and beat-wise variations that single-window encoding struggles to separate. Multiscale morphology modeling and training-sample diversity remain limited. We therefore propose Complementary Feature Fusion Dual-Path (CFF-DP), an ECG emotion recognition framework using beat-level complementary feature fusion, with three components: (1) a dual-path framework, where the morphology-stable path constructs representative beats with window-adaptive Gaussian weights, while the morphology-difference path combines beat-wise encoding, positional encoding, and additive attention; gated fusion integrates representations; (2) adaptive dilated convolution (ADConv), which extracts multiscale beat-morphology features using shared kernels and input-dependent scale weights; and (3) deviation-based beat-oriented augmentation (DBOA), which adjusts real-noise injection probability and target signal-to-noise ratio according to morphological deviation. CFF-DP achieved 43.39% mean Macro-F1, close to the best comparator, with the fewest multiply–accumulate operations in five-seed WESAD three-class leave-one-subject-out (LOSO) evaluation, although recognition mainly distinguishes stress, with limited amusement discrimination. With short-gap calibration and testing within the same recording, fine-tuning using 40 s per class achieved 76.72% Macro-F1, exceeding six comparators. Binary DREAMER LOSO retained WESAD hyperparameters: valence Macro-F1 exceeded six comparators, whereas arousal fell below four; both remained below uniform random baselines, indicating limited recognition under current experimental conditions. The framework combines computational efficiency with within-record personalization advantages, although cross-subject recognition remains limited. Full article

Macro-fibre composite (MFC) actuators offer improved flexibility and damage tolerance over monolithic piezoceramics, which have been widely utilised recently. However, direct bonding to a heritage substrate of MFC actuators may cause local stress concentration and complicate removal. To overcome this, a protective aluminium–concrete sandwich configuration is proposed, where the MFC patch is attached to a replaceable thin aluminium carrier layer that transfers a controlled deformation field to the concrete substrate through the bonded aluminium–concrete interface, enabling reversible, low-intrusion deformation control without direct modification of the protected substrate. A coupled electromechanical finite-element formulation is derived from the linear piezoelectric constitutive equations and Hamilton’s principle. The model is first is validated against benchmark deflection and modal data for a traditional MFC-bonded aluminium beam, achieving a maximum deflection error below 1% (0.9142 mm vs. 0.9141 mm at the free end) and excellent frequency agreement (19.7 Hz and 112.0 Hz). Laboratory cantilever tests under 400 V show a systematic amplitude reduction relative to the perfect-bond model: the measured free-end displacement is approximately 0.598 mm compared with 0.9142 mm numerically. A one-parameter effective actuation-transfer coefficient of 0.656 reduces the displacement-profile RMSE to 0.0072 mm (NRMSE 1.23%, R² = 0.9987), indicating that the dominant discrepancy is an amplitude loss associated with non-ideal strain transfer and boundary/electric-field effects rather than a change in deformation mode. For the proposed sandwich beam, simulations reveal a monotonic, voltage-dependent response: tip deflection rises from approximately 0.03 mm at 200 V to 0.115 mm at 800 V over a 200 mm span. Actuator placement near the fixed end yields higher bending authority, consistent with classical placement theory. The sandwich concept demonstrates that a replaceable protective layer can generate controllable curvature while maintaining moderate stress levels in the protected substrate, making it suitable for micro-crack suppression and temporary stabilisation of fragile components such as cultural relics or aged concrete. The validated model provides a foundation for future experimental calibration and distributed actuator optimisation. Full article

Extreme climate events increasingly amplify agricultural trade security risks across interconnected stages: agricultural production, commodity conditions, cold-chain storage, logistics distribution, and trade fulfillment. However, existing approaches predominantly focus on isolated stages and fail to model cross-stage risk propagation, enable real-time edge inference, or provide well-calibrated risk warnings. To address these challenges, a multimodal edge-intelligence framework, termed AgriClimate-EdgeNet, is proposed to jointly model climatic conditions, agricultural production, commodity imagery, cold-chain states, logistics trajectories, and trade records. An extreme-climate-aware cross-modal consistency mechanism is developed to capture normal inter-stage correspondence and identify abnormal information conflicts. Depthwise separable temporal convolutions, gated temporal units, lightweight attention, and Teacher–Student distillation are incorporated for efficient edge inference. Furthermore, dynamic modality reliability estimation and dual predictive uncertainty modeling (decoupling epistemic and heteroscedastic aleatoric uncertainties) are integrated to ensure decision trustworthiness under severe sensory noise and missing observations. On a 38,400-window agricultural trade dataset, AgriClimate-EdgeNet achieves an Accuracy of $0.914$, Recall of $0.896$, Macro-F1 of $0.902$, and AUC of $0.949$, while reducing expected calibration error to $0.028$ in routine single-pass Streaming Mode and $0.021$ under multi-sample Deep Audit Mode. In operational edge deployment on NVIDIA Jetson AGX Orin, the model requires $3.96$ M on-device parameters and $1.21$ G FLOPs, achieving an inference latency of $8.6$ ms ($116.3$ samples/s) in Streaming Mode and $38.4$ ms in Deep Audit Mode. These results demonstrate that AgriClimate-EdgeNet provides an accurate, robust, and low-latency solution for full-chain agricultural trade risk sensing under extreme climate shocks. Full article

Electroencephalography (EEG)-based Brain–Computer Interfaces offer direct insight into a driver’s mental state, relevant to autonomous-vehicle perception stacks. This work addresses vigilance/drowsiness specifically, one of several driver states relevant to Level 3 takeover readiness. Detecting drowsiness before handover in Level 3 vehicles requires balancing high-dimensional EEG features against the real-time demands of a lightweight safe-stop trigger mechanism. We present a complete pipeline: a standardized 22-feature-per-channel schema extracted across four driving-related EEG datasets, an autoencoder achieving over 95% dimensionality reduction with minimal performance loss, and a classifier driving a constant-deceleration kinematic profile used to obtain a measurable latency figure, evaluated under a fully subject-disjoint, leakage-free protocol. Leave-one-subject-out cross-validation on the vigilance dataset yields a mean macro-F1 of 0.793 ± 0.163 (0.763 ± 0.155 on an alternative run, within expected autoencoder-retraining variance). Extending compression evaluation to all four datasets shows preserved or improved performance on three, with a small cost on the fourth. The full decision chain completes in under 1 ms (mean 0.13 ms) on standard hardware; this covers the perception-to-trigger budget only, excluding solver latency that an optimization-based planner (MPC/CBF-QP) would add downstream. These findings show cognitive-state-aware safe-stop triggering is implementation-viable, while highlighting the calibration and cross-machine validation needed for deployment. Full article

The Internet of Medical Things (IoMT) environment is based on WiFi and MQTT communication, which results in highly imbalanced intrusion-detection data with a high degree of heterogeneity. In this study, XGB-AGMoE is proposed, a validation-adaptive XGBoost-anchored Granular Mixture-of-Experts framework that combines the quantile-derived granular descriptors and experts of CatBoost, LightGBM, XGBoost, class-wise post hoc sigmoid calibration, leakage-free logistic-regression stacking, and an XGBoost-anchored probability fusion stage. The official test partition was set aside for final evaluation after disjoint development subsets were used for preprocessing, calibration, stacking, and anchor selection. The value of 0.80 was chosen for the anchor weight for the calibrated XGBoost, and 0.20 was chosen for the calibrated stack for the canonical seed-42 experiment. The resulting model achieved 0.993887 accuracy, 0.993681 weighted-F1, 0.851256 macro-F1, 0.999634 macro-ROC-AUC, and 0.921352 macro-PR-AUC on a 150,000-record official test sample. The Brier score, negative log-likelihood, and expected calibration error were 0.007531, 0.015309, and 0.001775, respectively. Across seeds 42, 52, and 62, accuracy was $0.9926\pm 0.0012$ and macro-F1 was $0.8238\pm 0.0251$. Recall was higher for the top denial of service classes and lower for DDoS Publish Flood and Recon VulScan. The results corroborate with expert anchoring and validation-controlled evaluation; they also highlight that there are fusion gains that depend on the data split and that they should be evaluated in the light of robust component baselines. Full article

Dissolved gas analysis (DGA) is widely used for transformer condition assessment, but fixed-resolution classifiers may release overly specific fault labels when the available evidence is weak. This study develops a risk-controlled adaptive-resolution framework in which a lightweight hierarchical residual neural encoder generates fine-grained and electrical/thermal parent-level candidates, and separate credibility estimates determine whether the released result should remain fine-grained, fall back to the parent mechanism, or abstain. The 25-dimensional DGA representation is encoded by three residual fully connected blocks (64-32-24), yielding 9492 trainable parameters for the hierarchical B2 model. Credibility combines calibrated uncertainty and DGA physics-consistency evidence, while data quality and distributional support are treated separately as deployment admissibility safeguards. Under 10 × 5 repeated evaluation, the B2 backbone achieved a Macro-F1 of 0.723 ± 0.026 in ordinary cross-validation and 0.664 ± 0.086 under reference-grouped evaluation. In the adaptive-resolution analysis, hierarchical risk decreased from 0.212 to 0.094 and cross-parent risk from 0.062 to 0.026 while retaining approximately 91.5% answer coverage in the ordinary protocol. Physics augmentation was context-dependent: it was inferior to the calibrated margin under ordinary resampling but substantially improved AURC under reference-grouped evaluation. CPU-only profiling gave a median end-to-end latency of 0.713 ms per observation (P95 1.028 ms), far below the sampling cadence of the utility online DGA data. The study therefore emphasizes risk-controlled information release rather than universal superiority of the underlying classifier.Exploratory temporal results are presented solely as a transparency analysis and are not interpreted as evidence of validated early-warning capability. Full article

Continuous block missingness in condition monitoring data poses a major challenge to the reliability of Remaining Useful Life (RUL) prediction for industrial equipment. Existing frameworks, including direct prediction from incomplete observations and decoupled two-stage imputation–prediction pipelines, suffer from a noticeable performance decline under prolonged data voids. Although Generative Adversarial Network (GAN)-based generative models have shown promise in data repair, they remain limited by dimensionality contamination from fixed-dimension architectures, purely data-driven over-smoothing that does not explicitly account for physical degradation laws, the difficulty of capturing instance-specific heterogeneous degradation signatures, and limited characterization of downstream predictive epistemic uncertainty under reconstructed inputs. To address these limitations, we propose a two-stage reconstruction–prognosis framework with uncertainty-aware downstream prediction, integrating two core modules: a Macro-Anchored Physics-Informed Generative Adversarial Network (MAP-GAN) tailored for missing sequence reconstruction, and an Attention-Bidirectional Long Short-Term Memory network integrated with Monte Carlo Dropout (Attention-BiLSTM-MCD) for uncertainty-aware prognosis, which provides an empirical estimate of predictive uncertainty under reconstructed inputs and produces RUL estimates with probabilistic intervals. Specifically, a macro-anchored micro-window generator confines the receptive field to mitigate dimensionality contamination; degradation-informed physical constraints are embedded into the adversarial objective to encourage the synthesized trajectories to satisfy physical degradation constraints; an encoder-guided latent space inversion mechanism adaptively captures individual-specific degradation patterns; and the Attention-BiLSTM-MCD approximates epistemic uncertainty to support prognostic reliability. The effectiveness of the proposed method is examined on a lithium-ion battery dataset with continuous block missingness, and is further verified through missing-pattern sensitivity analyses (missing ratios of 20–60% and early/late missing positions), physics-constraint term-wise ablation with weight sensitivity, an uncertainty-calibration analysis, and an external validation on a second, independent battery dataset. Full article

This paper proposes a multi-expert unified forecasting architecture with simplex aggregation (MUFASA), a development-anchored forecast-combination framework for multi-step hourly solar irradiation forecasting. A nonlinear temporal expert and a regularized ridge expert provide complementary high-capacity and low-variance inductive biases. Development-calibrated shrinkage is followed by Euclidean projection onto the probability simplex, enforcing non-negative unit-sum weights and convex-hull-bounded forecasts. Site-specific hyperparameters are selected by Gaussian process Bayesian optimization using data through 2019 only; 2020 is retained as a held-out evaluation year and does not enter parameter, seed-weight, or aggregation-weight estimation. The framework is evaluated over 11 hourly forecast steps from 08:00 to 18:00 across six major metropolitan areas in South Korea. Under the controlled conditional/oracle-weather protocol, macro RMSE was 0.3430 MJ m⁻², macro MAE was 0.2444 MJ m⁻², and macro R² was 0.8740. MUFASA achieved the lowest site-level RMSE at all six sites and ranked first in 62 of 66 site–horizon comparisons against 18 trainable benchmark architectures. Dependence sensitivity remained stable across HAC lags 1–28 and moving-block lengths 3–28; strict simultaneous 11-horizon superiority was supported against 12 of 18 comparators. The findings describe controlled conditional forecasting performance rather than operational NWP-driven day-ahead accuracy. Full article

Particle crushing changes both particle size distribution and particle shape, which significantly affects the mechanical behavior of sand. This study uses a multi-scale discrete element method to investigate the direct-shear response of sand under prescribed post-crushing scenarios. Instead of modeling dynamic particle splitting during shearing, assemblies with predefined fractal particle size distributions and size-dependent angular shapes were assigned prior to testing. The performance of multi-shape assemblies was systematically compared against single-shape circular baselines. The numerical model was calibrated against plane-strain direct-shear laboratory experiments using granular analogs. Macro-scale findings reveal that shear strength and volumetric dilation exhibit non-monotonic trends, reaching a peak response at intermediate fractal dimensions. Angular fragments act as structural wedges that sustain interlocking. Micro-scale analysis demonstrates that peak shear strength is consistent with the evolution of contact force anisotropy. Conversely, particle rotations show an inverse relationship with shear strength and angular fragments impose anti-rotational constraints along the shear band. Full article