Search Results (506)

Free-jet altitude simulation requires coordinated regulation of the intake thermodynamic states and flexible-nozzle geometry under time-varying Mach number commands and possible actuator faults. This study proposes a cooperative dual-loop active disturbance rejection control (ADRC) framework for a multi-pivot semi-flexible nozzle and a dual-valve intake system. In the inner loop, a constrained cooperative allocation method determines the pivot forces required to match the reference nozzle profile, while a dual-valve allocation mechanism redistributes the control demand when a valve loses effectiveness. ADRC is employed for hydraulic-actuator position tracking, and an SMC-LADRC controller regulates the intake pressure and temperature. An outer Mach number feedback loop compensates for the residual error of the integrated system. The convergence of the projected-gradient allocation algorithm and the ultimate boundedness of the observer and tracking errors are established under bounded disturbance variations. Simulation results under Mach number transitions, external disturbances, and progressive valve failure show that the proposed method effectively maintains stable intake pressure and temperature, improves nozzle-profile tracking accuracy, suppresses Mach number fluctuations, and enhances the fault tolerance and overall control performance of the integrated free-jet system. Full article

Machine learning and statistical classification methods are widely applied to microbiome data for diagnostic, prognostic, and phenotypic insights. However, the complex, multivariate nature of microbiome communities makes it difficult to assess the relative performance of these methods. Most comparisons rely on a small number of published datasets, without considering their underlying ecological properties or how these properties may, in turn, influence classification performance. We introduced a coenocline-based simulation framework to generate synthetic microbiome datasets that incorporate realistic ecological variation arising from species’ responses to host-associated gradients such as disease severity. To evaluate the ecological fidelity of these simulations, we compared synthetic datasets to five widely used real-world microbiome datasets: Cirrhosis, Colorectal Cancer (CRC), Type 2 Diabetes (Chinese and Women cohorts), and the Human Microbiome Project (HMP). Comparisons across α-diversity (species richness), β-diversity (species composition and turnover), and abundance distributions demonstrated that coenocline simulations closely recapitulate the key ecological structures of empirical data. Synthetic datasets exhibited similar richness and abundance patterns to disease-associated microbiomes, with realistic distributions of few dominant and many rare taxa. Moreover, community composition analyses (Bray–Curtis index) revealed that the simulated datasets captured natural levels of compositional dissimilarity among samples, spanning the same variability range observed in real data. When compared against 100 independently simulated datasets, the coenocline model consistently reproduced empirical ranges of species diversity, relative abundance, and between-group compositional differences (ANOSIM-R values), confirming the model’s robustness and reproducibility. This coenocline-based simulation framework provides a novel, flexible, and ecologically grounded approach for generating synthetic microbiome data with controlled complexity. By reproducing realistic ecological gradients and community structures, the framework supplies the controlled test beds needed for systematic future benchmarking of machine learning and statistical classification methods across diverse and biologically meaningful scenarios. In doing so, it will help bridge the gap between ecological realism and computational modeling, thereby supporting more reliable and generalizable inference from microbiome data. Full article

The Jinsha River Basin on the eastern margin of the Qinghai–Xizang (Tibetan) Plateau is one of the most landslide-prone regions globally. The Batang reach (from Suwalong Township to Changbo Township) lies in the core of the Jinsha River Suture Zone, characterized by complex geological conditions and frequent landslide disasters. To address the limitations of traditional monitoring methods—including difficulty in full-area coverage, high costs, and low efficiency in mountainous canyon terrain—a systematic study of landslide monitoring was conducted using Sentinel-1A SAR data and Interferometric Synthetic Aperture Radar (InSAR) technology. The results demonstrate that SBAS-InSAR, via short-baseline combinations, achieves a monitoring point density of 697 points/km² (6.28 times that of PS-InSAR), offering significant advantages in mountainous canyon areas with dense vegetation and fragmented rock masses. Using this technical framework, a total of 38 active landslides were identified in the study area, including 5 newly detected rapidly deforming hazards, 17 with river blockage risk, and 10 with the potential to bury buildings. The maximum downslope deformation rate reaches −89.56 mm/yr for the landslide near Suwalong Hydropower Station. Landslides are concentrated within 500 m of faults, in weak rock zones, on steep slopes with gradients greater than 30°, and near road-cutting projects. Temporally, landslide deformation shows an evident correlation with rainfall; approximately 70% of annual cumulative deformation occurs within the rainy season. Engineering activities including hydropower station impoundment appear to be associated with elevated deformation rates of local landslides. This study provides a scientific basis for the safety of major infrastructure corridors (e.g., the Sichuan–Tibet Railway) and regional disaster prevention and mitigation. Full article

(1) Background: Climate change is reducing the climatic windows in which outdoor sport and physical activity can be safely practised. Yet, it remains unclear whether the resulting population exposure is distributed evenly across countries or is concentrated in countries with higher socio-economic vulnerability. Here, we assess the country-level distribution of population exposure to sport-relevant heat along a projected socio-economic vulnerability gradient. (2) Methods: We assess a screening-level, population-based proxy of sport-relevant heat exposure, defined by residents living in grid cells experiencing specified afternoon WBGT exceedance frequencies. We first quantify a population exposure burden by combining reconstructed afternoon wet-bulb globe temperature (WBGT) exceedance frequencies from bias-corrected CMIP6 projections (NEX-GDDP, five-model ensemble) with SSP-consistent gridded population. The resulting grid-cell exposure burden, expressed in person-days yr⁻¹, is aggregated to the country level. Countries are then ranked using a published national socio-economic vulnerability index (GVI), and the distribution of the exposure burden along this vulnerability gradient is quantified using a population-weighted concentration index (CI). The GVI is used as the ranking variable and is not incorporated into a multiplicative vulnerability-weighted risk metric. Two internally coherent futures (SSP1-2.6 and SSP2-4.5, each paired with its corresponding socio-economic pathway) are evaluated at mid-century (2055) and late century (2085). (3) Results: The population exposure burden is concentrated in more vulnerable countries in all scenario–horizon–threshold combinations (CI > 0; 0.09–0.25). Under SSP1-2.6 at mid-century, the five-model mean CI showed a descriptive increase from 0.17 for WBGT ≥ 28 °C to 0.25 for WBGT ≥ 32 °C, although the 95% paired hierarchical-bootstrap interval for the 32-minus-28 °C contrast included zero. By 2085, about 4.11 billion people live in cells experiencing at least 30 days yr⁻¹ with WBGT ≥ 32 °C under SSP2-4.5, compared with 2.14 billion under SSP1-2.6. (4) Conclusions: The results identify projected between-country inequalities in resident population exposure to sport-relevant heat conditions; they do not quantify actual sport participation, athlete exposure or health outcomes. The comparison between SSP1-2.6 and SSP2-4.5 further shows substantial differences in both climatic hazard and population exposure between these two integrated climate–demographic–socio-economic futures. Full article

The decarbonization of industrial steam production, representing up to 57% of energy use in the food industry, is critical for achieving EU climate neutrality goals. This study developed an integrated digital framework for the research project Hy4GreenSteam to optimize green-hydrogen integration through advanced predictive modeling. The employed LightGBM gradient-boosting algorithms were trained on 68,697 PV power measurements and 57,000 meteorological observations from 2020 to 2022. A “Production-Split” methodology was introduced for 24 h ahead forecasting, segmenting training into high (>2 kW) and low (≤2 kW) production regimes to manage solar heteroscedasticity. Results show the 15 min model achieved an R² of 0.868 and the 1 h model an R² of 0.832, while the day-ahead model—trained exclusively on information available at forecast issue time—achieved an R² of 0.701, a 70% relative improvement over same-time-yesterday persistence. A complementary regime analysis shows that the production regime is predictable with 90.7% accuracy and quantifies the accuracy headroom of regime-specialized models (oracle R² 0.794). These methods were integrated into a real-time React-based platform that calculates optimal H₂/CH₄ blending; for the reference pilot configuration, driven by measured on-site PV generation, the computed CO₂ emission reduction reaches 34% relative to natural-gas-only operation during high-solar operating intervals. Predictive modeling combined with a Digital Twin interface provides a TRL 6 decision-support solution, demonstrated in a relevant industrial environment, for managing renewable sources in industrial hydrogen applications. Full article

This paper considers priority-aware partial computation offloading in an uplink mobile edge computing (MEC) network. Devices assigned to different groups occupy orthogonal subbands, whereas devices within each group use power-domain non-orthogonal multiple access (NOMA) with successive interference cancellation. Task-input size determines the transmitted and processed workload, while queue backlog and application urgency determine the service weight. The Gaussian multiple-access-channel rate region is convex, but the complete allocation problem is not jointly convex in the adopted variables because the offloaded workload is coupled with reciprocal transmission rate and reciprocal edge-CPU allocation. A structure-exploiting block-coordinate projected-gradient method is developed. It combines exact finite-candidate offloading updates, an exact edge-CPU allocation bounded below by deadline feasibility and above by local-path saturation, and an analytical projected power step with Armijo backtracking. For eight users at 23 dBm, pairwise group-based NOMA reduces the weighted delay–energy cost and device energy by 6.18% and 23.44%, respectively, relative to orthogonal access. Queue-aware weighting reduces upper-backlog-quartile delay by 2.69 ms (95% confidence half-width: 0.78 ms) while increasing lower-quartile delay by 8.34 ms (half-width: 2.07 ms). In a paired 15-iteration ablation, generic projected block-coordinate updates have a cost ratio of 1.0098 (half-width: 0.0086) relative to the structured method. A hybrid deep deterministic policy-gradient policy, evaluated over five training seeds, has an 11.77% higher cost while requiring 0.84% of the median online decision time. Of 432 allocations, 392 satisfy the residual-qualified stopping tests and 40 are explicitly reported as iteration-safeguard terminations. Full article

Occupancy detection is fundamental to the operational intelligence of smart buildings, driving critical functions in energy management, HVAC automation, and physical security. While modern Deep Learning (DL) models have achieved high accuracy in parsing complex environmental sensor data, they remain highly vulnerable to adversarial examples, imperceptibly perturbed inputs designed to deceive neural networks. These vulnerabilities pose severe real-world risks, ranging from energy sabotage, in which systems heat empty rooms, to critical security breaches in which intruders go undetected. To address this security gap, we propose ADS-Guard, a novel Adversarial Detection and Sanitization (ADS) framework rooted in sequence-to-sequence autoencoder purification. Unlike standard denoising techniques, ADS-Guard incorporates a latent consistency regularization mechanism that encourages alignment between clean and adversarial representations in the latent feature space. We evaluated ADS-Guard using a comprehensive experimental pipeline comprising five distinct DL architectures (LSTM, GRU, 1D-CNN, MLP, and Transformer) across three diverse datasets: (1) The UCI Occupancy dataset (20,699 samples) for standard binary detection; (2) Building59 dataset (7200 samples) for three-class occupancy-level classification (Low, Medium, High); and (3) Room Occupancy dataset (10,129 samples), representing a highly imbalanced binary occupancy-detection task. We evaluate ADS-Guard against both Fast Gradient Sign Method (FGSM) and Projected Gradient Descent (PGD) attacks across diverse occupancy datasets and model architectures. We further assess the framework under adaptive white-box attacks and compare its performance with FGSM-based and PGD-based adversarial training baselines. Our results demonstrate that adversarial attacks can substantially degrade occupancy-detection performance across datasets and model architectures. ADS-Guard consistently improves robustness relative to undefended models against both FGSM and PGD attacks, recovering a substantial portion of the lost performance in binary occupancy tasks and providing meaningful gains in the more challenging multi-class setting. Furthermore, ADS-Guard remains effective under stronger adaptive threat models while providing a practical retraining-free defense that can be integrated with existing occupancy-detection systems without modifying downstream classifiers. Full article

Under the strategic goals of “peak carbon emissions and carbon neutrality”, traditional methods for solving economic dispatch problems involving carbon emission trading costs, wind power, and energy storage devices lack real-time performance and are difficult to support in real-time decision-making. This paper proposes an accelerated solution framework that expands the projection gradient descent method into a Deep Projected Gradient Network (D-PGNet). The network consists of K structured layers, each layer strictly embedding differentiable projection operations corresponding to physical constraints such as power balance, unit ramp-up, and energy storage timing dynamics. This paper systematically derived the projection closed-form solutions of each constraint set to the basic subset, designed an efficient differentiable projection layer based on Dykstra alternating projection, and analyzed the differentiability and convergence properties of the network. Multiple scenario tests have shown that the optimal gap of D-PGNet results is less than 1.08%, the carbon emission deviation is less than 0.2%, the solving speed is improved by more than 64 times at most, and the maximum violation of all physical constraints is below 1.3 × 10⁻⁵ p.u. Full article

In medium-voltage DC distribution systems (MVDC-DSs), feeder congestion may occur when multiple converter terminals sharing the same AC feeder experience fast source–load variations. Limited communication further challenges real-time converter coordination and secure system operation. To address these issues, this paper has proposed an adaptive online management strategy for multi-terminal feeder congestion in MVDC-DSs under limited communication. First, the economic operation objective and practical system constraints are formulated within a distributed optimization framework, where feeder congestion limits, voltage security, and load supply requirements are explicitly incorporated. Then, an adaptive online regulation mechanism is embedded into local converter controllers, enabling converter power to be coordinated in real time using only neighboring information. In this way, feeder congestion can be mitigated while reliable load supply and economic operation are maintained. Simulation studies and hardware-in-the-loop experimental results demonstrate the effectiveness, scalability, and real-time applicability of the proposed strategy under dynamic operating conditions. Full article

During deep excavation construction, deformation impacts on adjacent structures are inevitably induced, and field monitoring data are often contaminated by noise that degrades prediction accuracy. To address these issues, this study develops a joint denoising strategy combining CEEMDAN, sample entropy-based adaptive IMF screening, and an improved wavelet threshold (IWT) function, followed by a Bayesian optimization-based extreme gradient boosting (BO-XGBoost) model for surface settlement prediction. The developed method automatically identifies high-noise IMF components via sample entropy and processes them using an improved threshold function that overcomes the discontinuity of hard thresholding and the constant bias of soft thresholding, thereby preserving useful information while suppressing noise. Experimental results on a Wuhan metro deep excavation project demonstrate that the CEEMDAN-IWT method improves SNR by up to 4.09% and reduces RMSE by up to 8.00% compared with conventional CEEMDAN-wavelet threshold denoising. The BO-XGBoost model trained on denoised data achieves an RMSE of 0.09 mm and a MAPE of 3.54%, outperforming BP, LSTM, standard XGBoost, GRU, CNN-LSTM, TCN, and simple regression baselines. Feature importance analysis confirms that the denoised data retain physical interpretability consistent with soil deformation continuity. This framework provides a practical solution with promising accuracy for deformation monitoring and early warning in deep excavation engineering. Full article

Point cloud perception is important in autonomous driving, robotics, and other security-critical 3D systems, yet learned point cloud classifiers remain vulnerable to transferable adversarial perturbations. A central difficulty in transfer-based black-box attacks is surrogate overfitting: an update that is highly effective on an accessible source model may not generalize to an unknown target architecture. We introduce SpecEOT, a source-agnostic and graph-spectral expectation-over-transformation attack. A fixed graph Fourier transform (GFT) basis is constructed from each clean point cloud. At every optimization iteration, each non-identity view independently samples a frequency band and a perturbation sign from uniform distributions; the resulting view gradients are averaged with equal weights and used to update the adversarial point cloud through projected Adam ascent. We evaluate the stochastic method over repeated seeds, extend the ablation to two source architectures, and analyze the interaction between band count and randomization strength while reporting computational cost and assessing robustness to Gaussian jitter and point dropout. SpecEOT achieves strong transferability on ModelNet40 and ShapeNet. Full article

Background: Myopia represents a growing global public health concern, with prevalence projected to reach 4.9 billion people by 2050. Orthokeratology has emerged as a non-invasive optical intervention for myopia control in children. This systematic review and meta-analysis evaluated the efficacy of orthokeratology in reducing axial length (AL) growth compared to spectacles, atropine combinations, multifocal contact lenses, gradient soft contact lenses, modified orthokeratology with enhanced peripheral defocus (enhanced OK), and atropine therapy with single-vision spectacle correction. Methods: A systematic literature search was conducted in PubMed and Google Scholar for studies published in the past 20 years (2004–2024) using keywords “orthokeratology” OR “corneal reshaping” AND “myopia control.” Inclusion criteria comprised longitudinal clinical studies (randomized or non-randomized) with AL measurements, at least two groups including orthokeratology, and publication in English or Portuguese. Meta-analysis was performed using Review Manager 5.0.25 with random-effects models. Heterogeneity was assessed using I² statistics, and subgroup analyses were conducted by follow-up duration. Results: Thirty-one studies met inclusion criteria. Orthokeratology significantly reduced AL growth compared to spectacles (25 studies, 4267 patients; mean difference [MD] = −0.16 mm, 95% CI −0.19 to −0.14, p < 0.00001, I² = 97%). Combined orthokeratology with atropine showed superior efficacy to orthokeratology alone (8 studies, 888 patients; MD = 0.09 mm, 95% CI 0.06 to 0.12, p < 0.00001, I² = 98%). Orthokeratology demonstrated equivalent efficacy to multifocal contact lenses (3 studies, 208 patients; MD = 0.00 mm, 95% CI −0.04 to 0.05, p = 0.85, I² = 83%). Atropine therapy with single-vision spectacle correction and enhanced OK showed superior myopia control compared to conventional orthokeratology. Conclusions: Orthokeratology effectively slows axial elongation in myopic children compared to spectacles, with efficacy comparable to multifocal contact lenses. Combination therapies, particularly orthokeratology with atropine, demonstrate additive benefits. High heterogeneity across studies highlights the need for standardized protocols and long-term follow-up studies to optimize myopia control strategies. Full article

In constrained environments with no-fly zones (NFZs), bearing-only UAV target localization requires improving UAV-target relative geometry while avoiding NFZs. Inspired by the avoidance behavior observed in flying insects driven by perceptual stimuli, this paper proposes a bio-inspired anisotropic on-manifold guidance method for bearing-only UAV target localization under NFZ constraints. First, to quantify the measurement information with respect to the UAV-target relative geometry under unknown sensor bias, a projected uncertainty field is proposed. Its spatial gradient drives a perception-guided control law, and its components are further used to characterize local information sensitivity as a perceptual stimulus. Second, this sensitivity is incorporated into NFZ margin adaptation to construct an anisotropic dual-layer manifold field, enabling the UAV to adapt the avoidance margin according to the local information distribution. An on-manifold modulation mechanism is then applied to generate feasible motion under NFZ constraints. Simulations in NFZ-free, nonconvex NFZ, and dense NFZ scenarios show that the proposed method generates UAV trajectories that avoid NFZs while maintaining accurate and convergent target localization. Full article

In scrubbing tower (ST) systems, outlet vapor temperature regulation is subject to strict thermal constraints, requiring the bottom temperature to remain below a critical safety threshold. Transient overshoots can rapidly trigger pyrolytic coking, foul mass-transfer packed beds and spray nozzles, and ultimately force complete production shutdowns. Due to feedstock compositional drifts, high thermal inertia, and significant transport delays, high-fidelity predictive identification is essential for proactive early warning and for overcoming the limitations of reactive feedback control. To address these bottlenecks, this paper introduces an offset-free, hard-constrained, data-driven adaptive optimal control paradigm, designated as the improved GRU-coupled conjugate gradient linear quadratic regulator (IGRUCG-LQR). First, by constructing an augmented state space embedded with integral error, the proposed paradigm eliminates permanent tracking offsets induced by long-term nonstationary drifts. Second, automatic differentiation is used to extract the time-varying Jacobian matrix of a gated recurrent unit (GRU) online, thereby tracking the nonlinear evolution of the underlying thermodynamic baseline with high fidelity. To manage the critical trade-off between strict actuator saturation and short real-time sampling intervals, the conjugate gradient (CG) method is fused with a hard-boundary projection operator, enforcing physical constraints with high computational efficiency and without complex matrix inversions. Experimental validation on a real-world industrial dataset demonstrates that the proposed paradigm secures the safety baseline while achieving high-resolution transient tracking. Furthermore, it significantly suppresses high-frequency valve chattering to mitigate mechanical fatigue, establishing a solid theoretical and engineering foundation for the prolonged stable operation of safety-critical processes. The proposed framework achieves an Integral Absolute Error (IAE) of 86.66 and an Integral Time Absolute Error (ITAE) of 4064.49. Full article

Alzheimer’s disease (AD) is the most prevalent neurodegenerative disorder worldwide, affecting approximately 56.9 million people in 2021 and projected to reach 152 million by 2050. Its defining pathological features, amyloid-beta plaques and neurofibrillary tangles, accumulate for up to two decades before cognitive symptoms emerge, placing the preclinical and mild cognitive impairment (MCI) stages at the centre of the early detection problem. Despite this, current diagnostic practice in routine clinical settings remains unreliable, with post-mortem studies placing the specificity of clinical AD diagnosis between 44.3 and 70.8% even in specialist memory clinics. Machine learning (ML) and deep learning (DL) applied to neuroimaging and electrophysiological data have emerged as candidate tools for closing this diagnostic gap, yet whether the accuracy figures reported in published studies translate into clinically useful performance on independent data remains unresolved. This study presents a structured comparative review of machine learning and deep learning methods reported across four publicly available Alzheimer’s disease datasets, namely the Alzheimer’s Disease Neuroimaging Initiative (ADNI), the Open Access Series of Imaging Studies (OASIS), the OpenNeuro ds004504 electroencephalography (EEG) dataset, and the Kaggle Alzheimer’s magnetic resonance imaging (MRI) dataset. Thirteen model families are examined through the published literature rather than through new experiments, and for each model and dataset combination, the best accuracy reported in the source study is recorded alongside the model’s mathematical formulation. All performance figures reported in this abstract and throughout the paper are taken from the published studies reviewed, not from new experiments conducted by the authors. Across the reviewed studies, deep learning architectures pre-trained on ImageNet and fine-tuned on neuroimaging data are reported to produce the highest accuracy on MRI classification tasks. Residual Network (ResNet)-101 is reported at 98.21 percent on ADNI and 97.45 percent on OASIS, while the IncepRes fusion architecture reaches 98.35% on OASIS by combining multi-scale feature extraction from InceptionV3 with residual connectivity from ResNet152V2. Traditional machine learning classifiers remain competitive on tabular clinical and biomarker data, with Extreme Gradient Boosting (XGBoost) reaching 91% on ADNI multiclass features. Logistic Regression achieves 82 to 85% on binary ADNI classification and is the only classifier in this review that provides explicit per-feature prediction contributions without post hoc tooling. Gaussian Naïve Bayes achieves 80 to 83% on the same task. On the OpenNeuro EEG dataset, K-nearest neighbours (KNN) with singular value decomposition (SVD) entropy features achieves 91% binary accuracy, with feature engineering quality determining performance more reliably than classifier architecture. Eight principal findings emerge from the cross-dataset analysis. Binary classification consistently outperforms multiclass by 10 to 30% across all datasets, reflecting the genuine biological ambiguity of the mild cognitive impairment category. Dataset size and augmentation predict reported accuracy more reliably than model architecture. Ensemble methods outperform individual classifiers by 5 to 8% in nearly every imaging study. Deeper architectures can overfit small clinical cohorts. EEG models trail MRI models by approximately 10 to 15% on comparable binary tasks. Cross-dataset generalisation has not been systematically evaluated in most studies, and the few that have tested it report accuracy drops of 5 to 10% or more when models encounter data from different scanners or cohorts. Eight recurring limitations constrain the clinical utility of these findings. Small sample sizes and limited demographic diversity, severe class imbalance inflating raw accuracy metrics, poor cross-dataset generalisation driven by scanner heterogeneity, limited deep learning interpretability, the dominance of binary over multiclass tasks, the absence of longitudinal modelling despite available datasets, inadequate standardisation of preprocessing and evaluation protocols, and the signal-to-noise ratio constraints specific to EEG recordings of elderly patients collectively define the gap between benchmark performance and clinical readiness. Future work must prioritise multi-centre training cohorts, multimodal fusion architectures, longitudinal progression modelling, and standardised interpretability evaluation as non-optional requirements for any system intended for clinical deployment. Full article