Search Results (1,815)

The increasing penetration of nonlinear industrial loads, distributed renewable generation, and intelligent electrical infrastructures requires active power filters capable of simultaneously providing high-performance harmonic mitigation, reactive power compensation, coordinated operation of multiple converters, and deterministic real-time implementation. Conventional centralized shunt active power filters (SAPFs) exhibit limited scalability, while optimization-based approaches often compromise deterministic execution because of their computational complexity. To address these challenges, this paper proposes a Distributed Quantum Multi-Shunt Active Power Filter (Quantum Multi-SAPF) that combines deterministic-based local current control with asynchronous quantum-assisted supervisory optimization. The proposed architecture employs four distributed SAPF units operating under a hierarchical cyber–physical framework. The lower control layer, implemented on a MATLAB R2026a, includes all fast electrical functions—signal acquisition, SOGI-based synchronization, Clarke transformation, instantaneous $p$–$q$ current reference generation, current regulation, interleaved PWM modulation, and protection—which are executed deterministically at a switching frequency of 15 kHz. The upper supervisory layer operates asynchronously at 20 Hz and formulates converter coordination as a quadratic unconstrained binary optimization (QUBO) problem solved using Quantum Approximate Optimization Algorithm (QAOA) allocation together with Variational Quantum Eigensolver (VQE) predictive correction. This multi-rate architecture separates fast electrical dynamics from slow supervisory optimization, ensuring that uncertain optimization latency does not affect converter stability. The proposed controller is validated through comprehensive switching-level simulations on the MATLAB R2026a platform. Numerical results demonstrate a reduction in source current total harmonic distortion from 24.615% to 0.142%, corresponding to a 99.423% harmonic reduction, while improving the source power factor to 0.99999 and achieving 99.999% reactive power compensation. The distributed four-SAPF synchronization network maintains coherent phase alignment among all converter units throughout the simulation, thereby supporting coordinated compensation and balanced current sharing. This synchronized operation contributes to highly accurate compensation current tracking, with an RMS tracking error of only 0.026 A, while limiting the source current unbalance to 0.026%. These results confirm the effectiveness of the distributed synchronization and local control architecture in maintaining coordinated and balanced operation of the four parallel SAPFs. The proposed interleaved modulation strategy, combined with optimized current sharing, maintains balanced converter utilization while suppressing circulating currents without requiring a dedicated circulating current controller. The proposed Distributed Quantum Multi-SAPF establishes a scalable framework that combines deterministic industrial control with quantum-assisted supervisory optimization. The architecture provides high harmonic compensation capability, near-unity power factor, balanced converter utilization, comprehensive Safe Operating Area supervision, and practical industrial feasibility, making it a promising solution for future smart grids, renewable energy integration, electric vehicle charging infrastructures, and intelligent power quality conditioning systems. Full article

High-fidelity finite-element (FE) simulation of nonlinear full-field responses in AlSi10Mg tensile specimens is computationally expensive, while the available digital image correlation (DIC) data in this study consist only of engineering-strain screenshots rather than exportable numerical fields. This work therefore develops a simulation-to-experiment framework for rapid full-field stress prediction and weakly supervised DIC-informed spatial regularization. Two hundred independent Abaqus/Explicit loading cases generated 5600 graph samples, which were used to train an eight-layer FiLM-conditioned graph neural network for nodal von Mises stress and three-component displacement prediction, together with a decoupled structured force–displacement branch. On independent FE tests, von Mises stress prediction achieved R² = 0.9984 with an RMSE of 4.60 MPa, while the dense force–displacement curve achieved R² = 0.9957 and RMSE = 59.70 N. Five machine tensile tests gave mean pre-fracture curve NRMSEs of 4.40% for FE versus experiment and 5.31% for GNN versus experiment. On independent specimens L-7 and L-8, the constrained adapter reduced DIC spatial shape loss by 3.21% with a mean absolute stress correction of 0.232 MPa. The framework therefore provides rapid FE-surrogate prediction with conservative experiment-informed spatial regularization rather than experimental stress inversion. Full article

The main reason cageless bearings fail as protective bearings for magnetic levitation bearings is the uncontrolled motion of isolated rolling elements without a cage. This leads to sliding, friction, and collisions between the rolling element and the raceway, or among the rolling elements themselves. In this work, the universal nonlinear dynamic model of cageless bearings is established. The kinematic behavior of the rolling element in the cageless bearing is analyzed, and the sliding and discontinuous contact collision phenomena of the rolling element are explained from a kinematic perspective. Subsequently, the influence of different working conditions on the rolling-element contact behavior is demonstrated. In the model incorporating the randomness of rolling elements in cageless bearings, the numerical solution method is optimized. Experimental results confirm that the model is correct. The primary contact between adjacent rolling elements occurs in the transition zone between the loaded and unloaded zones. The higher the rotational speed, the less likely the rolling element is to collide and make contact. The larger the axial load, the more likely the rolling element is to rub continuously. The influence of radial load on the slip rate of the rolling elements is relatively small. These results provide a suitable operating range for cageless bearings. Full article

Reliable probabilistic prediction of dam-break wave run-up is important for screening downstream barriers and river infrastructures. However, it remains unclear whether propagation conditions exhibit stable statistical groups within the observed predictor domain and whether explicitly incorporating these groups improves prediction beyond the underlying continuous variables. This study develops a probabilistic framework integrating physics-guided self-tuning spectral clustering (PG-STSC), quantile regression forests (QRFs), and conformal calibration diagnostics. The framework was applied to 155 physical model observations characterized by relative propagation distance, Froude number, relative wave height, relative wavelength, and wave nonlinearity, with the relative maximum run-up as the response variable. Full-data PG-STSC identified two groups containing 66 and 89 observations, separating mainly shorter-distance, higher-$Fr$ conditions from longer-distance, lower-$Fr$ conditions. The partition showed moderate physical-space separation but high subsample stability, while the group-wise run-up distributions overlapped substantially. In fivefold out-of-fold evaluation, the group-free QRF achieved a continuous ranked probability score (CRPS) of 0.1537, corresponding to 22.8% CRPS skill relative to the unconditional empirical distribution. Global and group-conditioned conformal corrections produced identical pooled 90% coverage (0.942) and interval widths, with group-specific coverages of 0.922 and 0.952, respectively. Supported-domain exceedance maps translated the predicted conditional distributions into threshold screening information while masking unsupported extrapolation. Overall, physically coherent clustering improves interpretation and conditional diagnostics but does not necessarily enhance predictive performance. The resulting exceedance estimates are suitable for within-domain scenario screening rather than site-specific overtopping assessment. Full article

During service, conformal load-bearing antennas (CLBAs) are subjected to the coupled effects of aerodynamic and aerothermal loads. The resulting geometric distortion of the array surface, deflection of element pointing, and temperature drift of the material electromagnetic parameters lead to the degradation of radar cross-section (RCS) characteristics. To overcome the limitation of existing scattering models in uniformly describing the aforementioned multi-physics coupling effects, this paper proposes a comprehensive electromechanical–thermal coupled modeling method for the scattering field of CLBAs. This method establishes a complete mapping from flight conditions to the array RCS by incorporating geometric corrections for element-level pointing deflection and bending deformation, material corrections accounting for the temperature-dependent antenna efficiency, and phase corrections induced by aerodynamic displacements. Verification using a 9 × 9 cylindrical conformal array shows that, within a scanning range of ±30°, the model calculations agree with HFSS full-wave simulations with an absolute error of less than 1 dB, and the broadside RCS is reduced by 14.97 dB compared with that of a planar array. Furthermore, a BP neural network surrogate model is constructed to achieve accurate prediction of the array physical fields. Analyses across the Mach regime of 0.20–0.65 Ma indicate that structural deformation is the dominant cause of RCS distortion, with the trailing-edge array experiencing a rapid nonlinear increase in RCS peak increment, reaching up to 7 dB at 0.65 Ma. The proposed model provides an effective theoretical tool for the rapid evaluation of stealth performance for conformal antennas operating in complex environments. Full article

Background: We developed and externally validated a model for estimating the probability of clinical pregnancy after in vitro fertilization and fresh embryo transfer using routinely available laboratory and multimodal ultrasonographic measures. Methods: This two-center retrospective study included 451 women in the development cohort and 229 women in the external validation cohort. Eleven candidate predictors were evaluated using Group LASSO-penalized logistic regression with stratified 10-fold cross-validation. Continuous predictors were assessed for nonlinearity using restricted cubic splines, and the complete model-development procedure was repeated in 1000 bootstrap resamples. Model performance was assessed by discrimination, calibration, overall prediction error, and decision curve analysis. Results: The final model included female age, serum anti-müllerian hormone (AMH), endometrial thickness, endometrial volume, and vascularization flow index (VFI). The apparent area under the receiver operating characteristic curve (AUC) was 0.864 (95% confidence interval [CI], 0.826–0.898) in the development cohort; the optimism-corrected AUC was 0.842. The optimism-corrected calibration intercept, calibration slope, Brier score, and maximum absolute calibration error (Emax) were 0.002, 1.051, 0.163, and 0.045, respectively. In external validation, the AUC was 0.810 (95% CI, 0.750–0.868), the calibration intercept was −0.106, the calibration slope was 0.900, the Brier score was 0.177, and Emax was 0.070. Decision curve analysis indicated net benefit over selected threshold probabilities. Conclusions: The model showed good discrimination and generally satisfactory calibration. It may support individualized counseling before fresh embryo transfer, but prospective validation in larger and more diverse populations is required. Full article

Background/Objectives: Antibiotic residues in aquatic environments reflect interacting physicochemical, climatic, and microbiological processes. This study characterized selected antibiotics in wastewater and surface water associated with the Danube River near Novi Sad, Serbia, and evaluated explainable stacked machine-learning models for concentration prediction. Methods: Thirty-six samples collected during summer and autumn 2024 were analyzed using SPE-HPLC-MS/MS. Artificial neural network, random forest, support vector machine, XGBoost, stacked linear, and stacked random forest (STACK-RF) models were developed using environmental/physicochemical variables or presumptive resistant bacterial taxa. Models were evaluated by fivefold cross-validation, complementary error metrics, Holm-adjusted Diebold–Mariano tests, XGBoost Gain, and SHAP analysis. Results: All target antibiotics were detected at least once. Azithromycin was most prevalent (75.0%), followed by sulfamethoxazole (58.3%), trimethoprim, and ciprofloxacin (52.8% each), while wastewater generally exhibited broader antibiotic profiles and higher concentrations than surface water. Standalone algorithms showed weak-to-moderate performance, whereas STACK-RF achieved the highest numerical accuracy for all environmental/physicochemical models (R² = 0.745–0.913) and the available microbial-taxa models (R² = 0.819–0.945), with consistently lower prediction errors. However, most pairwise differences were not significant after Holm correction. Influential environmental predictors were compound-specific and included COD, BOD₅, pH, turbidity, electrical conductivity, water temperature, and relative humidity. Leading bacterial predictors included Klebsiella pneumoniae, Escherichia coli, Citrobacter freundii, and Aeromonas veronii. Conclusions: The results provide a proof-of-concept for machine-learning-assisted antibiotic prediction. Explainable STACK-RF modeling captured nonlinear, antibiotic-specific associations among residues, water-quality conditions, and microbial indicators. It may complement targeted chemical monitoring and support hypothesis generation, although larger, externally validated datasets are required before broader application. Full article

Existing top-down carbon emission estimation studies often rely on single models or conventional ensemble approaches, which may limit their ability to capture complex nonlinear relationships under small-sample conditions. To address this limitation, this paper proposes a heterogeneous ensemble stacking network (HESTNet) for city-level monthly CO₂ emission estimation using electricity-centered multi-source data. A candidate feature set integrating sector-specific electricity consumption, socioeconomic statistics, nighttime light data, and environmental and climatic variables is first constructed, from which 28 core features are selected using autoencoder reconstruction errors. A heterogeneous stacking ensemble comprising CatBoost, TabPFN v2, and TabM is then developed, with five-fold out-of-fold predictions fused by an L2-regularized Ridge meta-learner. Because reliable ground-truth city-level monthly CO₂ observations are generally unavailable, the model is trained and evaluated using 360 province-year samples from 30 provincial-level regions in mainland China during 2013–2024, with provincial annual CO₂ emissions used as supervised labels. At the supervised province-year evaluation scale, HESTNet achieves an R² of 0.914, RMSE of 0.0781, MAE of 0.0580, and MAPE of 9.38%. Compared with the conventional stacking baseline, HESTNet yields numerical improvements of 0.028 in R² and approximately 12.6%, 13.2%, and 14.0% in RMSE, MAE, and MAPE, respectively; however, the paired RMSE difference does not reach statistical significance after Holm–Bonferroni correction (adjusted p = 0.061). The trained model is subsequently applied to city-month-scale predictors to generate relative Emission Proxy Indices (EPIs). Under the constraint of official city-level annual CO₂ emissions, the EPIs are normalized into monthly allocation weights to derive model-derived, annual-constrained monthly CO₂ estimates. The proposed framework integrates electricity-centered multi-source proxies, heterogeneous ensemble learning, and annual-constrained temporal disaggregation, providing a data-driven approach for characterizing intra-annual variations in city-level CO₂ emissions. Full article

Torque characteristic surfaces in nonlinear hydraulic turbine models are typically derived from model tests. However, model–prototype discrepancies can reduce the accuracy of dynamic simulations of hydropower units. To improve hydraulic turbine modeling accuracy, this study uses measured operational data to perform a multi-stage correction of the torque characteristic surface. The method calibrates the input–output mapping of the original surface through sequential parameter estimation, with parameters fixed after each stage. Polynomial, Gaussian kernel and Sigmoid functions are combined with six port sequences to construct 18 correction schemes, with the parameters at each stage optimized using particle swarm optimization. The results show that correction accuracy and the preferred sequence depend on the function form. For the studied unit, the Gaussian kernel with the “guide-vane opening–unit torque–unit speed” sequence yields the lowest weighted composite error, reducing it by 80.76% relative to the original model. The corrected data in the normal operating region are further used to construct the zero-opening and zero-unit-speed boundaries, which are combined with the runaway-speed boundary to reconstruct the full-operating-range torque characteristic surface using a backpropagation neural network (BPNN). The resulting NRMSE and NMaxAE are 0.54% and 1.58%, respectively. The corrected model is then embedded in the hydropower unit for multi-condition validation under primary frequency regulation. The mean RMSE and MAE of active power decrease by 45.10% and 50.46%, respectively, while the mean accuracy of the response regulation magnitude increases to 99.27%. The prediction error of guide-vane opening is also reduced. These results demonstrate that the proposed method effectively reduces model–prototype discrepancies and improves the accuracy of dynamic prediction under primary frequency regulation. Full article

To address the problems of uneven illumination distribution, missing edge details, and noise interference in underground coal mine images, a mine image enhancement algorithm based on local perception and multi-domain guided fusion is proposed. First, the algorithm constructs a dynamic nonlinear luminance mapping function based on local luminance features to adaptively adjust the enhancement amplitude. This addresses the issues of overexposure in strong light areas and missing details in dark areas caused by uneven illumination. Second, an adaptive gradient enhancement strategy is introduced to construct a gradient weight matrix. This matrix dynamically allocates enhancement weights according to local luminance differences, thereby suppressing noise and sharpening edges while maintaining luminance balance, achieving the collaborative optimization of luminance and details. Finally, a saturation stretching module based on color drift perception is designed to correct color deviations. Combined with a non-local means (NLM) denoising mechanism in the YUV space, it further improves the overall perceptual quality and color naturalness of the images. Extensive experimental validations were conducted on the public Low-Light(LOL) test set and a self-built coal mine image dataset. Quantitative evaluation results show that the proposed method achieves the best overall performance in both full-reference metrics (e.g., Peak Signal-to-Noise Ratio(PSNR), Structural Similarity Index Measure(SSIM)) and no-reference metrics (e.g., Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE)). Compared with the evaluated methods, it more effectively improves image contrast, preserves structural details, and suppresses noise. Ultimately, this method effectively improves the luminance uniformity, contrast, and edge detail resolution of images in low-light environments, providing a feasible theoretical reference for downstream tasks such as image enhancement and target detection in coal mine intelligent monitoring systems. Full article

In order to improve power quality in low-voltage distribution networks with nonlinear, distorted, and unbalanced loads, this research suggests an innovative design for Shunt Active Power Filters (SAPFs). Compared to conventional single-inverter SAPF structures, the improved method uses a combination of two T-Type three-level inverters operating in a parallel configuration to improve compensator performance, leading to a higher current-carrying capacity as well as better harmonic reduction and system scalability. The Synchronous Reference Frame (SRF) algorithm is used to extract reference currents in order to achieve the required accuracy of harmonic cancellation. In order to ensure both a quick response and a suitable switch state selection for compensatory current references, the Model Predictive Current Control (MPCC) technique is employed. To maintain the DC-link voltage at a steady level and guarantee its correct operation under rapidly fluctuating loading conditions, a Proportional Integral (PI) controller-based DC–DC converter is also utilized. Four real-time operational circumstances are used to verify the performance of the proposed method using MATLAB/Simulink R2023a (i) SAPF activation under nonlinear loading conditions, (ii) dynamic load variation, (iii) distorted and unbalanced operation, and (iv) grid voltage disturbances including sag and swell conditions. The simulation study’s results show that, in all of the previously indicated scenarios, the source current Total Harmonic Distortion (THD) is reduced and an almost unity power factor is maintained while maintaining a constant DC-link voltage. Furthermore, the obtained performance meets IEEE-519-2022 requirements, demonstrating the feasibility of the suggested SAPF with two inverters under high-load circumstances. Full article

This study investigates the nonlinear relationship between sectoral loan diversification and credit risk in Vietnamese commercial banks. Using a balanced panel of 14 banks over 2012–2025, comprising 196 bank-year observations, the study measures diversification through a Shannon Entropy Index based on a harmonized ten-sector classification and measures credit risk using the reported non-performing loan ratio. The relationship is examined using conventional panel estimators, two-step System GMM, and bias-corrected LSDV models, together with alternative diversification measures. The results provide suggestive evidence of a U-shaped association: diversification is associated with lower credit risk at relatively low levels but with higher credit risk beyond a conditional turning point. This pattern is supported by the System GMM and small-sample bias-corrected estimates, although it is not statistically robust to the HHI-based measure. The findings therefore indicate that the effects of diversification depend on both its extent and measurement and should not be interpreted as identifying a universal optimal threshold. Banks and supervisors should assess sectoral diversification alongside cross-sector risk correlations, underwriting expertise, and monitoring capacity. Full article

Unsupervised feature selection is essential for high-dimensional data analysis, where irrelevant and redundant variables may obscure the intrinsic data structure, reduce interpretability, and degrade downstream learning performance. The key challenge is to identify informative features without label supervision while suppressing redundant selections under nonlinear dependencies. To address this issue, this paper proposes an unsupervised Hilbert–Schmidt Independence Criterion (HSIC)–Elastic Net (ENet) feature selection framework, termed U-HSIC-ENet. The proposed method reformulates unlabeled feature weighting as a self-supervised kernel alignment problem. Specifically, a target kernel is constructed directly from unlabeled data to encode global sample relationships, while each feature is represented by a centered and Frobenius-normalized feature-induced kernel. Feature relevance is then measured by centered kernel alignment (CKA) in a reproducing kernel Hilbert space (RKHS). The main novelty of U-HSIC-ENet lies in a unified relevance–redundancy–stability formulation. Feature–target alignment is used to estimate nonlinear relevance, whereas pairwise similarities between feature-induced kernels are used to characterize inter-feature redundancy in the same kernel alignment space. On this basis, an explicit off-diagonal redundancy penalty is incorporated into a nonnegative Elastic Net-type objective, which strengthens the suppression of co-selected similar features while preserving sparse and stable feature weighting. The resulting quadratic formulation clarifies how relevance promotion, redundancy control, sparsity, and numerical stabilization are coupled within a single optimization framework. Experiments on eight benchmark datasets under a fixed-budget evaluation protocol show that U-HSIC-ENet achieves the strongest average performance on Normalized Mutual Information (NMI), the Adjusted Rand Index (ARI), and clustering accuracy (ACC) compared with representative graph-, spectral-, and HSIC-based baselines. The advantage is the most pronounced on NMI, suggesting that the self-supervised target kernel and CKA-based relevance modeling are effective in preserving the clustering-relevant nonlinear structure. Friedman tests and Wilcoxon signed-rank tests with Holm correction provide statistical support for the observed improvements. Subsampling-based stability evaluation reveals a trade-off between clustering effectiveness and selection reproducibility: several baselines achieve higher stability scores despite the stronger average clustering performance of U-HSIC-ENet. These results indicate that the proposed framework is effective for unsupervised nonlinear feature weighting when relevance estimation, redundancy control, and stability are considered jointly. Full article

To address issues such as nonlinear strong coupling, time-varying parameters, and external wind disturbances in fixed-wing unmanned aerial vehicles (UAVs) operating in complex flight environments, this study develops a composite attitude control method integrating L₁ adaptive control with an extended state observer (ESO). First, pitch and roll attitude dynamic models considering nonlinear aerodynamic characteristics and channel coupling are established. Based on these models, a composite control architecture is developed in which the L₁ adaptive controller establishes the baseline prediction-error-driven command-tracking loop and introduces matched-uncertainty compensation through a low-pass-filtered adaptive channel, whereas the ESO estimates a generalized extended state and, after removal of the known nominal drift term, provides an independently scaled auxiliary feedforward correction based on the resulting lumped-disturbance estimate. Although the uncertainty contents observed by the two mechanisms may partially overlap, their control actions are coordinated through the residual closed-loop disturbance and the prediction-error-driven adaptation process rather than being independently superimposed at full amplitude. Since the L₁ adaptive law remains driven by the residual state-prediction error after the ESO action, the two mechanisms dynamically redistribute, rather than simply duplicate, the compensation effort. The control performance of the proportional–integral–derivative(PID) controller, the conventional L₁ adaptive controller, and the proposed method is comparatively evaluated through simulations under typical operating conditions, including step response, sinusoidal tracking, composite wind disturbances, and measurement noise. The results show improved transient response and disturbance/noise rejection relative to PID and conventional L₁ control under most of the tested conditions, while the high-frequency tracking benefit is channel-dependent. Overall, the proposed method improves transient response and disturbance/noise rejection while maintaining bounded tracking performance under the stated assumptions. The proposed method provides an effective approach for improving the attitude control performance of fixed-wing UAVs operating in complex environments. Full article

Conversion flight turns the aerodynamic versatility of a tiltrotor into a demanding propulsion-control problem. As the nacelles rotate and vertical load transfers from the proprotors to the wing, the two turboshaft engines must follow a rapidly changing shaft-power demand while respecting fuel command magnitude and rate limits, compressor-pressure limits and turbine temperature limits. A control-oriented conversion model is coupled to a component-level turboshaft engine through a long short-term memory (LSTM) dynamic surrogate embedded in a constrained receding-horizon controller. The aircraft model resolves wing force balance, blade-element/momentum rotor loads, forward acceleration, nacelle actuation and accessory power. The LSTM predicts six engine outputs from flight conditions, fuel command and previous-step spool speeds. Training and evaluation use 537 converged component model cases divided by complete simulation cases into 375 training, 80 validation and 82 held-out test cases. Against parameter-matched multilayer perceptron, temporal convolutional network and gated recurrent unit baselines, the LSTM gives the lowest power root-mean-square error (20.15 kW before online output correction). Its corrected 20–320-step forecasts outperform a linear autoregressive model and zero-order hold prediction, although the linear model remains slightly better at one step. In direct component-level closed-loop simulation, LSTM engine-surrogate nonlinear model predictive control (NMPC) reduces power RMSE from 29.80 to 16.23 kW relative to linear MPC and from 34.94 to 16.23 kW relative to PI control while reducing cumulative fuel command variation by 46.5% relative to linear MPC. Turbine temperature and compressor-pressure margins remain 119.0 K and 85.4 kPa, respectively. Mean optimization time is 29.6 ms for a 1.92 s update interval. The resulting framework connects conversion flight aerodynamic loading, multi-step engine prediction and constrained power control in a reproducible numerical validation chain. Full article