Search Results (57,653)

This study proposes a Gaussian Process Regression (GPR) method for estimating the compressive strength of soil mixtures containing bentonite and basalt fibers. GPR is a preferred probabilistic machine learning approach, especially for limited datasets, due to its high generalization ability and its ability to directly calculate prediction uncertainty. In this study, experimentally obtained bentonite and basalt fiber ratio values were used as input parameters of the model, and unconfined compressive strength (q_u) was determined as the output variable. Model performance was evaluated using the Leave-One-Out Cross-Validation (LOOCV) method. The performance of the proposed GPR model was evaluated with various metrics. Accordingly, the MAE and RMSE values of the model were calculated as 3.667 kPa and 4.756 kPa, respectively, while the R² value was 0.945. The results show that the GPR model provides high prediction accuracy and is a reliable prediction tool for small datasets. Furthermore, the prediction surfaces and uncertainty analyses obtained by the model contributed to a better understanding of the effect of mixture parameters on compressive strength. Full article

The gas atomization process for metal powder production is strongly influenced by the thermodynamic and fluid-dynamic conditions of the atomizing gas. In this study, stainless steel (316L) powders produced using a close-coupled gas atomizer (CCA) with heated nitrogen and argon were investigated. A systematic experimental campaign was conducted by varying gas type, gas pressure, and gas temperature to analyze their impact on the particle size distribution and morphology. Numerical gas-flow simulations were performed to characterize the dependence of the gas velocity field in the atomization region on the applied process parameters. Based on the experimental and numerical results, classical correlations for characteristic particle size prediction have been adapted for CCA hot gas atomization by introducing an additional temperature scaling term (T_G/T_R)^t. The modified correlations improved the prediction accuracy compared with the original formulations. For the Lubanska correlation, the mean prediction error could be significantly reduced. Furthermore, general linear models and a complementary analysis of covariance (ANCOVA) were used to evaluate the associations of gas temperature, gas type, and gas pressure and the derived flow parameters with the particle size distribution. The combined experimental–numerical approach provides further insight into hot gas atomization of metal melts and supports the optimization of process parameters for tailored metal powder production. Full article

Deceptive path planning enables autonomous agents to obscure their true goals from observers by deviating from an expected optimal path. Prior work largely solves full-horizon, end-to-end optimization for single agents, so planning complexity grows with the path length and size of the environment, and adaptation en route requires a new optimization. We propose a unified framework for deceptive path planning onboard the agent, computing over short-horizon candidate paths within a receding-horizon loop. The method processes an agent’s position sensor information and develops a policy for the next trajectory plan. By parameterizing a user-defined cost that captures optimal planning and deception (and optionally includes constraints, trajectory smoothness, and coupling terms between agents), a Boltzmann framework yields stochastic policies that balance the tradeoff between optimal paths and deceptive deviation. Policies are updated locally and do not require learning or training. The level of deception and adherence to constraints can be dynamically tuned, enabling online adaptation to changes in goals and constraints. This step-by-step tuning opens the door to new forms of dynamic deception. In the multi-agent case, we develop a joint planner that is also tunable and can be applied as desired among all or subsets of agents. Single- and multi-agent simulation studies demonstrate the flexibility of our approach, maintaining deception while adapting as desired, avoiding the recomputation required by full-horizon methods, and supporting intuitive tuning via a small set of parameters. Full article

Head and neck squamous cell carcinoma (HNSCC) remains clinically challenging because of its marked inter- and intra-tumour heterogeneity, the dynamic emergence of therapeutic resistance, and the limited ability of current biomarkers to guide treatment adaptation over time. Recent advances in digital twin (DT) technology have motivated the development of patient-specific, continuously updated computational models capable of integrating multi-scale data to support precision oncology. However, no current DT framework for HNSCC combines molecular stratification, longitudinal monitoring of resistance, pharmacodynamic modelling, and clinical decision support within a single adaptive system. In this review, we critically examine the current state of DT-enabled approaches for targeted therapy in HNSCC and propose a conceptual framework for their future clinical implementation. We discuss how genomic and multi-omic stratification, mechanistic imaging models, pharmacokinetic/pharmacodynamic modelling, longitudinal liquid biopsy (ctDNA and exosomes), ex vivo functional testing, toxicity prediction, and artificial intelligence could be integrated into a continuously updated patient-specific model. We further examine the biological mechanisms driving resistance to targeted therapies and immunotherapy, highlighting how these dynamic processes should inform adaptive therapeutic decision-making. We discuss existing DT-like approaches and evaluated them according to their capacity to fulfil the DT criteria, while also presenting DT-enabling technologies. Among the frameworks currently available, the deep reinforcement learning-based DITTO platform represents the closest approximation to a clinically relevant HNSCC digital twin. However, it does not yet incorporate molecular signalling networks, longitudinal resistance biomarkers, or multimodal biological data. We therefore identify the integration of these complementary data layers as the principal challenge and opportunity for the next generation of DTs. Collectively, this review provides a conceptual DT framework in which four main data domains could fulfil different roles within the DT. By sharing different parameters across these layers, the framework could forecast emerging resistance and update model predictions longitudinally. Such a DT could support biomarker-guided patient stratification, adaptive treatment selection, rational combination therapies, toxicity prediction, and future clinical trial design in HNSCC. We also highlight challenges and limitations that need to be addressed for future clinical translation, including data integration, interpretability, clinical validation, workflow integration, and ethical and regulatory considerations. Full article

Changes in modal order alter the predictor dimension, pole-label swaps disrupt frequency continuity, and time-varying envelopes affect phasor magnitude and phase. Estimating these quantities independently can propagate errors across successive processing stages. A coordinated estimator is developed for the modal subspace, pole states, and regression parameters. An order confidence index combines the spectral gap, cumulative energy, and noise separation to select the model order and reconstruct the signal in one low-rank subspace. Variable-order recursive prediction and frequency–damping state association then form continuous pole trajectories, followed by adaptive smoothing and class-dependent unit-circle projection. The associated oscillatory and decaying direct-current (DC) poles update the Maclaurin regression atoms. Finite-window coupling is handled by either modal initialization followed by Gram iteration or a direct joint regularized solution, avoiding repeated leakage compensation. Tests with modal-order changes, frequency dynamics, modal crossings, amplitude modulation, and decaying DC show that the coordinated parameter chain preserves pole identity and improves dynamic phasor measurement. In the main dynamic test case, the mean and 95th-percentile total vector errors (TVEs) are 3.2082% and 6.0672%; the 95% paired confidence interval for the mean-TVE difference between the proposed method and estimation of signal parameters via rotational invariance techniques (ESPRIT) remains below zero. A separate RK3568 bare-metal test of the standalone three-tone Prony kernel completed 800 frames without a processing failure. Its mean processing time was 18.621 ms per frame, with observed values from 18.537 to 19.070 ms. Full article

To exploit the operational flexibility of industrial parks in transformer capacity planning while addressing the expansion of controllable resources during system upgrades, this paper proposes a flexibility-aware bi-level planning framework based on cross-dimensional deep reinforcement transfer learning (DTRL). The upper level optimizes transformer capacity by minimizing the life-cycle cost, while the lower level formulates the coordinated operation of generation, flexible loads, and energy storage as a Markov decision process. To efficiently transfer scheduling knowledge from the existing system to an expanded configuration with additional control dimensions, a cross-dimensional DTRL framework integrating Monte Carlo scenario augmentation and elastic weight consolidation (EWC) is developed. Monte Carlo scenario augmentation enriches the limited operating samples available in the target domain, while the importance of pretrained network parameters is quantified using the Fisher information matrix. During fine-tuning, elastic regularization is introduced to preserve transferable scheduling knowledge and facilitate adaptation to newly introduced control dimensions. A case study of a typical industrial park shows that the proposed method achieves stable convergence under the expanded system configuration and reduces the planned transformer capacity from 1250 kVA to 1000 kVA while maintaining the maximum loading rate within the prescribed peak-loading constraint. Compared with the benchmark methods, the proposed approach reduces the annualized life-cycle cost by up to 1.9%, demonstrating the value of coordinated operational flexibility in balancing transformer capacity investment and operating economy. Full article

Background/Objectives: The development of multifaceted excipients is a requirement of the pharmaceutical industry. This study aimed to develop a granular co-processed excipient for tablets and to evaluate it using the SeDeM expert system. Methods: Four granule formulations were developed via the powder layering technique, varying binder concentrations (15% and 20%) and filler types (microcrystalline cellulose and lactose), using sugar as the core. The granules obtained were evaluated utilising the SeDeM expert system. The unloaded granules were compressed to yield uncoated compacted granules, which were verified for dimensional parameters, mechanical properties, and disintegration ability. Results: Varying binder concentrations (15% and 20%) together with differences in filler/disintegrant composition were associated with changes in particle-size distribution. During the SeDeM evaluation, E3 formulation exhibited good results in terms of parameter index (PI = 0.83), parameter profile index (PPI = 7.64), and Good Compressibility Index (GCI = 7.28). The recorded disintegration times were below 15 min for all compacted granules and complied with the Ph. Eur. 12 requirements for uncoated tablets. Conclusions: For granule development, variations in formulation composition, including binder concentration, were associated with differences in particle size and lubricity. The SeDeM expert system can be used in the preformulation studies to characterize powders or granule formulations from which uncoated tablets can be obtained, but there are several pharmacotechnical properties not included in the list of parameters that need to be analysed to comply with the industrial requirements. Full article

We have developed a new method for remote sensing of the ionosphere and neutral atmosphere based on the creation of artificial periodic irregularities (APIs) in ionospheric plasma. This method is used, in particular, to study atmospheric turbulence, which can affect GNSS signal delays by causing disturbances in the radio wave propagation medium. We applied the API method to determine the turbopause height and turbulent velocity at altitudes of 60–130 km in the lower ionosphere. Turbulence is one of the most important phenomena in the Earth’s atmosphere. This article presents the results of turbulence parameter measurements based on the resonant scattering of APIs. These irregularities are created in the ionosphere by powerful high-frequency radio waves. The altitude profile of turbulent velocity is obtained from measurements of the relaxation time of the API scattered signal after the end of the powerful radio emission on the ionosphere. The turbopause level is defined as the altitude at which turbulent mixing of atmospheric gases gives way to diffusive separation. During the transition from diffusive separation to turbulent mixing (below the turbopause level), irregularities decay more rapidly than during ambipolar diffusion, and their relaxation time is determined predominantly by turbulent diffusion. This fact allows us to determine the velocity of turbulent motion as a function of the diffusion time and the measured relaxation time of the scattered signal. The turbopause level is determined from the altitude profile of the relaxation time. Measurements were carried out using the SURA heating facility (56.15° N, 46.11° E). The altitude resolution of the API method is 0.15 km–1 km with a time resolution of 15 s, which allows us to study both fast and slow processes. According to our data, the turbopause level varied in the altitude range from 85 km to 110 km. Significant variability in the turbopause level was observed during the day and from day to day. The minimum turbopause level was located at mesospheric altitudes. The average turbulent velocity over a 5 min interval varied from near zero at the turbopause altitude to 5–6 m/s below it. Such high velocities were typically observed during natural ionospheric disturbances. The average velocity of regular vertical plasma motion, measured using the API method, reached 10 m/s and varied with altitude and direction. Temporal variations in turbulent parameters were observed, with periods ranging from 15 min to several hours. The effect of atmospheric wave propagation on the characteristics of scattered signals and environmental parameters was confirmed. Full article

Background: The ingestion of polyunsaturated fatty acids is vital for brain health, supporting cognitive development and helping to prevent chronic diseases, including neurodegenerative processes. Objective: This study aimed to investigate the effects of corn oil on the biochemical and neurochemical parameters of Drosophila melanogaster expressing human amyloid precursor protein β42 (APP-β₄₂). Methods: The flies were fed a diet supplemented with 37.8 mg/mL of corn oil from the larval stage until adulthood. Results: A diet supplemented with corn oil induced significant changes in biochemical markers, such as a decrease in head cholesterol levels (p < 0.001); decreased catalase activity and hydrogen peroxide levels in the heads (p < 0.0001) and thoracic muscles (p < 0.0001); reduced glutathione levels in the heads (p < 0.0001) and muscles (p < 0.01); and reduced formazan production (p < 0.01) and citrate synthase (CS) activity (p < 0.0001) in the head. However, in thoracic muscle, ingestion of corn oil led to an increase in formazan production (p < 0.01) and no significant change in CS activity. Lactate levels decreased in the heads (p < 0.0001) and thoraces (p < 0.001) after flies were fed corn oil. Finally, ingestion of corn oil resulted in a significant rise in acetylcholinesterase activity in the heads (p < 0.001) and thoracic muscles (p < 0.01). Conclusions: These results suggest that consuming corn oil reduces the oxidative stress typical of an Alzheimer’s disease model by enhancing antioxidant defenses. Full article

This paper presents a virtual sensor for the in-process prediction of cutting forces from feed drive current measurements in milling processes via inverse filtering. Components of the feed drive current in ball-screw drives that are related to inertia, friction, and gravity are separated from the cutting-force-related component via models or air-cutting experiments. Impact hammer tests are then used to identify the transfer function between forces at the tool tip and the corresponding response at the feed drive. The proposed inverse filtering approach completes the virtual sensor for online monitoring of cutting forces based on feed drive current signals. Compared to existing approaches, which are mainly based on Kalman-filter or deep learning models, this method avoids the additional effort for modeling, parameter identification and generation of training data. Experimental results are presented for cutting tests on a three-axis turn-milling center. The prediction error between the virtual sensor and the measured cutting forces lies between 6% and 17%, depending on the cutting parameters. In general, the virtual sensor can be implemented in any feed drive system with a minimal effort for parameter identification. Full article

Iron oxide-mediated abiotic humification is critical for long-term soil organic carbon (SOC) stabilization. Goethite (α-FeOOH), the most thermodynamically stable iron oxyhydroxide in terrestrial ecosystems, is proposed to catalyze coupled Maillard reaction and polyphenol oxidation via surface hydroxyl and Fe(III) active sites. Nevertheless, the divergent interfacial regulatory mechanisms of glycine (Gly, N source) and glucose (Glu, C source) concentration gradients in goethite-catalyzed polyphenol-Maillard systems remain poorly understood. Two independent 360-h gradient incubation experiments were performed using synthetic goethite as a catalyst and catechol as a model polyphenol precursor to probe their concentration-dependent effects on short-term interfacial transformation kinetics and humic-like product properties. Kinetics, product properties, and underlying pathways were characterized via Gaussian and first-order asymptotic modeling, humic-like acid (HLA)/fulvic-like acid (FLA) fractionation, elemental analysis, Fourier-transform infrared (FTIR) spectroscopy, and partial least squares structural equation modeling (PLS-SEM). Results showed that 0.12 mol/L was the optimal concentration for both precursors to accelerate intermediate transformation. Gly elevated the baseline aromaticity parameter y₀ by 108.87% (vs. 20.38% for Glu), while Glu increased the asymptotic maximum dissolved organic carbon (DOC) by 322.66% (vs. 78.02% for Gly). Moderate Gly (0.03 mol/L) yielded the highest C_HLA. Excessive precursors reduced the C_HLA/C_FLA ratio via interfacial competitive adsorption and site occupation. The two precursors exerted opposing effects on the hydroxyl stretching peak of humic products, but both induced goethite surface reconstruction and activated iron-bearing surface functional groups. PLS-SEM confirmed precursor concentration as the dominant driver of humification kinetics. This study clarifies the distinct roles of Gly in aromatic cyclization and Glu as an aliphatic C donor during short-term mineral–organic interfacial processes, providing quantitative benchmarks for optimizing artificial humus production. Full article

Gas hydrate-bearing formations are characterized by low temperature and high pressure, weak cementation, strong phase sensitivity, and pronounced heterogeneity. During drilling in such formations, the drilling fluid is required to maintain wellbore stability, transport cuttings, and control bottom-hole pressure; however, it may also alter near-wellbore hydrate stability through heat transfer, pressure-driven invasion, and component migration. Previous studies have shown that drilling-induced disturbances may trigger or intensify the dissociation of hydrates, sediment strength degradation, pore-pressure redistribution, seepage-channel development, and secondary hydrate formation in the wellbore. Under certain conditions, these processes may manifest as wellbore instability, gas invasion, changes in drilling-fluid properties, and wellbore blockage. This review summarizes recent progress in four aspects: risk manifestations, formation-response mechanisms, control strategies, and numerical modeling. The reviewed studies indicate that thermal disturbance, pressure disturbance, and component migration jointly control the phase, mechanical, and seepage responses of hydrate-bearing formations; temperature–pressure window management, inhibitors, plugging particles, functional materials, and dynamic load control act at different stages of risk evolution; and numerical models mainly include thermal–chemical, thermal–hydraulic–chemical, thermal–hydraulic–mechanical–chemical, wellbore multiphase flow and phase transition models and wellbore–formation coupled models. Current research still faces challenges in representing realistic formation heterogeneity, characterizing dynamic mud-cake evolution, integrating multiscale experiments with field data, and reducing the computational cost and improving the numerical robustness of fully coupled models. Future work should strengthen field-constrained multiscale experiments, dynamic parameter calibration, and integrated wellbore–formation control models, thereby providing a more reliable theoretical and technical basis for safe drilling in hydrate-bearing formations. Full article

The study investigated whether a dose of a commercial iron-based reagent determined under controlled laboratory conditions could be used as a starting point for full-scale hydrogen sulfide control in biogas. Laboratory tests were conducted in a reactor with a working volume of 30 dm³, and technical validation was performed in an operating anaerobic digester with a process volume of approximately 7200 m³. The common parameter used for scale transfer was the mass dose of reagent relative to the sulfur load supplied with the sludge (D_SF, kg reagent/kg S). An operational reference range of 300 ppm H₂S ±10% (270–330 ppm) was adopted for biogas supplied to combined heat and power units. In the final phase of the laboratory experiment, at a volumetric dose of 2.0 dm³/m³, the H₂S concentration stabilized at 312–318 ppm, corresponding to D_SF = 4.91 kg reagent/kg S. Full-scale validation was initiated at a similar dose of 4.75 kg/kg S, after which dosing was gradually reduced on successive days. The lowest tested dose was 1.26 kg/kg S, and the H₂S concentration on the final day of observation was 268 ppm. The CH₄ fraction at full scale remained stable at 62.3–64.0%. Reducing D_SF from 4.75 to 1.26 kg/kg S decreased the unit reagent cost from approximately 2.21 to 0.59 EUR/kg S. Within the 8-day full-scale validation, the laboratory-derived dose functioned as a suitable starting point for scale transfer, but the results do not establish a long-term optimum and do not support using the laboratory value directly as a fixed operating dose. Full article

Series arc faults in low-voltage electrical installations pose severe fire hazards due to their intermittent and load-dependent nature, making reliable detection across diverse conditions a persistent challenge. This paper presents Arc-FaultNet, a lightweight dual-branch convolutional neural network that jointly exploits temporal and spectral representations of the line current through complementary attention mechanisms. The temporal branch extracts four physically derived channels via a 1D convolutional stack enhanced with Squeeze-and-Excitation (SE) attention, while the spectral branch processes log-power Short-Time Fourier Transform (STFT) spectrograms through a learnable frequency gate. Both representations are fused via a cross-conditioned channel attention mechanism, enabling mutual temporal–spectral guidance. To assess generalization capacity and architectural stability, several training protocols were conducted—including strict GroupKFold cross-validation and single-model training—consistently yielding strong results: up to 98.77% accuracy, 98.68% F1-score, and 99% under single-model training, and 90.16% accuracy and 94.57% specificity under cross-validation, where cross-attention fusion outperforms naive concatenation by +5.62 percentage points (pp) in F1-score. An enhanced variant equipped with SE blocks and a deep classifier head further reduces performance variance by 28–51%. The consistently strong performance across all protocols confirms the generalization capacity and architectural robustness of Arc-FaultNet. With fewer than 365K parameters, Arc-FaultNet offers a practical pathway toward IEC 62606-compliant embedded deployment. Full article

This study investigated the effects of cotton-stalk-derived wood vinegar produced at different pyrolysis temperatures on nitrogen and phosphorus transformations and the associated microbial mechanisms during the aerobic composting of chicken manure. Wood vinegar produced at 300, 350, 400, 450, and 500 °C was applied in a 38-day composting experiment. Changes in physicochemical properties, nitrogen and phosphorus fractions, bacterial community succession, and potential functional profiles were analyzed using 16S rRNA gene sequencing, Spearman correlation analysis, and PICRUSt2-based COG functional prediction. The results showed that cotton-stalk-derived wood vinegar improved the composting process and regulated nutrient transformation. Compared with T0, the thermophilic phase was prolonged by 1–4 days in T1, T3, and T5. At the end of composting, the pH and electrical conductivity of all treatments ranged from 7.45 to 7.72 and from 1.37 to 1.75 mS/cm, respectively, which were within acceptable ranges for mature compost. The effects of wood vinegar on nitrogen and phosphorus transformations varied with pyrolysis temperature. All wood-vinegar-amended treatments maintained higher final TN concentrations than T0. Among them, T4 maintained a high final TN concentration of 22.90 g/kg, which was 11.17% higher than that of T0 and close to the highest value observed among all treatments. However, inorganic nitrogen responses differed among treatments: T1 showed more efficient conversion of NH₄⁺-N to NO₃⁻-N, whereas T3 had the lowest NH₄⁺-N/NO₃⁻-N ratio, indicating relatively higher compost maturity. Therefore, the nitrogen results should be interpreted on a concentration basis rather than as evidence of reactor-scale total nitrogen retention or nitrogen loss. Wood vinegar also affected phosphorus transformation, with T4 achieving a higher available phosphorus concentration. Bacterial community analysis showed that Bacillota and Actinomycetota dominated the composting process, and T4 maintained a more stable bacterial community structure during maturation. T4 was enriched in potential functional genera, including Saccharomonospora and Thermoactinomyces. Spearman correlation analysis indicated that Weissella and Pediococcus were significantly associated with nitrogen-related parameters in T4, whereas Saccharomonospora was closely linked to phosphorus transformation. PICRUSt2-based COG prediction further suggested that T4 maintained relatively stable nitrogen- and phosphorus-related functional potentials. Full article