Search Results (10,038)

In nuclear spectroscopy, a physical phenomenon known as the pile-up effect distorts direct measurements by causing temporal overlap of detector pulses. Existing deep learning-based pile-up correction methods rely heavily on supervised training with simulated data, which often generalize poorly to real measurements due to simulation–experiment discrepancies. In this work, we propose a contrastive learning framework to learn robust and transferable representations directly from large-scale unlabeled real nuclear pulse signals. The detector output is segmented into physically complete pulse aggregations using a zero-crossing-based strategy, which serve as semantically coherent instances for representation learning. Physics-inspired data augmentations are designed to realistically model detector noise and bandwidth effects while preserving pulse area. A one-dimensional ResNet encoder is employed for efficient representation learning. The learned representations are transferred to pile-up identification and counting-rate estimation tasks. Experimental results on real nuclear radiation detection systems demonstrate that our method achieves strong performance and robustness under high counting-rate conditions, with particularly pronounced advantages in challenging peak pile-up scenarios. Full article

A hybrid modelling framework is proposed to predict endpoint sulphur content in the ladle furnace (LF) refining process by embedding metallurgical expert knowledge into interpretable machine learning (ML). Industrial process data were extracted from the Level-2 (L2) system of a steel plant, and a desulphurisation dataset comprising 5169 heats with 29 process variables was constructed using a knowledge-guided time window from the joint satisfaction of refining conditions to the final argon-blowing stage. After data cleaning, normalisation and correlation-based feature selection, four algorithms—Random Forest (RF), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM) and Artificial Neural Network (ANN)—were trained and compared on a representative cluster of steel grades identified by K-means. The ANN model achieved a coefficient of determination (R²) of 0.7752, a root mean square error (RMSE) of 0.0027 wt%, a mean absolute error (MAE) of 0.0017 wt% and a hit rate (HR, ±0.0025 wt% for S) of 76.40% on the test set. SHapley Additive exPlanations (SHAP) indicate that limestone addition, slag basicity, argon flow rate, refining time and initial sulphur content dominantly govern sulphur removal. The expert-knowledge-guided, interpretable framework provides quantitative support for specification-conforming endpoint sulphur control while mitigating over-desulphurisation and reagent consumption. Full article

Railway bogies are typically engineered with conservative safety margins, which frequently results in the premature disposal of components retaining significant structural integrity. This study proposes a comprehensive 3D modelling framework designed to accurately predict and extend the Remaining Useful Life (RUL) of the bogie structure. To achieve this, a Building Information Modelling (BIM) approach was used, not only for the bogie, but for all train, using a rolling stock in Portugal as a case study. The use of both real and virtual sensors installed in the bogie, with data collected with a sampling rate according to the specificity of each sensor and, next, managed through machine learning tools, allows to implement a predictive maintenance (PdM) policy that aid to extend the RUL. The proposed approach demonstrates that extending the operational life of the bogie is both feasible and safe. This facilitates a strategic transition from the current practices to new approaches that improve the Availability of the Physical Assets, including through the metaverse. Full article

To address the current issues in soil organic carbon (SOC) content prediction where data preprocessing relies on expert experience to formulate fixed rules, resulting in a lack of uniform standards and insufficient consideration of regional soil heterogeneity; while hyperparameter tuning faces problems of high computational costs and excessively long runtimes, this study proposes an intelligent modeling workflow driven by Large Language Models (LLM). This workflow focuses on optimizing two key aspects of SOC Random Forest modeling: data preprocessing and hyperparameter tuning. Results: The LLM-defined rules achieved sample retention rates of 55.33% and 61.90% in the two regions, respectively, showing more significant differences compared to traditional hard-coded rules (56.2% and 59.3%), and the mean soil organic carbon content deviations (30.27% and 20.05%) were both lower than those of traditional hard-coding. At the same time, the mean soil organic carbon content values in both regions closely matched the effectiveness of other methods, indicating that the large language model has effectively captured regional soil differences. With only a single evaluation of hyperparameter optimization, the adaptive model achieved test set R² values of 0.394 and 0.694 in the black soil region and the aeolian sandy soil region, respectively, with root mean square error values of 8.76 g/kg and 6.07 g/kg—its performance is comparable to that of Grid Search and Random Search, while computational efficiency improved by over 95%. Performance comparisons with eXtreme Gradient Boosting (XGBoost) and Partial Least Squares Regression (PLSR) show that the LLM-optimized Random Forest achieved R² = 0.394 and RMSE = 8.76 g/kg in the black soil region, and R² = 0.694 and RMSE = 6.07 g/kg in the windblown sandy soil region, demonstrating practical application value. Full article

With the continuous integration of a large amount of renewable energy sources such as wind and solar power into the power system, the economic and secure scheduling of the power grid, as a crucial carrier for electricity transmission, becomes of paramount importance. However, issues such as voltage fluctuations at grid nodes, low renewable energy consumption rates, and increased active power losses, caused by the widespread integration of high proportions of renewable energy, urgently need to be addressed. To effectively solve these problems, this paper proposes a multi-objective coordinated optimization scheduling method for the economy and security of source–grid–load–storage based on an effective scenario-screening approach. Firstly, an iterative self-organizing data analysis algorithm based on density noise application spatial clustering is designed to efficiently generate typical output scenarios for renewable energy sources such as wind and solar power. Meanwhile, to achieve low-carbon scheduling objectives, green certificate and carbon trading mechanisms are introduced. A multi-objective coordinated scheduling and trading model for the economy and security of large power grids, sources, loads, and storage is constructed with the goal of enhancing renewable energy consumption, and it is solved using the weight assignment method and an improved particle swarm optimization algorithm. Finally, the effectiveness and feasibility of the proposed method are validated and illustrated based on an improved IEEE standard node test system. Full article

Background/Objectives: Appropriate nutritional management constitutes one of the key elements of conservative treatment and renal replacement therapy in patients with chronic kidney disease (CKD). The level of patients’ nutritional knowledge may significantly influence adherence to dietary recommendations, the rate of disease progression, and the frequency of complications. The aim of this study was to assess the level of nutritional knowledge, dietary habits, adherence to dietary recommendations, and nutritional status of patients with CKD according to disease stage. Methods: This cross-sectional study was conducted among 98 adult patients diagnosed with CKD. A questionnaire assessing nutritional knowledge and dietary behaviors was administered. An overall nutritional knowledge score was calculated based on eight questionnaire items assessing nutritional knowledge. Nutritional status was evaluated using the Subjective Global Assessment (SGA) and the Simplified Nutritional Appetite Questionnaire (SNAQ). Anthropometric, clinical, and biochemical data were collected. Statistical analysis was performed using tests appropriate to the data distribution. Results: The level of nutritional knowledge varied and was dependent on CKD stage. Patients in more advanced stages of the disease demonstrated significantly higher awareness of dietary recommendations compared with those in earlier stages. The median nutritional knowledge score was 6 points, with 46.9% of participants demonstrating insufficient knowledge (<6 points) and 53.1% achieving adequate knowledge (≥6 points). The greatest knowledge deficits concerned the control of phosphorus, potassium, sodium, and fluid intake. Discrepancies were also observed between declared knowledge and actual dietary behaviors. Good nutritional status (SGA A) was identified in 73 patients, risk of malnutrition or moderate malnutrition (SGA B) in 22 individuals, and severe malnutrition (SGA C) in 3 patients. SNAQ indicated good appetite in the study population, with an average consumption of three meals per day, and identified a risk of weight loss in 6% of patients. Overweight and obesity were present in more than half of the study population, while underweight was observed in 4%. Conclusions: Nutritional knowledge among patients with CKD remains insufficient, particularly in the early stages of the disease. The findings highlight the necessity of early and systematic implementation of individualized nutritional education as an integral component of slowing disease progression. Full article

Convolutional neural networks (CNNs) excel in tasks such as image, speech, and video recognition, as well as pattern analysis. However, their reliance on large training datasets, often sourced from third-party providers, exposes them to security risks, particularly poisoning attacks. Targeted poisoning attacks, also known as backdoor attacks, enable a CNN model to correctly classify normal data while misclassifying inputs containing specific triggers. In contrast, untargeted poisoning attacks aim to degrade the overall performance of the model. This research introduces an invisible targeted poisoning attack characterized by low implementation complexity and high computational efficiency due to its computationally inexpensive LSB-based embedding mechanism, without requiring complex optimization procedures against a basic CNN model and a residual network (ResNet-18) model. By embedding trigger images within poisoned samples, the attack remains covert, evading detection. The model is then trained on a dataset comprising both original and poisoned samples. The expected outcome is that the model will classify regular images correctly, but will misclassify those containing the embedded trigger as belonging to a target class. Experimental results on the CIFAR-10 dataset demonstrate the effectiveness of this approach, achieving a 99.32% Adversarial Success Rate (ASR) against ResNet-18 with only a 0.02% reduction in accuracy on benign test samples. Full article

Background: Robotic-assisted thoracic surgery is increasingly recognized as an advanced minimally invasive technique for treating non-small cell lung cancer, offering technical advantages such as enhanced precision and visualization. Although numerous studies have been published worldwide, there are no comparable data from Poland. Therefore, evidence on the perioperative safety and oncologic adequacy of robotic-assisted lobectomy during early phase of program implementation within the Polish healthcare system remains limited. Methods: This retrospective, single-institution observational study included 81 consecutive patients who underwent robotic-assisted lobectomy for primary NSCLC between January 2022 and December 2024. All procedures were carried out using the da Vinci Xi system with a standardized four-arm portal approach. Clinical, perioperative, and pathologic parameters were prospectively collected and analyzed descriptively. Postoperative complications were classified according to Clavien-Dindo. Results: The median patient age was 70 years (IQR: 65–74), 52% were male, and 67% had a history of smoking. Adenocarcinoma was the predominant histologic subtype (51%). The median operative time was 176 min (IQR: 149–220). There were no conversions to thoracotomy and no 30-day mortalities. Postoperative complications occurred in 24% of cases, with prolonged air leak being most common (17%). The median hospital stay was 8 days (IQR: 6–10). R0 resection was achieved in 96% of patients, with a median of 14 lymph nodes dissected across 5 nodal stations. Conclusions: Robotic-assisted lobectomy performed during the early implementation phase of a national program demonstrated low morbidity, high rates of complete (R0) resection, and adequate lymph node yields consistent with international benchmarks. These results support the feasibility of robotic lobectomy within the Polish healthcare setting; however, the single-surgeon, single-center design limits generalizability. Further multicenter prospective studies are needed to confirm reproducibility, assess learning curves, and evaluate long-term oncologic outcomes. Full article

Satellite observations of methane are frequently compromised by extensive data gaps caused by cloud cover and aerosol contamination, limiting their utility for continuous regional monitoring. To reconstruct these spatiotemporal discontinuities, this study developed the Stacked Ensemble Reconstruction Framework for Methane (SERF-XCH₄). By integrating Sentinel-5P TROPOMI retrievals with 25 multi-source environmental covariates, we generated a spatiotemporally continuous, high-resolution (0.1°) monthly dataset (SERF-XCH₄-IM) for Inner Mongolia spanning 2019 to 2023. Comprehensive validation demonstrates that the framework achieves exceptional predictive fidelity with a Coefficient of Determination ($R^2$) of 0.93 and a Root Mean Square Error (RMSE) of 7.89 ppb, significantly surpassing the performance of individual base learners and traditional interpolation methods. Furthermore, spatial block cross-validation confirmed robust generalization capabilities ($R^2=0.90$) in data-void regions. To unravel the “black box” of the model, SHapley Additive exPlanations (SHAP) analysis was employed, revealing that temporal factors (contributing 63.9%), air temperature, and elevation are the dominant drivers governing XCH₄ variability. Spatiotemporal analysis further identified the Hulunbuir region as a significant growth “hotspot” with an annual increase rate exceeding 18.5 ppb/yr, a trend primarily driven by intensified emissions during the autumn and winter seasons. Consequently, this framework establishes a high-precision, interpretable paradigm for regional methane monitoring and geo-information reconstruction. Full article

The Landslide susceptibility assessment in complex mountainous terrain is frequently limited by static modelling frameworks that inadequately capture nonlinear deformation characteristics and temporally evolving hazard processes. To bridge this gap, a continuous-time dynamic assessment framework is proposed for Shazhou Town, Sichuan Province, integrating slowly moving scatterogram interferometric radar (S(BAS-InSAR))-derived deformation time series with Liquid Neural Networks (LNN). By incorporating a liquid time-constant architecture, the model accommodates irregular temporal sampling and captures non-stationary environmental responses through adaptive multimodal feature fusion. Analysis of long-term SBAS-InSAR observations (January 2021–May 2025) reveals distinctive deformation patterns, identifying eight active zones with maximum annual displacement rates of 107 mm yr⁻¹ and cumulative subsidence of 535.7 mm, which serve as critical dynamic inputs for the susceptibility model. Comparative experiments demonstrate that the LNN framework outperforms benchmark models (including LSTM, GRU, Random Forest, and SVM), achieving a coefficient of determination (R²) of 0.95 and an RMSE of 0.50. Furthermore, multi-temporal validation against 189 historical landslide records (2008–2025) confirms the model’s robustness, yielding a 91.5% capture rate within high-susceptibility zones. Interpretability analyses via SHAP and Layer-wise relevance propagation identify rainfall and vegetation cover as dominant dynamic controls, while characterising a distinct slope threshold effect at approximately 20°. These findings demonstrate that explicit continuous-time neural modelling enables physically consistent representation of irregular satellite acquisition intervals and delayed hydro-mechanical responses, thereby advancing landslide susceptibility assessment from static spatial classification toward dynamic state evolution inference under asynchronous Earth observation data streams. Full article

Although the genus Rhododendron is globally distributed and rich in bioactive constituents, the metabolomic landscapes of most species remain unexplored, hampering elucidation of their adaptive strategies and pharmaceutical potential. Objectives: This study sought to construct comprehensive metabolic atlases of four representative yet understudied Rhododendron species—R. triflorum, R. faucium, R. nivale, and R. strigillosum—and to quantify inter-specific metabolic divergence by UPLC-MS/MS-based, widely targeted metabolomics. Methods: The petals of four Rhododendron species were freeze-dried, pulverised, and extracted with 70% methanol (containing an internal standard). Metabolites were separated on an SB-C18 column (2.1 × 100 mm, 1.8 µm) using a 0–95% acetonitrile gradient (flow rate 0.35 mL min⁻¹, 40 °C) and analysed by tandem mass spectrometry. Reliable quantification was ensured by molecular weight database matching, ion source standardisation, and quality control (QC), achieving a coefficient of variation (CV) < 15%. Principal component analysis (PCA) and optimised partial least squares discriminant analysis (OPLS-DA) were performed on standardised data with unit variance. Results: A total of 3705 metabolites were confidently identified, dominated by flavonoids (870), terpenoids (572), phenolic acids (394), and amino-acid derivatives (332). PCA and OPLS-DA models revealed clear species-specific clustering (R²Y ≥ 0.98, Q² ≥ 0.95; permutation test p < 0.01). Comparative analysis yielded 1495 significantly differential metabolites; R. triflorum exhibited the highest cumulative abundance, followed by R. faucium, R. nivale, and R. strigillosum. KEGG enrichment highlighted “metabolic pathways” as the most significantly over-represented, together with flavonoid biosynthesis, phenylpropanoid metabolism, and terpenoid backbone biosynthesis. Conclusions: The study delivers the first high-coverage metabolomic reference for four neglected Rhododendron species, evidencing profound inter-specific metabolic differentiation centred on flavonoids, terpenoids, and phenolic acids. The data provide a robust foundation for understanding molecular adaptation to alpine environments and for accelerating targeted drug discovery from Rhododendron resources. Full article

Mathematical and numerical models for Packed Bed Thermal Energy Storage (PBTES) systems are essential to predict the different parameters that influence their thermodynamic behavior and then optimize their performance and efficiency. In this research paper, an industrial-scale sensible thermocline Packed Bed Thermal Energy Storage system (9.17 m high and 4.72 m in diameter) was modeled and simulated during the heat charging process, based on FEM, CFD one-dimensional, and two-phase analysis. The model rigorously couples the Local Thermal Non-Equilibrium (LTNE) energy formulation with Darcy–Forchheimer hydrodynamics. The developed model was verified and validated using experimental data from the literature. The model was in close agreement with the experiment, with a global mean relative error of 3.62%. The two-dimensional velocity and temperature fields were presented to describe flow and temperature distributions in the hybrid medium (free and porous). The effect of varying flow rates (8–15 kg/s), porosities (0.35–0.55), and particle diameters (5–20 cm) on the thermal behavior of the heat storage system, temperature fields for solid and fluid, thermocline behavior, and charge efficiency were evaluated and presented. The simulation results demonstrate that the system achieves a high charge efficiency of 92.3% at a nominal charging rate of 15 kg/s. Increasing mass flow rate accelerates charging but widens the thermocline thickness and thermal stratification. Furthermore, increasing the porosity from 0.35 to 0.55 reduced charging time, decreased the temperature difference between the HTF and the storage medium by 10 °C, and increased the final heat charging efficiency by 8%. On the contrary, an increase in particle size from 5 to 20 cm leads to a slower rise in temperature within the solid phase, creating an important LTNE lag of ≈34 °C, thereby reducing the final heat charge efficiency by 16%, and prolonging the time required to charge the tank. Full article

Large-scale composite equipment manufacturing imposes stringent requirements on the lean management of multi-specification fiber prepreg sheet storage, while existing optimization methods suffer from poor process adaptability, insufficient multi-objective collaborative optimization capability, and low space utilization of static layouts. This study constructs a symmetric optimization framework for multi-objective composite sheet storage to address these critical bottlenecks. Specifically, the multi-dimensional process value of fiber sheets is quantified, and the layered storage optimization problem is transformed into a 0–1 knapsack problem with symmetric constraints. An improved Dynamic Programming–Backtracking (DP-BT) material selection algorithm and an adaptive dynamic programming iterative space optimization algorithm are proposed to achieve a symmetric balance of inter-layer space utilization and global optimization. Experimental validation with actual production data of 17 fiber sheet types verifies that the proposed method enables space optimization for specified layer counts to maximize average space utilization, with the rate rising from 79.4% (initial 4-layer layout) to 95.7% (3-layer) and 99.9% (2-layer), and a peak single-layer utilization of 100%. This framework achieves favorable optimization performance in the target production scenario and provides a referenceable symmetric optimization approach for the lean storage management of similar fiber sheet storage scenarios in composite manufacturing. Full article

The clinical diagnosis of sleep apnea conventionally necessitates resource-intensive Polysomnography (PSG). We propose a weakly supervised framework to detect apnea using non-invasive Ballistocardiography (BCG), thereby addressing the critical scarcity of labeled BCG data. Instead of manual annotation, our pipeline transfers knowledge from a synchronized ECG signal, using it as a “teacher” to generate pseudo-labels for the BCG model. We formulated a User-Defined Function (UDF) that combines Heart Rate Variability and ECG-Derived Respiration to autonomously label the BCG windows. These pseudo-labels were subsequently employed to train a 1D Convolutional Neural Network. Testing on a public dataset, the CNN model achieved 71.8% accuracy against the pseudo-labels. When projected against the clinical ground truth, we estimate a true accuracy of 77.7%. These results validate that ECG-based supervision can effectively train low-cost home sensors without the bottleneck of manual medical annotation. Full article

Satellite remote sensing offers a cost-effective solution for the continuous monitoring of black and odorous water bodies (BOWs). However, limitations in spatial and spectral resolution hinder the quantitative inversion of water quality parameters and the precise assessment of risk levels using satellite data alone. To address this challenge, this study proposes a synergistic approach combining satellite and Unmanned Aerial Vehicle (UAV) remote sensing to rapidly identify potentially polluted water bodies and quantitatively assess their risk levels. First, a Black and Odorous Water Index (MBOWI) was constructed based on reflectance characteristics in the visible to near-infrared bands to screen for potential black and odorous water bodies using satellite imagery. Subsequently, high-resolution multispectral UAV imagery, integrated with in situ sampling data, was employed to develop machine learning models for inverting key water quality parameters, including Chemical Oxygen Demand (COD), Dissolved Oxygen (DO), Total Phosphorus (TP) and Ammonia Nitrogen (NH₃-N). Comparative analysis of Polynomial Regression (PR), Random Forest (RF), and Simulated Annealing-optimized Support Vector Regression (SA-SVR) revealed that RF and SA-SVR exhibited superior performance in inverting four non-optically active water quality parameters due to their robust nonlinear fitting capabilities, with the mean Adjusted Coefficient of Determination ($R_{adj}^2$) ranging from 0.57 to 0.69. Water quality classification based on the single-factor worst-case method achieved an overall accuracy of 0.70 across validation samples. Notably, for Class V (heavily polluted) water bodies, both classification accuracy and recall rate reached 0.89, demonstrating the model’s high precision in identifying high-risk waters. Finally, the proposed framework was applied to northern Zhejiang Province to assess seven potential black and odorous water bodies, successfully identifying four as high-risk and one as low-risk. This study validates satellite and UAV synergistic remote sensing for the hierarchical risk management of black and odorous water bodies. Full article