Search Results (263)

In this study, data on the characteristics of overhead power lines of high voltage was used in a classification task to predict power supply outages by means of a supervised machine learning technique. In order to choose the most optimal features for power outage prediction, an Exploratory Data Analysis on power line parameters was carried out, including statistical and correlational methods. For the given task, five classifiers were considered as machine learning algorithms: Support Vector Machine, Logistic Regression, Random Forest, and two gradient-boosting algorithms over decisive trees LightGBM Classifier and CatBoost Classifier. To automate the process of data conversion and eliminate the possibility of data leakage, Pipeline and Column Transformers (builder of heterogeneous features) were applied; data for the models was prepared using One-Hot Encoding and standardization techniques. The data were divided into training and validation samples through cross-validation with stratified separation. The hyperparameters of the classifiers were adjusted using optimization methods: randomized and exhaustive search over specified parameter values. The results of the study demonstrated the potential for predicting power failures on 110 kV overhead power lines based on data on their parameters, as can be seen from the derived quality metrics of tuned classifiers. The best quality of outage prediction was achieved by the Logistic Regression model with quality metrics ROC AUC equal to 0.78 and AUC-PR equal to 0.68. In the final phase of the research, an analysis of the influence of power line parameters on the failure probability was made using the embedded method for determining the feature importance of various models, including estimating the vector of regression coefficients. It allowed for the evaluation of the numerical impact of power line parameters on power supply outages. Full article

Electrification introduces additional NVH (noise, vibration and harshness) challenges during the development of powertrain mounting systems due to high-frequency excitations from the powertrain and the absence of masking effects from the combustion engine. In these frequency ranges, engine mounts can stiffen up to a factor of five due to continuum resonances, reducing their structure-borne sound isolation properties and negatively impacting the customer’s NVH perception. Common hardening factors used during elastomer mount development are therefore limited in terms of their applicable validation frequency range. This study presents a methodology for determining decoupled permissible stiffness ranges for a double-isolated mounting system up to 1500 Hz, based on solution space engineering. Instead of optimizing for a single best design, we seek to maximize solution boxes, resulting in robust stiffness ranges that ensure the fulfillment of the formulated system requirements. These ranges serve as NVH requirements at the component level, derived from the sound pressure level at the seat location. They provide tailored guidelines for mount development, such as geometric design or optimal resonance placement, while simultaneously offering maximum flexibility by spanning the solution space. The integration of machine learning approaches enables the application of large-scale finite-element models within the framework of solution space analysis by reducing the computational time by a factor of $7.19·{10}^3$. From a design process standpoint, this facilitates frontloading by accelerating the evaluation phase as suppliers can directly benchmark their mounting concepts against the permissible ranges and immediately verify compliance with the defined targets. Full article

Mechanical vibrations induced by electromagnetic forces during transformer operation can lead to winding deformation or failure, an issue responsible for over 12% of all transformer faults. While previous studies have predominantly relied on accelerometers for vibration monitoring, this study explores the use of an optical sensor for real-time vibration measurement in a dry-type transformer. Experiments were conducted using a custom-designed single-phase transformer model specifically developed for laboratory testing. This experimental setup offers a unique advantage: it allows for the interchangeable simulation of healthy and deformed winding sections without causing permanent damage, enabling controlled and repeatable testing scenarios. The transformer’s secondary winding was short-circuited, and three levels of current (low, intermediate, and high) were applied to simulate varying stress conditions. Vibration displacement data were collected under load to assess mechanical responses. The primary goal was to classify this vibration data to localize potential winding deformation faults. Five supervised learning algorithms were evaluated: Random Forest, Support Vector Machine, K-Nearest Neighbors, Logistic Regression, and Decision Tree classifiers. Hyperparameter tuning was applied, and a comparative analysis among the top four models yielded average prediction accuracies of approximately 60%. These results, achieved under controlled laboratory conditions, highlight the promise of this approach for further development and future real-world applications. Overall, the combination of optical sensing and machine learning classification offers a promising pathway for proactive monitoring and localization of winding deformations, supporting early fault detection and enhanced reliability in power transformers. Full article

This study presents an integrated approach combining bibliometric analysis and machine learning to explore research trends and predict the performance of cement pastes containing bio-based phase change materials. A bibliometric review of 5928 articles from the Web of Science Core Collection was conducted using CiteSpace (v.6.3.R1) to identify research hotspots. A dataset of 100 experimental samples was compiled, including nine input variables and three output properties identified as thermal conductivity (Tc), latent heat capacity (LH) and compressive strength (CS). Four machine learning algorithms (SVR, RF, XGBoost, and CatBoost) were optimized using five metaheuristic algorithms (GA, PSO, WOA, GWO, and FFA), resulting in 24 optimized hybrid models. Of all the models considered, CatBoost-WOA achieved the best overall performance, with R² values of 0.927, 0.955, and 0.944, and RMSEs of 0.0057 W/m·K, 1.84 J/g, and 2.91 MPa for Tc, LH, and CS. Additionally, SVR-GWO and XGBoost-WOA also showed strong generalization and low error dispersion. The developed models provide a transferable and data-driven modeling pipeline for predicting the coupled thermal and mechanical behavior of cement pastes containing bio-based phase change materials. Full article

Oil spill simulation models are essential for predicting the oil spill behavior and movement in marine environments. In this study, we comprehensively reviewed a large and diverse body of peer-reviewed literature obtained from Scopus and Web of Science. Our initial analysis phase focused on examining trends in scientific publications, utilizing the complete dataset derived after systematic screening and database integration. In the second phase, we applied elements of a systematic review to identify and evaluate the most influential contributions in the scientific field of oil spill simulations. Our analysis revealed a steady and accelerating growth of research activity over the past five decades, with a particularly notable expansion in the last two. The field has also experienced a marked increase in collaborative practices, including a rise in international co-authorship and multi-authored contributions, reflecting a more global and interdisciplinary research landscape. We cataloged the key modeling frameworks that have shaped the field from established systems such as OSCAR, OIL-MAP/SIMAP, and GNOME to emerging hybrid and Lagrangian approaches. Hydrodynamic models were consistently central, often integrated with biogeochemical, wave, atmospheric, and oil-spill-specific modules. Environmental variables such as wind, ocean currents, and temperature were frequently used to drive model behavior. Geographically, research has concentrated on ecologically and economically sensitive coastal and marine regions. We conclude that future progress will rely on the real-time integration of high-resolution environmental data streams, the development of machine-learning-based surrogate models to accelerate computations, and the incorporation of advanced biodegradation and weathering mechanisms supported by experimental data. These advancements are expected to enhance the accuracy, responsiveness, and operational value of oil spill modeling tools, supporting environmental monitoring and emergency response. Full article

In the architectural design phase, the inspection of design deliverables is critical, yet traditional manual checking methods are time-consuming, labor-intensive, and inefficient, with numerous drawbacks. With the development of BIM technology, automated rule compliance checking has become a trend. This paper presents a method combining semantic web technology and IFC data to enhance human–machine collaborative inspection capabilities. First, a five-step process integrated with domain specifications is designed to construct a building object ontology, covering most architectural objects in the AEC domain. Second, a set of mapping rules is developed based on the expression mechanisms of IFC entities to establish a semantic bridge between IfcOWL and the building object ontology. Then, by analyzing regulatory codes, query rule templates for major constraint types are developed using semantic web SPARQL. Finally, the feasibility of the method is validated through a case study based on the Jena framework. Full article

Air pollution, particularly fine (PM_2.5) and coarse (PM₁₀) particulate matter, poses significant risks to public health and environmental sustainability. This study aims to develop robust predictive and forecasting models for hourly PM concentrations in Craiova, Romania, using advanced hybrid Artificial Intelligence (AI) approaches. A five-year dataset (2020–2024), comprising 20 meteorological and pollution-related variables recorded by four air quality monitoring stations, was analyzed. The methodology consists of three main phases: (i) data preprocessing, including anomaly detection and missing value handling; (ii) exploratory analysis to identify trends and correlations between PM concentrations (PMs) and predictor variables; and (iii) model development using 23 machine learning and deep learning algorithms, enhanced by 50 feature selection techniques. A deep Nonlinear AutoRegressive Moving Average with eXogenous inputs (Deep-NARMAX) model was employed for multi-step-ahead forecasting. The best-performing models achieved R² values of 0.85 for PM_2.5 and 0.89 for PM₁₀, with low RMSE and MAPE scores, demonstrating high accuracy and generalizability. The GEO-based feature selection method effectively identified the most relevant predictors, while the Deep-NARMAX model captured temporal dynamics for accurate forecasting. These results highlight the potential of hybrid AI models for air quality management and provide a scalable framework for urban pollution monitoring, predicting, and forecasting. Full article

Among the existing challenges in the field of hyperspectral imaging, the need to optimize memory usage and computational capacity in material detection methods stands out, given the vast amount of data associated with the hundreds of reflectance bands. In line with this, this article proposes a comparative study on the effectiveness and efficiency of five computational methods for detecting composite material asbestos cement (AC) in hyperspectral images: correlation, spectral differential similarity (SDS), Fourier phase similarity (FPS), area under the curve (AUC), and decision trees (DT). The novelty lies in the comparison between the first four methods, which represent the spectral proximity method and a machine learning method, such as DT. Furthermore, SDS and FPS are novel methods proposed in the present document. Given the accuracy that detection methods based on supervised learning have demonstrated in material identification, the results obtained from the DT model were compared with the percentage of AC detected in a hyperspectral image of the Manga neighborhood in the city of Cartagena by the other four methods. Similarly, in terms of computational efficiency, a 20 × 20 pixel region with 380 bands was selected for the execution of multiple repetitions of each of the five computational methods considered, in order to obtain the average processing time of each method and the relative efficiency of the methods with respect to the method with the best effectiveness. The decision tree (DT) model achieved the highest classification accuracy at 99.4%, identifying 11.44% of asbestos cement (AC) pixels in the reference image. However, the correlation method, while detecting a lower percentage of AC pixels (9.72%), showed the most accurate visual performance and had no spectral overlap, with a 1.4% separation between AC and non-AC pixels. The SDS method was the most computationally efficient, running 23.85 times faster than the DT model. The proposed methods and results can be applied to other hyperspectral imaging tasks involving material identification in urban environments, especially when balancing accuracy and computational efficiency is essential. Full article

This paper presents an advanced time-series forecasting methodology that integrates multiple machine learning models to improve data prediction in enhanced long-range navigation (eLoran) systems. The analysis evaluates five forecasting approaches: multivariate linear regression, long short-term memory (LSTM) networks, random forest (RF), a fusion model combining LSTM and RF, and a dynamic weighting (DW) model. The results demonstrate that the DW model achieves the highest prediction accuracy while maintaining strong computational efficiency, making it particularly suitable for real-time applications with stringent performance requirements. Although the LSTM model effectively captures temporal dependencies, it demands considerable computational resources. The hybrid model utilises the strengths of LSTM and RF to enhance the accuracy but involves extended training times. By contrast, the DW model dynamically adjusts the relative contributions of LSTM and RF on the basis of the data characteristics, thereby enhancing the accuracy while significantly reducing the computational demands. Demonstrating exceptional performance on the ASF2 dataset, the DW model provides a well-balanced solution that combines precision with operational efficiency. This research offers valuable insights into optimising additional secondary phase factor (ASF) prediction in eLoran systems and highlights the broader applicability of real-time forecasting models. Full article

Yield prediction in wheat cultivated under semi-arid climatic conditions is gaining increasing importance for sustainable production strategies and decision support systems. In this study, a time-series-based modeling approach was implemented using sensor-based data (SPAD, NSPAD, NDVI, INSEY, and plant height measurements collected at four different Zadoks growth stages (ZD24, ZD30, ZD31, and ZD32). Five different machine learning algorithms (Random Forest, Gradient Boosting, AdaBoost, LightGBM, and XGBoost) were tested individually for each stage, and the model performances were evaluated using statistical metrics such as R²%, RMSE t/ha, and MAE t/ha. Modeling results revealed that the ZD31 stage (first node detectable) was identified as the most successful phase for prediction accuracy, with the XGBoost model achieving the highest R²% score (81.0). In the same model, RMSE and MAE values were calculated as 0.49 and 0.37, respectively. The LightGBM model also showed remarkable performance during the ZD30 stage, achieving an R²% of 78.0, an RMSE of 0.52, and an MAE of 0.40. The SHAP (SHapley Additive exPlanations) method used to interpret feature importance revealed that the NDVI and INSEY indices contributed the most significant values to prediction accuracy for yield. This study demonstrates that phenology-sensitive yield prediction approaches offer high potential for sensor-based digital applications. Furthermore, the integration of timing, model selection, and explainability provided valuable insights for the development of advanced decision support systems. Full article

Objectives: The early detection of low bone mineral density (BMD) is essential for preventing osteoporosis and related complications. While dual-energy X-ray absorptiometry (DXA) remains the gold standard for diagnosis, its cost and limited availability restrict its use in large-scale screening. This study investigated whether raw bioelectrical impedance analysis (BIA) data combined with explainable machine learning (ML) models could accurately classify osteopenia in women aged 40 to 55. Methods: In a cross-sectional design, 138 women underwent same-day BIA and DXA assessments. Participants were categorized as osteopenic (T-score between −1.0 and −2.5; n = 33) or normal (T-score ≥ −1.0) based on DXA results. Overall, 24.1% of the sample were classified as osteopenic, and 32.85% were postmenopausal. Raw BIA outputs were used as input features, including impedance values, phase angles, and segmental tissue parameters. A sequential forward feature selection (SFFS) algorithm was employed to optimize input dimensionality. Four ML classifiers were trained using stratified five-fold cross-validation, and SHapley Additive exPlanations (SHAP) were applied to interpret feature contributions. Results: The neural network (NN) model achieved the highest classification accuracy (92.12%) using 34 selected features, including raw impedance measurements, derived body composition indices such as regional lean mass estimates and the edema index, as well as a limited number of categorical variables, including self-reported physical activity status. SHAP analysis identified muscle mass indices and fluid distribution metrics, features previously associated with bone health, as the most influential predictors in the current model. Other classifiers performed comparably but with lower precision or interpretability. Conclusions: ML models based on raw BIA data can classify osteopenia with high accuracy and clinical transparency. This approach provides a cost-effective and interpretable alternative for the early identification of individuals at risk for low BMD in resource-limited or primary care settings. Full article

Interest in explainable machine learning has grown, particularly in healthcare, where transparency and trust are essential. We developed a semi-automated evaluation framework within a clinical decision support system (CDSS-EQCM) that integrates LIME and SHAP explanations with multi-criteria decision-making (TOPSIS and Borda count) to rank model interpretability. After two-phase preprocessing of 2934 COVID-19 patient records spanning four epidemic waves, we applied five classifiers (Random Forest, Decision Tree, Logistic Regression, k-NN, SVM). Five infectious disease physicians used a Streamlit interface to generate patient-specific explanations and rate models on accuracy, separability, stability, response time, understandability, and user experience. Random Forest combined with SHAP consistently achieved the highest rankings in Borda count. Clinicians reported reduced evaluation time, enhanced explanation clarity, and increased confidence in model outputs. These results demonstrate that CDSS-EQCM can effectively streamline interpretability assessment and support clinician decision-making in medical diagnostics. Future work will focus on deeper electronic medical record integration and interactive parameter tuning to further enhance real-time diagnostic support. Full article

In order to achieve the optimal stator–rotor combinations of conventional flux-switching permanent magnet (FSPM) machines, this paper proposes and analyzes a general principle with prime phase numbers. Based on the coil complementarity concept, the proposed methodology specifically addresses the phase symmetry of back electromotive force (back-EMF) and electromagnetic torque optimization, with comprehensive analysis conducted for two-phase, three-phase, and five-phase configurations. Firstly, the coil-EMF vectors and the concept of coil pairs of conventional FSPM machines are introduced. Then, based on the coil-EMF vectors, an analytical model determining the stator pole and rotor teeth combinations is proposed. Further, the combinations for conventional FSPM machines with prime phase numbers are synthesized and summarized on the basis of the results obtained by the proposed model. To validate the model and combination principles, the FSPM machines satisfying the principles have been verified to exhibit a symmetrical phase back-EMF waveform by finite element analysis (FEA) and experiments on prototypes. In addition, the winding factors of the conventional FSPM machines with different stator pole and rotor teeth combinations are calculated. Full article

Flight behavior in pigeons is governed by intricate neural mechanisms that regulate movement patterns and flight dynamics. Among various kinematic parameters, flight acceleration provides critical information for the brain to modulate movement intensity, speed, and direction. However, the neural representation mechanisms underlying flight acceleration remain insufficiently understood. To address this, we conducted outdoor free-flight experiments in homing pigeons, during which GPS data, flight posture, and eight-channel local field potentials (LFPs) were synchronously recorded. Our analysis revealed that gamma-band activity in the dorsal intermediate arcopallium (AId) region was more prominent during behaviorally demanding phases of flight. In parallel, local functional network analysis showed that the clustering coefficient of gamma-band activity in the AId followed a nonlinear, U-shaped relationship with flight acceleration—exhibiting the strongest and most widespread connectivity during deceleration, moderate connectivity during acceleration, and the weakest network coupling during steady flight. This pattern likely reflects the increased neural demands associated with flight phase transitions, where greater cognitive and sensorimotor integration is required. Furthermore, using LFP signals from five distinct frequency bands as input, machine learning models were developed to decode flight acceleration, further confirming the role of gamma-band dynamics in motor regulation during natural flight. This study provides the first evidence that gamma-band activity in the avian AId region encodes flight acceleration, offering new insights into the neural representation of motor states in natural flight and implications for bio-inspired flight control systems. Full article

Characterising and quantifying complex multiphase steels is a challenging and time-consuming process, which is often open to subjectivity when based on image analysis of optical metallographic or SEM images. The properties of multiphase steels are highly sensitive to their individual phase properties and fractions, necessitating the development of robust characterisation tools. This paper presents a method for classifying nanoindentation maps into proportional fractions of up to five distinct microstructural regions in dual-phase and complex-phase steels. The phases/regions considered are ferrite, ferrite containing mobile dislocations, bainite, tempered martensite, and untempered martensite. A range of microstructures with varying fractions of phases were evaluated using both SEM/EBSD and nanoindentation. A machine learning (ML) approach applied to EBSD data showed good consistency in characterising a two-phase system. However, as the microstructural system complexity increased, variations were observed between different analysts and the sensitivity to the ML training data increased when four phases were present (reaching up to ~11% difference in the ferrite phase fraction determined). The proposed nanoindentation mapping technique does not show operator sensitivity and enables the quantification of additional microstructural features, such as identifying and quantifying ferrite regions with a high density of mobile dislocations and the degree of martensite tempering. Full article