Search Results (6,119)

One of the most common problems in drilling operations is lost circulation, which can significantly increase well costs and lead to issues such as pipe sticking, blowouts, and even well closures. Identifying thief zones using analytical models is especially difficult, and there are no robust equations available in the literature due to a wide range of influential parameters, both controllable and uncontrollable. These parameters include operational factors, as well as the physical properties of the rock and drilling fluid. This study presents an artificial intelligence-based model designed to predict lost circulation zones. It investigates the underexplored potential of WV-curves for feature selection. Traditionally used to represent the spectral characteristics of training data, their role in feature selection has not been widely examined in the literature. The presentation of WV-curves is modified, and their effectiveness in identifying the optimal number of input and hidden neurons is evaluated. In this research study, a total of 15,000 data points were used and collected from oil wells in the Middle East. The artificial neural network (ANN) model exhibited a remarkable ability to accurately predict the locations of lost circulation zones based on the collected data, achieving an impressive accuracy of 94.5%. This is a significant achievement when compared to existing ANN models in the literature. The results highlight the strength of the ANN model in predicting lost circulation locations across a wide range of data collected from various wells in the Middle East. In addition, this model takes into account a diverse set of drilling operational parameters, as well as rock characteristics and fluid properties, offering a broader approach compared to other available ANN models. This advancement will also greatly facilitate future studies, enabling the prediction of lost circulation zones, and enabling advanced planning of appropriate prevention and remediation methods during the well planning phase to reduce the risk of lost circulation. Nevertheless, it should be noted that one limitation of the proposed methodology relates to data availability, as comprehensive formation parameters were not fully accessible; the inclusion of additional formation data may offer opportunities for further improvement in future studies. Full article

Artificial intelligence (AI) has transformed orthodontic diagnosis, yet its application to orthodontic materials science remains critically underexplored. This perspective identifies and characterizes the AI–materials integration gap as the central unresolved problem in digital orthodontics: AI-optimized treatment plans are currently executed through empirically selected materials whose mechanical behavior is never modeled by the planning system. We examine four domains where this gap is consequential: thermoplastic aligner polymers (PETG vs. TPU), where supervised ANNs can predict force decay from polymer composition; NiTi archwire alloys, where Bayesian optimization and Gaussian process regression are accelerating alloy design; additive manufacturing of orthodontic devices, where supervised ML reduced print-parameter optimization burden in a 2025 five-variable surface roughness study; and AI-driven biological response prediction, where FEA-surrogate neural networks reduced biomechanical computation from minutes to milliseconds per patient query. A scoping review of clear aligner AI identified 41 studies—none addressing aligner material properties as a primary outcome. We argue that closing the AI–materials gap requires standardized open material-performance datasets; FEA-surrogate models integrating polymer stiffness as a treatment-planning input; patient-specific digital twins with defined material, mechanical, and biological parameter layers; and federated learning infrastructure spanning clinics and manufacturers. Full article

Designing lightweight composite sandwich structures is challenging due to the conflicting objectives of minimizing structural weight and cost while satisfying strength and stiffness requirements. The optimization procedure becomes more complex when multiple discrete design variables and nonlinear material behavior are involved. This study presents a newly developed optimization methodology for a sandwich structure composed of Fiber Reinforced Polymer (FRP) laminated facesheets and an aluminum honeycomb core. To reduce the computational cost associated with repeated high-fidelity Finite Element (FE) analyses, a surrogate modeling strategy based on Artificial Neural Networks (ANNs) is employed to approximate the structural response. The applied dataset is generated using Monte Carlo simulation in which combinations of design variables are used as inputs, and the corresponding structural responses obtained from the analytical formulation are used as outputs for training the ANN surrogate model. The trained ANN model is integrated with a Multi-Objective Niching Memetic Particle Swarm Optimization (MO-NMPSO) algorithm to simultaneously minimize structural weight and material cost while satisfying constraints on facesheet strength, wrinkling, intra-cell buckling, deflection, core shear failure and structural thickness. The resulting Pareto-optimal solutions are validated through detailed FE simulations, demonstrating the reliability of the newly elaborated optimization framework. The results of the newly developed computationally efficient optimization procedure provide a diverse set of optimal design solutions for the investigated sandwich structure. Full article

Accurate fault diagnosis of power transformers using Dissolved Gas Analysis (DGA) depends on effective feature extraction to reduce redundancy and improve classification performance. This study compares linear and nonlinear feature extraction methods viz. Principal Component Analysis (PCA) and bottleneck Autoencoders (AE) to determine whether nonlinear representations provide diagnostic advantages for transformer fault classification. A dataset of 595 IEC 60599-labeled DGA samples covering six fault classes (PD, D1, D2, T1, T2, T3) was used. A 15-dimensional feature space was constructed from gas concentrations, total hydrocarbon content, and IEC-aligned gas ratios. PCA and AE were applied for dimensionality reduction across latent dimensions (k = 1–15), followed by an identical Artificial Neural Network (ANN) classifier. Performance was evaluated using test accuracy, cross-validation stability, and per-class F1-scores. The PCA+ANN model achieved a maximum accuracy of 68.9% at k = 11, outperforming AE+ANN, which achieved 66.4% at k = 4. PCA also demonstrated greater cross-validation stability (62 ± 3.5%) compared to AE (62 ± 6.6%). However, AE improved F1-scores for discharge faults (D1 and D2) by enhancing nonlinear separation of overlapping samples. PCA provides superior overall accuracy and stability for transformer fault diagnosis, while AE offers targeted advantages in distinguishing discharge-related faults. These findings establish a consistent benchmark for future studies and highlight the complementary roles of linear and nonlinear feature extraction in DGA-based diagnostic systems. Full article

This paper introduces a cutting-edge, real-time fault detection and classification method powered by artificial neural networks (ANNs), designed to significantly boost the reliability and sustainability of medium voltage (MV) underground cable distribution systems. The research analyzes the electrical and physical properties of MV underground cables and common fault types, including line-to-line, line-to-ground, and double line-to-ground faults. A simulation model is developed using MATLAB/Simulink R2025b to generate fault scenarios under various operating conditions. Raw data in the form of Voltage and current signals are generated and processed to extract significant features, which are then fed into the ANN model. The ANN is trained using a supervised learning approach, using a dataset of labeled fault instances. Key parameters like hidden layers, activation functions, and learning rates are optimized to improve the model’s performance. The results show that the proposed ANN-based fault detection technique achieves over 95% accuracy in detecting and classifying faults in real-time, with minimal computational delay. Comparative analysis with conventional fault classification techniques demonstrates the superiority of the ANN model in handling noisy and non-linear data. Full article

Coal combustion in large-scale power plants is a major source of atmospheric pollution, including SO₂, NO_x, particulate matter, and the halogen acids HCl and HF. Predicting HCl and HF emissions is challenging due to interactions among fuel composition, fly ash chemistry, combustion conditions, and flue gas dynamics. In this study, artificial neural network (ANN) models are developed from field experiments at the lignite-fired TPP “Kostolac B”. The models incorporate operational parameters (flue gas temperature and flow rate) and fuel/ash characteristics (moisture and total sulphur in coal and CaO content in ash) to estimate HCl and HF emissions. SHAP analysis identified key variables affecting halogen acid release. The developed ANN models achieved satisfactory predictive accuracy, with the test-set performances of RMSE = 2.05 mg/Nm³, R² = 0.80, and MAPE = 18.7% for HCl prediction, and RMSE = 3.23 mg/Nm³, R² = 0.83, and MAPE = 18.7% for HF prediction. SHAP analysis indicated that CaO content in fly ash and coal moisture are the primary drivers of HCl and HF emissions, while operating conditions and coal sulphur content influence emissions through non-linear interaction effects. The proposed ANN-SHAP framework provides a data-driven approach for emission prediction and interpretation, supporting decision-making in emission management. Full article

As the demand for sustainable construction materials grows, wood ash and nanosilica have emerged as promising components for eco-friendly mortars, whose optimization requires advanced analytical techniques capable of capturing their complex linear and nonlinear interactions, making frameworks such as response surface methodology and artificial neural networks essential for effective mix design. This study examines the mechanical performance of eco-friendly mortar incorporating wood ash (WA) as a partial cement replacement and nanosilica solution (NSS) as a strength-enhancing additive, with the aim of optimizing compressive and flexural behaviour. Wood ash was substituted at levels of 5–25%, while NS (0.265 moL⁻¹) was substituted at levels of 0–1.7%. Twenty-one mortar samples were produced and tested at multiple curing ages. Two modelling techniques, response surface methodology (RSM) and artificial neural networks (ANNs), were employed to evaluate the individual and interactive effects of WA and NSS on strength development at curing ages of 28 and 180 days. While RSM provided insight into factor significance and linear interactions, ANN more effectively captured nonlinear behaviour, achieving superior predictive accuracy (R² = 1.000 for 28-day strength). Experimental results revealed that nanosilica substantially enhanced strength up to an optimal dosage of approximately 2.5 g, beyond which performance declined due to particle agglomeration or matrix over-refinement. In contrast, higher WA contents produced strength reductions attributable to dilution effects. Optimization showed that mixtures containing low WA (≤30 g) combined with moderate NSS (2.0–2.5 g) exhibited the highest mechanical performance. Collectively, the findings confirm that ANN-based models outperform RSM and multilinear regression, underscoring their effectiveness for mix design optimization and performance forecasting in sustainable cementitious systems. Full article

Solar chimneys represent an effective passive ventilation technology capable of improving indoor thermal comfort while reducing building energy consumption. In this study, the thermal and fluid dynamic performance of a solar chimney integrated into a residential building located in Bordj Bou Arréridj (Eastern Algeria) was investigated through a comprehensive numerical, predictive, and optimization framework. A transient mathematical model was developed to evaluate the influence of key geometric parameters, including chimney width and inlet opening width, as well as environmental factors such as solar radiation intensity and wind speed, on the system performance. The generated simulation database was subsequently employed to develop and compare four machine learning models, namely, Artificial Neural Networks with Bayesian Regularization (ANN-BR), Deep Neural Networks optimized by Improved Grey Wolf Optimization (DNN-IGWO), k-Nearest Neighbors (KNN), and Extreme Gradient Boosting (XGBoost), for predicting eight output parameters including glazing temperature, fluid temperature, absorber temperature, outlet temperature, thermal efficiency, air change rate (ACH), mass flow rate, and outlet velocity. The results demonstrated that increasing chimney and inlet widths significantly enhances ventilation performance by increasing airflow rate and ACH. Weather conditions and wind speed were also found to strongly affect thermal efficiency and buoyancy-driven airflow. Among the predictive models, XGBoost and DNN-IGWO exhibited the highest predictive accuracy, achieving coefficients of determination ($R^2$) close to unity and very low prediction errors for all output variables, confirming their robustness and generalization capability. The proposed methodology provides a reliable tool for rapid performance prediction and design optimization of solar chimney systems under different climatic and operating conditions, thereby supporting the development of energy-efficient passive ventilation strategies for residential buildings. Full article

The dairy industry produces large amounts of dairy wastewater containing ammonia nitrogen (NH₃-N). Sustainable treatment technologies are needed which can reduce the environmental pollution caused by NH₃-N emissions from dairy wastewater. Chemical coagulation combined with the photo-electro-Fenton (PEF) treatment process has been considered a promising technology that can effectively remove NH₃-N from dairy wastewater. In this study, Taguchi design was used first to narrow down the operating factors from five to three. The three most influential factors were then further optimized for an optimum NH₃-N removal efficiency using response surface methodology (RSM) coupled with Box–Behnken design. Both RSM and artificial neural network (ANN) models were developed to predict the NH₃-N removal efficiency. Under the optimal conditions of 0.51 mM Fe²⁺, 49.44 mA/cm² current density, and 118.60 min treatment time, removal of 92.13% NH₃-N from dairy wastewater with 90% N₂ selectivity was achieved during validation experiments. The ANN model showed a superior predictive performance to the RSM model. The NH₃-N degradation rate was calculated at 0.0229 min⁻¹ based on a pseudo-first-order kinetic model. These findings demonstrate the applicability of the integrated chemical coagulation and PEF process for significantly reducing ammonia nitrogen in dairy wastewater. Full article

This study addresses the challenge of modeling the complex non-linear relationship between process parameters and relative density in selective laser melting (SLM) of IN718 nickel-based superalloy under small-sample conditions. A data-driven prediction framework integrating data augmentation, physics-informed feature engineering, machine learning, and model interpretability analysis was developed and systematically validated. Fourteen sets of experimental data covering both vertical and horizontal building directions were collected by varying laser power (P), scan speed (v), and hatch spacing (h). To overcome the small-sample limitation, three augmentation strategies—radial basis function (RBF) interpolation, generative adversarial network (GAN), and K-nearest neighbors (KNN)—were systematically compared under unified physical constraints combining local perturbation and volumetric energy density (E_vol) filtering, with Pearson correlation coefficient consistency used to select the optimal strategy. Eight physically meaningful input features were constructed, including E_vol and line energy density (E_line), explicitly embedding SLM process physics into the learning framework. Support vector regression (SVR), random forest (RF), and artificial neural network (ANN) models were trained and their hyperparameters were systematically optimized via exhaustive grid search combined with leave-one-out cross-validation (LOO-CV), ensuring robust model selection under small-sample constraints. A physics-based baseline model (E_vol quadratic fitting, LOO-CV average R² = 0.2534) was established to quantify the gain of machine learning over empirical formulas. LOO-CV results show that ANN achieves the highest average R² of 0.9269, followed by SVR (0.9148) and RF (0.8393), all of which substantially outperform the physical baseline. Feature importance analysis reveals that E_vol accounts for 51.58% of the predictive power, and ablation experiments confirm that introducing physics-derived features improves the average R² by 0.0246 compared with raw process parameters alone. To further elucidate the predictive mechanism of the optimal ANN model, Partial Dependence Plot (PDP) analysis was conducted for all eight input features, visualizing their marginal effects on predicted density and confirming physical consistency with SLM mechanisms. This framework provides a reliable, interpretable, data-driven solution for intelligent SLM process optimization with limited experimental data. Full article

Inverse analysis from indentation experiments has been a challenging problem due to the nonlinear relationship between indentation response and material parameters. In this work, a data-driven method is proposed that integrates an artificial neural network (ANN) and evolutionary optimization for the reliable and efficient inverse parameter identification. A large dataset is generated by simulating the indentation process based on different combinations of material parameters in a systematic way. Then, by using the simulated data, a set of ANN models is trained that can efficiently predict the indentation responses, i.e., the displacement–time curve, the indentation force, and the surface profile, as a function of material parameters. These trained models exhibit the potential to replace the computationally expensive numerical simulations for the identification of material parameters by inverse analysis. In this way, the surrogate models make the numerical evaluation of the loss function, which is minimized during the inverse analysis, orders of magnitude faster. This enables the use of the powerful genetic algorithm for the minimization of the loss function, which would be impossible without numerically efficient surrogate models, as this algorithm requires many iterations to produce robust results. In this work, we systematically investigate which mathematical loss function leads to robust and unique results in determining the material parameters through inverse analysis of indentation results. The results show that such an inverse analysis can be successfully performed for simulation data. In forthcoming work, this method will be generalized to experimental indentation data, which will allow the characterization of the mechanical behaviour of materials by micro- or nano-indentation tests. Full article

Accurate design and performance assessment of solar thermal domestic hot water systems coupled with a heat pump auxiliary typically requires transient simulation, as the system’s behavior depends on multiple interactions among collector characteristics, storage stratification, control logic, weather, and draw-off timing. Monthly methods such as the f-chart are useful for first-pass estimates, but they do not resolve stratification, thermostat operation, or demand timing, and they may become inaccurate for stratified thermostat-controlled systems. Direct comparisons of locally inspectable symbolic and black-box surrogate families for this system class remain limited. A 10,982-case development dataset was generated from minute-resolved annual MATLAB simulations, parameterized by collector area, optical efficiency, and first- and second-order loss coefficients. Three surrogate families were benchmarked under a unified protocol, random forest-assisted shape-constrained symbolic regression (SR), feed-forward artificial neural network (ANN) models, and Automatic Learning of Algebraic Models for Optimization (ALAMO), with the f-chart used as a monthly reference method. The targets were the 12 monthly solar fractions under the direct solar heat definition and the corresponding annual mean solar fraction, evaluated on the same independent 991-case test set. SR achieved the lowest average error (mean absolute percentage error, MAPE = 0.82%; root mean square error, RMSE = 0.006), followed by the ANN (MAPE = 2.07%, RMSE = 0.028) and ALAMO (MAPE = 3.67%, RMSE = 0.060), with Nash–Sutcliffe efficiency (NSE) values above 0.98 for all models. Evaluation times were 0.0026–0.124 s per target, compared with about 1000 s for one full-year simulation. These results define the study as a common protocol benchmark within the studied simulator-defined envelope. SR gives the strongest accuracy with local symbolic inspectability, the ANN remains the flexible retrainable option, and ALAMO provides compact algebraic evaluation with the shortest learned model runtime. Full article

Maximum power point tracking (MPPT) is a critical function for maximizing energy extraction in photovoltaic (PV) systems. Due to the inherently dynamic nature of the maximum power point under varying irradiance conditions, achieving fast convergence, low steady-state oscillations, and high tracking efficiency remains a challenging research problem. This paper proposes a hybrid ANN-based MPPT strategy for photovoltaic systems operating under rapidly changing environmental conditions. The proposed approach integrates a rule-based operating-condition estimation stage with a recurrent ANN-based control stage, enabling adaptive duty-cycle generation using measured PV voltage and current signals. Unlike conventional MPPT techniques, the proposed method utilizes operating-region estimation together with an extended ANN input feature vector and a recurrent backpropagation neural network to improve dynamic tracking performance under abrupt irradiance variations. In addition, a composite loss function is adopted to enhance tracking accuracy, guidance consistency, and control smoothness. The ANN is initially trained offline and subsequently refined online using lightweight incremental adaptation to maintain effective operation with a low computational burden. The proposed MPPT strategy is evaluated against P&O, FLC, and SMC. Simulation results demonstrate improved tracking performance, faster dynamic response, and reduced steady-state oscillations under abrupt irradiance variations. Full article

Artificial intelligence is rapidly entering higher education, yet many students experience anxiety when learning to use it. This study examines how performance expectations, perceived explainability, and perceived ethical risks shape two algorithm attitudes, algorithm aversion and algorithm appreciation, and how these attitudes influence artificial intelligence learning anxiety. Using a hybrid partial least squares structural equation modeling–artificial neural network (PLS-SEM–ANN) approach, this study analyzed survey data from 409 university students. Results show that both algorithm aversion and algorithm appreciation significantly increase artificial intelligence learning anxiety, although the effect of algorithm aversion is much stronger, supporting an approach–avoidance account. Perceived ethical risk is the strongest predictor of algorithm aversion but has no significant effect on algorithm appreciation. By contrast, performance expectations and perceived explainability strengthen algorithm appreciation while also showing weaker positive effects on algorithm aversion. These findings suggest that, in educational settings, stronger performance value and greater explainability do not simply reassure students; they can also increase pressure by making errors, responsibility, and the need to use artificial intelligence effectively more salient. The artificial neural network results corroborate these patterns. This study extends research on algorithm attitudes and offers guidance for creating more supportive artificial intelligence learning environments. Full article

There are significant difficulties with power quality and stability as a result of active cooperation between renewable energy sources and load demand. To maintain power stability between renewable energy supplies and the microgrid/utility grid, novel solutions must be implemented. By using an artificial and computational intelligence controller to schedule power from multiple sources (photovoltaic, wind, grid, and battery) under a set of constraints, such as weather, load-shedding hours, and peak pricing hours, this paper introduces a novel approach for power management in grid-connected hybrid renewable systems with PV–wind and energy storage systems. The approach involves using an artificial neural network (ANN) to process all of the inputs and creating an ANN rule set from a modelled hybrid renewable system. A rule-based power scheduler is developed, and simulations are run for a full day. The suggested fuzzy control approach can detect ongoing variations in grid load-shedding patterns, PV–wind power generation, load demands, and battery state-of-charge to enable prompt and accurate decision-making. The proposed ANN rule-based scheduler can handle nonlinearity by integrating metaheuristics into computer-assisted decision-making and can function effectively with imprecise inputs, negating the need for an exact numerical model. The MATLAB/Simulink R2023a software was used for simulation, and the system operated as efficiently as possible. The simulation results suggested that an ANN offers a foundation for extension to handle numerous particular scenarios. Full article