Artificial Neural Network Model Research Articles

The topology structure of the artificial neural network is an intelligent control model, which is used for the intelligent vehicle control system and household sweeping robot. When setting the intelligent control system, the connection point of each network is regarded as a neuron in the nervous system, and each connection point has input and output functions. Only when the input of nodes reaches a certain threshold can the output function of nodes be stimulated. Using the networking mode of the artificial neural network model, the mobile node can output in multiple directions. If the input direction of a certain path is the same as that of other nodes, it can choose to avoid and choose another path. The weighted value of each path between nodes is different, which means that the influence of the front node on the current node varies. The control method based on the artificial neural network model can be applied to vehicle control, household sweeping robots, and other fields, and a relatively optimized scheme can be obtained from the aspect of time and energy consumption.

Glycolipid biosurfactants (BSs) are multifunctional biomolecules with potential applications in therapeutics and industry due to their biocompatibility and biodegradability. In this study, we report the isolation and characterization of a novel glycolipid biosurfactant from a Bacillus species, with emphasis on its anticancer properties and production optimization using a machine learning (ML) algorithm. Water samples from the Ganga River were screened for biosurfactant-producing strains, and the most efficient isolate was cultivated under optimized conditions. The purified biosurfactant was structurally characterized using Fourier-transform infrared (FTIR) spectroscopy, proton and carbon nuclear magnetic resonance (1H and 13C NMR), and liquid chromatography–mass spectrometry (LC–MS), confirming glycolipid moieties. A multilayer perceptron artificial neural network (MLP–ANN) model was employed to optimize medium composition and growth conditions, resulting in improved biosurfactant yield. The findings highlight both the anticancer activity and production efficiency of the newly identified glycolipid biosurfactant, supporting its potential in biomedical and biotechnological applications.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-20759-3.

Microplastic pollution in marine environments represents a significant ecological threat due to its persistence and harmful effects on biodiversity and human health. In Colombia, coastal ecosystems (particularly in La Guajira) have exhibited increasing microplastic concentrations, but systematic monitoring remains limited. This study explored the application of remote sensing, including multispectral satellite imagery (Sentinel-2) and machine learning algorithms, to detect and monitor microplastics in the coastal zone of Riohacha, La Guajira. To inform the model selection and ensure methodological relevance, a focused systematic literature review was conducted, serving as a foundational step in identifying effective remote sensing strategies and machine learning algorithms previously applied to microplastic detection in aquatic environments. Moreover, microplastic samples were collected from four coastal sites on Riohacha’s coast and analyzed via Fourier transform infrared spectroscopy (FTIR), while environmental parameters were recorded in situ. The remote sensing data were processed and integrated with field observations to train linear regression, random forest, and artificial neural network (ANN) models. The ANN model achieved the highest accuracy (MAE = 0.040; RMSE = 0.071), outperforming the other models in estimating the microplastic concentrations. Based on these results, environmental risk maps were generated, identifying critical zones of pollution. The findings support the integration of remote sensing tools and field data for scalable, cost-efficient microplastic monitoring, offering a methodological framework for marine pollution assessment in Colombia and other developing coastal regions.

With the increasing adoption of rooftop photovoltaic (PV) systems, accurate power output forecasting has become essential for effective energy management and grid integration. This study proposes a hybrid Artificial Neural Network (ANN) model optimized using the Salp Swarm Algorithm (SSA) to enhance prediction accuracy for rooftop PV output. SSA was selected for its strong exploration and exploitation capabilities, which complement the ANN’s learning strengths. Historical PV data from two university campuses in Malaysia, representing varied climatic conditions, were used over a one-year period to ensure model robustness. Key input variables influencing PV output were identified through correlation analysis, enabling more focused ANN training. SSA was used to optimize the ANN’s initial weights and biases, accelerating convergence and improving accuracy. Across three test cases, the SSA-ANN model achieved Mean Squared Error (MSE) values as low as 0.0155 and correlation coefficients (R) up to 0.98069, significantly outperforming standalone ANN approaches. These results demonstrate the model's effectiveness in improving PV forecasting accuracy, offering practical benefits for urban energy planning and sustainable power systems.

Abstract Bone scaffolds require tailored mechanical and structural properties to support tissue regeneration. This study optimizes fused deposition modeling (FDM)-printed commercial poly-lactic acid plus (PLA +) scaffolds enhanced with calcium carbonate (CaCO₃) and barium sulfate (BaSO₄) by evaluating printing temperature (190–230 °C), infill density (20–60%), and raster angle (30°–90°). Using response surface methodology (RSM) and analysis of variance (ANOVA), infill density was identified as the most statistically significant parameter (p < 0.05), followed by raster angle and temperature. Optimal parameters (230 °C, [0,90]° raster angle, 60% infill density) achieved an elastic modulus of 1537 MPa and compressive strength of 128 MPa, surpassing conventional PLA scaffolds. Artificial neural networks (ANN) outperformed RSM in predictive accuracy, with single-output ANN models yielding high correlation coefficients (training, testing, validation). Differential scanning calorimetry (DSC) confirmed elevated crystallinity at 230 °C, while X-ray diffraction (XRD) identified semi-crystalline PLA, presence of BaSO4 and CaCO3, and crystallinity shifts under compression. Fourier transform infrared spectroscopy (FTIR) revealed molecular interactions linked to strength variations, and thermogravimetric analysis (TGA)/derivative thermogravimetry (DTG) demonstrated improved thermal stability with CaCO₃/BaSO₄ additives (Weight loss: 70–74% at ≤ 210 °C vs. 92.21% at 230 °C). Scanning electron microscopy (SEM) and XRD correlated crystallinity with mechanical performance. By integrating ANN with RSM-driven validation, this work advances predictive modeling for FDM-based scaffold design, positioning commercial PLA + as a candidate for patient-specific bone tissue engineering.

Enhancing interlayer bond strength remains a critical challenge in the extrusion-based 3D printing of cementitious materials. This study investigates the optimisation of interlayer bond strength in extrusion-based 3D-printed cementitious materials through a combined application of Response Surface Methodology (RSM) and Artificial Neural Networks (ANNs). Using a concise yet comprehensive dataset, RSM provided interpretable main effects, curvature, and interactions, while the ANN captured non-linearities beyond quadratic forms. Comparative analysis revealed that the RSM model achieved higher predictive accuracy (R2=0.95) compared to the ANN model (R2=0.87). Desirability-based optimisation confirmed the critical importance of minimising casting delays to mitigate interlayer weaknesses, with RSM suggesting a water-to-cement (W/C) ratio of approximately 0.45 and a minimal time gap of less than 5 min, while ANN predicted slightly lower optimal W/C values but with reduced reliability due to the limited dataset. Sensitivity analysis using partial dependence plots (PDPs) further highlighted that ordinary Portland cement (OPC) content and W/C ratio are the dominant factors, contributing approximately 2.0 and 1.8 MPa respectively to the variation in predicted bond strength, followed by superplasticiser dosage and silica content. Variables such as water content, viscosity-modifying agent, and time gap exhibited moderate influence, while sand and fibre content had marginal effects within the tested ranges. These results demonstrate that RSM provides robust predictive performance and interpretable optimisation guidance, while ANN offers flexible non-linear modelling but requires larger datasets to achieve stable generalisation. Integrating both methods offers a complementary pathway to advance mix design and process control strategies in 3D concrete printing.

Introduction: Anaerobic digestion of livestock waste represents a sustainable alternative for energy generation, with methane serving as the main energy carrier. However, the experimental quantification of the biochemical methane potential (BMP) is time-consuming and costly. Objective: This study aimed to develop an Artificial Neural Network (ANN) model to estimate the methane content in biogas from livestock residues, using the substrates' elemental composition (C, H, and N) as input variables. Theoretical Framework: ANNs are applied to model nonlinear relations between livestock waste composition and methane yield, overcoming the limits of costly BMP assays. Method: Three ANN architectures were tested, with the 3-7-7-1 structure showing the best performance. The models were evaluated using statistical metrics (R² and RMSE) and a physical plausibility filter to exclude meaningless predictions. Results and Discussion: The network accurately predicted methane content (error <10%) in 66.67% of validated samples, with no valid prediction exceeding 17% error. Dataset limitations and waste heterogeneity caused some instability, although all three ANN configurations showed similar trends. Research Implications: The application of ANN proves promising for estimating methane content, delivering satisfactory performance despite experimental limitations. When combined with physical filters and critical analysis, ANNs can complement laboratory methods and provide a rapid tool for biogas assessment. Originality/Value: This study pioneers methane content prediction with ANN and a plausibility filter, offering a rapid alternative to laboratory methods.

Artificial Neural Network Modeling of Natural Convection Williamson Fluid Flow with Magnetic and Soret Effects in a Vertical Channel

RETRACTION: N.H. Abu‐Hamdeh, “Improving Energy Storage And Performance Through Implementing Artificial Neural Network Modeling To Forecast Heat Transfer And Entropy Generation Within A Wavy‐Wall Microchannel Under Discontinuous‐Boundary Condition–Hybrid Nanofluid Utilization”, Mathematical Methods in the Applied Sciences (Early View): https://doi.org/10.1002/mma.7319.The above article, published online on 31 May 2021 in Wiley Online Library (wileyonlinelibrary.com), has been retracted by agreement between the journal Editor‐in‐Chief, Wolfgang Sprößig; and John Wiley & Sons Ltd. The article was submitted as part of a guest‐edited special issue. Following publication, it has come to the attention of the journal that the article was accepted solely on the basis of compromised editorial handling and peer review processes. As a result, the data and conclusions are considered unreliable, therefore the article must be retracted. When informed of the decision, the authors have disagreed with the decision to retract.

A Hybrid Model of Artificial Neural Network and SARIMA Models for Predicting Inflation Rate Change in Nigeria's Economy

Continuous Removal of Fluoride by Fe/Al-Anchored Bamboo Charcoal in Fixed Bed Column and Prediction of Breakthrough Curve using Artificial Neural Network Model

In this study, a guar gum-based adsorbent was developed and evaluated for the removal of Ni(II) ions from aqueous solutions through a combined experimental and machine learning (ML) approach. The adsorbent was characterized using FTIR, SEM, XRD, TGA, and BET analyses to confirm surface functionality and porous morphology suitable for metal binding. Batch adsorption experiments were conducted to optimize the effects of pH, adsorbent dosage, contact time, temperature, and initial metal concentration. The adsorption efficiency increased with higher pH and adsorbent dosage, achieving a maximum Ni(II) removal of 97% (qₘ = 86.0 mg g−1) under optimal conditions (pH 6.0, dosage 1.0 g L−1, contact time 60 min, and initial concentration 50 mg L−1). The process followed the pseudo-second-order kinetic and Langmuir isotherm models. Thermodynamic results revealed the spontaneous, endothermic, and physical nature of the adsorption process. To complement the experimental findings, artificial neural network (ANN) and k-nearest neighbor (KNN) models were developed to predict Ni(II) removal efficiency based on process parameters. The ANN model yielded a higher prediction accuracy (R2 = 0.97) compared to KNN (R2 = 0.95), validating the strong correlation between experimental and predicted outcomes. The convergence of experimental optimization and ML prediction demonstrates a robust framework for designing eco-friendly, biopolymer-based adsorbents for heavy metal remediation.

Electrocoagulation (EC) employing aluminum–aluminum (Al–Al) electrodes was investigated for hospital wastewater treatment, targeting the removal of turbidity, soluble chemical oxygen demand (sCOD), and total dissolved solids (TDS). A hybrid modeling framework integrating response surface methodology (RSM) and artificial neural networks (ANN) was developed to enhance predictive reliability and identify energy-efficient operating conditions. A Box–Behnken design with 15 experimental runs evaluated the effects of pH, current density, and electrolysis time. Multi-response optimization determined the overall optimal conditions at pH 7.0, current density 20 mA/cm2, and electrolysis time 75 min, achieving 94.5% turbidity, 69.8% sCOD, and 19.1% TDS removal with a low energy consumption of 0.34 kWh/m3. The hybrid RSM–ANN model exhibited high predictive accuracy (R2 > 97%), outperforming standalone RSM models, with ANN more effectively capturing nonlinear relationships, particularly for TDS. The results confirm that EC with Al–Al electrodes represent a technically promising and energy-efficient approach for decentralized hospital wastewater treatment, and that the hybrid modeling framework provides a reliable optimization and prediction tool to support process scale-up and sustainable water reuse.

Abstract Evidence theory offers a flexible framework for characterizing both aleatory and epistemic uncertainties. However, uncertainty propagation under the evidence theory framework is computationally tedious due to the combinatorial explosion of input focal elements and frequent evaluations of the system response function for extremum analysis. To address these issues, this article proposes an active sampling approach that accurately and efficiently constructs a metamodel of the system response function, thereby reducing the frequency of system response evaluations. The proposed metamodeling strategy effectively balances exploration, exploitation, and robustness, while also establishing an optimal maximin distance strategy to generate well-distributed candidate sample points. Additionally, an artificial neural network (ANN) model is introduced to replace the extremum calculation of evidential variables. In constructing the ANN model, a centroid-based farthest point sampling method is developed to select training focal elements, with joint focal elements of inputs and response focal elements serving as input and output features of the ANN model, respectively. Furthermore, multiple stopping criteria based on the Hartley measure and Jousselme distance are applied to the iterative training process to ensure convergence. Numerical and engineering case studies demonstrate that the proposed method achieves high accuracy and efficiency when handling engineering applications with a large number of focal elements and high nonlinearity features.

Growing energy demand and the need to reduce the emission of greenhouse gases have created greater interest in alternative fuels such as diesel substitutes, with biodiesel, biogas, and bio-hydrogen being rated as the viable alternatives. Biodiesel improves combustion and reduces CO and HC emissions, biogas is economically viable utilization but its efficiency is impacted by the loss resulting from the presence of CO?, and bio-hydrogen supports the development of flame, thermal efficiency, and reduces carbon-based emissions. However, issues with abnormal combustion, reduced efficiency, and high levels of NOx with high levels of substitution necessitate optimization of the parameters. In this study, Response Surface Methodology (RSM) and Artificial Neural Networks (ANN) were used to evaluated and optimize the effect of load, compression ratio, ignition pressure, and gas flow rates on engine performance and emission. RSM outputs reported load, ignition pressure, and bio-hydrogen to have strong effects on BTE, BSFC, CO, and NOx with a maximum of 40.55% BTE, 303.48 g/kWh BSFC, 2.35 g/kWh CO, and 869.78 ppm NOx. ANN models reported a good predictive capability with R² > 0.99 and were better at predicting emission trends compared to RSM. The integration of RSM and ANN offers a highly effective tool for optimizing dual-fuel diesel engines to attain improved efficiency, improved fuel utilization, and reduced emissions for green energy use.

Hybrid numerical–artificial neural network modeling of bioconvection in MHD nanofluids with gyrotactic microorganisms

ABSTRACT The combined operation of multiple particulate matter (PM) emission sources in industrial and port areas creates major environmental threats and serious public health risks. Current methods of monitoring and predictive models lack sufficient capability to detect PM emission sources in real time. This study developed an integrated framework that uses Artificial Neural Networks (ANNs) and Computational Fluid Dynamics (CFD) to precisely locate PM emission sources in flat terrain. The CFD model was validated through experimental data analysis and the Monin-Obukhov similarity theory to precisely represent the particulate matter transport and atmospheric profiles. We created a simulation dataset containing 243 runs that tested different wind speed and direction combinations with variations in emission height and emission interval. The dataset served as training material for two deep learning models which used Long Short-Term Memory (LSTM) and a one-dimensional Convolutional Neural Network (CNN1D) to perform PM emission location classification. Both models achieved high accuracy levels with F1-scores above 0.95. The time needed to optimize hyperparameters proved the difference between models because LSTM required 4 h and 15 min and CNN1D needed 4 h and 43 min. This study proves that using CFD-generated data with ANN models allows reliable emission source localization which shows promise for environmental regulation, industrial accountability, and public health protection. The proposed framework represents a major breakthrough in real-time PM source localization in industrial and port environments.

The main aim of this study was to evaluate the optimum conditions for extracting the total reducing sugar content for bioethanol production using spirulina algae. The spirulina algae was pretreated using microwave-assisted acid hydrolysis, and the parameters were optimized using response surface methodology (RSM). The selected independent parameters were microwave power (250-350 W), sulfuric acid concentration (1-7%), and time duration (1-5min). The results showed that a maximum reducing sugar concentration of 3.8mg/mL was produced at optimum conditions. ANOVA and R-squared (R2) value (99.87%) show the model was significant (p value is < 0.0001). Additionally, a study on optimization and modeling was conducted utilizing response surface methodology (RSM) as well as artificial neural networks (ANN) to evaluate the impact of temperature (30-40°C), concentration of inoculum (1-5g/L), and fermentation duration (12-45h). This comparative assessment showed that the highest ethanol concentration of 1.824g/L was achieved under optimal conditions of 30°C, 5g/L inoculum concentration, and 28.5h duration, as determined by the high-performance liquid chromatography method. Finally, it is suggested that the RSM approach demonstrated superior performance with a higher R2 value (97.42%), p value is < 0.0001 (significant), and a lower mean square error (MSE) of 0.0065 compared to the ANN model.

This study experimentally investigates the effects of adding different amounts (1-5 wt.%) of Al2O3 particles on the wear behavior of glass fiber-reinforced epoxy composites to improve their tribological performance. Composite laminates produced using the hand-lay up method were subjected to wear tests using a ball-on-disc test setup under dry sliding conditions. Among all tested compositions, the composite containing 3 wt.% Al2O3 exhibited the highest wear resistance. Compared to the neat composite, the specific wear rate was reduced by up to 70%. In contrast, 4% and 5% Al2O3 additions resulted in a decrease in wear resistance due to particle agglomeration. While the highest specific wear rate was 260×10⁻⁶ mm³/Nm, this value decreased to 80×10⁻⁶ mm³/Nm in the 3% added sample. Furthermore, wear rate predictions were performed using models such as artificial neural network and different machine learning regressors. Random Forest (17.62%), Ridge regressor (18.46) and artificial neural network (19.92%) achieved the lowest MAPE values, indicating strong predictive performance for Al2O3-reinforced glass fiber composites. The artificial neural network model optimized with grid search achieved a mean squared error of 0.90 and a coefficient of determination of 0.92, while the random forest regressor demonstrated strong generalization with a coefficient of determination of 0.91. The results demonstrated the critical roles of both particle ratio and data-driven models in wear performance analysis.

Soil contamination remains a critical environmental concern, necessitating efficient techniques for site characterization and remediation. The radon-deficit technique (RDT) offers a non-invasive approach to identifying organic contamination, relying on the behavior of radon-222 (222Rn) as a tracer. However, RDT results are influenced by environmental variables such as soil moisture, temperature, and atmospheric pressure, potentially leading to uncertainties. This study evaluates the application of machine learning (ML) models-including linear regression (LR), random forest (RF), artificial neural network (ANN), and gradient boosting machine (GBM)-to predict 222Rn activity in soil gas based on environmental parameters. A year-long dataset of continuous measurements was collected from an uncontaminated granite-based site in Madrid, encompassing variables such as soil moisture, ambient and soil temperatures, and atmospheric conditions. ANN and RF models exhibited superior performance in predicting 222Rn variability, identifying soil moisture and ambient temperature as the most influential predictors. The findings demonstrate that ML can significantly enhance the reliability of RDT by accounting for environmental variability, enabling more accurate identification of contamination hotspots. While the application of these models requires substantial datasets, they offer a promising tool for improving the efficacy of contamination screening and long-term remediation monitoring. Further studies are recommended to explore ML's predictive capacity in contaminated sites and expand the approach to diverse geological contexts.