Artificial Neural Network Approach Research Articles

A once-through artificial neural network approach for used nuclear fuel inverse depletion analysis: A comparative study

This work addresses the challenge of identifying the origins and burnup history of used nuclear fuel, which is crucial for safeguarding and non-proliferation analysis. It presents a comparative evaluation of two distinct computational approaches to bundles at the same in-core position: the first entails building a forward Artificial Neural Networks (ANN) surrogate model for fuel inventory prediction followed by solving the inverse depletion problem using an optimization solver. The second approach is to build an inverse ANN model that relates the used nuclear fuel inventory to its initial and burnup conditions (a once-through approach), eliminating the need for the inverse optimization solver step which results in better accuracy, less computational costs, and renders the application of such approach practical. KENO VI depletion model is utilized to deplete a VVER assembly model generating the necessary datasets for the ANN inverse model training and validation. The inverse model is then employed to estimate the initial composition and the burnup for several test cases. The results are then compared to the known initial and burnup conditions as well as to the estimations made by benchmark results from an existing ANN forward surrogate model combined with the Particle Swarm Optimization (PSO) solver. Comparative analyses across eight scenarios consistently reveal that the once-through ANN inverse model has superior accuracy and consistency in terms of estimating the actual burnup and initial fuel enrichment with a relative root mean squared error (RMSE) of 1.20×10-2 compared to the benchmark approach (ANN-PSO) which yielded a relative RMSE of 2.18×10-2. Moreover, the once-through inverse approach has a significantly lower computational cost as it requires 1 inverse model run compared to the 9,450,000 model runs required by the benchmark approach (ANN-PSO). Overall, the results demonstrate that the proposed once-through approach simplifies and enhances the efficiency of solving the inverse fuel depletion problem while outperforming the benchmark approach in terms of predictive accuracy computational cost.

Measuring BIM Implementation: A Mathematical Modeling and Artificial Neural Network Approach

Measuring BIM Implementation: A Mathematical Modeling and Artificial Neural Network Approach

Prediction of Delay Occurrence in Disputed Real Estate Projects: An Artificial Neural Network Approach

Prediction of Delay Occurrence in Disputed Real Estate Projects: An Artificial Neural Network Approach

Orthogonal array and artificial neural network approach for sustainable cutting optimization machining of 17-4 PH steel under CNC wet turning operations

Sustainable manufacturing strives to increase output while minimizing resource usage, costs, and environmental impact. Tool longevity is crucial, considering material, power, and resource consumption. Challenges like chip removal, heat, and friction necessitate effective solutions. Cutting fluids, like rice bran oil, reduce temperatures, friction, and enhance tool life and surface quality. Research focuses on using rice bran oil with a TiAlN tipped tool to minimize friction and cutting forces. Taguchi's robust design and L-9 orthogonal array reduce experimental trials. Controllable factors include speed, feed, depth of cut, and rice bran oil flow rate, with surface roughness as the performance metric. Taguchi analysis improves turning performance, particularly with 17-4 PH stainless steel. Rice bran oil reduces frictional forces, enabling improved surface quality across all samples, promoting sustainability. Surface roughness (Ra) is notably influenced by feed rate and rice bran oil, contributing 58.06% and 49.5%, respectively. This research underscores the potential of optimizing cutting fluid and process parameters to enhance sustainability in manufacturing.

Prediction of pressure evolution in non-venting self-pressurized liquid hydrogen tanks using artificial neural network approach

Liquid hydrogen (LH2) is a future energy carrier compared to most conventional fuels. One of the practical bottlenecks in non-venting LH2 tanks is to accurately predict the pressure evolution. Accordingly, many thermal models and correlations were developed; however significant deviations arose due to inherent complex thermal phenomena which have not been addressed yet. A general artificial neural network (ANN) model has been first developed based on 448 datasets amassed from different literature sources to predict the pressure evolution under different heat fluxes, initial filling ratios, as well as tank volumes and shapes. The model has been configured with a typical multi-layer feed-forward structure using a back-propagation learning algorithm and Levenberg-Marquardt training algorithm. For the ANN model, a preliminary optimization process has been conducted for reaching the optimal values for training-to-testing datasets ratio, learning parameters and optimum network structure. Besides, some criteria were set to evaluate the ANN model performance. The ANN model with the configuration of 5-11-11-1 was found to be adopted for such a prediction task. Compared to experimental results and literature models and correlations, the ANN model has been proven to predict the pressure evolution in non-venting self-pressurized LH2 tanks with a mean square error of 1.1779 × 10−5 s2, root mean square error of 0.0034 s, mean absolute error of 6%, mean relative error of 0.2% as well as a satisfactory correlation coefficient of 0.99934 between the network outputs and corresponding targets. This research work provides an effective modeling approach based on artificial neural network to solve one complex problem related with LH2 tanks with fast, accurate and consistent results. Also, the results of the present work can offer some practical guidance for the design of LH2 tanks. Further, the present work gives some insights into new application areas of ANN which will be the emphasis of future research work.

Comparative analysis of sliding wear behaviour of A356 nanocomposites using response surface methodology and artificial neural network approaches

Abstract The present study is aimed at analysing the predictive capacity of response surface methodology and artificial neural network of wear behaviour of A356/Al2O3 nanocomposites. In order to develop nanocomposites with different Al2O3 content the mechanical milling and powder metallurgy routes were adopted. The wear testing experiments were conducted using pin on disc tribometer to study the influence of parameters such as Al2O3 content, load, sliding speed and distance on wear loss. The testing was conducted based on the experimental design generated through Taguchi’s L27 technique. The response surface methodology and artificial neural network were used to predict the wear loss of A356 nanocomposites and comparative analysis was performed to analyse the predictive capability of these two techniques. Analysis of variance results showed significant influence of sliding speed on the wear loss while impact of sliding distance was minimal. The average relative error between the artificial neural network predicted and experimental value was 4.861% while for response surface methodology it was 9.307%. This comparative analysis indicated better predicting capacity for artificial neural network model. Worn surface analysis showed dominant abrasion and mild delamination as wear mechanisms for both unreinforced and nanocomposite samples.

An Artificial Neural Network Approach for Construction Project Risk Management

An Artificial Neural Network Approach for Construction Project Risk Management

Estimation of friction and wear properties of additively manufactured recycled-ABS parts using artificial neural network approach: effects of layer thickness, infill rate, and building direction

Abstract This investigation aims to elucidate friction and wear features of additively manufactured recycled-ABS components by utilizing neural network algorithms. In that sense, it is the first initiative in the technical literature and brings fused deposition modeling (FDM) technology, recycled filament-based products, and artificial neural network strategies together to estimate the friction coefficient and volume loss outcomes. In the experimental stage, to provide the required data for five different neural algorithms, dry-sliding wear tests, and hardness measurements were conducted. As FDM printing variables, layer thickness (0.1, 0.2, and 0.3 mm), infill rate (40, 70, and 100 %), and building direction (vertical, and horizontal) were selected. The obtained results pointed out that vertically built samples usually had lower wear resistance than the horizontally built samples. This case can be clarified with the initially measured hardness levels of horizontally built samples and optical microscopic analyses. Besides, the Levenberg Marquard (LM) algorithm was the best option to foresee the wear outputs compared to other approaches. Considering all error levels in this paper, the offered results by neural networks are notably acceptable for the real industrial usage of material, mechanical, and manufacturing engineering areas.

The Effect of Biotic Stress in Plant Species Induced by ‘Candidatus Phytoplasma solani’—An Artificial Neural Network Approach

Infections with phytoplasma present one of the most significant biotic stresses influencing plant health, growth, and production. The phytoplasma ‘Candidatus Phytoplasma solani’ infects a variety of plant species. This pathogen impacts the physiological and morphological characteristics of plants causing stunting, yellowing, leaf curling, and other symptoms that can lead to significant economic losses. The aim of this study was to determine biochemical changes in peony (Paeonia tenuifolia L.), mint (Mentha × piperita L.), and dill (Anethum graveolens L.) induced by ‘Ca. Phytoplasma solani’ in Serbia as well as to predict the impact of the biotic stress using artificial neural network (ANN) modeling. The phylogenetic position of the Serbian ‘Ca. Phytoplasma solani’ strains originated from the tested hosts using 16S rRNA (peony and carrot strains) and plsC (mint and dill strains) sequences indicated by their genetic homogeneity despite the host of origin. Biochemical parameters significantly differed in asymptomatic and symptomatic plants, except for total anthocyanidins contents in dill and the capacity of peony and mint extracts to neutralize superoxide anions and hydroxyl radicals, respectively. Principal Component Analysis (PCA) showed a correlation between different chemical parameters and revealed a clear separation among the samples. Based on the ANN performance, the optimal number of hidden neurons for the calculation of TS, RG, PAL, LP, NBT, •OH, TP, TT, Tflav, Tpro, Tant, DPPH, and Car was nine (using MLP 8-9-13), as it produced high r2 values (1.000 during the training period) and low SOS values. Developing an effective early warning system for the detection of plant diseases in different plant species is critical for improving crop yield and quality.

Optimasi Jaringan Syaraf Tiruan Dalam Memprediksi Jumlah Penduduk Nasional

Predictions or forecasting are important parameters in determining policy that will be implemented in the future. The higher the population of a country, the more it will affect the balance of that country. This research aims to predict the national population using the Artificial Neural Network (ANN) approach. The sample for this research is population data for each province obtained from the National Central Statistics Agency (BPS). The data is iterated from 2011 to 2022. The approach method uses backpropagation Artificial Neural Networks (ANN). The architectural model is determined with 2 hidden layers and 1 output, namely 5-20-10-1 with learning rate variations of 0.01, 0.02, and 0.03. ANN prediction results are able to predict the national population. The success of the three models shows that the ANN model with a Learning rate of 0.01 has the highest accuracy level of 96.9697, equivalent to 97%. This is proven by the model's success in predicting 32 out of 33 data samples with a time duration of 117 seconds. The prediction results showed that the highest population numbers were in three provinces, namely Central Java, West Java, and East Java. Meanwhile, the lowest population is in West Papua province. Keywords: BPS, JST, Learning Rate, Optimization.

Artificial neural networks for inverse design of a semi-auxetic metamaterial

This study introduces an artificial neural network approach for the inverse design of a novel semi-auxetic mechanical metamaterial to achieve a specified stress-strain curve and/or Poisson's ratio-strain curve. To accomplish this, after presenting the metamaterial and assessing its characteristics, 1500 structures of the same metamaterial with various parameters are generated using a parametric model. The metamaterials are then gone through a compression test simulation using Finite Element (FE) analysis; accordingly, each metamaterial's stress-strain and Poisson's ratio curves are derived. The results of FE simulations are validated using mesh convergence check and experimental compression tests on a 3D printed specimen of the proposed metamaterial. In the next step, 80 % of the data are randomly selected to be used as training data for the artificial neural networks (ANN), while the remaining 20 % is employed to evaluate the performance of the ANNs using different metrics. The capability of the ANNs to predict the design parameters of the proposed metamaterial is assessed by providing different kinds of inputs, including the stress-strain curve, Poisson's ratio curve, and both. The observations reveal that the ANNs achieve more accurate results when both the stress-strain and Poisson's ratio-strain curves are provided as the inputs. The presented ANN in this study serves as a robust tool for precisely designing the parameters of the proposed metamaterial, allowing for the attainment of the desired stress-strain and/or Poisson's ratio-strain behavior. It is shown that the proposed metamaterial owns potential applications in crawling soft robotics, automotive, and construction industries.

Persistence of disaggregate energy RD&D expenditures in top-five economies: Evidence from artificial neural network approach

The motivation of this paper is to investigate the resistance of countries' energy research and development (RD&D) expenditures to random shocks. The analysis includes the five countries (France, Germany, Japan, the United States, and the United Kingdom) that are the biggest investors in RD&D in fossil fuels, renewables, energy efficiency, and nuclear energy. Thus, economies will be able to take precautions against global shocks in producing sustainable energy policies. To achieve this aim, this paper uses the artificial neural network approach, which uses a distributed computing model that can store and generalize data after a learning period. Empirical results widely depend on countries and distinct energy technology policy fields. The key findings provide evidence that RD&D expenditures, except for renewables, do not tend to mean revert. The economies of Japan, Germany, and the United States should make more renewable investments to reduce the resistance of renewable energy sources to shocks and thus produce policies for both environmental sustainability and energy security. The economies of France and the United Kingdom can ensure energy security by continuing their sustainable energy policies.

Artificial neural network-based direct power control to enhance the performance of a PMSG-wind energy conversion system under real wind speed and parameter uncertainties: An experimental validation

With the increasing emphasis on embedding advanced technology into system controls, the Direct Power Control (DPC) approach has garnered considerable attention due to its simple and highly adaptable algorithm. This approach has been increasingly recognized in numerous applications. However, the variable frequency, harmonic distortion of the currents, and power ripples caused by Hysteresis controllers and switching tables decrease its effectiveness and robustness, affecting the system’s performance. For this reason, this paper proposed a new DPC based on Artificial Neural Network (ANN) approaches. In this approach, the hysteresis comparator and the switching table are substituted with ANN controllers and then applied on both sides: machine-side converter (MSC) and grid-side converter (GSC) of a Permanent Magnet Synchronous Generator based Wind Energy Conversion System (PMSG-WECS). Moreover, to make the system more efficient in varying wind conditions, this study expands the utilization of the artificial neural networks (ANN) to encompass the maximum power point tracking (MPPT) control strategy. To demonstrate the effectiveness of the proposed approach on the system behaviors, a simulation test was carried out in the Matlab/Simulink environment, using a real wind profile of a Moroccan city (Essaouira). In comparison to the classical DPC control, the simulation results showed the superior performance of the proposed ANN-DPC control in terms of reference tracking, response time, overshoot, precision, and its capacity to reduce the rate of power ripples and total harmonic distortion (THD) in the injected currents. Furthermore, a robustness test was also included in this work to check the robustness of the proposed control against parameters variation. In conclusion, the feasibility and effectiveness of the ANN-DPC control approach were confirmed through experimental validation using the dSPACE DS1104 board.

Analysis of the friction performance of deep-drawing steel sheets using network models

AbstractThis article presents the results of pilot studies on the lubrication of the blankholder zone in sheet metal forming using a pressurised lubricant. The authors invented a method and built a special tribometer for pressure-assisted lubrication. This approach reduces friction in sheet metal forming processes compared to conventional lubrication. Moreover, the artificial neural network approach combined with a force-directed Fruchterman-Reingold graph algorithm and Spearman’s correlation was used for the first time to analyse the relationships between the friction process parameters and the output parameters (the coefficient of friction and the resulting surface roughness of the sheet metal). The experimental tests were conducted utilising strip drawing on four grades of steel sheets known to be outstanding for deep-drawing quality. Different oils, oil pressures and contact pressures were used. Artificial neural network models were used for the first time to determine these relationships in a strip drawing test where every parameter is represented by one node, and all nodes are connected by edges with each other. R Software version 4.2.3 was used to construct the network using the ‘qgraph’ and ‘networktools’ packages. It was found that friction conditions had a highly significant negative correlation with coefficient of friction (COF) and a moderately significant negative correlation with the final surface roughness. However, the initial surface roughness of the as-received sheets had a negative correlation with the COF and a positive one with the resulting surface roughness of the sheet metal. The parameters most related to the COF are the strength coefficient, the ultimate tensile strength and the friction conditions (dry friction or pressurised lubrication). Spearman’s correlation coefficients showed a strong correlation between the kinematic viscosity and the friction conditions.

A Final Cost Estimating Model for Building Renovation Projects

The construction sector in Greece has been developing radically in the field of building renovations. The foremost problem for projects in the building construction industry is producing an accurate and reliable cost estimate at the onset of construction. The artificial neural network (AΝΝ) approach, using data available at the early stages of the project, can help resolve or prevent any kind of difficulty that could make the successful completion of a building less likely. ANNs have been highly efficient in gaining results which could prevent the failure of building constructions projects. The ultimate goal is to highlight the usefulness of the adoption of ANNs models to predict the final cost of a building renovation project. Thus, construction companies could avoid financial failure, provided that the gap between cost prediction and final cost for renovation projects is minimized. This paper presents an artificial neural network (ANN) approach for predicting renovation costs in Greek construction projects. The study, based on a comprehensive literature review and real renovation data from construction companies, employs IBM SPSS Statistics software to build, train, and test the ANN model. The model, which uses initial cost, estimated time, and initial demolition cost as inputs, is based on the radial basis function procedure. The model presents high performance with up to 2% sum of squares error and near zero relative error, demonstrating the ANN’s effectiveness in estimating total renovation costs.

Combustion Characteristic Prediction of Dual Direct Injection Fuel (Diesel-Propane) on RCEM Based on an Artificial Neural Network Approach

Studies from around the world show that engines using biofuel, LPG, and CNG emit fewer pollutants than those using conventional fuels. Experimental research has focused on a rapid compression and expansion machine (RCEM) that resembles a compression ignition (CI) engine. It uses dual direct injection fuel, diesel and propane (DP), with propane injection timing varying from 0 to 40 before top dead center (BTDC) and diesel injection timing remaining at 10 BTDC. The compression ratio was changed at points 17 and 19 by adjusting the RCEM connecting rod. A converge simulation program was used to run the simulation model, which was used to examine how the fire and inflow inside the chamber developed. The ANN method was used to predict pressure, temperature, power, TKE, and ITE data output based on propane energy fraction, compression ratio, and SOI of propane as input data parameters. It was noticed that the ANN prediction on experimental data has a higher accuracy compared to the simulation prediction. The R and MSE values were used to identify the accuracy of the prediction on output parameter data. ANN generalization capability is comparatively high when trained with large enough databases. The highest accuracy of prediction was produced on TKE, which had an MSE of 0.003715 and R value of 0.99981 from 287900 sample data. This shows that the ANN model is quite accurate in forecasting output experimental data.

Artificial neural network and ANFIS approaches for mechanical properties prediction and optimization of a turbine blade treated by laser shock peening

This research article covers a comprehensive investigation on the prediction and optimization of residual stresses, plastic deformations and damage induced by laser shock peening (LSP) on a thin Ti-6Al-4 V leading edge sample of a turbine blade. The study is based on a two-part approach: the first part involves a numerical simulation of the LSP process and the characterization of its effects on the material. The finite element analysis (FEA) is used to simulate the complex interactions between laser parameters, material properties and mechanical response. This numerical analysis generates valuable insights into the residual stress distribution, the plastic deformation and the potential damage within the sample being treated. Based on the numerical results, the second part highlights the novel application of the artificial neural network (ANN) and the adaptive neuro-fuzzy inference system (ANFIS) methods. By taking advantage of the strength of ANN and ANFIS, the models are able to learn accurately from all the data generated by numerical simulations, allowing them to precisely predict and optimize the effects induced by LSP: residual stresses, plastic strains and damage. This paper provides a significant contribution to the field of materials science and engineering, by offering an in-depth comprehension of the LSP process impact on Ti-6Al-4 V leading-edge turbine blade specimens. The combined utilization of numerical analysis and artificial intelligence AI-driven techniques offers a comprehensive approach to optimize the LSP process, leading to the fatigue resistance improvement and service life extension of turbine blades and other critical aerospace components. The proven capability of ANN and ANFIS enables innovative solutions in aerospace engineering and surface treatment technologies, promoting the AI implementation in materials processing and design.

A machine-learning based approach to estimate acoustic macroscopic parameters of porous concrete

Porous concrete with expanded clay inherent porosity makes it an interesting and effective acoustic material, applied in numerous scenarios such as highways, airports and architectural structures, due to its capacity to mitigate noise pollution, by absorbing and damping sound waves. It is usually accepted that macroscopic properties such as open porosity, tortuosity or airflow resistivity of such materials play a fundamental role in the definition of the internal absorption process. This study explores the application of tailored artificial neural networks (ANNs) for predicting first the macroscopic properties (open porosity, tortuosity and airflow resistivity) and then the sound absorption coefficient (α) of these porous concrete mixtures, using only two input parameters (size class of the expanded clay and density of the test specimens). The results demonstrate the efficacy of the proposed ANN approach in accurately predicting macroscopic properties and the sound absorption coefficient of these mixtures, making it possible to obtain such important parameters in an effective and much simpler way.

Predictive Modeling and Optimization of Plywood Drying: An Artificial Neural Network Approach

Introduction: This investigation delves into the optimization of the plywood drying process through the development of predictive models for output moisture content (MC_Out) and waviness. It focuses on bridging the gap in current methodologies by employing artificial neural networks (ANNs), optimized with genetic algorithms, to enhance prediction accuracy and process efficiency. Materials and Methods: A comprehensive experimental design was employed, analyzing the effects of three wood types (Doncel, Tamburo, and Zapote), two thickness levels, and three drying speeds on MC_Out and waviness. Data collected were subjected to both traditional statistical analysis and ANNs. The ANNs were fine-tuned through genetic algorithms, exploring different network architectures to achieve optimal predictive performance. Results: Statistical models revealed the significant influence of wood type, thickness, and drying speed on MC_Out and waviness, explaining 95.9% and 84.3% of the variations, respectively. The optimized ANN models, however, demonstrated superior accuracy, with the MC_Out model achieving fitted R-squared values of 0.940 and 0.757 for training and validation sets, respectively, thus outperforming traditional models in predicting drying outcomes. Discussion: The study underscores the effectiveness of ANNs in capturing complex nonlinear relationships within the plywood drying data, which traditional statistical models might not fully elucidate. The successful optimization of ANN architecture via genetic algorithms further highlights the potential of machine learning approaches in industrial applications, offering a more precise and reliable method for predicting drying process outcomes. Conclusion: The integration of artificial neural networks, optimized through genetic algorithms, represents a significant advancement in the predictive modeling of plywood drying processes. This approach not only offers enhanced prediction accuracy for key variables such as MC_Out and waviness but also paves the way for more efficient and controlled drying operations, ultimately contributing to the production of higher-quality plywood.

Application of Artificial Neural Networks in Soil Science Research

Artificial Neural Networks utilize high-performance computation and large data technology, allowing research to generate new prospects in agriculture. ANN is currently a preferred technique for crop yield prediction, forecasting, and classification in biological science domains. Among different agriculture fields, soil science research plays a vital role in understanding and managing the complex processes occurring within the soil environment as oil is a complex system with dynamic surface layers that differ from the other parts of the matrix. Due to the increasing accessibility of innovative computing techniques, Artificial Neural Networks (ANNs) have developed into useful tools for modeling and forecasting soil-related activities. The numerous applications of ANNs in soil science research, with a focus on how well they can classify soils, assess soil fertility, forecast soil erosion, and estimate soil moisture. They are vital tools for identifying soil types, evaluating fertility levels, predicting erosion, and soil moisture estimation. ANN models were effective at predicting soil characteristics like pH, organic carbon concentration, and clay content. By training on vast datasets that contain the chemical, biological, and physical properties of soil, ANNs are able to accurately predict different soil types and enable land-use planning, precision farming, and environmental management. This mini-review focuses on ANN approaches that possess the potential to increase our understanding of soil science and encourage informed decisions for soil management and conservation.