Search

Structures made of heterogeneous materials, such as composites, often require a multiscale approach when their behavior is simulated using the finite element method. By solving the boundary value problem of the macroscale model, for previously homogenized material properties, the resulting stress maps can be obtained. However, such stress results do not describe the actual behavior of the material and are often significantly different from the actual stresses in the heterogeneous microstructure. Finding high-accuracy stress results for such materials leads to time-consuming analyses in both scales. This paper focuses on the application of machine learning to multiscale analysis of structures made of composite materials, to substantially decrease the time of computations of such localization problems. The presented methodology was validated by a numerical example where a structure made of resin epoxy with randomly distributed short glass fibers was analyzed using a computational multiscale approach. Carefully prepared training data allowed artificial neural networks to learn relationships between two scales and significantly increased the efficiency of the multiscale approach. [ABSTRACT FROM AUTHOR]

No study has been reported to use machine learning methods to predict the full-range test curves of circular CFST columns. In this paper, the long short-term memory (LSTM) network was introduced to calculate the axially compressive load–strain curves of the circular CFST columns according to an experiment database of limited scale. To improve the feasibility of input data for the recurrent neural network algorithm, data preprocessing methods and data configurations were discussed. The prediction results indicate that the LSTM network provides more accurate estimations compared with the artificial neural networks, random forest and support vector regression. Meanwhile, this method can be used to calculate the mechanical properties including the elastic modulus, ultimate bearing capacity, and the ductility of the columns with acceptable accuracy for engineering practice (the prediction error within 20%). For future research, it is expected that the machine learning method will be applied to predict the structural response of different members under various loading conditions. [ABSTRACT FROM AUTHOR]

Developing accurate and interpretable models to forecast concrete's self-healing behavior is of interest to material engineers, scientists, and civil engineering contractors. Machine learning (ML) and artificial intelligence are powerful tools that allow constructing high-precision predictions, yet often considered "black box" methods due to their complexity. Those approaches are commonly used for the modeling of mechanical properties of concrete with exceptional accuracy; however, there are few studies dealing with the application of ML for the self-healing of cementitious materials. This paper proposes a pioneering study on the utilization of ML for predicting post-fire self-healing of concrete. A large database is constructed based on the literature studies. Twelve input variables are analyzed: w/c, age of concrete, amount of cement, fine aggregate, coarse aggregate, peak loading temperature, duration of peak loading temperature, cooling regime, duration of cooling, curing regime, duration of curing, and specimen volume. The output of the model is the compressive strength recovery, being one of the self-healing efficiency indicators. Four ML methods are optimized and compared based on their performance error: Support Vector Machines (SVM), Regression Trees (RT), Artificial Neural Networks (ANN), and Ensemble of Regression Trees (ET). Monte Carlo analysis is conducted to verify the stability of the selected model. All ML approaches demonstrate satisfying precision, twice as good as linear regression. The ET model is found to be the most optimal with the highest prediction accuracy and sufficient robustness. Model interpretation is performed using Partial Dependence Plots and Individual Conditional Expectation Plots. Temperature, curing regime, and amounts of aggregates are identified as the most significant predictors. [ABSTRACT FROM AUTHOR]

Stencil printing is the most crucial process in reflow soldering for the mass assembly of electronic circuits. This paper investigates different machine learning-based methods to predict the essential process characteristics of stencil printing: the area, thickness, and volume of deposited solder paste. The training dataset was obtained experimentally by varying the printing speed (from 20 to 120 mm/s), the size (area ratio from 0.35 to 1.7) of stencil apertures, and the particle size (characterized by a log-normal distribution) in the solder paste. Various machine learning-based methods were assessed; ANFIS–adaptive neuro-fuzzy inference systems; ANN artificial neural networks (with different learning methods); boosted trees, regression trees, SVM–support vector machines. Each method was optimized and fine-tuned with hyperparameter optimization, and the overfitting phenomenon was also prevented with cross-validation. The regression tree was the best performing approach for modelling the stencil printing, while ANN with the Bayesian regularization learning method was only slightly worse. The presented methodology for fine-tuning, parameter optimization, and the comparison of different machine learning-based methods can easily be adapted to any application field in electronics manufacturing. [ABSTRACT FROM AUTHOR]

Manufacturing processes in industry applications are often controlled by the evaluation of surface topography. Topography, in its overall performance, includes form, waviness, and roughness. Methods of measurement of surface roughness can be roughly divided into tactile and contactless techniques. The latter ones are much faster but sensitive to external disturbances from the environment. One type of external source error, while the measurement of surface topography occurs, is a high-frequency noise. This noise originates from the vibration of the measuring system. In this study, the methods for reducing high-frequency errors from the results of contactless roughness measurements of turned surfaces were supported by machine learning methods. This research delves into optimizing filtration methods for surface topography measurements through the application of machine learning models, focusing on enhancing the accuracy of surface roughness assessments. By examining turned surfaces under specific machining conditions and employing a variety of digital filters, the study identifies the Gaussian regression filter and spline filter as the most effective methods at a 22.5 µm cut-off. Utilizing neural networks, support vector machines, and decision trees, the research demonstrates the superior performance of SVMs, achieving remarkable accuracy and sensitivity in predicting optimal filtration methods. [ABSTRACT FROM AUTHOR]

The reaction–diffusion equation approach, which solves differential equations of the development of density distributions of mobile and immobile dislocations under mutual interactions, is a method widely used to model the dislocation structure formation. A challenge in the approach is the difficulty in the determination of appropriate parameters in the governing equations because deductive (bottom-up) determination for such a phenomenological model is problematic. To circumvent this problem, we propose an inductive approach utilizing the machine-learning method to search a parameter set that produces simulation results consistent with experiments. Using a thin film model, we performed numerical simulations based on the reaction–diffusion equations for various sets of input parameters to obtain dislocation patterns. The resulting patterns are represented by the following two parameters; the number of dislocation walls ( p 2 ), and the average width of the walls ( p 3 ). Then, we constructed an artificial neural network (ANN) model to map between the input parameters and the output dislocation patterns. The constructed ANN model was found to be able to predict dislocation patterns; i.e., average errors in p 2 and p 3 for test data having 10% deviation from the training data were within 7% of the average magnitude of p 2 and p 3 . The proposed scheme enables us to find appropriate constitutive laws that lead to reasonable simulation results, once realistic observations of the phenomenon in question are provided. This approach provides a new scheme to bridge models for different length scales in the hierarchical multiscale simulation framework. [ABSTRACT FROM AUTHOR]

The mechanical properties of concrete are the important parameters in a design code. The amount of laboratory trial batches and experiments required to produce useful design data can be decreased by using robust prediction models for the mechanical properties of concrete, which can save time and money. Portland cement is frequently substituted with metakaolin (MK) because of its technical and environmental advantages. In this study, three mechanical properties of concrete with MK, i.e., compressive strength ( f ′ c ), splitting tensile strength ( f st ), and flexural strength (FS) were modelled by using four machine learning (ML) techniques: gene expression programming (GEP), artificial neural network (ANN), M5P model tree algorithm, and random forest (RF). For this purpose, a comprehensive database containing detail of concrete mixture proportions and values of f ′ c , f st , and FS at different ages was gathered from peer-reviewed published documents. Various statistical metrics were used to compare the predictive and generalization capability of the ML techniques. The comparative study of ML techniques revealed that RF has better predictive and generalization capability as compared with GEP, ANN, and M5P model tree algorithm. Moreover, the sensitivity and parametric analysis (PA) was carried out. The PA showed that the most suitable proportions of MK as partial cement replacement were 10% for FS and 15% for both f ′ c and f st . [ABSTRACT FROM AUTHOR]

Sustainable concrete is gaining in popularity as a result of research into waste materials, such as recycled aggregate (RA). This strategy not only protects the environment, but also meets the demand for concrete materials. Using advanced artificial intelligence (AI) approaches, this study anticipates the split tensile strength (STS) of concrete samples incorporating RA. Three machine-learning techniques, artificial neural network (ANN), decision tree (DT), and random forest (RF), were examined for the specified database. The results suggest that the RF model shows high precision compared with the DT and ANN models at predicting the STS of RA-based concrete. The high value of the coefficient of determination and the low error values of the mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE) provided significant evidence for the accuracy and precision of the RF model. Furthermore, statistical tests and the k-fold cross-validation technique were used to validate the models. The importance of the input parameters and their contribution levels was also investigated using sensitivity analysis and SHAP analysis. [ABSTRACT FROM AUTHOR]

As machining processes are complex in nature due to the involvement of large plastic strains occurring at higher strain rates, and simultaneous thermal softening of material, it is necessary for manufacturers to have some manner of determining whether the inputs will achieve the desired outputs within the limitations of available resources. However, finite element simulations—the most common means to analyze and understand the machining of high-performance materials under various cutting conditions and environments—require high amounts of processing power and time in order to output reliable and accurate results which can lead to delays in the initiation of manufacture. The objective of this study is to reduce the time required prior to fabrication to determine how available inputs will affect the desired outputs and machining parameters. This study proposes a hybrid predictive methodology where finite element simulation data and machine learning are combined by feeding the time series output data generated by Finite Element Modeling to an Artificial Neural Network in order to acquire reliable predictions of optimal and/or expected machining inputs (depending on the application of the proposed approach) using what we describe as a backwards training model. The trained network was then fed a test dataset from the simulations, and the results acquired show a high degree of accuracy with regards to cutting force and depth of cut, whereas the predicted/expected feed rate was wildly inaccurate. This is believed to be due to either a limited dataset or the much stronger effect that cutting speed and depth of cut have on power, cutting forces, etc., as opposed to the feed rate. It shows great promise for further research to be performed for implementation in manufacturing facilities for the generation of optimal inputs or the real-time monitoring of input conditions to ensure machining conditions do not vary beyond the norm during the machining process. [ABSTRACT FROM AUTHOR]

Additive manufacturing (AM), widely known as 3D-printing, builds parts by adding material in a layer-by-layer process. This tool-less procedure enables the manufacturing of porous sound absorbers with defined geometric features, however, the connection of the acoustic behavior and the material's micro-scale structure is only known for special cases. To bridge this gap, the work presented here employs machine-learning techniques that compute acoustic material parameters (Biot parameters) from the material's micro-scale geometry. For this purpose, a set of test specimens is used that have been developed in earlier studies. The test specimens resemble generic absorbers by a regular lattice structure based on a bar design and allow a variety of parameter variations, such as bar width, or bar height. A set of 50 test specimens is manufactured by material extrusion (MEX) with a nozzle diameter of 0.2 mm and a targeted under extrusion to represent finer structures. For the training of the machine learning models, the Biot parameters are inversely identified from the manufactured specimen. Therefore, laboratory measurements of the flow resistivity and absorption coefficient are used. The resulting data is used for training two different machine learning models, an artificial neural network and a k-nearest neighbor approach. It can be shown that both models are able to predict the Biot parameters from the specimen's micro-scale with reasonable accuracy. Moreover, the detour via the Biot parameters allows the application of the process for application cases that lie beyond the scope of the initial database, for example, the material behavior for other sound fields or frequency ranges can be predicted. This makes the process particularly useful for material design and takes a step forward in the direction of tailoring materials specific to their application. [ABSTRACT FROM AUTHOR]

In this study, a method for the prediction of cyclic stress–strain properties of ferrite-pearlite steels was proposed. At first, synthetic microstructures were generated based on an anisotropic tessellation from the results of electron backscatter diffraction (EBSD) analyses. Low-cycle fatigue experiments under strain-controlled conditions were conducted in order to calibrate material property parameters for both an anisotropic crystal plasticity and an isotropic J2 model. Numerical finite element simulations were conducted using these synthetic microstructures and material properties based on experimental results, and cyclic stress-strain properties were calculated. Then, two-point correlations of synthetic microstructures were calculated to quantify the microstructures. The microstructure-property dataset was obtained by associating a two-point correlation and calculated cyclic stress-strain property. Machine learning, such as a linear regression model and neural network, was conducted using the dataset. Finally, cyclic stress-strain properties were predicted from the result of EBSD analysis using the obtained machine learning model and were compared with the results of the low-cycle fatigue experiments. [ABSTRACT FROM AUTHOR]