Prediction Of Physical Properties Research Articles

Development of ANN models based on combined UV-vis-NIR spectra for rapid quantification of physical and chemical properties of industrial hemp extracts.

The aim of this study was to develop artificial neural network (ANNs) models for prediction of physical (total dissolved solids, extraction yield) and chemical (total polyphenolic content, antioxidant activity) properties of industrial hemp extracts, prepared by two different extraction methods (solid-liquid extraction and microwave-assisted extraction) based on combined UV-VIS-NIR spectra. Spectral data were gathered for 46 samples per extraction method. The PCA analysis ensured efficient separation of the samples based on the amount of ethanol in extraction solvent using NIR spectra for both conventional and microwave-assisted extraction. Results showed that reliable ANN models (R2 >0.7000) for describing physical, chemical, and simultaneously physical and chemical characteristics can be developed based on combined UV-VIS-NIR spectra of industrial hemp extracts without spectra pre-processing.

First-principles insights on the stability, linear and nonlinear optical properties of new potential thioborates AgBS2, CuBS2

Thioborates compounds are receiving increasing attention due to their numerous optical applications. In this work, we present a theoretical prediction of physical properties of CuBS2 and AgBS2 compounds by mean the density functional theory using the pseudo potential plane-wave method implemented in ABINIT and CASTEP codes. We investigated the structure stability and other ground state properties. The dynamical stability is also discussed from the phonon dispersion for the new hypothetical phase. We used the hybrid functional to estimate the band gap. Moreover, we present the optical anisotropy and nonlinear optical properties, both studied compounds have wide band gap, adequate optical birefringence and good second order NLO coefficients, which make them potential candidates for NLO applications.

Combining phonon accuracy with high transferability in Gaussian approximation potential models.

Machine learning driven interatomic potentials, including Gaussian approximation potential (GAP) models, are emerging tools for atomistic simulations. Here, we address the methodological question of how one can fit GAP models that accurately predict vibrational properties in specific regions of configuration space while retaining flexibility and transferability to others. We use an adaptive regularization of the GAP fit that scales with the absolute force magnitude on any given atom, thereby exploring the Bayesian interpretation of GAP regularization as an "expected error" and its impact on the prediction of physical properties for a material of interest. The approach enables excellent predictions of phonon modes (to within 0.1 THz-0.2 THz) for structurally diverse silicon allotropes, and it can be coupled with existing fitting databases for high transferability across different regions of configuration space, which we demonstrate for liquid and amorphous silicon. These findings and workflows are expected to be useful for GAP-driven materials modeling more generally.

Phase transitions, percolation, fracture of materials, and deep learning.

Percolation and fracture propagation in disordered solids represent two important problems in science and engineering that are characterized by phase transitions: loss of macroscopic connectivity at the percolation threshold p_{c} and formation of a macroscopic fracture network at the incipient fracture point (IFP). Percolation also represents the fracture problem in the limit of very strong disorder. An important unsolved problem is accurate prediction of physical properties of systems undergoing such transitions, given limited data far from the transition point. There is currently no theoretical method that can use limited data for a region far from a transition point p_{c} or the IFP and predict the physical properties all the way to that point, including their location. We present a deep neural network (DNN) for predicting such properties of two- and three-dimensional systems and in particular their percolation probability, the threshold p_{c}, the elastic moduli, and the universal Poisson ratio at p_{c}. All the predictions are in excellent agreement with the data. In particular, the DNN predicts correctly p_{c}, even though the training data were for the state of the systems far from p_{c}. This opens up the possibility of using the DNN for predicting physical properties of many types of disordered materials that undergo phase transformation, for which limited data are available for only far from the transition point.

Extending the ASS1ST Active Space Selection Scheme to Large Molecules and Excited States

Multireference electronic structure methods based on the CAS (complete active space) Ansatz are well established as means to provide reliable predictions of physical properties of strongly correlated systems. A critical aspect of every CAS calculation is the selection of an adequate active space, in particular as the boundaries for tractable active spaces have been shifted significantly with the emergence of efficient approximations to the Full-CI problem like the Density Matrix Renormalization Group and Full-CI Quantum Monte Carlo. Recently, we proposed an Active Space Selection based on 1st order perturbation theory (ASS1ST) that yields satisfactory results for the electronic ground state of a variety strongly correlated systems. In this work we present a state-averaged extension of ASS1ST (SA-ASS1ST) that determines suitable active spaces when electronically excited state are targeted. Furthermore, the computational cost of the single state and state-averaged variants are significantly reduced by a simple approximation that avoids the most expensive step of the original method, the evaluation of active space four-electron reduced density matrices, altogether. After the applicability of the approximation is established test calculations on a biomimetic \ce{Mn4O4} cluster demonstrate the enhanced range of ASS1ST in terms of system size and complexity. Furthermore, calculations on \ce{[VOCl4]2-}, \ce{MeMn(CO)3}-$\alpha$-diimine and anthracene show that SA-ASS1ST suggests well-suited active spaces to describe d$\rightarrow$d and charge-transfer excitations in transition metal complexes as well as $\pi\rightarrow\pi^*$ excitations in aryl compounds. Finally the application of ASS1ST on multiple points of the potential energy surface of Cr 2 illustrates the applicability of the method even when extremely complicated bonding patterns are met. More importantly, however, it highlights the necessity to use special strategies when different points of a potential energy surface are investigated, e.g. during chemical reactions.

Prediction of physical properties of thermosetting resin by using machine learning and structural formulas of raw materials

AbstractThermosetting resins are one of the most widely used functional materials in industrial applications. Although some of the physical properties of thermosetting resins are controlled by changing the functional groups of the raw materials or adjusting their mixing ratios, it was conventionally challenging to construct machine learning (ML) models, which include both mixing ratio and chemical information such as functional groups. To overcome this problem, we propose a machine learning approach based on extended circular fingerprint (ECFP) in this study. First, we predicted the classification of raw materials by the random forest, where ECFP was used as the explanatory variable. Then, we aggregated ECFP for each classification predicted by the random forest. After that, we constructed the prediction model by using the aggregated ECFP, feature quantities of reaction intermediates, and curing conditions of resin as explanatory variables. As a result, the model was able to predict in high accuracy (R^2 = 0.8), for example, the elastic modulus of thermosetting resins. Furthermore, we also show the result of verification of prediction accuracy in first step, such as using the one-hot-encording. Therefore, we confirmed that the properties of thermosetting resins could be predicted using mixed raw materials by the proposed method.

Three-dimensional multiscale fusion for porous media on microtomography images of different resolutions.

Accurately acquiring the three-dimensional (3D) image of a porous medium is an imperative issue for the prediction of multiple physical properties. Considering the inherent nature of the multiscale pores contained in porous media such as tight sandstones, to completely characterize the pore structure, one needs to scan the microstructure at different resolutions. Specifically, low-resolution (LR) images cover a larger field of view (FOV) of the sample, but are lacking small-scale features, whereas high-resolution (HR) images contain ample information, but sometimes only cover a limited FOV. To address this issue, we propose a method for fusing the spatial information from a two-dimensional (2D) HR image into a 3D LR image, and finally reconstructing an integrated 3D structure with added fine-scale features. In the fusion process, the large-scale structure depicted by the 3D LR image is fixed as background and the 2D image is utilized as training image to reconstruct a small-scale structure based on the background. To assess the performance of our method, we test it on a sandstone scanned with low and high resolutions. Statistical properties between the reconstructed image and the target are quantitatively compared. The comparison indicates that the proposed method enables an accurate fusion of the LR and HR images because the small-scale information is precisely reproduced within the large one.

Predictive Modeling of Aerospace Fuel Properties Using Comprehensive Two-Dimensional Gas Chromatography with Time-Of-Flight Mass Spectrometry and Partial Least Squares Analysis

Increasingly stringent requirements for aerospace propulsion system performance, reliability, and operability motivate quantitative connections between fuel composition, physical characteristics, and system performance. Chemically accurate assessment of aviation turbine fuels (Jet-A, JP-8, etc.) and kerosene-based rocket propellants (RP-1 and RP-2) is requisite to mature these models. Comprehensive two-dimensional gas chromatography with time-of-flight mass spectrometry (GC × GC–TOFMS) is an excellent analytical tool for measuring detailed chemical information contained in complex fuels. Additionally, multivariate data analysis methods, referred to as chemometrics, are ideally suited to relate detailed chemical information contained within the GC × GC–TOFMS data to fuel properties and performance in a predictive manner. Herein, we apply these techniques to a chemically diverse set of 74 distillate and multicomponent aerospace fuels, resulting in an improved understanding of the chemical compositional basis for physical and thermochemical behavior. Informed by GC × GC–TOFMS data, highly reliable partial least squares (PLS) models are developed and employed in the prediction of physical properties (measured separately using conventional test methods). Root-mean-square errors of cross-validation (RMSECV) were relatively low: values of 0.0450 cSt, 41.3 Btu/lbm, 0.130 mass %, and 0.0064 g/mL were obtained for viscosity, heat of combustion, hydrogen content, and density, respectively. The corresponding normalized root-mean-square errors of cross-validation (NRMSECV) were 6.01, 10.3, 8.71, and 7.12%, respectively. Investigation of the linear regression vectors (LRVs) provides valuable insight into the relationship between the chemical composition and physical properties, enabling, in principle, the model-informed selection of fuel chemical composition to achieve desired performance criteria.

Identification and Resolution of Unphysical Multielectron Excitations in the Real-Time Time-Dependent Kohn-Sham Formulation.

We resolve a fundamental issue associated with the conventional Kohn-Sham formulation of real-time time-dependent density functional theory. We show that unphysical multielectron excitations, generated during time propagation of the Kohn-Sham equations due to fixation of the total number of Kohn-Sham orbitals and their occupations, result in incorrect electron density and, therefore, wrong predictions of physical properties. A new formulation is proposed in that the number of Kohn-Sham orbitals and their occupations are updated on the fly, the unphysical multielectron excitations are removed, and the correct electron density is determined. The correctness of the new formulation is demonstrated by simulations of Rabi oscillation, as analytical results are available for comparison in the case of noninteracting electrons.

The expert system for improving the technological processes of composite manufacturing

An expert system for improving the technological processes of composite manufacturing is presented. The domain of the expert system is the technology of pressure treatment of structurally inhomogeneous materials, including powder materials and composites. The purpose of the expert system is the interpretation of experimental data and the prediction of physical and mechanical properties of blanks and products. The function of the expert system is to issue recommendations for improving the technological processes of manufacturing blanks of non-compact metal raw materials, produced on the basis of predicting the properties and behavior of the object of the study. The problems solved by the expert system are planning natural and computational experiments, statistical processing of experimental data, regression analysis, checking the adequacy of mathematical models, multiparameter optimization of the systems and processes under study.

Prediction of Physical and Mechanical Properties for Metallic Materials Selection Using Big Data and Artificial Neural Networks

In this work, a computer-aided tool is developed to predict relevant physical and mechanical properties that are involved in the selection tasks of metallic materials. The system is based on the use of artificial neural networks supported by big data collection of information about the technological characteristics of thousands of materials. Thus, the volume of data exceeds 43k. The system can access an open online material library (a website where material data are recorded), download the required information, read it, filter it, organise it and move on to the step based on artificial intelligence. An artificial neural network (ANN) is built with thousands of perceptrons, whose topology and connections have been optimised to accelerate the training and predictive capacity of the ANN. After the corresponding training, the system is able to make predictions about the material density and Young’s modulus with average confidences greater than 99% and 98%, respectively.

Representations and descriptors unifying the study of molecular and bulk systems

AbstractEstablishing a unified framework for describing the structures of molecular and periodic systems is a long‐standing challenge in physics, chemistry, and material science. With the rise of machine learning methods in these fields, there is a growing need for such a method. This perspective aims to discuss the development and use of three promising approaches—topological, atom‐density, and symmetry‐based—for the prediction and rationalization of physical, chemical, and mechanical properties of atomistic systems across different scales and compositions.

Process optimization of applying heat and wear proof coating to machine elements in complex acoustic field

The process of obtaining expected physical and mechanical properties of protective coating, which is applied on machine elements in complex acoustic field by electroacoustic method, is conducted with the use of highly concentrated sources of electric energy. Prediction of physical and mechanical properties of the reinforced surface of machine parts is made on the basis of the experimental design technique with nonorthogonal planning matrices. Mathematical models were tested for adequacy on the basis of Fisher’s variance ratio. A structural scheme for microprocessor control system of electroacoustic spray coating was developed. The physical basis of the mass transfer of materials to the substrate in a complex acoustic field is developed, which is done by transforming longitudinal vibrations into longitudinal-torsional ones, while the amplitude of the acoustic system oscillates from 5 to 15 micrometers. The use of longitudinal-torsional ultrasonic waveguides in this process and ultrafast cooling temperatures allow the formation of such elements as carbides, carbonitrides, intermetallics on the surface of the conductive material. The subsequent shock produced by the waveguide-electrode with a shift to the surface leads to irreversible microflow of the surface layer, which provide the specified parameters of the surface quality and its physical and mechanical properties (microhardness and roughness). The thickness of the sprayed layer is 6-8 micrometers, and the thickness of the plastically deformed layer is 30-45 micrometers.

Prediction of Physical Properties of Steels Using Artificial Neural Networks for Numerical Simulation of Electrical Installations

For modern electric installations to be competitive, they should be supported by numerical simulations of high quality and accuracy. In turn, to improve the accuracy of numerical simulation of electric installations, it is required to take into account the thermal-physical, electromagnetic, and mechanical properties of materials as much as possible. This article discusses the possibility of predicting the properties of carbon steels on the basis of their chemical composition for a wide temperature range using a set of artificial neural networks as exemplified by electrical resistivity. Training and verification of neural networks have been based on experimental data from reference books and databases. The methods have been applied so as to avoid retraining of neural networks, as well as to use initial data more completely. The training results of neural networks are given, the error level is evaluated, and the potential is determined of applying neural networks to production of material properties and, as a consequence, improving simulation quality.

Deep Learning for Deep Chemistry: Optimizing the Prediction of Chemical Patterns.

Computational Chemistry is currently a synergistic assembly between ab initio calculations, simulation, machine learning (ML) and optimization strategies for describing, solving and predicting chemical data and related phenomena. These include accelerated literature searches, analysis and prediction of physical and quantum chemical properties, transition states, chemical structures, chemical reactions, and also new catalysts and drug candidates. The generalization of scalability to larger chemical problems, rather than specialization, is now the main principle for transforming chemical tasks in multiple fronts, for which systematic and cost-effective solutions have benefited from ML approaches, including those based on deep learning (e.g. quantum chemistry, molecular screening, synthetic route design, catalysis, drug discovery). The latter class of ML algorithms is capable of combining raw input into layers of intermediate features, enabling bench-to-bytes designs with the potential to transform several chemical domains. In this review, the most exciting developments concerning the use of ML in a range of different chemical scenarios are described. A range of different chemical problems and respective rationalization, that have hitherto been inaccessible due to the lack of suitable analysis tools, is thus detailed, evidencing the breadth of potential applications of these emerging multidimensional approaches. Focus is given to the models, algorithms and methods proposed to facilitate research on compound design and synthesis, materials design, prediction of binding, molecular activity, and soft matter behavior. The information produced by pairing Chemistry and ML, through data-driven analyses, neural network predictions and monitoring of chemical systems, allows (i) prompting the ability to understand the complexity of chemical data, (ii) streamlining and designing experiments, (ii) discovering new molecular targets and materials, and also (iv) planning or rethinking forthcoming chemical challenges. In fact, optimization engulfs all these tasks directly.

Prediction Model of Organic Molecular Absorption Energies based on Deep Learning trained by Chaos-enhanced Accelerated Evolutionary algorithm

As an important physical property of molecules, absorption energy can characterize the electronic property and structural information of molecules. Moreover, the accurate calculation of molecular absorption energies is highly valuable. Present linear and nonlinear methods hold low calculation accuracies due to great errors, especially irregular complicated molecular systems for structures. Thus, developing a prediction model for molecular absorption energies with enhanced accuracy, efficiency, and stability is highly beneficial. By combining deep learning and intelligence algorithms, we propose a prediction model based on the chaos-enhanced accelerated particle swarm optimization algorithm and deep artificial neural network (CAPSO BP DNN) that possesses a seven-layer 8-4-4-4-4-4-1 structure. Eight parameters related to molecular absorption energies are selected as inputs, such as a theoretical calculating value Ec of absorption energy (B3LYP/STO-3G), molecular electron number Ne, oscillator strength Os, number of double bonds Ndb, total number of atoms Na, number of hydrogen atoms Nh, number of carbon atoms Nc, and number of nitrogen atoms NN; and one parameter representing the molecular absorption energy is regarded as the output. A prediction experiment on organic molecular absorption energies indicates that CAPSO BP DNN exhibits a favourable predictive effect, accuracy, and correlation. The tested absolute average relative error, predicted root-mean-square error, and square correlation coefficient are 0.033, 0.0153, and 0.9957, respectively. Relative to other prediction models, the CAPSO BP DNN model exhibits a good comprehensive prediction performance and can provide references for other materials, chemistry and physics fields, such as nonlinear prediction of chemical and physical properties, QSAR/QAPR and chemical information modelling, etc.

Application of multivariate regression and artificial neural network modelling for prediction of physical and chemical properties of medicinal plants aqueous extracts

In recent years, multivariate modelling techniques have been employed with the aim of analysing, describing, and generally interpreting multidimensional data obtained from experiments. The objective of this study was to evaluate the applicability of multiple linear regression, nonlinear regression, piecewise linear regression, and artificial neural network modelling for the prediction of the physical properties (total dissolved solids, extraction yield), and chemical properties (total phenolic content and antioxidant activity) of the aqueous extracts of nine medicinal plants (dandelion, camomile, lavender, lemon balm, marigold, mint, nettle, plantain, and yarrow), prepared in dynamic experiments based on the extraction conditions (time and temperature), and plant species. Results indicated that simple multivariate regression models could be used for prediction of physical and chemical properties of medicinal plants aqueous extracts (the highest R2 were obtained for total phenolic content), while the artificial neural network proved a very effective tool (R2 > 0.9) for simultaneous prediction of both physical and chemical properties of medicinal plants aqueous extracts.

Prediction of physical properties for the design of processes in the oil & gas industry using molecular simulation

Prediction of physical properties for the design of processes in the oil & gas industry using molecular simulation

Physicochemical Properties of analogs α-Aminophosphonates Drugs Determined via Molecular Dynamics Simulation

Objective: Cyclin-dependent kinases (CDKs, play important roles in cell cycle regulation. Since deregulation of cyclins and/or alteration or absence of inhibitors has been associated with many cancers, there is strong interest in CDKs inhibitors that could play an necessary role in the discovery of a new family of antitumor agents. Material and Methods: Molecular modeling is used to design new materials, which the correct prediction of physical properties of realistic systems is required. Gromacs is an tool to perform molecular dynamics simulations and energy minimization of bimolecular systems which commonly consists of several tens to thousands of amino acid residues. From the simulations the first 100 ps were regarded as equilibration, leaving 900 ps for analysis purposes. Results: Thermodynamic properties density, potential energy, temperature, and pressure are given. In this results tetramethyl ((1,4-phenylenebis (azanediyl)) bis ((4-chlorophenyl) methylene)) bis (phosphonate) was found to be the better selective known inhibitor for cyclin-dependent kinase2 because it shown lowest energy.Conclusion: The phosphoric acid moiety is considered to bind to the affected protein more actively than the corresponding carboxylic acid because of its di anionic character.

Comparative investigation of Shore, Schmidt, and Leeb hardness tests in the characterization of rock materials

Prediction of physical and mechanical properties of rock materials using rebound-based hardness test methods is widely preferred in many fields of engineering and in the characterization of rock materials, because they are non-destructive, practical, and economical. In this study, 40 types of rocks with magmatic, metamorphic, and sedimentary origins, represented by travertine, limestone, marble, dolomite, granite, syenite, dunite, andesite, schist, gabbro, tuff, and ignimbrite were selected. First, dry unit weight (γd), open porosity (no), water absorption by weight (WAW), wide wheel abrasion (WA), and uniaxial compressive strength values were determined. After that, Shore C-2 scleroscope (HSC), L-type Schmidt hammer (HSL), and Leeb (HLD) rebound-based hardness tests were carried out on all samples, and then, hardness values by three methods were compared with the obtained parameters. The Leeb hardness test, which is more recent and innovative than the Shore and Schmidt hardness tests, was initially developed for metallic materials. However, the method has become increasingly popular in the determination of hardness of rock materials in laboratory as well as in field. In this study, the Leeb hardness test was found to be more useful due to its quick and precise measurement capabilities compared to Shore and Schmidt hardness tests. The results of the study reveal that the prediction of physical and mechanical properties of rocks can more precisely be determined by the HLD method than the HSL and HSC methods using the proposed equations.