Model For Prediction Of Properties Research Articles

A model predicting mechanical properties of asphalt mixtures considering void ratio and loading conditions

The void ratio is an essential parameter in asphalt mixture design, which significantly affects asphalt mixtures' mechanical properties and road performance. The prediction model for mechanical properties will be established based on the Computed tomography (CT) scanning, strength, and fatigue test results in this research to reveal the relationship between the asphalt mixtures' mechanical properties and void ratio. Firstly, the voids' transverse and longitudinal distribution characteristics in asphalt mixtures of different gradations were investigated based on CT scanning. Then, the influences of void ratio, loading rates on the strength, and the impacts of void ratio and stress levels on the fatigue life were investigated by conducting indirect tensile tests. Finally, the mechanical properties prediction model was established for void ratio and loading conditions. Further, the feasibility of the model is verified in conjunction with existing research results. The results show that the void ratio calculated based on CT scanning is about 1.15 times the measured void ratio in the water weight method. The number of voids was uniformly distributed in the range of 15 mm∼50 mm in height and 10 mm∼40 mm in radius of the specimen. The asphalt mixtures' strength decreases with increasing void ratio and increases with increasing loading rate. The asphalt mixtures' fatigue life declines with increasing void ratio and stress level. The correlation coefficients of the established strength and fatigue prediction models were 0.963 and 0.976, respectively. The relative error between the measured strength, fatigue life, and predicted results is about 10.75% and 11.27%, respectively. This research provides a basis for designing high-performance asphalt mixtures by establishing the relationship between asphalt mixtures' microscopic and mechanical characteristics.

MDs-NP: a property prediction model construction procedure for naphtha based on molecular dynamics simulation

In the context of carbon neutrality and carbon peaking, molecular management has become a focus of the petrochemical industry. The key to achieving molecular management is molecular reconstruction, which relies on rapid and accurate calculation of oil properties. Focusing on naphtha, we proposed a novel property prediction model construction procedure (MDs-NP) employing molecular dynamics simulations for property collections and gamma distribution from real analytical data for calculating mole fractions of simulation mixtures. We calculated 348 sets of mixture properties data in the range of 273 K–300 K by molecular dynamics simulations. Molecular feature extraction was based on molecular descriptors. In addition to descriptors based on open-source toolkits (RDKit and Mordred), we designed 12 naphtha knowledge (NK) descriptors with a focus on naphtha. Three machine learning algorithms (support vector regression, extreme gradient boosting and artificial neural network) were applied and compared to establish models for the prediction of the density and viscosity of naphtha. Mordred and NK descriptors + support vector regression algorithm achieved the best performance for density. The selected RDKFp and NK descriptors + artificial neural network algorithm achieved the best performance for viscosity. Using ablation studies, T, P_w and CC(C)C are three effective descriptors in NK that can improve the performance of the property prediction models. MDs-NP has the potential to be extended to more properties as well as more-complex petroleum systems. The models from MDs-NP can be used for rapid molecular reconstruction to facilitate construction of data-driven models and intelligent transformation of petrochemical processes.

A novel spectral model for prediction of nongray radiative properties in gaseous combustion

Radiative heat transfer during gaseous combustion displays very strong spectral or “nongray” behavior, which is both very difficult to characterize and to evaluate. A novel spectral model based on the full-spectrum correlated k-distribution (FSCK) method combined with the multilayer perceptron (MLP), i.e., NFM model, is developed for the prediction of nongray radiative properties in gaseous combustion. Followed by the nature of FSCK method, the NFM model is designed as a simple MLP structure that has only two outputs to directly give both correlated k-value and corresponding ka-value. This not only avoids on-the-fly calculations of a-values but also saves storage by the simple structure. Moreover, by constructing training data with a number of small mole fractions, the errors caused by nonlinear effects are found to be mitigated during the use of NFM model. Several test cases have been carried out to compare the developed NFM model and other spectral tools including FSCK look-up tables and traditional FSCK MLP (TFM) model. Results show that the NFM model can achieve excellent accuracy that is even better than FSCK look-up tables at a tiny computational cost that is far less than that of TFM model.

PointGAT: A Quantum Chemical Property Prediction Model Integrating Graph Attention and 3D Geometry.

Predicting quantum chemical properties is a fundamental challenge for computational chemistry. While the development of graph neural networks has advanced molecular representation learning and property prediction, their performance could be further enhanced by incorporating three-dimensional (3D) structural geometry into two-dimensional (2D) molecular graph representation. In this study, we introduce the PointGAT model for quantum molecular property prediction, which integrates 3D molecular coordinates with graph-attention modeling. Comparison with other current models in molecular prediction tasks showed that PointGAT could provide higher predictive accuracy in various benchmark data sets from MoleculeNet, including ESOL, FreeSolv, Lipop, HIV, and 6 out of 12 tasks of the QM9 data set. To further examine PointGAT prediction of quantum mechanical (QM) energies, we constructed a C10 data set comprising 11,841 charged and chiral carbocation intermediates with QM energies calculated at the DM21/6-31G*//B3LYP/6-31G* levels. Notably, PointGAT achieved an R2 value of 0.950 and an MAE of 1.616 kcal/mol, outperforming even the best-performing graph neural network model with a reduction of 0.216 kcal/mol in MAE and an improvement of 0.050 in R2. Additional ablation studies indicated that incorporating molecular geometry into the model resulted in markedly higher predictive accuracy, reducing the MAE value from 1.802 to 1.616 kcal/mol. Moreover, visualization of PointGAT atomic attention weights suggested its predictions were interpretable. Findings in this study support the application of PointGAT as a powerful and versatile tool for quantum chemical property prediction that can facilitate high-accuracy modeling for fundamental exploration of chemical space as well as drug design and molecular engineering.

Durability of concrete-encapsulated GFRP bars subjected to seawater immersion

Glass fiber reinforce polymer (GFRP) bars have better durability than traditional steel bars, but there are few studies on their long-term performance in actual service environments especially the marine environment. To investigate the durability of GFRP bars used in cross-sea bridges, this research delves into the performance of 132 such bars, spanning diameters of 10, 12, 16, and 25 mm, in an environment that closely mimics real-world condition. The GFRP bars, encapsulated in concrete, were submerged in seawater solution, maintained at either room temperature or 60°C, for periods spanning from 0 to 183 days. Then the tensile strength and elastic modulus degradation of GFRP bars were investigated by tensile tests. Scanning electron microscope (SEM) was also used to analyze the micro-degradation mechanism of GFRP bars. In addition, this paper compares the four existing models for predicting long-term mechanical properties of GFRP bars with the experimental data, so as to determine the formulas and methods to be used in the design. The result shows that the elastic modulus of GFRP bars after 183 days of immersion in concrete pore solution was almost unchanged while the tensile strength decreased significantly. Among them, the tensile strength of 10 mm diameter GFRP bars decreased by 41.52 %, after 183 days of immersion at high temperature (60°C). The SEM method revealed that the main cause of such phenomena was the weakening of the bond between the fibers and matrix, which led to a decrease in the interface property. Finally, by fitting and comparing the relationship between the test data and the existing prediction models for the long-term mechanical properties of GFRP bars, the residual tensile strength of GFRP bars used in the actual project of Yanzhou Bridge for 100 years was predicted according to the optimal prediction model, i.e. the Serb model, and the results showed that the residual strength of GFRP bars is able to meet the current standard requirement.

Study on phase transition heat transfer of nanofluids in porous media based on LBM

In this paper, a lattice Boltzmann method is used to simulate the phase transition heat transfer of nanofluid in a miniature heat tube in a porous medium. Based on this, the existing research basis is summarized, the influence of nanoparticles on the thermal properties of the base solution is analyzed, the relevant thermal property prediction model is constructed, and the thermal property parameters of nanofluid are fitted. We generated porous media with different porosity with particle size of 100 micron to 150 microns. Combining the force form proposed by Gong with the exact difference method, the interparticle force is introduced. Based on the improved LBM model and the coupled energy equation, so as to analyze the phase transition heat transfer situation of nanofluids with different thermal properties in porous media.The study shows that nanofluidics can effectively enhance the heat transfer performance of heat pipe equipment.In a certain concentration limit, with the increase of volume concentration, the heat transfer performance of heat pipe is enhanced. The greater the temperature difference between the cold and heat source, that is, the higher the temperature of the heat source, the lower the cold source temperature, the better the heat transfer performance of the heat pipe. For the fluid flow in the porous media region, that is, the turbulent phenomenon that may occur in the porous media, and more vortices occur in the porous media with 40% and 50% porosity. Higher porosity has better heat pipe temperature uniformity.

Modeling study on the mechanical performance of CFRP/Al single-lap rivet joints under hygrothermal environmental conditions

Carbon Fiber Reinforced Polymers (CFRP) and aluminum alloys, owing to their exceptional mechanical properties and lightweight attributes, are extensively used in aircraft manufacturing. However, prolonged exposure to a hygrothermal environment can compromise the mechanical integrity of composite joint structures. In this paper, accelerated aging experiments were designed to test the mechanical properties of CFRP after hygrothermal aging, a novel mechanical property prediction model tailored for the hygrothermal coupled environment is presented. The model accurately predicts the modulus and strength of CFRP after hydrothermal aging. A three-dimensional finite element model for the CFRP interference riveted structure, considering a tri-coupled state of moisture, temperature, and force was established by the application of subroutine and field superposition. The precision of this finite element model has been affirmed through accelerated aging tests. By integrating finite element analysis with experimental methods, this research delves into the failure modes and mechanisms of CFRP and its joints under hygrothermal conditions. It was discerned that the hygrothermal environment undermines the bond between fibers and the matrix, resulting in pronounced interlaminar delamination and shear failure in CFRP. The failure forms gradually change from “flaky” at lower levels of aging to “filamentary” at higher levels of aging. For CFRP riveted structures, the hygrothermal conditions influence their loadbearing capability and shift the primary positions of failure.

A node graph using SEM images for material properties prediction with a case study of composite material

Establishing a mapping model between the microstructure and material properties of composite materials is crucial for material development. Scanning electron microscope (SEM) images are widely used for the prediction of material properties. However, the prediction from a single SEM image is independent and does not fully reflect the microstructure characteristics. To address this issue, this paper proposes a node graph construction strategy for SEM images and establishes a multi-graph-based graph attention network (GAT) material property prediction model to achieve the convergence of mutual complementation in microstructure features by using GAT. Firstly, multiple SEM images are constructed into node graphs by a microstructure feature encoder. Next, the microstructure features of multiple SEM images on the node graphs are mutually complemented and converged by using GAT. Finally, the prediction is carried out by using multiple SEM images. The experimental results show that the proposed method shows better performance than other methods.

Prediction and comparative analysis of friction material properties using a GA-SVM optimization model

PurposeWith the rapid advancement in the automotive industry, the friction coefficient (FC), wear rate (WR) and weight loss (WL) have emerged as crucial parameters to measure the performance of automotive braking systems, so the FC, WR and WL of friction material are predicted and analyzed in this work, with an aim of achieving accurate prediction of friction material properties.Design/methodology/approachGenetic algorithm support vector machine (GA-SVM) model is obtained by applying GA to optimize the SVM in this work, thus establishing a prediction model for friction material properties and achieving the predictive and comparative analysis of friction material properties. The process parameters are analyzed by using response surface methodology (RSM) and GA-RSM to determine them for optimal friction performance.FindingsThe results indicate that the GA-SVM prediction model has the smallest error for FC, WR and WL, showing that it owns excellent prediction accuracy. The predicted values obtained by response surface analysis are closed to those of GA-SVM model, providing further evidence of the validity and the rationality of the established prediction model.Originality/valueThe relevant results can serve as a valuable theoretical foundation for the preparation of friction material in engineering practice.

Application of a property prediction model based on the structure oriented lumping method in the fluid catalytic cracking process

Based on the Structure Oriented Lumping (SOL) method and the Artificial Neural Network (ANN) algorithm, a SOL-ANN property prediction model was constructed to predict the properties of molecules and products in the fluid catalytic cracking (FCC) process. The properties of each structural vector in the molecular composition matrices of gasoline and diesel were calculated. The influences of reaction temperature on the properties of gasoline and diesel were investigated from the perspective of molecular composition. When the reaction temperature increased from 490 °C to 510 °C, the content of aromatics and olefins in gasoline and the content of aromatics in diesel increased, resulting in the research octane number (RON) of gasoline increasing by 2.96 units and the cetane number (CN) of diesel decreasing by 1.37 units. Using the molecular composition information of products to calculate the properties of molecules and products could guide the product quality evaluation and process optimization.

Abstract 3884: Artificial intelligence (AI)-enabled discovery of GLX1546, a novel and potent inhibitor of ULK1/2 kinases and autophagy

Abstract Background: Autophagy is an evolutionarily conserved process, critical for the recycling of damaged or excess cellular components, which promotes cell survival and growth under stress conditions. Cancer cells, especially those with RAS mutations, can activate the ULK1/2 (UNC-51-like kinase 1/2)-initiated autophagy pathway for survival and diminish RAS pathway inhibitors’ efficacy. The combination of RAS pathway and ULK1/2 kinase inhibitors may provide synergistic or additive effects for RAS mutated solid tumors. We have identified a structurally small molecule, GLX1546, which shows strong inhibition against ULK1/2, as well as autophagy induced by RAS pathway inhibitors in cells. Methods: Multiple dual-target activity predictive models were constructed and trained using extensive protein-ligand interaction data. These models, in conjunction with in-house built physicochemical property predictive models, were integrated into an AI-driven computational pipeline to perform the in silico screening on both ULK1/2 inhibitory activities and drug-like properties. ULK1/2 in vitro activities were assessed by enzymatic assays and cellular assays based on the phosphorylation of ULK substrate. The in vivo anti-tumor efficacy of GLX1546 was evaluated in A549 CDX mouse model both as monotherapy and in combination with trametinib. Results: An AI-driven high-throughput virtual screening strategy using deep learning models was implemented for the discovery of novel and potent ULK1/2 small-molecule inhibitors. GLX1546 was developed as a potent ULK1/2 inhibitor with IC50 values of 1.13 nM for ULK1 and 4.75 nM for ULK2. Moreover, to address trametinib-induced activation of ULK1/2 in KRAS and BRAF mutated solid tumors, GLX1546 provided IC50 values of 54 nM in A549, 21 nM in HCT 116, 67 nM in MIA PaCa-2, and 76 nM in A375 cells against both basal and trametinib-induced pATG13. Furthermore, GLX1546 exhibited good exposure, sound safety, and high oral bioavailability (~100%) in mice. The combination of trametinib (PO, QD) with GLX1546 (PO, BID) in 21-day cycles also led to inhibition of ATG13 phosphorylation and greater tumor growth inhibition than trametinib alone in A549 CDX mouse model. Conclusions: These results indicate that GLX1546 can inhibit ULK-mediated autophagy induced by trametinib. The combination of GLX1546 with RAS pathway inhibitors may offer a novel cancer therapy for RAS mutated cancers, highlighting AI as an effective tool for drug discovery. Keywords: AI-driven drug discovery, autophagy, small-molecule inhibitors, ULK Citation Format: Guonan Cui, Xulun Sun, Wenyu Zhu, Qi Wang, Jinchen Li, Zhongzong Pan, Chengtao Li. Artificial intelligence (AI)-enabled discovery of GLX1546, a novel and potent inhibitor of ULK1/2 kinases and autophagy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3884.

Gaussian Process Regression-Based Near-Infrared d-Luciferin Analogue Design Using Mutation-Controlled Graph-Based Genetic Algorithm.

Molecular discovery is central to the field of chemical informatics. Although optimization approaches have been developed that target-specific molecular properties in combination with machine learning techniques, optimization using databases of limited size is challenging for efficient molecular design. We present a molecular design method with a Gaussian process regression model and a graph-based genetic algorithm (GB-GA) from a data set comprising a small number of compounds by introducing mutation probability control in the genetic algorithm to enhance the optimization capability and speed up the convergence to the optimal solution. In addition, we propose reducing the number of parameters in the conventional GB-GA focusing on efficient molecular design from a small database. We generated a target-specific database by combining active learning and iterative design in the evolutionary methodologies and chose Gaussian process regression as the prediction model for molecular properties. We show that the proposed scheme is more efficient for optimization toward the target properties from goal-directed benchmarks with several drug-like molecules compared to the conventional GB-GA method. Finally, we provide a demonstration whereby we designed D-luciferin analogues with near-infrared fluorescence for bioimaging, which is desirable for effective in vivo light sources, from a small-size data set.

Discovering novel halide perovskite alloys using multi-fidelity machine learning and genetic algorithm.

Expanding the pool of stable halide perovskites with attractive optoelectronic properties is crucial to addressing current limitations in their performance as photovoltaic (PV) absorbers. In this article, we demonstrate how a high-throughput density functional theory (DFT) dataset of halide perovskite alloys can be used to train accurate surrogate models for property prediction and subsequently perform inverse design using genetic algorithm (GA). Our dataset consists of decomposition energies, bandgaps, and photovoltaic efficiencies of nearly 800 pure and mixed composition ABX3 compounds from both the GGA-PBE and HSE06 functionals, and are combined with ∼100 experimental data points collected from the literature. Multi-fidelity random forest regression models are trained on the DFT + experimental dataset for each property using descriptors that one-hot encode composition, phase, and fidelity, and additionally include well-known elemental or molecular properties of species at the A, B, and X sites. Rigorously optimized models are deployed for experiment-level prediction over >150 000 hypothetical compounds, leading to thousands of promising materials with low decomposition energy, band gap between 1 and 2eV, and efficiency of >15%. Surrogate models are further combined with GA using an objective function to maintain chemical feasibility, minimize decomposition energy, maximize PV efficiency, and keep bandgap between 1 and 2eV; thus, hundreds more optimal compositions and phases are discovered. We present an analysis of the screened and inverse-designed materials, visualize ternary phase diagrams generated for many systems of interest using machine learning predictions, and suggest strategies for further improvement and expansion in the future.

SG-ATT: A Sequence Graph Cross-Attention Representation Architecture for Molecular Property Prediction.

Existing formats based on the simplified molecular input line entry system (SMILES) encoding and molecular graph structure are designed to encode the complete semantic and structural information of molecules. However, the physicochemical properties of molecules are complex, and a single encoding of molecular features from SMILES sequences or molecular graph structures cannot adequately represent molecular information. Aiming to address this problem, this study proposes a sequence graph cross-attention (SG-ATT) representation architecture for a molecular property prediction model to efficiently use domain knowledge to enhance molecular graph feature encoding and combine the features of molecular SMILES sequences. The SG-ATT fuses the two-dimensional molecular features so that the current model input molecular information contains molecular structure information and semantic information. The SG-ATT was tested on nine molecular property prediction tasks. Among them, the biggest SG-ATT model performance improvement was 4.5% on the BACE dataset, and the average model performance improvement was 1.83% on the full dataset. Additionally, specific model interpretability studies were conducted to showcase the performance of the SG-ATT model on different datasets. In-depth analysis was provided through case studies of in vitro validation. Finally, network tools for molecular property prediction were developed for the use of researchers.

Development of Property Prediction Model for Hot Strip Mill Using Machine Learning Algorithms

Development of Property Prediction Model for Hot Strip Mill Using Machine Learning Algorithms

Inverse design machine learning model for metallic glasses with good glass-forming ability and properties

The design of metallic glasses (MGs) with good properties is one of the long-standing bottlenecks in materials science and engineering, which has been relying mostly on far less efficient traditional trial-and-error methods. Even the currently popular machine learning-based forward designs, which use manual input to navigate high dimensional compositional space, often become inefficient with the increasing compositional complexity in MGs. Here, we developed an inverse design machine learning model, leveraging the variational autoencoder (VAE), to directly generate the MGs with good glass-forming ability (GFA). We demonstrate that our VAE with the property prediction model is not only an expressive generative model but also able to do accurate property prediction. Our model allows us to automatically generate novel MG compositions by performing simple operations in the latent space. After randomly generating 3000MG compositions using the model, a detailed analysis of four typical metallic alloys shows that unreported MG compositions with better glass-forming ability can be predicted. Moreover, our model facilitates the use of powerful optimization algorithms to efficiently guide the search for MGs with good GFA in the latent space. We believe that this is an efficient way to discover MGs with excellent properties.

Prediction Models for Mechanical Properties of Cement-Bound Aggregate with Waste Rubber

The high stiffness of cement-bound aggregate (CBA) is recognized as its main drawback. The stiffness is described by the modulus of elasticity, which is difficult to determine precisely in CBA. Incorporating rubber in these mixtures reduces their stiffness, but mathematical models of the influence of rubber on the mechanical characteristics have not previously been defined. The scope of this research was to define a prediction model for the compressive strength (fc), dynamic modulus of elasticity (Edyn) and static modulus of elasticity (Est) based on the measured ultrasonic pulse velocity as a non-destructive test method. The difference between these two modules is based on the measurement method. Within this research, the cement and waste rubber content were varied, and the mechanical properties were determined for three curing periods. The Edyn was measured using the ultrasonic pulse velocity (UPV), while the Est was determined using three-dimensional digital image correlation (3D DIC). The influence of the amount of cement and rubber and the curing period on the UPV was determined. The development of prediction models for estimating the fc and Est of CBA modified with waste rubber based on the non-destructive test results is highlighted as the most significant contribution of this work. The curing period was statistically significant for the prediction of the Est, which points to the development of CBA elastic properties through different stages during the cement-hydration process. By contrast, the curing period was not statistically significant when estimating the fc, resulting in a simplified, practical and usable prediction model.

Predicting small molecules solubility on endpoint devices using deep ensemble neural networks.

Aqueous solubility is a valuable yet challenging property to predict. Computing solubility using first-principles methods requires accounting for the competing effects of entropy and enthalpy, resulting in long computations for relatively poor accuracy. Data-driven approaches, such as deep learning, offer improved accuracy and computational efficiency but typically lack uncertainty quantification. Additionally, ease of use remains a concern for any computational technique, resulting in the sustained popularity of group-based contribution methods. In this work, we addressed these problems with a deep learning model with predictive uncertainty that runs on a static website (without a server). This approach moves computing needs onto the website visitor without requiring installation, removing the need to pay for and maintain servers. Our model achieves satisfactory results in solubility prediction. Furthermore, we demonstrate how to create molecular property prediction models that balance uncertainty and ease of use. The code is available at https://github.com/ur-whitelab/mol.dev, and the model is useable at https://mol.dev.

Realistic material property prediction using domain adaptation based machine learning

Scientists aim to predict properties for a known subset of related out-of-distribution materials rather than universally distributed samples. Introducing target material information enhances machine learning models.

Influence of seawater sea sand concrete environment on the long‐term tensile properties of pultruded basalt and basalt‐carbon‐hybrid fiber reinforced polymer profiles

AbstractTo reveal the durability of fiber‐reinforced polymer (FRP) profiles, which differ from FRP bars due to the clear variations in the internal structure of the fibers. This study experimentally investigated the tensile properties of FRP profiles in the seawater sea sand concrete (SSC) environment at 25, 45, and 60°C. Unidirectional basalt FRP profiles, multidirectional basalt FRP profiles, and basalt and carbon hybrid FRP profiles were considered. SEM analysis was performed to evaluate the corrosion mechanism. Results show that the strength degradation of FRP profiles accelerated with rising ambient temperature due to resin corrosion in hyperthermia. After 63 days of immersion in the SSC environment solution, multidirectional basalt FRP (BFRP) profiles exhibited more severe corrosion than unidirectional BFRP profiles (82.5% and 88.2%, respectively). The hybrid FRP profiles exhibited a strength retention of 86.4%, which was higher than that of multidirectional BFRP profiles due to the superior alkali resistance of carbon fibers compared to basalt fibers. In addition, prediction models of FRP profiles based on the Arrhenius relation were proposed with a high correlation coefficient of R2.Highlights Durability of FRP profiles in SSC environment was studied. Basalt and carbon fibers hybrid reinforced multidirectional polymer profiles. Corrosion mechanism of FRP profiles was revealed. Tensile property prediction model of FRPs in SSC environment was proposed.