Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “training data”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,477 records · Page 82
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Speeding up and reducing memory usage for scientific machine learning via mixed precision

Scientific machine learning (SciML) has emerged as a versatile approach to address complex computational science and engineering problems. Within this field, physics-informed neural networks (PINNs) and deep operator networks (DeepONets) stand out as the leading techniques for solving partial differential equations by incorporating both physical equations and experimental data. However, training PINNs and DeepONets require significant computational resources, including long computational times and large amounts of memory. In search of computational efficiency, training neural networks using half precision (float16) rather than the conventional single (float32) or double (float64) precision has gained substantial interest, given the inherent benefits of reduced computational time and memory consumed. However, we find that float16 cannot be applied to SciML methods, because of gradient divergence at the start of training, weight updates going to zero, and the inability to converge to a local minima. To overcome these limitations, we explore mixed precision, which is an approach that combines the float16 and float32 numerical formats to reduce memory usage and increase computational speed. Our experiments showcase that mixed precision training not only substantially decreases training times and memory demands but also maintains model accuracy. Here, we also reinforce our empirical observations with a theoretical analysis. The research has broad implications for SciML in various computational applications.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Audacity of huge: overcoming challenges of data scarcity and data quality for machine learning in computational materials discovery

Machine learning (ML)-accelerated discovery requires large amounts of high-fidelity data to reveal predictive structure–property relationships. For many properties of interest in materials discovery, the challenging nature and high cost of data generation has resulted in a data landscape that is both scarcely populated and of dubious quality. Data-driven techniques starting to overcome these limitations include the use of consensus across functionals in density functional theory, the development of new functionals or accelerated electronic structure theories, and the detection of where computationally demanding methods are most necessary. When properties cannot be reliably simulated, large experimental data sets can be used to train ML models. In the absence of manual curation, increasingly sophisticated natural language processing and automated image analysis are making it possible to learn structure–property relationships from the literature. Finally, models trained on these data sets will improve as they incorporate community feedback.

36 MATERIALS SCIENCE↗
DOE OSTI2018

Predictive modeling of dynamic fracture growth in brittle materials with machine learning

We use simulation data from a high delity Finite-Discrete Element Model to build an e cient Machine Learning (ML) approach to predict fracture growth and coalescence. Our goal is for the ML approach to be used as an emulator in place of the computationally intensive high delity models in an uncertainty quanti cation framework where thousands of forward runs are required. The failure of materials with various fracture con gurations (size, orientation and the number of initial cracks) are explored and used as data to train our ML model. This novel approach has shown promise in predicting spatial (path to failure) and temporal (time to failure) aspects of brittle material failure. Predictions of where dominant fracture paths formed within a material were ~85% accurate and the time of material failure deviated from the actual failure time by an average of ~16%. Additionally, the ML model achieves a reduction in computational cost by multiple orders of magnitude.

36 MATERIALS SCIENCE↗
DOE OSTI2025

A systematic review of machine learning in groundwater monitoring

With increasing concerns about water scarcity, groundwater has become crucial since this resource provides most of the freshwater needs. However, various human and natural activities often contaminate the groundwater, making it unsuitable for use. Over the years, scientists and engineers have used many methods to predict and track groundwater contamination as part of environmental monitoring. Consequently, there is an urgent need for improved methods, particularly in the face of increasing contamination. Machine learning has sometimes been used to monitor groundwater, air quality, and climate. Traditional methods must be improved due to the complexity and large amount of environmental data. This includes using hybrid models that combine traditional and new techniques. Despite the use of machine learning in many scientific areas, there is a lack of comprehensive reviews focusing on its use in environmental monitoring, especially groundwater monitoring. We aim to fill this gap by exploring machine-learning applications in groundwater monitoring. We discuss relevant methods, their limitations, and future potential. We summarize research on automating data processing and model training using groundwater sensor data. Our research underscores the transformative potential of machine learning to revolutionize long-term groundwater monitoring and contamination detection, providing valuable insights for future research and practical applications.

AI/ML↗
DOE OSTI2016

Gaussian processes with built-in dimensionality reduction: Applications to high-dimensional uncertainty propagation

Uncertainty quantification (UQ) tasks, such as model calibration, uncertainty propagation, and optimization under uncertainty, typically require several thousand evaluations of the underlying computer codes. To cope with the cost of simulations, one replaces the real response surface with a cheap surrogate based, e.g., on polynomial chaos expansions, neural networks, support vector machines, or Gaussian processes (GP). However, the number of simulations required to learn a generic multivariate response grows exponentially as the input dimension increases. This curse of dimensionality can only be addressed, if the response exhibits some special structure that can be discovered and exploited. A wide range of physical responses exhibit a special structure known as an active subspace (AS). An AS is a linear manifold of the stochastic space characterized by maximal response variation. The idea is that one should first identify this low dimensional manifold, project the high-dimensional input onto it, and then link the projection to the output. If the dimensionality of the AS is low enough, then learning the link function is a much easier problem than the original problem of learning a high-dimensional function. The classic approach to discovering the AS requires gradient information, a fact that severely limits its applicability. Furthermore, and partly because of its reliance to gradients, it is not able to handle noisy observations. The latter is an essential trait if one wants to be able to propagate uncertainty through stochastic simulators, e.g., through molecular dynamics codes. In this work, we develop a probabilistic version of AS which is gradient-free and robust to observational noise. Our approach relies on a novel Gaussian process regression with built-in dimensionality reduction. In particular, the AS is represented as an orthogonal projection matrix that serves as yet another covariance function hyper-parameter to be estimated from the data. To train the model, we design a two-step maximum likelihood optimization procedure that ensures the orthogonality of the projection matrix by exploiting recent results on the Stiefel manifold, i.e., the manifold of matrices with orthogonal columns. The additional benefit of our probabilistic formulation, is that it allows us to select the dimensionality of the AS via the Bayesian information criterion. We validate our approach by showing that it can discover the right AS in synthetic examples without gradient information using both noiseless and noisy observations. We demonstrate that our method is able to discover the same AS as the classical approach in a challenging one-hundred-dimensional problem involving an elliptic stochastic partial differential equation with random conductivity. Finally, we use our approach to study the effect of geometric and material uncertainties in the propagation of solitary waves in a one dimensional granular system.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2019

A hybrid point-particle force model that combines physical and data-driven approaches

Here this study improves upon the physics-based pairwise interaction extended point-particle (PIEP) model. The PIEP model leverages our physical understanding to predict fluid mediated interactions between solid particles. By considering the relative location of neighboring particles, the PIEP model is able to provide better predictions than existing drag models. While the current physical PIEP model is a powerful tool, its assumptions lead to increased error in flows with higher particle volume fractions. To reduce this error, a regression algorithm makes direct use of the results of direct numerical simulations (DNS) of an array of monodisperse solid particles subjected to uniform ambient flow at varying Reynolds numbers. The resulting statistical model and the physical PIEP model are superimposed to construct a hybrid, physics-based data-driven PIEP model. It must be noted that the performance of a pure data-driven approach without the model-form provided by the physical PIEP model is substantially inferior. The hybrid model's predictive capabilities are analyzed using additional DNS data that was not part of training the data-driven model. In every case tested, the hybrid models resulting from the regression were capable of (1) improving upon the physical PIEP model's prediction and (2) recovering underlying relevant physics from the DNS data. As the particle volume fraction increases, the physical PIEP model loses the ability to approximate the forces experienced by the particles, but the statistical model continues to produce accurate approximations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Multitask Machine Learning of Collective Variables for Enhanced Sampling of Rare Events

Computing accurate reaction rates is a central challenge in computational chemistry and biology because of the high cost of free energy estimation with unbiased molecular dynamics. In this work, a data-driven machine learning algorithm is devised to learn collective variables with a multitask neural network, where a common upstream part reduces the high dimensionality of atomic configurations to a low dimensional latent space and separate downstream parts map the latent space to predictions of basin class labels and potential energies. Here, the resulting latent space is shown to be an effective low-dimensional representation, capturing the reaction progress and guiding effective umbrella sampling to obtain accurate free energy landscapes. This approach is successfully applied to model systems including a 5D Müller Brown model, a 5D three-well model, the alanine dipeptide in vacuum, and an Au(110) surface reconstruction unit reaction. It enables automated dimensionality reduction for energy controlled reactions in complex systems, offers a unified and data-efficient framework that can be trained with limited data, and outperforms single-task learning approaches, including autoencoders.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Condensed Phase Water Molecular Multipole Moments from Deep Neural Network Models Trained on Ab Initio Simulation Data

Ionic solvation phenomena in liquids involve intense interactions in the inner solvation shell. For interactions beyond the first shell, the ion–solvent interaction energies result from the sum of many smaller-magnitude contributions that can still include polarization effects. Deep neural network (DNN) methods have recently found wide application in developing efficient molecular models that maintain near-quantum accuracy. In this work, we extend the DeePMD-kit code to produce accurate molecular multipole moments in the bulk and near interfaces. The new method is validated by comparing the DNN moments with those generated by ab initio simulations. The moments are used to compute the electrostatic potential at the center of a molecular-sized hydrophobic cavity in water. The results show that the fields produced by the DNN models are in quantitative agreement with the AIMD-derived values. These efficient methods will open the door to more accurate solvation models for large solutes such as proteins.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Machine Learning-Guided Identification of PET Hydrolases from Natural Diversity

The enzymatic depolymerization of poly(ethylene terephthalate) (PET) is emerging as a leading chemical recycling technology for waste polyester. As part of this endeavor, new candidate enzymes identified from natural diversity can serve as useful starting points for enzyme evolution and engineering. In this study, we improved upon HMM searches by applying an iterative machine learning strategy to identify 400 putative PET-degrading enzymes (PET hydrolases) from naturally occurring homologs. Using high-throughput (HTP) experimental techniques, we successfully expressed and purified >200 enzyme candidates and assayed them for PET hydrolysis activity as a function of pH, temperature, and substrate crystallinity. From this library, we discovered 91 previously unknown PET hydrolases, 35 of which retain activity at pH 4.5 on crystalline material, which are conditions relevant to developing more efficient commercial processes. Notably, four enzymes showed equal to or higher activity than LCC-ICCG, a benchmark PET hydrolase, at this challenging condition in our screening assay, and 11 of which have pH optima <7. Using these data, we identified regions of PETases statistically correlated to activity at lower pH. We additionally investigated the effect of condition-specific activity data on trained machine learning predictors and found a precision (putative hit rate) improvement of up to 30% compared to a Hidden Markov Model alone. Our findings show that by pointing enzyme discovery toward conditions of interest with multiple rounds of experimental and machine learning, we can discover large sets of active enzymes and explore factors associated with activity at those conditions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2022

Heuristic Computational Model for Predicting Lignin Solubility in Tailored Organic Solvents

Lignin is a random heteropolymer that has been extensively studied as a renewable source of aromatic precursors for high-value chemicals, biofuels, and bioplastics. A key challenge in lignin valorization is the structural and compositional heterogeneity of lignin feedstocks. Solvent-based approaches are commonly used to fractionate lignin to reduce this heterogeneity, but solvent selection can be challenging due to variability in lignin composition. In this work, we developed computational methods to predict good and poor organic solvents as a function of lignin composition. We analyzed 28 different linear pentamer structures, 18 from known libraries and 10 hypothetical polymers, and calculated their activity coefficients in 50 different organic solvents by using the conductor-like screening model for realistic solvents. We used these data to train a regression model that enabled the extensive investigation of the impact of solvent and monolignol compositions on predicted lignin solubility. The exhaustive exploration of solubility trends using model predictions revealed sets of solvents, identified using Kamlet–Taft parameters, that are predicted to promote lignin dissolution regardless of lignin composition. We further identified solvents expected to selectively isolate lignin fractions enriched in certain subunits. Furthermore, these results establish heuristic guidelines for solvent selection that can be used to tailor fractionation processes for lignin feedstocks of distinct composition or to design new processes that isolate fractions with higher proportions of selected subunits.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

Internal Variability Increased Arctic Amplification During 1980–2022

Abstract Since 1980, the Arctic surface has warmed four times faster than the global mean. Enhanced Arctic warming relative to the global average warming is referred to as Arctic Amplification (AA). While AA is a robust feature in climate change simulations, models rarely reproduce the observed magnitude of AA, leading to concerns that models may not accurately capture the response of the Arctic to greenhouse gas emissions. Here, we use CMIP6 data to train a machine learning algorithm to quantify the influence of internal variability in surface air temperature trends over both the Arctic and global domains. Application of this machine learning algorithm to observations reveals that internal variability increases the Arctic warming but slows global warming in recent decades, inflating AA since 1980 by 38% relative to the externally forced AA. Accounting for the role of internal variability reconciles the discrepancy between simulated and observed AA.

58 GEOSCIENCES↗
DOE OSTI2022

Predicting solid state material platforms for quantum technologies

Semiconductor materials provide a compelling platform for quantum technologies (QT). However, identifying promising material hosts among the plethora of candidates is a major challenge. Therefore, we have developed a framework for the automated discovery of semiconductor platforms for QT using material informatics and machine learning methods. Different approaches were implemented to label data for training the supervised machine learning (ML) algorithms logistic regression, decision trees, random forests and gradient boosting. We find that an empirical approach relying exclusively on findings from the literature yields a clear separation between predicted suitable and unsuitable candidates. In contrast to expectations from the literature focusing on band gap and ionic character as important properties for QT compatibility, the ML methods highlight features related to symmetry and crystal structure, including bond length, orientation and radial distribution, as influential when predicting a material as suitable for QT.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Million-scale data integrated deep neural network for phonon properties of heuslers spanning the periodic table

Existing machine learning potentials for predicting phonon properties of crystals are typically limited on a material-to-material basis, primarily due to the exponential scaling of model complexity with the number of atomic species. We address this bottleneck with the developed Elemental Spatial Density Neural Network Force Field, namely Elemental-SDNNFF. The effectiveness and precision of our Elemental-SDNNFF approach are demonstrated on 11,866 full, half, and quaternary Heusler structures spanning 55 elements in the periodic table by prediction of complete phonon properties. Self-improvement schemes including active learning and data augmentation techniques provide an abundant 9.4 million atomic data for training. Deep insight into predicted ultralow lattice thermal conductivity (<1 Wm –1 K –1 ) of 774 Heusler structures is gained by p–d orbital hybridization analysis. Additionally, a class of two-band charge-2 Weyl points, referred to as “double Weyl points”, are found in 68% and 87% of 1662 half and 1550 quaternary Heuslers, respectively.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Machine learning for endoleak detection after endovascular aortic repair

Diagnosis of endoleak following endovascular aortic repair (EVAR) relies on manual review of multi-slice CT angiography (CTA) by physicians which is a tedious and time-consuming process that is susceptible to error. We evaluate the use of a deep neural network for the detection of endoleak on CTA for post-EVAR patients using a novel data efficient training approach. 50 CTAs and 20 CTAs with and without endoleak respectively were identified based on gold standard interpretation by a cardiovascular subspecialty radiologist. The Endoleak Augmentor, a custom designed augmentation method, provided robust training for the machine learning (ML) model. Predicted segmentation maps underwent post-processing to determine the presence of endoleak. The model was tested against 3 blinded general radiologists and 1 blinded subspecialist using a held-out subset (10 positive endoleak CTAs, 10 control CTAs). Model accuracy, precision and recall for endoleak diagnosis were 95%, 90% and 100% relative to reference subspecialist interpretation (AUC = 0.99). Accuracy, precision and recall was 70/70/70% for generalist1, 50/50/90% for generalist2, and 90/83/100% for generalist3. The blinded subspecialist had concordant interpretations for all test cases compared with the reference. In conclusion, our ML-based approach has similar performance for endoleak diagnosis relative to subspecialists and superior performance compared with generalists.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2021

Machine learning assisted prediction of the Young’s modulus of compositionally complex alloys

We identify compositionally complex alloys (CCAs) that offer exceptional mechanical properties for elevated temperature applications by employing machine learning (ML) in conjunction with rapid synthesis and testing of alloys for validation to accelerate alloy design. The advantages of this approach are scalability, rapidity, and reasonably accurate predictions. ML tools were implemented to predict Young’s modulus of refractory-based CCAs by employing different ML models. Our results, in conjunction with experimental validation, suggest that average valence electron concentration, the difference in atomic radius, a geometrical parameter λ and melting temperature of the alloys are the key features that determine the Young’s modulus of CCAs and refractory-based CCAs. The Gradient Boosting model provided the best predictive capabilities (mean absolute error of 6.15 GPa) among the models studied. Our approach integrates high-quality validation data from experiments, literature data for training machine-learning models, and feature selection based on physical insights. It opens a new avenue to optimize the desired materials property for different engineering applications.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Forecasting induced seismicity in Oklahoma using machine learning methods

Oklahoma earthquakes in the past decade have been mostly associated with wastewater injection. Here we use a machine learning technique—the Random Forest to forecast induced seismicity rate in Oklahoma based on injection-related parameters. We split the data into training (2011.01–2015.05) and test (2015.06–2020.12) periods. The model forecasts seismicity rate during the test period based on input features, including operational parameters (injection rate and pressure), geological information (depth to basement), and modeled pore pressure and poroelastic stress. The results show overall good match with observed seismicity rate (adjusted R 2 of 0.75). The model shows that pore pressure rate and poroelastic stressing rates are the two most important features in forecasting. The absolute values of pore pressure and poroelastic stress, and the injection rate itself, are less important than the stressing rates. These findings further emphasize that temporal changes of stressing rates would lead to significant changes in seismicity rates.

58 GEOSCIENCES↗
DOE OSTI2024

Innovative dead-time correction and background subtraction for neutron multiplicity measurements using neural networks

Abstract The number of neutrons emitted from a nuclear reaction plays a crucial role in various fields, including nuclear theory, nuclear nonproliferation, nuclear energy and nuclear criticality safety. Accurate determination of neutron multiplicities requires the application of several corrections, with dead-time correction and background subtraction being particularly significant. These corrections become more challenging for neutron detectors with time-dependent neutron capture. In this work, we perform a comprehensive study of three existing methods used for dead-time correction and background subtraction in neutron detectors with time-dependent neutron capture. The methods were tested for dead-times in the range from 0 to 1 μs using a Monte Carlo model simulating the dead-time and background effects in the standard neutron multiplicity probability distribution of $$^{252}$$ 252 Cf. The previous methods showed larger than desired uncertainty or systematic trade off. Those uncertainties prompted the development of a novel approach using neural networks trained with data from Monte Carlo simulations. The Neural Network method enabled the correction of neutron multiplicity probabilities more accurately than the other methods with fractional errors smaller than 3% for multiplicities around the peak of $$^{252}$$ 252 Cf. A similar approach using neural networks could be applied to problems where the system being studied can be accurately simulated without having an accurate analytical description available. The neural network method presented in this paper can be easily expanded if multiplicities greater than 10 are expected.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗