SEARCH · Engineering Papers

Results for “Feature importance”

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.

SourceResearch connectionFrom yearThrough year

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

At least 271 records · Page 15

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Opportunities and Challenges of Computational Electromagnetics Methods for Superconducting Circuit Quantum Device Modeling: A Practical Review

High-fidelity numerical methods that model the physical layout of a device are essential for the design of many technologies. For methods that characterize electromagnetic effects, these numerical methods are referred to as computational electromagnetics (CEM) methods. Although the CEM research field is mature, emerging applications can still stress the capabilities of the techniques in use today. The design of superconducting circuit quantum devices falls in this category due to the unconventional material properties and important features of the devices covering nanometer to centimeter scales. Such multiscale devices can stress the fundamental properties of CEM tools which can lead to an increase in simulation times, a loss in accuracy, or even cause no solution to be reliably found. While these challenges are being investigated by CEM researchers, knowledge about them is limited in the broader community of users of these CEM tools. This review is meant to serve as a practical introduction to the fundamental aspects of the major CEM techniques that a researcher may need to choose between to model a device, as well as provide insight into what steps they may take to alleviate some of their challenges. Our focus is on highlighting the main concepts without rigorously deriving all the details, which can be found in many textbooks and articles. After covering the fundamentals, we discuss more advanced topics related to the challenges of modeling multiscale devices with specific examples from superconducting circuit quantum devices. We conclude with a discussion on future research directions that will be valuable for improving the ability to successfully design increasingly more sophisticated superconducting circuit quantum devices. Although our focus and examples are taken from this area, researchers from other fields will still benefit from the details discussed here.

Elkin, Samuel T. [Virginia Tech., Blacksburg]↗

DOE OSTI2024

Data Efficiency Assessment of Generative Adversarial Networks for Critical Heat Flux Synthetic Data Generation

This study investigates the application of generative artificial intelligence techniques, particularly conditional generative adversarial networks (cGAN), in real-world engineering contexts, with a specific focus on synthetic data generation for critical heat flux (CHF). Utilizing a dataset comprising more than 20,000 real experimental CHF measurements, we conduct a series of experiments to examine cGAN’s behavior. These experiments encompass varying sizes of the training dataset, training cGAN on data from diverse experimental sources to generate new data on unseen experimental setups, and assessing the impact of excluding various input features on cGAN’s data generation accuracy. Our findings underscore the pronounced data dependency of cGAN for reliable performance, with decreased efficacy observed with smaller training dataset sizes. Notably, cGAN exhibits varying performance when trained on data from different experiments, with superior predictive capabilities observed for certain experiment sources compared to others. For instance, when cGAN was trained on data from Smolin et al.’s experiments or Zenkevich et al., it exhibited relatively good performance in generating the data from Becker et al., Kirillov et al., and Alekseev et al. experiments. In contrast, when trained with Alekseev et al.’s data and tasked with generating other experimental setups, cGAN showed notably poor performance. In both scenarios, cGAN’s performance was inferior compared to training on samples from all experiments concurrently. A feature importance analysis highlights the significant influence of parameters such as mass flux and heated length on accurate CHF generation, while other parameters like diameter and pressure have less impact. Inlet temperature is identified as a moderating factor by cGAN.

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2025

Implementation and Demonstration of the Digital Twin Certification System Remote Operations Framework

Microreactors are one promising advanced-reactor concept being pursued by the nuclear industry. They are distinguished by a relatively low power output of 20 MWth or less. These microreactors are intended for deployment in applications where conventional small-capacity power solutions, such as diesel generators, are either economically unfeasible or logistically challenging. Such applications include providing electric power and/or heat for remote communities, mining sites, defense installations, and humanitarian and disaster-relief missions. An important feature for the successful deployment of microreactors is their capability to be operated remotely. This capability can significantly reduce staffing costs by eliminating the need for licensed operators to be physically present at each reactor site. Instead, operators can be centralized in a single remote operations center placed in an economically advantageous location, thereby optimizing resources by consolidating expertise and enhancing operational efficiency. However, the implementation of a remote operation system for nuclear reactors raises new concerns regarding the security, reliability, and resilience of such a system. One way in which remote operations can be supported in a manner that maintains system security, reliability, and resilience is through the use of digital twins in a novel framework designed to verify and validate sensor data and commands communicated between the remote operations center and reactor. This framework, known as the Digital Twin Certification System (DTCS), has previously been proposed as an operations architecture that can bring security and resiliency levels of remote nuclear-reactor operations to a level acceptable for commercial deployment. This paper moves the proposed DTCS architecture from concept to reality by presenting the implementation and testing of the system. The rationale and implementation of the DTCS using tools such as DeepLynx and Apache Airflow, is covered in-depth. This is followed by a demonstration of the DTCS by applying the implemented system architecture to the Single Primary Heat Extraction and Removal Emulator, a small-scale non-nuclear test bed that emulates thermal behavior of a microreactor. The demonstration includes both normal and abnormal operating scenarios to highlight how the DTCS can increase the security, reliability, and resilience of a remote operations system.

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2025

Implementation and Demonstration of the Digital Twin Certification System Remote Operations Framework

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2023

Accelerated Materials Design for Molten Salt Technologies Using Innovative High-Throughput Methods

The focus of the project is on building an innovative accelerated materials design platform for molten salt technologies using novel high-throughput methods coupled to data analytics. The main objectives is to predict a FeCrMnNi alloy compositional space with better corrosion resistance than stainless steel 316 and identify new molten salt corrosion mechanisms. The project demonstrates the feasibility to use high-throughput methods coupled to data analytics to accelerate alloy design for extreme environments applications. Using a trained and tested machine learning (ML) model, 2000 FeCrMnNi alloy corrosion rate in molten chloride salts were predicted and a compositional field with corrosion rate lower than 316 stainless steel was identified. The ML model interpretability unveiled multiple features of importance in the model prediction. Some features were expected to be of relative significance, such as work function, surface energy and alloy electronegativity, and the ML model interpretability analysis confirmed those. On the other hand, the most important feature is the diffusion coefficient of Ni in the bulk alloy which indicates that a surface diffusion mechanism plays an important role in the overall corrosion mechanism in molten salts.

36 MATERIALS SCIENCE↗

DOE OSTI2021

The Challenge of Disproportionate Importance of Temporal Features in Predicting HPC Power Consumption

In this work, we demonstrate the challenges in predicting HPC cluster power consumption in the face of significant temporal skew in power consumption behavioral patterns. Predicting large power swings that extend several megawatts has significant operational value for HPC centers, however, prediction is challenging due to the relative rarity of such events and also due to the abrupt or disjoint deviation from the average power consumption levels. To study the impact of this challenge, we have trained a recurrent neural network (RNN) as a reasonably sophisticated model to predict power consumption of the one-year worth of node power consumption data from the Summit supercomputer located in the Oak Ridge Leadership Computing Facility. By studying the prediction results, we have found that although simple usage of RNN models can provide good results on average power consumption levels, it would fail at predicting the power swings that have more operational value. With such results, we discuss potential next steps in addressing such issues aiming towards a robust usage of power prediction techniques in HPC operations.

Li, Chengcheng↗

DOE OSTI2023

Explainable machine learning for hydrogen diffusion in metals and random binary alloys

Hydrogen diffusion in metals and alloys plays an important role in the discovery of new materials for fuel cell and energy storage technology. While analytic models use hand-selected features that have clear physical ties to hydrogen diffusion, they often lack accuracy when making quantitative predictions. Machine learning models are capable of making accurate predictions, but their inner workings are obscured, rendering it unclear which physical features are truly important. To develop interpretable machine learning models to predict the activation energies of hydrogen diffusion in metals and random binary alloys, we create a database for physical and chemical properties of the species and use it to fit six machine learning models. Our models achieve root-mean-squared errors between 98–119 meV on the testing data and accurately predict that elemental Ru has a large activation energy, while elemental Cr and Fe have small activation energies. By analyzing the feature importances of these fitted models, we identify relevant physical properties for predicting hydrogen diffusivity. While metrics for measuring the individual feature importances for machine learning models exist, correlations between the features lead to disagreement between models and limit the conclusions that can be drawn. Instead grouped feature importance, formed by combining the features via their correlations, agree across the six models and reveal that the two groups containing the packing factor and electronic specific heat are particularly significant for predicting hydrogen diffusion in metals and random binary alloys. In conclusion, this framework allows us to interpret machine learning models and enables rapid screening of new materials with the desired rates of hydrogen diffusion.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Environmental and ecological signals predict food shortages for subtropical populations of Australian flying foxes, reservoirs of Hendra virus

Food availability determines where and how animals use space across a landscape and, therefore, affects the risk of encounters leading to zoonotic spillover. This relationship is evident in Australian flying foxes (Pteropus spp.; fruit bats), where acute food shortages precede clusters of Hendra virus spillovers. Using machine learning, we predicted months of food shortages from climatological and ecological covariates (1996–2022) in subtropical Australia. Overall accuracy in predicting months of low food availability on a test set from 2018 up to 2022 reached 93.33 and 92.59% based on climatological and bat-level features, respectively. Seasonality and the Oceanic El Niño Index were the most important environmental features, while the number of bats in rescue centres and their body weights were the most important bat-level features. These models support predictive signals up to nine months in advance, facilitating action to mitigate spillover risk.

Pteropus↗

DOE OSTI2021

Predictive Models of Genetic Redundancy in Arabidopsis thaliana

Abstract Genetic redundancy refers to a situation where an individual with a loss-of-function mutation in one gene (single mutant) does not show an apparent phenotype until one or more paralogs are also knocked out (double/higher-order mutant). Previous studies have identified some characteristics common among redundant gene pairs, but a predictive model of genetic redundancy incorporating a wide variety of features derived from accumulating omics and mutant phenotype data is yet to be established. In addition, the relative importance of these features for genetic redundancy remains largely unclear. Here, we establish machine learning models for predicting whether a gene pair is likely redundant or not in the model plant Arabidopsis thaliana based on six feature categories: functional annotations, evolutionary conservation including duplication patterns and mechanisms, epigenetic marks, protein properties including posttranslational modifications, gene expression, and gene network properties. The definition of redundancy, data transformations, feature subsets, and machine learning algorithms used significantly affected model performance based on holdout, testing phenotype data. Among the most important features in predicting gene pairs as redundant were having a paralog(s) from recent duplication events, annotation as a transcription factor, downregulation during stress conditions, and having similar expression patterns under stress conditions. We also explored the potential reasons underlying mispredictions and limitations of our studies. This genetic redundancy model sheds light on characteristics that may contribute to long-term maintenance of paralogs, and will ultimately allow for more targeted generation of functionally informative double mutants, advancing functional genomic studies.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2021

Roles of Dynamic State Estimation in Power System Modeling, Monitoring and Operation

—Power system dynamic state estimation (DSE) remains an active research area. This is driven by the absence of accurate models, the increasing availability of fast-sampled, time synchronized measurements, and the advances in the capability, scalability, and affordability of computing and communications. This paper discusses the advantages of DSE as compared to static state estimation, and the implementation differences between the two, including the measurement configuration, modeling framework and support software features. The important roles of DSE are discussed from modeling, monitoring and operation aspects for today’s synchronous machine dominated systems and the future power electronics-interfaced generation systems. Several examples are presented to demonstrate the benefits of DSE on enhancing the operational robustness and resilience of 21st century power system through time critical applications. Future research directions are identified and discussed, paving the way for developing the next generation of energy management systems.

Zhao, Junbo↗

DOE OSTI2023

A detailed study of interpretability of deep neural network based top taggers

Abstract Recent developments in the methods of explainable artificial intelligence (XAI) allow researchers to explore the inner workings of deep neural networks (DNNs), revealing crucial information about input–output relationships and realizing how data connects with machine learning models. In this paper we explore interpretability of DNN models designed to identify jets coming from top quark decay in high energy proton–proton collisions at the Large Hadron Collider. We review a subset of existing top tagger models and explore different quantitative methods to identify which features play the most important roles in identifying the top jets. We also investigate how and why feature importance varies across different XAI metrics, how correlations among features impact their explainability, and how latent space representations encode information as well as correlate with physically meaningful quantities. Our studies uncover some major pitfalls of existing XAI methods and illustrate how they can be overcome to obtain consistent and meaningful interpretation of these models. We additionally illustrate the activity of hidden layers as neural activation pattern diagrams and demonstrate how they can be used to understand how DNNs relay information across the layers and how this understanding can help to make such models significantly simpler by allowing effective model reoptimization and hyperparameter tuning. These studies not only facilitate a methodological approach to interpreting models but also unveil new insights about what these models learn. Incorporating these observations into augmented model design, we propose the particle flow interaction network model and demonstrate how interpretability-inspired model augmentation can improve top tagging performance.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Machine learning reveals factors that control ion mobility in anti-perovskite solid electrolytes

Solid-state batteries are projected to exhibit improved energy densities and safety compared to liquid-electrolyte-based systems. Consequently, the development of solid electrolytes that can conduct cations at rates comparable to that of liquid electrolytes is an active area of research. To facilitate this search it is desirable to understand the chemical and structural features that control ion transport through a crystalline lattice. Here, machine learning (ML) is used to identify features that influence ion mobility in solids, quantify the relative importance of those features, and demonstrate how the variation of individual features alters ion mobility. Adopting the anti-perovskite lattice as a model system, ML algorithms were trained to predict migration barriers from a data set containing hundreds of barriers calculated using density functional theory. In this study, more than 10 6 feature sets of non-redundant descriptors were evaluated for their ability to predict migration barriers. The present analysis correctly reiterates the importance of several features that are known to influence mobility – for example, anion polarizability – but also identifies others whose importance is not widely recognized. Lattice properties such as hopping distance and channel width exhibit the greatest influence on cation mobility. Individual conditional expectation analyses performed on these features shows that barriers decrease as hopping distance decreases and as channel width increases. The defect formation energy – which is commonly associated with the concentration of defects, but not with their mobility – is identified as an important feature for interstitial migration. In sum, this analysis aids in the design of optimal solid electrolytes by simplifying the multi-dimensional design space to a sub-set of properties that are the most important.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2022

PyTREES

PyTREES (Python tool for Training/Testing Robust Explainable Ensembles on Spectra) is software that implements a data-driven approach to predicting the amount of specific oxides present in materials samples of laser-induced breakdown spectroscopy (LIBS); such as from the ChemCam instrument suite onboard the NASA Curiosity rover. PyTREES is designed to input LIBS data in the format provided by the ChemCam team [1]. PyTREES then applies appropriate pre-processing to this data [2], and implements several regression methods for predicting oxides from spectra. The regression methods include: ensemble methods (random forest, extra trees, and gradient boosting regression) and blended submodels using the “double blending” technique. PyTREES additionally implements methods for quantifying the importance of features in regression model: (1) mean decrease in impurity (MDI) and (2) permutation importance to investigate the wavelengths used by the regression methods. [1] Gasda et al. (2021). Spectrochim Acta B, 181, 106223. [2] Clegg et al. (2017). Spectrochim Acta B , 129, 64–85.

Oyen, Diane↗

DOE OSTI2022

Extracting structural motifs from pair distribution function data of nanostructures using explainable machine learning

Characterization of material structure with X-ray or neutron scattering using e.g. Pair Distribution Function (PDF) analysis most often rely on refining a structure model against an experimental dataset. However, identifying a suitable model is often a bottleneck. Recently, automated approaches have made it possible to test thousands of models for each dataset, but these methods are computationally expensive and analysing the output, i.e. extracting structural information from the resulting fits in a meaningful way, is challenging. Our Machine Learning based Motif Extractor (ML-MotEx) trains an ML algorithm on thousands of fits, and uses SHAP (SHapley Additive exPlanation) values to identify which model features are important for the fit quality. We use the method for 4 different chemical systems, including disordered nanomaterials and clusters. ML-MotEx opens for a type of modelling where each feature in a model is assigned an importance value for the fit quality based on explainable ML.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Effective parameterization of phase-field models of fission gas bubble growth

Fission gas bubbles are one of the most important microstructural features of ceramic nuclear fuels. As gas bubbles grow and interconnect, they allow release of gases, with important consequences for fuel performance. Phase-field modeling has been increasingly used to simulate the evolution of fission gas bubble microstructural because of its capability to capture complex microstructural features. However, computational performance limitations have made it difficult to simulate all the defects present in fuels during operation. For this reason, phase-field models have often simulated only vacancies and used multiple approaches to include the effect of vacancy-interstitial recombination and sinks in a simplified way. Here, we compare some of the most prevalent approaches, including source-only and source/sink. The kinetics of bubble growth using these approaches are analyzed analytically, and simulations with these approaches are compared to a full vacancy-interstitial model.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

DOE OSTI2021

Data Science Meets Physical Organic Chemistry

At the heart of synthetic chemistry is the holy grail of predictable catalyst design. In particular, researchers involved in reaction development in asymmetric catalysis have pursued a variety of strategies toward this goal. This is driven by both the pragmatic need to achieve high selectivities and the inability to readily identify why a certain catalyst is effective for a given reaction. While empiricism and intuition have dominated the field of asymmetric catalysis since its inception, enantioselectivity offers a mechanistically rich platform to interrogate catalyst-structure response patterns that explain the performance of a particular catalyst or substrate. In the early stages of an asymmetric reaction development campaign, the overarching mechanism of the reaction, catalyst speciation, the turnover limiting step, and many other details are unknown or posited based on related reactions. Considering the unclear details leading to a successful reaction, initial enantioselectivity data are often used to intuitively guide the ultimate direction of optimization. However, if the conditions of the Curtin-Hammett principle are satisfied, then measured enantioselectivity can be directly connected to the ensemble of diastereomeric transition states (TSs) that lead to the enantiomeric products, and the associated free energy difference between competing TSs (ΔΔ G ‡ = - RT ln[( S )/( R )], where ( S ) and ( R ) represent the concentrations of the enantiomeric products). We, and others, speculated that this important piece of information can be leveraged to guide reaction optimization in a quantitative way. Although traditional linear free energy relationships (LFERs), such as Hammett plots, have been used to illuminate important mechanistic features, we sought to develop data science derived tools to expand the power of LFERs in order to describe complex reactions frequently encountered in modern asymmetric catalysis. Specifically, we investigated whether enantioselectivity data from a reaction can be quantitatively connected to the attributes of reaction components, such as catalyst and substrate structural features, to harness data for asymmetric catalyst design. In this context, we developed a workflow to relate computationally derived features of reaction components to enantioselectivity using data science tools. The mathematical representation of molecules can incorporate many aspects of a transformation, such as molecular features from substrate, product, catalyst, and proposed transition states. Statistical models relating these features to reaction outputs can be used for various tasks, such as performance prediction of untested molecules. Perhaps most importantly, statistical models can guide the generation of mechanistic hypotheses that are embedded within complex patterns of reaction responses. Overall, merging traditional physical organic experiments with statistical modeling techniques creates a feedback loop that enables both evaluation of multiple mechanistic hypotheses and future catalyst design. In this Account, we highlight the evolution and application of this approach in the context of a collaborative program based on chiral phosphoric acid catalysts (CPAs) in asymmetric catalysis.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

Materials characterization: Can artificial intelligence be used to address reproducibility challenges?

Material characterization techniques are widely used to characterize the physical and chemical properties of materials at the nanoscale and, thus, play central roles in material scientific discoveries. However, the large and complex datasets generated by these techniques often require significant human effort to interpret and extract meaningful physicochemical insights. Artificial intelligence (AI) techniques such as machine learning (ML) have the potential to improve the efficiency and accuracy of surface analysis by automating data analysis and interpretation. In this perspective paper, we review the current role of AI in surface analysis and discuss its future potential to accelerate discoveries in surface science, materials science, and interface science. We highlight several applications where AI has already been used to analyze surface analysis data, including the identification of crystal structures from XRD data, analysis of XPS spectra for surface composition, and the interpretation of TEM and SEM images for particle morphology and size. We also discuss the challenges and opportunities associated with the integration of AI into surface analysis workflows. These include the need for large and diverse datasets for training ML models, the importance of feature selection and representation, and the potential for ML to enable new insights and discoveries by identifying patterns and relationships in complex datasets. Most importantly, AI analyzed data must not just find the best mathematical description of the data, but it must find the most physical and chemically meaningful results. In addition, the need for reproducibility in scientific research has become increasingly important in recent years. The advancement of AI, including both conventional and the increasing popular deep learning, is showing promise in addressing those challenges by enabling the execution and verification of scientific progress. By training models on large experimental datasets and providing automated analysis and data interpretation, AI can help to ensure that scientific results are reproducible and reliable. Although integration of knowledge and AI models must be considered for the transparency and interpretability of models, the incorporation of AI into the data collection and processing workflow will significantly enhance the efficiency and accuracy of various surface analysis techniques and deepen our understanding at an accelerated pace.

Materials Science↗

DOE OSTI2020

3D Printing of Liquid Crystal Elastomer Foams for Enhanced Energy Dissipation Under Mechanical Insult

Polymer foams are an essential class of lightweight materials used to protect assets against mechanical insults, such as shock and vibration. Two features are important to enhance their energy absorption characteristics: the foam structure and the matrix phase mechanical behavior. This study investigates novel approaches to control both of these features to enhance the energy absorption capability of flexible lattice foams. First, we consider 3D printing via digital light processing (DLP) as a method to control the foam mesostructure across a suite of periodic unit cells. Second, we introduce an additional energy dissipation mechanism in the solid matrix phase material by 3D printing the lattice foams with polydomain liquid crystal elastomer (LCE), which undergo a mechanically induced phase transition under large strains. This phase transition is associated with LC mesogen rotation and alignment and provides a second mechanism for mechanical energy dissipation in addition to the viscoelastic relaxation of the polymer network. Here, we contrast the 3D printed LCE lattices with conventional, thermomechanically near-equivalent elastomer lattice foams to quantify the energy-absorbing enhancement the LCE matrix phase provides. Under cyclic quasi-static uniaxial compression conditions, the LCE lattices show dramatically enhanced energy dissipation in uniaxial compression compared to the non-LCE equivalent foams printed with a commercially available photocurable elastomer resin. The lattice geometry also plays a prominent role in determining the energy dissipation ratio between the LCE and non-LCE foams. We show that when increasing the lattice connectivity, the foam deformation transitions from bending-dominated to stretching-dominated deformations, which generates higher axial strains in the struts and higher energy dissipation in the lattice foam, as stretching allows greater mesogen rotation than bending. The LCE foams demonstrate superior energy absorption during the repeated dynamic loading during drop testing compared with the non-LCE equivalent foams, demonstrating the potential of LCEs to enhance physical protection systems against mechanical impact.

3D printing↗