Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “BERT”

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 91 records · Page 5

What Went Wrong: A Survey of Wildfire UAS Mishaps through Named Entity Recognition

Increasingly, unmanned aircraft systems (UAS) are being applied to wildfire incidents for tasks such as mapping, aerial ignition, and delivery. As a result, incident reporting systems for wildfires are beginning to accumulate data related to UAS mishaps in wildfire response. In this research, we apply state-of-the-art natural language processing (NLP) techniques to develop a custom Named Entity Recognition (NER) model which extracts a Failure Modes and Effects Analysis (FMEA)-style survey of wildfire UAS mishaps reported in SAFECOM. The custom NER model is built by fine-tuning an existing (BERT) model, resulting in a generalizable NER model that can extract engineering relevant entities including failure modes, causes, effects, control processes, and recommendations from any failure-relevant text. Similar mishaps are clustered and reported as single rows within the FMEA. For each cluster, frequency, severity, and overall risk are computed. The methodology can be applied as part of a broader safety management system to track trends in mishaps and discover knowledge that can be utilized to improve safety outcomes and system performance.

Machine Learning↗

What Went Wrong: A Survey of Wildfire UAS Mishaps through Named Entity Recognition

Increasingly, unmanned aircraft systems (UAS) are being applied to wildfire incidents for tasks such as mapping, aerial ignition, and delivery. As a result, aviation incident reporting systems for wildfires are beginning to accumulate data related to UAS mishaps in wildfire response. In this research, we apply state-of-the-art natural language processing (NLP) techniques to develop a custom Named Entity Recognition (NER) model which extracts entities relevant to safety analysts. The custom NER model is built by fine-tuning an existing Bidirectional Encoder Representations from Transformers (BERT) model, resulting in a generalizable NER model that can extract engineering relevant entities including failure modes, causes, effects, control processes, and recommendations from failure-relevant text. This model performs passably, with a weighted average f1 score of 0.33 across entity types, indicating more labeled training data is needed. Extracted entities are used to form a Failure Modes and Effects Analysis (FMEA)-style survey of wildfire UAS mishaps reported using the SAFECOM system. Similar mishaps are manually clustered and reported as single rows within an FMEA. Foreach cluster, we compute frequency, severity, and overall riskin accordance with FAA standards. This methodology can beapplied as part of a broader safety management system totrack trends in mishaps (e.g., likelihood, severity) and discoverknowledge (e.g., causes, effects) that can be utilized to improvesafety outcomes and system performance.

Machine Learning↗

What Went Wrong: A Survey of Wildfire UAS Mishaps through Named Entity Recognition

Increasingly, unmanned aircraft systems (UAS) are being applied to wildfire incidents for tasks such as mapping, aerial ignition, and delivery. As a result, incident reporting systems for wildfires are beginning to accumulate data related to UAS mishaps in wildfire response. In this research, we apply state-of-the-art natural language processing (NLP) techniques to develop a custom Named Entity Recognition (NER) model which extracts a Failure Modes and Effects Analysis (FMEA)-style survey of wildfire UAS mishaps reported in SAFECOM. The custom NER model is built by fine-tuning an existing (BERT) model, resulting in a generalizable NER model that can extract engineering relevant entities including failure modes, causes, effects, control processes, and recommendations from any failure-relevant text. Similar mishaps are clustered and reported as single rows within the FMEA. For each cluster, frequency, severity, and overall risk are computed. The methodology can be applied as part of a broader safety management system to track trends in mishaps and discover knowledge that can be utilized to improve safety outcomes and system performance.

Machine Learning↗

ForQint

ForQint is a Fortran software library to adaptively integrate multivariable functions versus well known probability distributions (e.g. normal, lognormal, etc.) using quadrature rules. Such integration is known to be more efficient and accurate especially when the function evaluation is expensive. The library also FQint enables the creation of both full and sparse anisotropic multi-dimensional quadrature rules. These latter encompass different such as Gauss-Hermite, Gauss-Legendre, etc. The functions Maher Salloum, Bert Debusschere, Kenny Chowdhary and subroutines rely on the C++ Stokhos package of Trilinos (trilinos.sandia.gov) where the appropriate classes were wrapped into Fortran.

Source record↗

Evaluating the Use of Foundational Chemical Language Models in Multimodal Graph Fusion

Rapid and accurate prediction of the physicochemical properties of molecules given their structures remains a key challenge in cheminformatics. Machine learning approaches offer high-throughput options, but the optimality of inductive biases and data representations are up for debate. For example, BERT-based masked language models (MLMs) can be trained in a self-supervised way on hundreds of millions to billions of readily available SMILES strings. Another option is graph neural networks (GNNs), which can operate directly on molecular structures. Yet, generating accurate molecular geometry is computationally expensive, leading to a relative scarcity in data compared to SMILES strings. It is attractive to combine these two paradigms by pre-training an LM on a large corpus of SMILES strings and embedding these representation into a geometric graph neural network. Despite the promise of such an approach, and contrary to previous studies, we find mixed results with the combination of the LMs and GNNs on several molecule datasets. In particular, we found evidence for improvement on the FreeSolv and QM7 benchmarks, but degraded performance on the ESOL, LIPO and QM9 datasets compared to a GNN baseline.

Francel, Collin [University of Alabama]↗

>24% screen printed Cu contacted n-TOPCon solar cells with successful implementation of LECO process

In this paper, we report the successful fabrication of >24.0 % efficiency n-TOPCon Si solar cells with screen-printed, fire-through Cu contact to n-TOPCon on the rear side and Ag contacted boron emitter on the front side by implementing optimized firing and LECO conditions. The highest efficiency (24.3%) Cu contacted n-TOPCon cell in this study showed excellent cell performance parameters with V oc >730 mV, J sc of 41.1 mA/cm 2 and FF of 80.8%, resulting in an absolute efficiency gap of 0.2% between Cu-contacted and fully Ag contacted n-TOPCon cells (24.5%). The mini-module fabricated with the Cu contacted n-TOPCon cell showed excellent reliability and durability of open-circuit voltage (V oc ), pseudo fill factor (pFF) and efficiency after prolonged damp-heat tests. Such high efficiency screen printed Cu contacted n-TOPCon cells provide unique opportunity to replace very expensive Ag contact on n-TOPCon with cheaper screen printable Cu metal pastes.

14 SOLAR ENERGY↗

Thermally Stable and Regenerable Platinum–Tin Clusters for Propane Dehydrogenation Prepared by Atom Trapping on Ceria

Abstract Ceria (CeO 2 ) supports are unique in their ability to trap ionic platinum (Pt), providing exceptional stability for isolated single atoms of Pt. The reactivity and stability of single‐atom Pt species was explored for the industrially important light alkane dehydrogenation reaction. The single‐atom Pt/CeO 2 catalysts are stable during propane dehydrogenation, but are not selective for propylene. DFT calculations show strong adsorption of the olefin produced, leading to further unwanted reactions. In contrast, when tin (Sn) is added to CeO 2 , the single‐atom Pt catalyst undergoes an activation phase where it transforms into Pt–Sn clusters under reaction conditions. Formation of small Pt–Sn clusters allows the catalyst to achieve high selectivity towards propylene because of facile desorption of the product. The CeO 2 ‐supported Pt–Sn clusters are very stable, even during extended reaction at 680 °C. Coke formation is almost completely suppressed by adding water vapor to the feed. Furthermore, upon oxidation the Pt–Sn clusters readily revert to the atomically dispersed species on CeO 2 , making Pt–Sn/CeO 2 a fully regenerable catalyst.

Xiong, Haifeng↗

On the Limited Role of Electronic Support Effects in Selective Alkyne Hydrogenation: A Kinetic Study of Au/MO x Catalysts Prepared from Oleylamine‐Capped Colloidal Nanoparticles

Abstract We report a quantitative kinetic evaluation and study of support effects for partial alkyne hydrogenation using oleylamine‐capped Au colloids as catalyst precursors. The amine capping agents can be removed under reducing conditions, generating supported Au nanoparticles of ∼2.5 nm in diameter. The catalysts showed high alkene selectivity (>90 %) at all conversions during alkyne partial hydrogenation. Catalytic activity, observed rate constants, and apparent activation energies (25–40 kJ/mol) were similar for all Au catalysts, indicating support effects are relatively small. Alkyne adsorption, probed with FTIR and DFT, showed adsorption on the support was associated with hydrogen‐bonding interactions. DFT calculations indicate strong alkyne adsorption on Au sites, with the strongest adsorption sites at the metal‐support interface (MSI). The catalysts had similar hydrogen reaction orders (0.7–0.9), and 1‐octyne reaction orders (∼−0.2), suggesting a common mechanism. The reaction kinetics are most consistent with a mechanism involving the non‐competitive activated adsorption of H 2 on an alkyne‐covered Au surface.

Bruno, James E.↗

Recommendations for improving rigor and reproducibility in site specific characterization

Heterogeneous catalysis is driven by the interaction of reactant molecules and the catalyst surface. The locus of this interaction as well as the surrounding ensemble of atoms is referred to as the catalyst active site. Active site characterization attempts to distinguish active catalytic sites from inactive surface sites, to elucidate the structural and chemical nature of active sites, and to quantify active site concentration. Numerous techniques have been demonstrated to provide compositional and structural information about the active sites within a catalyst. However, each technique has its own limitations and experimental pitfalls that can lead to data misinterpretation or irreproducible results. Further, this work aims to provide an overview of the types of data that can be collected, to outline common experimental challenges and how to avoid them, and to assemble relevant references for the most used active site characterization techniques. More broadly, we aim to outline best practices for researchers to collect, interpret, and report active site characterization data in a way that provides the most benefit to the broader catalysis community. Increasing the rigor and reproducibility of active site characterization offers a strategy to better link properties with catalytic performance and to enable the community to develop consensus concerning these relationships.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Hybrid discrete/continuum algorithms for stochastic reaction networks

Direct solutions of the Chemical Master Equation (CME) governing Stochastic Reaction Networks (SRNs) are generally prohibitively expensive due to excessive numbers of possible discrete states in such systems. To enhance computational efficiency we develop a hybrid approach where the evolution of states with low molecule counts is treated with the discrete CME model while that of states with large molecule counts is modeled by the continuum Fokker–Planck equation. The Fokker–Planck equation is discretized using a 2nd order finite volume approach with appropriate treatment of flux components. The numerical construction at the interface between the discrete and continuum regions implements the transfer of probability reaction by reaction according to the stoichiometry of the system. The performance of this novel hybrid approach is explored for a two-species circadian model with computational efficiency gains of about one order of magnitude.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Impact of Ionic Liquid on Lithium Ion Battery with a Solid Poly(Ionic Liquid) Pentablock Terpolymer as Electrolyte and Separator

In this work, a ternary blend solid polymer electrolyte (SPE) consisting of a poly(ionic liquid) (PIL) multiblock polymer, lithium salt, and ionic liquid (IL) was investigated with the goal of understanding the influence of IL concentration on ion transport mechanisms and electrochemical stability. The physical, transport, mechanical, morphological, and electrochemical properties of the ternary blend SPE were systematically investigated as a function of IL concentration (r = [IL]/[PIL] mole ratio between 0.1 and 0.7). The results indicate that with increasing IL concentration, the continuous conductive domain increases along with the polymer chain segmental mobility, as well as facilitate ionic conductivity, while the mechanical modulus exhibits a percolation threshold (from 52.17 MPa to 0.55 MPa at r = 0.2 to 0.3). Surprisingly, at higher IL concentrations, there is a reduction in the lithium cation mobility (i.e., 9.3 × 10-12 m2 s-1 versus 3.8 × 10-12 m2 s-1 for r = 0.2 versus r = 0.5, respectively) as evidenced by pulsed-field gradient nuclear magnetic resonance (PFG-NMR), which coincides with an increased overpotential (i.e., 50 mV versus 150 mV for r = 0.2 versus r = 0.5, respectively) evidenced by lithium metal stripping and plating. This work provides valuable insights into the impact of IL on lithium ion conducting PIL block polymers, which may enable the design of new SPEs with both high ionic conductivity and improved stability for the future lithium ion batteries.

Chen, Tzu-Ling↗