Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Domain adaptation”

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 181 records · Page 10
All sourcesNASA NTRSDOE OSTI
NASA NTRS1997

Multilevel methods for elliptic problems on unstructured grids

Multilevel methods on unstructured grids for elliptic problems are reviewed. The advantages of these techniques are the flexible approximation of the boundaries of complicated physical domains and the ability to adapt the grid to the resolution of fine scaled structures. Multilevel methods, which include multigrid methods and domain decomposition methods, depend on the correct splitting of appropriate finite element spaces. The standard splittings used in the structured grid case cannot be directly extended to unstructured grids due to their requirement for a hierarchical grid structure. Issues related to the application of multilevel methods to unstructured grids are discussed, including how the coarse spaces and transfer operators are defined and how different types of boundary conditions are treated. An obvious way to generate a coarse mesh is to regrid the physical domain several times. Several alternatives are proposed and discussed: node nested coarse spaces, agglomerated coarse spaces and algebraically generated coarse spaces.

Chan, Tony F.↗
DOE OSTI2020

Data-driven materials research enabled by natural language processing and information extraction

Given the emergence of data science and machine learning throughout all aspects of society, but particularly in the scientific domain, there is increased importance placed on obtaining data. Data in materials science are particularly heterogeneous, based on the significant range in materials classes that are explored and the variety of materials properties that are of interest. This leads to data that range many orders of magnitude, and these data may manifest as numerical text or image-based information, which requires quantitative interpretation. The ability to automatically consume and codify the scientific literature across domains - enabled by techniques adapted from the field of natural language processing - therefore has immense potential to unlock and generate the rich datasets necessary for data science and machine learning. This review focuses on the progress and practices of natural language processing and text mining of materials science literature and highlights opportunities for extracting additional information beyond text contained in figures and tables in articles. Here, we discuss and provide examples for several reasons for the pursuit of natural language processing for materials, including data compilation, hypothesis development, and understanding the trends within and across fields. Current and emerging natural language processing methods along with their applications to materials science are detailed. We, then, discuss natural language processing and data challenges within the materials science domain where future directions may prove valuable.

36 MATERIALS SCIENCE↗
NASA NTRSDate not supplied

Towards an Aviation Large Language Model by Fine-tuning and Evaluating Transformers

In the aviation domain, there are many applications for machine learning and artificial intelligence tools that utilize natural language. For example, there is a desire to know the commonalities in written safety reports such as voluntary post incidents reports or aerial wildfire operations reports to better understand the risks present. Another use-case is the possibility of extracting airspace procedures and constraints currently written in documents such as Letters of Agreement. These applications can benefit from the use of state-of-the-art natural language processing techniques when adapted to the language/phraseology specific to the aviation domain. This paper evaluates the viability of adaptation of NLP tools to the aviation domain by fine-tuning transformer based models using aviation data sets. In 2018, a novel language model based on neural units (also called transformers) was created and became known as “Bidirectional Encoder Representations from Transformers” or BERT. This architecture combined with large amounts of English training data and innovative semi-supervised training tasks set the standard for what would later emerge as Large Language Models. The performance of these models was further improved by hyperparameter tuning and refinement of the semi-supervised training task and resulted in “Robustly Optimized BERT Pre-training Approach through hyperparameter tuning” or RoBERTa models. These pre-trained Large Language Models proved to be useful for a wide variety of natural language processing tasks such as text classification and question answering through a process called fine-tuning. The transformer architecture with pre-trained weights served as the basis with the last few layers replaced with layers fine-tuned to perform a new task e.g., a layer that provides a label for the entire input text. This process of fine-tuning can also be used to adapt the models to new domains; e.g., BioBERT started with the pre-trained BERT model and was completed by additional fine-tuning and training on biomedical documents. Transformer-based architectures can also be used to create rich representations of text called embeddings which can serve as the input to other machine learning models. This allows simpler algorithms such as logistic regression to use context-rich representations of the text while still remaining quick to train and evaluate. In the world of aviation, there is a growing demand for natural language processing and understanding but the domain presents unique challenges. Due to the technical content (and specialized language) of most aviation documents, fine-tuning pre-trained Large Language Models to specific tasks has not met the benchmark on natural language processing tasks set by simpler models trained from scratch on the data. To address this deficiency, this paper evaluates the improvements from fine-tuning a Large Language Model on a large set of aviation documents using the original semi-supervised training tasks before performing specific natural language tasks. In fine-tuning, a domain-specific dataset is used on the original training task but with the pre-trained Large Language Model instead of starting from a random initialization. This approach allows the model to be adapted to the specific domain language without discarding the information gained from training on general English data. This paper utilized two major dataset types to train and assess the RoBERTa fine-tuning performance. The first are 7,057 Letters of Agreement which are Federal Aviation Administration (FAA) documents that formalize airspace operations across the national airspace system. They contain many examples of ‘aviation English’ using domain specific terminology and phrasing which serves as a representative basis to perform the semi-supervised fine-tuning. The second type is the 494 document classification labels to be used for evaluation. This down-stream evaluation aims to show the performance of the fine-tuned model, better understand how much data is needed for an effective fine-tuning, and how fine-tuning can be adapted for different applications in-the domain. After semi-supervised training, evaluation begins by encoding the documents for classification using the fine-tuned RoBERTa model. Then a logistic regression classifier is trained to label the document type and compared against our ground truth labels. This currently leads to a 82.8% accuracy on 10-fold cross validation showing improvement over baseline RoBERTa which achieved 81.0%. We plan to measure the improvements on additional tasks and it is expected that these improvements will lead to more robust models that can tackle the natural language processing challenges present in aviation datasets.

ATM↗
DOE OSTI2024

Investigating explainable transfer learning for battery lifetime prediction under state transitions

Battery lifetime prediction at early cycles is crucial for researchers and manufacturers to examine product quality and promote technology development. Machine learning has been widely utilized to construct data-driven solutions for high-accuracy predictions. However, the internal mechanisms of batteries are sensitive to many factors, such as charging/discharging protocols, manufacturing/storage conditions, and usage patterns. These factors will induce state transitions, thereby decreasing the prediction accuracy of data-driven approaches. Transfer learning is a promising technique that overcomes this difficulty and achieves accurate predictions by jointly utilizing information from various sources. Hence, we develop two transfer learning methods, Bayesian Model Fusion and Weighted Orthogonal Matching Pursuit, to strategically combine prior knowledge with limited information from the target dataset to achieve superior prediction performance. From our results, our transfer learning methods reduce root-mean-squared error by 41% through adapting to the target domain. Furthermore, the transfer learning strategies identify the variations of impactful features across different sets of batteries and therefore disentangle the battery degradation mechanisms and the root cause of state transitions from the perspective of data mining. These findings suggest that the transfer learning strategies proposed in our work are capable of acquiring knowledge across multiple data sources for solving specialized issues.

25 ENERGY STORAGE↗
DOE OSTI2022

Computational Estimation by Scientific Data Mining with Classical Methods to Automate Learning Strategies of Scientists

Experimental results are often plotted as 2-dimensional graphical plots (aka graphs) in scientific domains depicting dependent versus independent variables to aid visual analysis of processes. Repeatedly performing laboratory experiments consumes significant time and resources, motivating the need for computational estimation. The goals are to estimate the graph obtained in an experiment given its input conditions, and to estimate the conditions that would lead to a desired graph. Existing estimation approaches often do not meet accuracy and efficiency needs of targeted applications. We develop a computational estimation approach called AutoDomainMine that integrates clustering and classification over complex scientific data in a framework so as to automate classical learning methods of scientists. Knowledge discovered thereby from a database of existing experiments serves as the basis for estimation. Challenges include preserving domain semantics in clustering, finding matching strategies in classification, striking a good balance between elaboration and conciseness while displaying estimation results based on needs of targeted users, and deriving objective measures to capture subjective user interests. These and other challenges are addressed in this work. The AutoDomainMine approach is used to build a computational estimation system, rigorously evaluated with real data in Materials Science. Our evaluation confirms that AutoDomainMine provides desired accuracy and efficiency in computational estimation. It is extendable to other science and engineering domains as proved by adaptation of its sub-processes within fields such as Bioinformatics and Nanotechnology.

Computer Science↗
DOE OSTI2025

Modular Autonomous Experimentation for Biological Applications (Full Report)

The Modular Autonomous Research System (MARS) was developed to address the pressing need for faster, more reliable, and more adaptable scientific discovery. Traditional experimentation is limited by manual labor, long cycle times, and fragmented data streams, which constrain the ability to explore complex chemical and materials design spaces. To overcome these limitations, we created an integrated, modular platform that combines laboratory robotics, diverse measurement instruments, and a central data infrastructure with artificial intelligence–driven decision-making. The system links liquid handling robots, robotic arms, and optical plate readers into a closed loop where experiments are executed automatically, data is analyzed in real time, and subsequent experimental conditions are adaptively chosen to maximize information gain. Over the course of the project, MARS was validated on two primary test cases—spectroscopic metal–ligand binding assays and peptide-directed mineralization—which highlighted the system’s ability to handle uncertainty and variability in experimental measurements. To further demonstrate modularity and extensibility, we also established additional testbeds in electrochemistry for catalyst discovery and electrolyte formulation for advanced batteries. The results show that MARS can reliably conduct autonomous campaigns with minimal human intervention, adapt to distinct scientific domains, and provide a scalable model for future self-driving laboratories. This work establishes new capabilities for modular, uncertainty-aware automation and directly supports the need for advanced, data-driven research platforms capable of accelerating discovery across a wide range of scientific and national security missions.

59 BASIC BIOLOGICAL SCIENCES↗
NASA NTRS1987

Understanding natural language for spacecraft sequencing

The paper describes a natural language understanding system, START, that translates English text into a knowledge base. The understanding and the generating modules of START share a Grammar which is built upon reversible transformations. Users can retrieve information by querying the knowledge base in English; the system then produces an English response. START can be easily adapted to many different domains. One such domain is spacecraft sequencing. A high-level overview of sequencing as it is practiced at JPL is presented in the paper, and three areas within this activity are identified for potential application of the START system. Examples are given of an actual dialog with START based on simulated data for the Mars Observer mission.

Katz, Boris↗
NASA NTRS1992

Active control of sound transmission/radiation from elastic plates by vibration inputs. II - Experiments

Actively controlled harmonic force inputs were applied experimentally to reduce the sound transmitted through an elastic circular plate. The control implementation used a time domain least mean square adaptive algorithm with two error sensors. The control forces were applied directly to the plate by point force vibration inputs, while the error information and performance were measured in the radiated acoustic field by microphones. Test cases were also performed in which the error sensors were accelerometers mounted on the plate. When accelerometers were used as error sensors, the controller performance was degraded; leading to the conclusion that minimizing plate motion does not necessarily lead to an associated decrease in radiated sound levels. In contrast, the results show excellent attenuation of the transmitted sound field when microphone error sensors were used. This result was consistent over a range of frequencies. In addition, the experimental results are compared to previously derived analytical results and the effect of using a point or global minimization scheme is discussed.

Metcalf, V. L.↗
NASA NTRS1994

An Approach for Dynamic Grids

An approach is presented for the generation of two-dimensional, structured, dynamic grids. The grid motion may be due to the motion of the boundaries of the computational domain or to the adaptation of the grid to the transient, physical solution. A time-dependent grid is computed through the time integration of the grid speeds which are computed from a system of grid speed equations. The grid speed equations are derived from the time-differentiation of the grid equations so as to ensure that the dynamic grid maintains the desired qualities of the static grid. The grid equations are the Euler-Lagrange equations derived from a variational statement for the grid. The dynamic grid method is demonstrated for a model problem involving boundary motion, an inviscid flow in a converging-diverging nozzle during startup, and a viscous flow over a flat plate with an impinging shock wave. It is shown that the approach is more accurate for transient flows than an approach in which the grid speeds are computed using a finite difference with respect to time of the grid. However, the approach requires significantly more computational effort.

Slater, John W.↗
NASA NTRS1999

Flight Test of ASAC Aircraft Interior Noise Control System

A flight test is described in which an active structural/acoustic control system reduces turboprop induced interior noise on a Raytheon Aircraft Company 1900D airliner. Control inputs to 21 inertial force actuators were computed adaptively using a transform domain version of the multichannel filtered-X LMS algorithm to minimize the mean square response of 32 microphones. A combinatorial search algorithm was employed to optimize placement of the force actuators on the aircraft frame. Both single frequency and multi-frequency results are presented. Reductions of up to 15 dB were obtained at the blade passage frequency (BPF) during single frequency control tests. Simultaneous reductions of the BPF and next 2 harmonics of 10 dB, 2.5 dB and 3.0 dB, were obtained in a multi-frequency test.

Palumbo, Dan↗
NASA NTRS2007

Blurring the Inputs: A Natural Language Approach to Sensitivity Analysis

To document model parameter uncertainties and to automate sensitivity analyses for numerical simulation codes, a natural-language-based method to specify tolerances has been developed. With this new method, uncertainties are expressed in a natural manner, i.e., as one would on an engineering drawing, namely, 5.25 +/- 0.01. This approach is robust and readily adapted to various application domains because it does not rely on parsing the particular structure of input file formats. Instead, tolerances of a standard format are added to existing fields within an input file. As a demonstration of the power of this simple, natural language approach, a Monte Carlo sensitivity analysis is performed for three disparate simulation codes: fluid dynamics (LAURA), radiation (HARA), and ablation (FIAT). Effort required to harness each code for sensitivity analysis was recorded to demonstrate the generality and flexibility of this new approach.

Kleb, William L.↗
NASA NTRS2012

A Comparison of Declarative and Hybrid Declarative-Procedural Models for Rover Operations

The MAPGEN [2] (Mixed-initiative Activity Plan GENerator) planning system is a great example of a hybrid procedural/declarative system where the advantages of each are leveraged to produce an effective planner/scheduler for Mars Exploration Rover tactical planning. We explore the adaptation of the same domain to an entirely declarative planning system (ASPEN [4] Activity Scheduling and Planning ENvironment), and demonstrate that, with some translation, much of the procedural knowledge encoding is amenable to a declarative knowledge encoding.

declarative domain representations↗
NASA NTRS2018

Robust Metric-Aligned Quad-Dominant Meshing Using L(sub p) Centroidal Voronoi Tessellation

We introduce a meshing algorithm that can be used to both generate and adapt meshes for bounded domains in an anisotropic manner. This is particularly beneficial when anisotropic flow features like shock waves or contact discontinuities are present in the computational domain. The algorithm presented in this paper is based upon meshing under the imposed Riemannian metric tensor, which controls the orientation and size of the mesh elements. In this way there is no need for user intervention to recognize these features. We demonstrate that the method indeed aligns the elements with the underlying metric and produces right-angled simplices that can be recombined into quadrilateral elements. The aim is to eventually incorporate this meshing strategy in the monolithic high-order spectral element solver that is currently being developed at NASA Ames. This paper has two main contributions: First, we demonstrate that we can generate quad-dominant metric-aligned meshes for bounded domains using a generalized form of L(sub p)-Centroidal Voronoi Tessellation (L(sub p)-CVT). Unlike previous works, we do not rely on a background mesh and discretize the bounded domain in a hierarchical way by first discretizing the boundaries and then the volume using the underlying metric. Second, we present an alternative for clipping the Voronoi cells on the boundary, which is common practice in CVT-based meshing algorithms, by reconstructing the Voronoi cells using the defined metric field. In this way we avoid the geometrical complexity of the clipping procedure and we show that we evaluate the energy and its gradient correctly. We show that the reconstruction of the computational domain is consistent with the Lloyds’ algorithm that is used to compute the L(sub p)-CVT.

Centroidal↗
NASA NTRS2020

Famine Early Warning Systems Network (FEWS NET) Land Data Assimilation System (LDAS) and Other Assimilated Hydrological Data at NASA GES DISC

The NASA Goddard Earth Sciences Data and Information Services Center (GES DISC) provides science support for several data sets relevant to agriculture and food security, including the Famine Early Warning Systems Network (FEWS NET) Land Data Assimilation System (LDAS), or FLDAS data set. The GES DISC is one of twelve NASA Earth Observing System (EOS) data centers that process, archive, document, and distribute data from Earth science missions and related projects. The GES DISC hosts a wide range of remote sensing and model data, and provides reliable and robust data access and other services to users worldwide. Beyond data archive and access, the GES DISC offers many services to visualize and analyze the data. This presentation provides a summary of the hydrological data available at the GES DISC, along with an overview of related data services. Specifically, the FLDAS data set has been adapted to work with domains, data streams, and monitoring and forecast requirements associated with food security assessment in data-sparse, developing country settings. The FLDAS global monthly data have a 0.1 x 0.1 degree spatial resolution covering the period from January 1982 to present. Global FLDAS monthly anomaly and monthly climatology data are also available at the GES DISC to evaluate how current conditions compare to averages over the FLDAS 35-year period. Several case studies using the FLDAS soil moisture, evapotranspiration, rainfall, runoff, and surface temperature data will be presented.

Loeser, Carlee↗
DOE OSTI2020

Single Episode Policy Transfer in Reinforcement Learning

Transfer and adaptation to new unknown environmental dynamics is a key challenge for reinforcement learning (RL). An even greater challenge is performing near-optimally in a single attempt at test time, possibly without access to dense rewards, which is not addressed by current methods that require multiple experience rollouts for adaptation. To achieve single episode transfer in a family of environments with related dynamics, we propose a general algorithm that optimizes a probe and an inference model to rapidly estimate underlying latent variables of test dynamics, which are then immediately used as input to a universal control policy. This modular approach enables integration of state-of-the-art algorithms for variational inference or RL. Moreover, our approach does not require access to rewards at test time, allowing it to perform in settings where existing adaptive approaches cannot. In diverse experimental domains with a single episode test constraint, we report our method significantly outperforms existing adaptive approaches and shows favorable performance against baselines for robust transfer.

42 ENGINEERING↗
NASA NTRS1999

Performing Verification and Validation in Reuse-Based Software Engineering

The implementation of reuse-based software engineering not only introduces new activities to the software development process, such as domain analysis and domain modeling, it also impacts other aspects of software engineering. Other areas of software engineering that are affected include Configuration Management, Testing, Quality Control, and Verification and Validation (V&V). Activities in each of these areas must be adapted to address the entire domain or product line rather than a specific application system. This paper discusses changes and enhancements to the V&V process, in order to adapt V&V to reuse-based software engineering.

Addy, Edward A.↗
NASA NTRS2008

Parallel Anisotropic Tetrahedral Adaptation

An adaptive method that robustly produces high aspect ratio tetrahedra to a general 3D metric specification without introducing hybrid semi-structured regions is presented. The elemental operators and higher-level logic is described with their respective domain-decomposed parallelizations. An anisotropic tetrahedral grid adaptation scheme is demonstrated for 1000-1 stretching for a simple cube geometry. This form of adaptation is applicable to more complex domain boundaries via a cut-cell approach as demonstrated by a parallel 3D supersonic simulation of a complex fighter aircraft. To avoid the assumptions and approximations required to form a metric to specify adaptation, an approach is introduced that directly evaluates interpolation error. The grid is adapted to reduce and equidistribute this interpolation error calculation without the use of an intervening anisotropic metric. Direct interpolation error adaptation is illustrated for 1D and 3D domains.

Park, Michael A.↗