Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Domain knowledge”

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 631 records · Page 35
All sourcesNASA NTRSDOE OSTI
NASA NTRS1992

Advanced training systems for the next decade and beyond

Autonomous training systems grouped under the term intelligent computer-aided training (ICAT) are described with references to their intelligent technologies. An ICAT system was developed for the deployment of the payload-assist module satellites from the Space Shuttle, fault detection in vacuum-vent lines on Spacelab, main-propulsion pneumatics, and for instrument pointing systems. The basic ICAT architecture is delineated with descriptions of elements such as the user interface, domain expert, traning-session manager, trainee model, and training-scenario generator. The ICAT systems employ advanced technologies including knowledge acquisition, virtual environments, and fuzzy logic. The general architecture of the ICAT systems is shown to be fully developed and applicable to the construction of seven different tasks, and the ICAT systems are concluded to enhance the training of aerospace personnel.

Loftin, R. B.↗
NASA NTRS1993

Frequency-Channelized Recovery Of Timing

Method of extracting bit or baud rate of excess bandwidth pulse-amplitude-modulated signal based on frequency-domain equivalent of time-domain delay-and-multiply operation. Involves use of outputs of Fourier-transform-based, channelized receiver. Very little priori knowledge of parameters of signal needed, and processing of signal by method yields spectral line at pulse-repetition rate.

Satorius, Edgar H.↗
NASA NTRS1993

Management of Knowledge Representation Standards Activities

Ever since the mid-seventies, researchers have recognized that capturing knowledge is the key to building large and powerful AI systems. In the years since, we have also found that representing knowledge is difficult and time consuming. In spite of the tools developed to help with knowledge acquisition, knowledge base construction remains one of the major costs in building an Al system: For almost every system we build, a new knowledge base must be constructed from scratch. As a result, most systems remain small to medium in size. Even if we build several systems within a general area, such as medicine or electronics diagnosis, significant portions of the domain must be represented for every system we create. The cost of this duplication of effort has been high and will become prohibitive as we attempt to build larger and larger systems. To overcome this barrier we must find ways of preserving existing knowledge bases and of sharing, re-using, and building on them. This report describes the efforts undertaken over the last two years to identify the issues underlying the current difficulties in sharing and reuse, and a community wide initiative to overcome them. First, we discuss four bottlenecks to sharing and reuse, present a vision of a future in which these bottlenecks have been ameliorated, and describe the efforts of the initiative's four working groups to address these bottlenecks. We then address the supporting technology and infrastructure that is critical to enabling the vision of the future. Finally, we consider topics of longer-range interest by reviewing some of the research issues raised by our vision.

Patil, Ramesh S.↗
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

Understanding the Physics Representation of Deep Learning Models in Environmental Applications

Deep learning (DL) models have been popular in earth and environmental modeling and analysis, which exhibit huge potential in capturing and reconstructing the non-linearity of relevant environmental processes. They are extensively used as analytical tools or emulators for multiple domains (atmosphere, land surface, ocean, and biogeochemistry). Despite their success, their internal working mechanism remains largely unknown. Such a lack of knowledge hinders the identification of physically consistent models that are fully adaptive to non-stationary climate, as well as the development of physics-informed machine learning such as physics-informed neural network (PINN). To establish preliminary knowledge and framework of such physics representation evaluation, this project focuses on an improved understanding of DL models in the environmental applications. DL models are increasingly applied to environmental modeling and prediction. However, they have been evaluated mostly from a performance perspective, and there is a gap in understanding how they represent the known physics internally. Such knowledge is especially critical when applying DL models under climate change conditions, where new inputs are likely outside the ranges of the training datasets. In this project, we reveal how the known physical processes are represented within DL models from both statistical and mechanistic perspectives. Leveraging the traditional model evaluations that focus more on the accuracies of predictions, we establish a framework that examines both the accuracy and physics representation of DL models. This analysis framework can identify DL models that make the correct predictions based on correct physics, thus enhancing the existing explainable artificial intelligence (explainable-AI) portfolio. It lays a foundation for developing novel metrics to evaluate the emerging DL models in environmental applications. This knowledge also informs the development of physics-informed DL models by revealing the direct connections between the known physical processes and specific model components or structures.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS1996

Program Simulates A Modular Manufacturing System

SSE computer program provides simulation environment for modeling manufacturing systems containing relatively small numbers of stations and operators. Designed to simulate manufacturing of apparel, also used in other manufacturing domains. Excellent for small or medium-size firms including those lacking expertise to develop detailed models or have only minimal knowledge in describing manufacturing systems and in analyzing results of simulations on mathematical models. User does not need to know simulation language to use SSE. Used to design new modules and to evaluate existing modules. Originally written in Turbo C v2.0 for IBM PC-compatible computers running MS-DOS and successfully implemented by use of Turbo C++ v3.0.

Schroer, Bernard J.↗
NASA NTRS1996

Induction as Knowledge Integration

Two key issues for induction algorithms are the accuracy of the learned hypothesis and the computational resources consumed in inducing that hypothesis. One of the most promising ways to improve performance along both dimensions is to make use of additional knowledge. Multi-strategy learning algorithms tackle this problem by employing several strategies for handling different kinds of knowledge in different ways. However, integrating knowledge into an induction algorithm can be difficult when the new knowledge differs significantly from the knowledge the algorithm already uses. In many cases the algorithm must be rewritten. This paper presents Knowledge Integration framework for Induction (KII), a KII, that provides a uniform mechanism for integrating knowledge into induction. In theory, arbitrary knowledge can be integrated with this mechanism, but in practice the knowledge representation language determines both the knowledge that can be integrated, and the costs of integration and induction. By instantiating KII with various set representations, algorithms can be generated at different trade-off points along these dimensions. One instantiation of KII, called RS-KII, is presented that can implement hybrid induction algorithms, depending on which knowledge it utilizes. RS-KII is demonstrated to implement AQ-11, as well as a hybrid algorithm that utilizes a domain theory and noisy examples. Other algorithms are also possible.

Smith, Benjamin D.↗
NASA NTRS2018

Toward Justifiable Trust in Autonomous Systems Incorporating Human Knowledge in Autonomous Systems through Machine Learning

Trust in Autonomous Systems is largely about humans trusting the decisions made by autonomous systems. This trust can be increased through learning from domain experts. In particular, autonomous systems can learn offline from past mission operations before conducting any operations of its own. Additionally, autonomous systems can learn online by obtaining human feedback during operations. We will discuss several classes of machine learning methods and our application of them to autonomous systems. The first class of methods is anomaly detection, which uses operations data to identify examples of anomalous operations. The second class of methods is inverse reinforcement learning, also known as apprenticeship learning, that takes past operations data as input and yields a controller that is able to duplicate the operations described by the data. The third class is active learning, which identifies examples on which the model is most uncertain and requests domain expert feedback.

Oza, Nikunj C.↗
NASA NTRS2004

Designing Domain-Specific HUMS Architectures: An Automated Approach

The HUMS automation system automates the design of HUMS architectures. The automated design process involves selection of solutions from a large space of designs as well as pure synthesis of designs. Hence the whole objective is to efficiently search for or synthesize designs or parts of designs in the database and to integrate them to form the entire system design. The automation system adopts two approaches in order to produce the designs: (a) Bottom-up approach and (b) Top down approach. Both the approaches are endowed with a Suite of quantitative and quantitative techniques that enable a) the selection of matching component instances, b) the determination of design parameters, c) the evaluation of candidate designs at component-level and at system-level, d) the performance of cost-benefit analyses, e) the performance of trade-off analyses, etc. In short, the automation system attempts to capitalize on the knowledge developed from years of experience in engineering, system design and operation of the HUMS systems in order to economically produce the most optimal and domain-specific designs.

Mukkamala, Ravi↗
NASA NTRS2026

Polarized Resonant Soft X-ray Scattering (P-RSoXS) as a New Technique for Characterizing Amorphous Astromaterials

In the coming years, samples will be returned from several asteroids, the lunar surface, and the first material returned directly from the Martian surface. Previous in-situ and remote studies indicate these samples will contain abundant amorphous or weakly crystalline materials. However, detailed characterization and quantification of these amorphous materials remains challenging. Conventional techniques, including electron microscopy and X-ray diffraction, provide important information on material structure, but are generally limited to crystalline materials. Polarized resonant soft X-ray scattering (P-RSoXS) is a synchrotron-based X-ray scattering technique that has been used to characterize and quantify weakly crystalline systems, including soft materials. This research aims to translate this technique to geologic and extraterrestrial materials, building off the knowledge of using P-RSoXS to interrogate soft materials. P-RSoXS is well-suited to interrogate geologic materials, which are often multiphase, heterogeneous systems with crystalline and amorphous components, and domain sizes on the order of tens of nanometers. It is anticipated that future work will help reveal new chemical and structural information in geologic materials. Maturation of this preliminary work to develop P-RSoXS to characterize astromaterial-relevant samples will provide new understanding of the secondary processes responsible for the development of amorphous astromaterials and further elucidate our knowledge of the geologic history and past alteration processes.

Joshua H Litofsky↗
NASA NTRS1994

Knowledge-based processing for aircraft flight control

This Contractor Report documents research in Intelligent Control using knowledge-based processing in a manner dual to methods found in the classic stochastic decision, estimation, and control discipline. Such knowledge-based control has also been called Declarative, and Hybid. Software architectures were sought, employing the parallelism inherent in modern object-oriented modeling and programming. The viewpoint adopted was that Intelligent Control employs a class of domain-specific software architectures having features common over a broad variety of implementations, such as management of aircraft flight, power distribution, etc. As much attention was paid to software engineering issues as to artificial intelligence and control issues. This research considered that particular processing methods from the stochastic and knowledge-based worlds are duals, that is, similar in a broad context. They provide architectural design concepts which serve as bridges between the disparate disciplines of decision, estimation, control, and artificial intelligence. This research was applied to the control of a subsonic transport aircraft in the airport terminal area.

Painter, John H.↗
NASA NTRS1998

Fuzzy Versions of Epistemic and Deontic Logic

Epistemic and deontic logics are modal logics, respectively, of knowledge and of the normative concepts of obligation, permission, and prohibition. Epistemic logic is useful in formalizing systems of communicating processes and knowledge and belief in AI (Artificial Intelligence). Deontic logic is useful in computer science wherever we must distinguish between actual and ideal behavior, as in fault tolerance and database integrity constraints. We here discuss fuzzy versions of these logics. In the crisp versions, various axioms correspond to various properties of the structures used in defining the semantics of the logics. Thus, any axiomatic theory will be characterized not only by its axioms but also by the set of properties holding of the corresponding semantic structures. Fuzzy logic does not proceed with axiomatic systems, but fuzzy versions of the semantic properties exist and can be shown to correspond to some of the axioms for the crisp systems in special ways that support dependency networks among assertions in a modal domain. This in turn allows one to implement truth maintenance systems. For the technical development of epistemic logic, and for that of deontic logic. To our knowledge, we are the first to address fuzzy epistemic and fuzzy deontic logic explicitly and to consider the different systems and semantic properties available. We give the syntax and semantics of epistemic logic and discuss the correspondence between axioms of epistemic logic and properties of semantic structures. The same topics are covered for deontic logic. Fuzzy epistemic and fuzzy deontic logic discusses the relationship between axioms and semantic properties for these logics. Our results can be exploited in truth maintenance systems.

Ramasamy S Gounder↗
DOE OSTI2025

AI-Powered Knowledge Graphs for Neuromorphic and Energy-Efficient Computing

The surge in scientific literature obscures breakthroughs and hinders the discovery of new research paths. We propose an artificial intelligence (AI) powered framework using large language models (LLMs) and knowledge graphs (KGs) to automate parts of scientific discovery, focusing on energy-efficient AI circuits. Our hybrid approach combines LLMs, structured data, and ontology-based reasoning to construct a comprehensive knowledge graph that integrates insights across computational neuroscience, spiking neuron models, learning rules, architectural motifs, and neuromorphic device technologies. This multi-domain representation enables the generation of hypotheses that connect biological function with implementable, energy-efficient hardware architectures. Using KG embeddings and graph neural networks, the framework generates hypotheses for novel circuits, validates them through optimization on exascale HPC systems, and with tools like SuperNeuro and Fugu, the most promising designs will be prototyped in hardware. This open-source system aims to accelerate discoveries and bridging neuroscience with hardware innovation, drive collaboration, and unlock new opportunities in low-power AI computing.

Gautam, Ashish [ORNL]↗
DOE OSTI2025

FAIR to WISE (F2W) v1.0.0

FAIR to WISE (F2W) is an iterative, large-language model (LLM) driven pipeline that turns unstructured research PDFs into structured, queryable knowledge graphs (KGs). Core features include schema-driven extraction to a LinkML model; full provenance capture; ontology-grounded enrichment (e.g., chemical validation and ChEBI lookup); graph construction to JSON-LD with stable IDs; and KG-RAG question answering with evidence-aware retrieval. The system is engineered for reproducibility and accessibility (open-source Ollama models, temperature=0, NVTX/Nsight profiling) with robust QA (relation verification, deduplication, and deterministic outputs). Primary uses are literature-to-KG automation, knowledge-grounded Q&A, and experimental steering support. We demonstrate the approach in organic photovoltaics, where the pipeline ingests papers, builds a domain KG, and evaluates answers against expert competency questions to guide experimental planning and interpretation. Compared with off-the-shelf LLMs and ad-hoc NLP tools, F2W addresses ontology gaps and reduces hallucination risk by grounding responses in extracted evidence and enforcing schema constraints; it also offers deterministic, provenance-linked outputs and open, cost-aware deployment. Evidence-aware ranking further improves answer quality over pure vector search.

Abramov, David [Lawrence Berkeley National Laborat↗
NASA NTRS1987

Temporal and contextual knowledge in model-based expert systems

A basic paradigm that allows representation of physical systems with a focus on context and time is presented. Paragon provides the capability to quickly capture an expert's knowledge in a cognitively resonant manner. From that description, Paragon creates a simulation model in LISP, which when executed, verifies that the domain expert did not make any mistakes. The Achille's heel of rule-based systems has been the lack of a systematic methodology for testing, and Paragon's developers are certain that the model-based approach overcomes that problem. The reason this testing is now possible is that software, which is very difficult to test, has in essence been transformed into hardware.

Toth-Fejel, Tihamer↗
NASA NTRS1989

The role of topography in geodetic gravity field modelling

Masses associated with the topography, bathymetry, and its isostatic compensation are a dominant source of gravity field variations, especially at shorter wavelengths. On global scales the topographic/isostatic effects are also significant, except for the lowest harmonics. In practice, though, global effects need not be taken into account as such effects are included in the coefficients of the geopotential reference fields. On local scales, the short-wavelength gravity variations due to the topography may, in rugged terrain, be an order of magnitude larger than other effects. In such cases, explicit or implicit terrain reduction procedures are mandatory in order to obtain good prediction results. Such effects may be computed by space-domain integration or by fast Fourier transformation (FFT) methods. Numerical examples are given for areas of the Canadian Rockies. In principle, good knowledge of the topographic densities is required to produce the smoothest residual field. Densities may be determined from sample measurements or by gravimetric means, but both are somewhat troublesome methods in practice. The use of a standard density, e.g., 2.67 g/cu cm, may often yield satisfactory results and may be put within a consistent theoretical framework. The independence of density assumptions is the key point of the classical Molodensky approach to the geodetic boundary value problem. The Molodensky solutions take into account that land gravity field observations are done on a non-level surface. Molodensky's problem may be solved by integral expansions or more effective FFT methods, but the solution should not be intermixed with the use of terrain reductions. The methods are actually complimentary and may both be required in order to obtain the smoothest possible signal, least prone to aliasing and other effects coming from sparse data coverage, typical of rugged topography.

Forsberg, R.↗
NASA NTRS1992

Representation and presentation of requirements knowledge

An approach to representation and presentation of knowledge used in the ARIES, an experimental requirements/specification environment, is described. The approach applies the notion of a representation architecture to the domain of software engineering and incorporates a strong coupling to a transformation system. It is characterized by a single highly expressive underlying representation, interfaced simultaneously to multiple presentations, each with notations of differing degrees of expressivity. This enables analysts to use multiple languages for describing systems and have these descriptions yield a single consistent model of the system.

Johnson, W. L.↗
NASA NTRS1996

Another Program Simulates A Modular Manufacturing System

SSE5 computer program provides simulation environment for modeling manufacturing systems containing relatively small numbers of stations and operators. Designed to simulate manufacturing of apparel, also used in other manufacturing domains. Valuable for small or medium-size firms, including those lacking expertise to develop detailed mathematical models or have only minimal knowledge in describing manufacturing systems and in analyzing results of simulations on mathematical models. Two other programs available bundled together as SSE (MFS-26245). Each program models slightly different manufacturing scenario. Written in Turbo C v2.0 for IBM PC-series and compatible computers running MS-DOS and successfully compiled using Turbo C++ v3.0.

Schroer, Bernard J.↗