Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Learning with errors”

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 307 records · Page 17
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

An Analysis of Grid Operator Survey Responses: Inexperience, Workload and Fatigue in the Control Room

Although a wide array of tools and technologies have been developed over the last decade to support power grid operators, deployment of these tools has been less successful. One reason for unsuccessful deployment may be an inadequate understanding of the factors that contribute to operator error in the control room. An analysis of operators’ current vulnerabilities may provide the baseline understanding needed to inform new technology integration. In an attempt to learn more about these vulnerabilities and their perceived impact on human error we collected and analyzed survey data from 20 electric grid control room operators. We asked survey respondents to consider the various operator, technology and interaction vulnerabilities that may arise during work in the control room and record their attitudes and experiences toward each. Results suggest operator inexperience, high mental workload and fatigue are the most common vulnerabilities experienced during a shift. Survey results were analyzed to explore these vulnerabilities in greater depth.

Inexperience, Workload, Fatigue↗
DOE OSTI2020

Reinforcement learning based automated history matching for improved hydrocarbon production forecast

History matching aims to find a numerical reservoir model that can be used to predict the reservoir performance. An engineer and model calibration (data inversion) method are required to adjust various parameters/properties of the numerical model in order to match the reservoir production history. In this study, we develop deep neural networks within the reinforcement learning framework to achieve automated history matching that will reduce engineers’ efforts, human bias, automatically and intelligently explore the parameter space, and remove the need of large set of labeled training data. To that end, a fast-marching-based reservoir simulator is encapsulated as an environment for the proposed reinforcement learning. The deep neural-network-based learning agent interacts with the reservoir simulator within reinforcement learning framework to achieve the automated history matching. Reinforcement learning techniques, such as discrete Deep Q Network and continuous Deep Deterministic Policy Gradients, are used toth, used to train the learning agents. The continuous actions enable the Deep Deterministic Policy Gradients to explore more states at each iteration in a a learning episode; consequently, a better history matching is achieved using this algorithm as compared to Deep Q Network. For simplified dual-target composite reservoir models, the best history-matching performances of the discrete and continuous learning methods in terms of normalized root mean square errors are 0.0447 and 0.0038, respectively. Furthermore, our study shows that continuous action space achieved by the deep deterministic policy gradient drastically outperforms deep Q network.

42 ENGINEERING↗
DOE OSTI2020

RLScheduler: An Automated HPC Batch Job Scheduler Using Reinforcement Learning

Today’s high-performance computing (HPC) platforms are still dominated by batch jobs. Accordingly, effective batch job scheduling is crucial to obtain high system efficiency. Existing HPC batch job schedulers typically leverage heuristic priority functions to prioritize and schedule jobs. But, once configured and deployed by the experts, such priority functions can hardly adapt to the changes of job loads, optimization goals, or system settings, potentially leading to degraded system efficiency when changes occur. To address this fundamental issue, we present RLScheduler, an automated HPC batch job scheduler built on reinforcement learning. RLScheduler relies on minimal manual interventions or expert knowledge, but can learn high-quality scheduling policies via its own continuous ‘trial and error’. We introduce a new kernel-based neural network structure and trajectory filtering mechanism in RLScheduler to improve and stabilize the learning process. Through extensive evaluations, we confirm that RLScheduler can learn high-quality scheduling policies towards various workloads and various optimization goals with relatively low computation cost. Moreover, we show that the learned models perform stably even when applied to unseen workloads, making them practical for production use.

Zhang, Di↗
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 OSTI2026

Active learning emulators for nuclear two-body scattering in momentum space

In this work we extend the active learning emulators for two-body scattering in coordinate space with error estimation, recently developed by Maldonado et al. [Phys. Rev. C 112, 024002], to coupled-channel scattering in momentum space. Our full-order model (FOM) solver is based on the Lippmann-Schwinger integral equation for the scattering t-matrix as opposed to the radial Schrödinger equation. We use (Petrov-)Galerkin projections and high-fidelity calculations at a few snapshots across the parameter space of the interaction to construct efficient reduced-order models (ROMs), trained by a greedy algorithm for locally optimal snapshot selection. Both the FOM solver and the corresponding ROMs are implemented efficiently in Python using Google's JAX library. We present results for emulating scattering phase shifts in coupled and uncoupled channels and cross sections, and assess the accuracy of the developed ROMs and their computational speedup factors. We also develop emulator error estimation for both the t-matrix and the total cross section. The software framework for reproducing and extending our results is publicly available. Together with our recent advances in developing active-learning emulators for three-body scattering, these emulator frameworks set the stage for full Bayesian calibrations of chiral nuclear interactions and optical models against scattering data with quantified emulator errors.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2025

GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning

Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language often provides a much richer learning medium for LLMs, compared to policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA (Genetic-Pareto), a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing one or more LLM prompts, GEPA samples trajectories (e.g., reasoning, tool calls, and tool outputs) and reflects on them in natural language to diagnose problems, propose and test prompt updates, and combine complementary lessons from the Pareto frontier of its own attempts. As a result of GEPA's design, it can often turn even just a few rollouts into a large quality gain. Across six tasks, GEPA outperforms GRPO by 6% on average and by up to 20%, while using up to 35x fewer rollouts. GEPA also outperforms the leading prompt optimizer, MIPROv2, by over 10% (e.g., +12% accuracy on AIME-2025), and demonstrates promising results as an inference-time search strategy for code optimization. We release our code at https://github.com/gepa-ai/gepa.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Learning together: Towards foundation models for machine learning interatomic potentials with meta-learning

Abstract The development of machine learning models has led to an abundance of datasets containing quantum mechanical (QM) calculations for molecular and material systems. However, traditional training methods for machine learning models are unable to leverage the plethora of data available as they require that each dataset be generated using the same QM method. Taking machine learning interatomic potentials (MLIPs) as an example, we show that meta-learning techniques, a recent advancement from the machine learning community, can be used to fit multiple levels of QM theory in the same training process. Meta-learning changes the training procedure to learn a representation that can be easily re-trained to new tasks with small amounts of data. We then demonstrate that meta-learning enables simultaneously training to multiple large organic molecule datasets. As a proof of concept, we examine the performance of a MLIP refit to a small drug-like molecule and show that pre-training potentials to multiple levels of theory with meta-learning improves performance. This difference in performance can be seen both in the reduced error and in the improved smoothness of the potential energy surface produced. We therefore show that meta-learning can utilize existing datasets with inconsistent QM levels of theory to produce models that are better at specializing to new datasets. This opens new routes for creating pre-trained, foundation models for interatomic potentials.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2023

Autonomous Inversion of In Situ Deformation Measurement Data for Injection-Induced Stress Change

Geologic carbon storage (GCS) is likely to play a key part of the global effort to dramatically reduce CO2 emissions and perhaps even reduce atmospheric CO2 concentrations through carbon negative operations. A critical part of effort to commercialize and widely deploy this technology is developing the capability to rapidly assimilate real-time monitoring data into a form that will enable site operators to make decisions to manage the safe and efficient operations. Two of the risks associate with GCS are the risk of inducing fractures in the sealing formations that can create leakage pathways and the risk of inducing earthquakes of sufficient magnitude to cause public concern, property damage, or safety risks. To properly manage these risks the site operator needs to know the initial state of stress, the change in stress induced by injection, and the relationship between operational parameters such as injection rate and pressure and the change in stress. Current methods of estimating the change in stress require choosing the type of constitutive model and the model parameters based on core, log, and geophysical data during the characterization phase, with little feedback from operational observations to validate or refine these choices. These characterization methods interrogate the geologic formations using length scales, loading rates or magnitudes that are quite different from those encountered by the actual storage system. It is shown that errors in the assumed constitutive response, even when informed by laboratory tests on core samples, are likely to be common, large, and underestimate the magnitude of stress change caused by injection. Recent advances in borehole-based strain instruments and borehole and surface-based tilt and displacement instruments have now enabled monitoring of the deformation of the storage system throughout its operational lifespan. This data can enable validation and refinement of the knowledge of the geomechanical properties and state of the system, but brings with it a challenge to transform the raw data into actionable knowledge. We demonstrate a method that uses automatic differentiation and a finite-element based geomechanical model perform a gradient-based deterministic inversion of geomechanical monitoring data. This approach allows autonomous integration of the instrument data without the need for time consuming manual interpretation and selection of updated model parameters. Furthermore, only isotropic linear elasticity is considered in this paper, the approach presented is very flexible as to what type of geomechanical constitutive response can be used. The approach is easily adaptable to nonlinear physics-based constitutive models to account for common rock behaviors such as creep and plasticity. The approach also enables training of machine learning-based constitutive models by allowing back propagation of errors through the finite element calculations. This enables strongly enforcing known physics, such as conservation of momentum and continuity, while allowing data-driven models to learn the truly unknown physics such as the constitutive or petrophysical responses.

Burghardt, Jeffrey A.↗
DOE OSTI2025

Reducing Frequency Bias of Fourier Neural Operators in 3D Seismic Wavefield Simulations Through Multistage Training

The recent development of neural operator (NeurOp) learning for solutions to the elastic wave equation shows promising results and provides the basis for fast large-scale simulations for different seismological applications. In this article, we use the Fourier neural operator (FNO) model to directly solve the 3D Helmholtz wave equation for fast seismic ground-motion simulations on different frequencies and show the frequency bias of the FNO model, that is, it learns the lower frequencies better comparing to the higher frequencies. To reduce the frequency bias, we adopt the multistage FNO training, that is, after training a stage 1 FNO model for estimating the ground motion, we use a second FNO model as the stage 2 to learn from the residual, which greatly reduced the errors on the higher frequencies. By adopting this multistage training, the FNO models show reduced biases on higher frequencies, which enhanced the overall results of the ground-motion simulations. Thus the multistage training FNO improves the accuracy and realism of the ground-motion simulations.

earthquakes↗
DOE OSTI2022

Toward Forecasting Geomagnetic Storms

We present a selection of attempts to infer geomagnetic storm properties from time series data. Initially, we unsuccessfully attempt to predict the duration of a geomagnetic storm based on data preceding the onset of a geomagnetic storm. Similar techniques are also used to attempt to predict the maximum depth of a geomagnetic storm. Additionally, we present a case study in Gaussian Process regression, where we are able to accurately interpolate between geomagnetic storm parameters, but are unable to predict future storm behavior. Lastly, we describe attempts at using random forest regression with sliding windows to infer geomagnetic storm behavior at future times. These results did not provide robust forecasts of geomagnetic storm behavior, but did occasionally correctly predict a storm within the margin of error. Techniques presented in this paper include machine learning classification and regression, Gaussian Process regression, and machine learning regression with sliding windows. Many of the attempts presented in this document did not result in meaningful predictions of geomagnetic storm behavior. We hope this document can be used as a guide for future readers attempting to predict geomagnetic storm properties.

58 GEOSCIENCES↗
NASA NTRS1997

Quality Issues in Propulsion

Propulsion system quality is defined in this paper as having high reliability, that is, quality is a high probability of within-tolerance performance or operation. Since failures are out-of-tolerance performance, the probability of failures and their occurrence is the difference between high and low quality systems. Failures can be described at 3 levels: the system failure (which is the detectable end of a failure), the failure mode (which is the failure process), and the failure cause (which is the start). Failure causes can be evaluated & classified by type. The results of typing flight history failures shows that most failures are in unrecognized modes and result from human error or noise, i.e. failures are when engineers learn how things really work. Although the study based on US launch vehicles, a sampling of failures from other countries indicates the finding has broad application. The parameters of the design of a propulsion system are not single valued, but have dispersions associated with the manufacturing of parts. Many tests are needed to find failures, if the dispersions are large relative to tolerances, which could contribute to the large number of failures in unrecognized modes.

McCarty, John P.↗
NASA NTRS2001

Discovering Communicable Scientific Knowledge from Spatio-Temporal Data

This paper describes how we used regression rules to improve upon a result previously published in the Earth science literature. In such a scientific application of machine learning, it is crucially important for the learned models to be understandable and communicable. We recount how we selected a learning algorithm to maximize communicability, and then describe two visualization techniques that we developed to aid in understanding the model by exploiting the spatial nature of the data. We also report how evaluating the learned models across time let us discover an error in the data.

Schwabacher, Mark↗
NASA NTRS2002

Discovering Communicable Models from Earth Science Data

This chapter describes how we used regression rules to improve upon results previously published in the Earth science literature. In such a scientific application of machine learning, it is crucially important for the learned models to be understandable and communicable. We recount how we selected a learning algorithm to maximize communicability, and then describe two visualization techniques that we developed to aid in understanding the model by exploiting the spatial nature of the data. We also report how evaluating the learned models across time let us discover an error in the data.

Schwabacher, Mark↗
NASA NTRS2014

UPSS and G2

The Universal Propellant Servicing System (UPSS) is a dedicated mobile launcher propellant delivery method that will minimize danger and complexity in order to allow vehicles to be serviced and ultimately launched from a variety of locations previously not seen fit for space launch. The UPPS/G2 project is the development of a model, simulation, and ultimately a working application that will control and monitor the cryogenic fluid delivery to the rocket for testing purposes. To accomplish this, the project is using the programming language/environment Gensym G2. The environment is an all-inclusive application that allows development, testing, modeling, and finally operation of the unique application through graphical and programmatic methods. We have learned G2 through classes and trial-and-error, and are now in the process of building the application that will soon be able to be tested on apparatuses here at Kennedy Space Center, and eventually on the actual unit. The UPSS will bring near-autonomous control of launches to those that need it, as well it will be a great addition to NASA and KSC's operational viability and the opportunity to bring space launches to parts of the world, and in time constraints, once not thought possible.

cryogenic fluids↗
DOE OSTI2022

Procedure Parsing: A Method for Parsing Handwritten Documents into Computer-Based Procedures

The nuclear industry is heavily procedure driven, where almost everything has a step-by-step instruction that is expected to be followed in detail. Historically, these procedures were printed on paper copies. Recently, the industry transitioned towards electronic copies (i.e., PDFs on tablets). One major drive for this transition is the introduction of human error and loss of situation awareness when using paper copies. However, electronic copies of documents inherently have the same error traps as their paper cousins. Therefore, there is an increased interest in a way to utilize the information in the step-by-step guidance, but to present it in a dynamic manner that guides the user and adapts to any encountered conditions. Researchers at Idaho National Laboratory propose a flexible, automated method based on document parsing and augmented by natural language processing (NLP) techniques, to address these shortcomings and capitalize on these recent advancements in machine learning. The proposed method provides a cost-effective solution for computer-assisted procedure parsing of hand-written control room procedures, originally authored in Word or PDF formats, into instructions that can be displayed as computer-based procedures (CBP) in a modern graphical user interface. The researchers devised, implemented and demonstrated the Operating Procedure Extender for Novel Systems (OPENS) method in 2020. The key to OPENS is to map the original procedure text into a context-free grammar, tying content to equipment, locations, and other steps, actions, etc. This formal grammar is then used to isolate and define keywords and actions verbs, such as “measure” or “evaluate” and tie them to specific equipment referenced within that step or located in other steps, substeps, actions, subactions and tables throughout the procedure. OPENS generates an abstract syntax tree from the document which it uses to store a copy of this information in the open-standard, machine-readable and human-readable file formats XML and JSON. The XML is useful to preserve the relational aspects of the procedure for referencing tables and branching information so the user can be directed to the next appropriate active step based on the values entered for that step and previous steps. The JSON is useful for storing and exchanging data objects used to track responses to previous steps and state changes in simulated environments. In future iterations, these formats can also be used for storing more detailed information about input during plant operation or simulation. The techniques the researcher developed could further be improved by integration of recent advancements in machine learning. NLP methods could standardize documents, correct for grammatical error, and provide automated semantic validation. The researcher expects that self-supervised techniques applied to collections of natural language instructions could strengthen the model with broader context. All these methods together give us a practical way to automatically extract protocols from documents and user interactions, empowering researchers, procedure writers and nuclear operators while moving the industry forward.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2026

Deep nonparametric estimation of operators between infinite dimensional spaces

Learning operators between infinitely dimensional spaces is an important learning task arising in machine learning, imaging science, mathematical modeling and simulations, etc. This paper studies the nonparametric estimation of Lipschitz operators using deep neural networks. Non-asymptotic upper bounds are derived for the generalization error of the empirical risk minimizer over a properly chosen network class. Under the assumption that the target operator exhibits a low dimensional structure, our error bounds decay as the training sample size increases, with an attractive fast rate depending on the intrinsic dimension in our estimation. Our assumptions cover most scenarios in real applications and our results give rise to fast rates by exploiting low dimensional structures of data in operator estimation. We also investigate the influence of network structures (e.g., network width, depth, and sparsity) on the generalization error of the neural network estimator and propose a general suggestion on the choice of network structures to maximize the learning efficiency quantitatively.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2007

Adaptive Modeling of the International Space Station Electrical Power System

Software simulations provide NASA engineers the ability to experiment with spacecraft systems in a computer-imitated environment. Engineers currently develop software models that encapsulate spacecraft system behavior. These models can be inaccurate due to invalid assumptions, erroneous operation, or system evolution. Increasing accuracy requires manual calibration and domain-specific knowledge. This thesis presents a method for automatically learning system models without any assumptions regarding system behavior. Data stream mining techniques are applied to learn models for critical portions of the International Space Station (ISS) Electrical Power System (EPS). We also explore a knowledge fusion approach that uses traditional engineered EPS models to supplement the learned models. We observed that these engineered EPS models provide useful background knowledge to reduce predictive error spikes when confronted with making predictions in situations that are quite different from the training scenarios used when learning the model. Evaluations using ISS sensor data and existing EPS models demonstrate the success of the adaptive approach. Our experimental results show that adaptive modeling provides reductions in model error anywhere from 80% to 96% over these existing models. Final discussions include impending use of adaptive modeling technology for ISS mission operations and the need for adaptive modeling in future NASA lunar and Martian exploration.

Thomas, Justin Ray↗
NASA NTRS1991

Rotational wind indicator enhances control of rotated displays

Rotation by 108 deg of the spatial mapping between a visual display and a manual input device produces large spatial errors in a discrete aiming task. These errors are not easily corrected by voluntary mental effort, but the central nervous system does adapt gradually to the new mapping. Bernotat (1970) showed that adding true hand position to a 90 deg rotated display improved performance of a compensatory tracking task, but tracking error rose again upon removal of the explicit cue. This suggests that the explicit error signal did not induce changes in the neural mapping, but rather allowed the operator to reduce tracking error using a higher mental strategy. In this report, we describe an explicit visual display enhancement applied to a 108 deg rotated discrete aiming task. A 'wind indicator' corresponding to the effect of the mapping rotation is displayed on the operator-controlled cursor. The human operator is instructed to oppose the virtual force represented by the indicator, as one would do if flying an airplane in a crosswind. This enhancement reduces spatial aiming error in the first 10 minutes of practice by an average of 70 percent when compared to a no enhancement control condition. Moreover, it produces adaptation aftereffect, which is evidence of learning by neural adaptation rather than by mental strategy. Finally, aiming error does not rise upon removal of the explicit cue.

Cunningham, H. A.↗