Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Model error”

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 1,207 records · Page 67
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

A Neural Optimizer With Decision-Focused Learning for Optimal Energy Storage Operation

Here, this article introduces a neural optimizer-based framework for optimizing battery energy storage system (BESS) control for grid services, including demand charge and energy cost reduction. By leveraging decision-focused learning (DFL), the proposed framework ensures seamless integration and adaptation, significantly enhancing control performance. A patch time-series transformer is employed for peak load forecasting, incorporating aleatoric uncertainty quantification to account for forecasting uncertainties within the decision-making process. The framework utilizes a solver-in-the-loop approach to generate optimal BESS actions, which are then used to train the neural optimizer-based agent. By co-optimizing both BESS operational modes and output power within the NN, the system achieves improved performance and robustness. After initial training, the forecasting and control models are jointly fine-tuned to account for forecasting errors, further improving decision precision and efficiency through DFL. Case studies are performed to validate the performance of the framework using multiple real-world datasets, demonstrating superior performance in monthly peak load forecasting compared to state-of-the-art models. In addition, the results are compared against existing decision-making approaches. The results demonstrate a reduction in monthly peak forecasting error by approximately 15% across various performance measures and achieve an optimization gap for BESS operation that is about three times smaller compared to existing methods.

Kim, Hyeonjin [Pacific Northwest National Laborato↗
DOE OSTI2021

Predicting impurity spectral functions using machine learning

The Anderson Impurity Model (AIM) is a canonical model of quantum many-body physics. Here we investigate whether machine learning models, both neural networks (NN) and kernel ridge regression (KRR), can accurately predict the AIM spectral function in all of its regimes, from empty orbital, to mixed valence, to Kondo. To tackle this question, we construct two large spectral databases containing approximately 410 000 and 600 000 spectral functions of the single-channel impurity problem. We show that the NN models can accurately predict the AIM spectral function in all of its regimes, with pointwise mean absolute errors down to 0.003 in normalized units. We find that the trained NN models outperform models based on KRR and enjoy a speedup on the order of 10 5 over traditional AIM solvers. Finally, the required size of the training set of our model can be significantly reduced using farthest point sampling in the AIM parameter space, which is important for generalizing our method to more complicated multichannel impurity problems of relevance to predicting the properties of real materials.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2024

Three-Stage Adjusted Regression Forecasting for Software Defect Prediction

In this paper, a three-stage adjusted regression forecasting model is proposed to forecast the local regression model [6]. This is a growth curve approximation model that predicts the parameters for a future linear model based on a sliding window of previous linear models. The three stages of the model are as follows: • Initial fit: train regression models on a sliding window of the date and record model coefficients. • Prediction: fit new regression models to the coefficient lists and predict the value of the next coefficient. • Error correction: correct coefficient prediction error using the residual of the last point of the coefficient list and moving average. The resulting model from the multi-stage process is a forecast of the local regression model that represents the future window of data and is referred to as the predicted line. Results suggest the three-stage model demonstrates better prediction capability compared to existing solutions.

Pritchard, Shadow↗
DOE OSTI2024

Understanding Mixed Precision GEMM with MPGemmFI: Insights into Fault Resilience

Emerging deep learning workloads urgently need fast general matrix multiplication (GEMM). Thus, one of the critical features of machine-learning-specific accelerators such as NVIDIA Tensor Cores, AMD Matrix Cores, and Google TPUs is the support of mixed-precision enabled GEMM. For DNN models, lower-precision FP data formats and computation offer acceptable correctness but significant performance, area, and memory footprint improvement. While promising, the mixed-precision computation on error resilience remains unexplored. To this end, we develop a fault injection framework that systematically injects fault into the mixed-precision computation results. We investigate how the faults affect the accuracy of machine learning applications. Based on the characteristics of error resilience, we offer lightweight error detection and correction solutions that significantly improve the overall model accuracy by 75% if the models experience hardware faults. The solutions can be efficiently integrated into the accelerator's pipelines.

Fang, Bo↗
DOE OSTI2026

DC Cryogenic Modeling of Open-Source SkyWater 130 nm MOSFETs at 77 K Using BSIM4

Cryogenic applications in high-energy physics (HEP) demand reliable, low-power CMOS electronics capable of operating at liquid nitrogen temperatures (77 K). The open-source SkyWater 130nm (SKY130) CMOS process has previously been shown to operate at temperatures as low as 4 K making it a promising candidate for HEP applications. In this work, we characterize and model SKY130 low-threshold voltage transistors at 77 K, which is a temperature commonly used in modeling applications for liquid argon detectors. DC characteristic measurements were performed at both room temperature and liquid nitrogen temperature. We created a cryogenic modeling approach to produce a SPICE-compatible, isothermal BSIM4-based model for select transistor sizes at 77 K. The resulting model agrees with data at 77 K with an average error on the order of 20% (relative RMS) and shows no dependence on drain voltage. Due to the open-source nature of SKY130, we have made our models publicly available on Github. We hope this work will continue the trend for democratizing circuit design at cryogenic temperatures in high-energy physics by enabling open access to accurate cryogenic CMOS device models at 77 K.

Beall, F. [Texas U., Arlington]↗
DOE OSTI2021

A mathematical framework for ejecta cloud dynamics with application to source models and piezoelectric mass measurements

We present a mathematical framework for describing the dynamical evolution of an ejecta cloud generated by a generic ejecta source model. We consider a piezoelectric sensor fielded in the path of an ejecta cloud, for experimental configurations in which the ejecta are created at a singly shocked planar surface and fly ballistically through vacuum to the stationary sensor. To do so, we introduce the concept of a time- and velocity-dependent ejecta “areal mass function.” We derive expressions for the analytic (“true”) accumulated ejecta areal mass at the sensor and the measured (“inferred”) value obtained via the standard method for analyzing piezoelectric voltages. In this way, we derive an exact expression and upper bound for the error imposed upon a piezoelectric ejecta mass measurement (in a perfect system) by the assumption of instantaneous creation, which is commonly required for momentum diagnostic analyses. This error term is zero for truly instantaneous source models; otherwise, the standard piezoelectric analysis is guaranteed to overestimate the true mass. When combined with a piezoelectric dataset, this framework provides a unique solution for the ejecta particle velocity distribution, subject to the assumptions inherent in the data analysis. The framework also leads to strong boundary conditions that any ejecta source model must satisfy in order to be consistent with apparently global properties of piezoelectric measurements from a wide range of experiments. We demonstrate this methodology by applying it to the Richtmyer–Meshkov instability+self-similar velocity distribution ejecta source model currently under development at Los Alamos National Laboratory.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Predicting Device Parameters for Dye-Sensitized Solar Cells from Electronic Structure Calculations to Reproduce Experiment.

Given that improvements to the power-conversion efficiency (PCE) of dye-sensitized solar cells (DSSCs) have slowed in recent years, a means to accurately predict the device parameters yielded by trial dyes in silico, without having to synthesize them, would be extremely valuable to speed up the design process. Currently, the best-performing methods of calculating device parameters rely on a set of experimentally determined kinetic coefficients. In practice, it is very difficult to measure these kinetic parameters accurately, limiting the overall accuracy of such predictive methods. This work proposes a model to obtain key parameters such as J(SC), V-OC, and PCE using only the results from density functional theory (DFT) and time-dependent DFT calculations, noting that rates of electron-transfer steps are ultimately linked to the electronic structure of the dye center dot center dot center dot TiO2 working electrode. Six organic DSSC dyes from dissimilar chemical classes (L0, L1, L2, WS-2, WS-92, and C281) were chosen to demonstrate the power of this approach. Their a priori known experimentally determined device performance metrics served to validate our predictions. The greatest absolute error in our predicted PCE values was 0.36% relative to the experiment, while the greatest fractional error was 0.042. This indicates that the proposed model offers a dramatic improvement on previous predictive methods for DSSC device parameters, both in accuracy and in consistency. Moreover, such a predictive model has great potential to be applied to other photovoltaic applications, further enabling the design of novel, highly efficient photoactive materials.

density functional theory↗
DOE OSTI2025

SIDDA: SInkhorn Dynamic Domain Adaptation for image classification with equivariant neural networks

Modern neural networks (NNs) often do not generalize well in the presence of a ‘covariate shift’; that is, in situations where the training and test data distributions differ, but the conditional distribution of classification labels given the data remains unchanged. In such cases, NN generalization can be reduced to a problem of learning more robust, domain-invariant features. Domain adaptation (DA) methods include a broad range of techniques aimed at achieving this; however, these methods have struggled with the need for extensive hyperparameter tuning, which then incurs significant computational costs. In this work, we introduce SInkhorn Dynamic Domain Adaptation (SIDDA), an out-of-the-box DA training algorithm built upon the Sinkhorn divergence, that can achieve effective domain alignment with minimal hyperparameter tuning and computational overhead. We demonstrate the efficacy of our method on multiple simulated and real datasets of varying complexity, including simple shapes, handwritten digits, real astronomical observations, and remote sensing data. These datasets exhibit covariate shifts due to noise, blurring, differences between telescopes, and variations in imaging wavelengths. SIDDA is compatible with a variety of NN architectures, and it works particularly well in improving classification accuracy and model calibration when paired with symmetry-aware equivariant NNs (ENNs). We find that SIDDA consistently enhances the generalization capabilities of NNs, achieving up to a ${\approx}40\%$ improvement in classification accuracy on unlabeled target data, while also providing a more modest performance gain of $\lesssim 1\%$ on labeled source data. We also study the efficacy of DA on ENNs with respect to the varying group orders of the dihedral group DN, and find that the model performance improves as the degree of equivariance increases. Finally, if SIDDA achieves proper domain alignment, it also enhances model calibration on both source and target data, with the most significant gains in the unlabeled target domain—achieving over an order of magnitude improvement in the expected calibration error and Brier score. SIDDA’s versatility across various NN models and datasets, combined with its automated approach to domain alignment, has the potential to significantly advance multi-dataset studies by enabling the development of highly generalizable models.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

Incorporation of aerosol into the COSPv2 satellite lidar simulator for climate model evaluation

Abstract. Atmospheric aerosol has substantial impacts on climate, air quality and biogeochemical cycles, and its concentrations are highly variable in space and time. A key variability to evaluate within models that simulate aerosol is the vertical distribution, which influences atmospheric heating profiles and aerosol–cloud interactions, to help constrain aerosol residence time and to better represent the magnitude of simulated impacts. To ensure a consistent comparison between modeled and observed vertical distribution of aerosol, we implemented an aerosol lidar simulator within the Cloud Feedback Model Intercomparison Project (CFMIP) Observation Simulator Package version 2 (COSPv2). We assessed the attenuated total backscattered (ATB) signal and the backscatter ratios (SRs) at 532 nm in the U.S. Department of Energy's Energy Exascale Earth System Model version 1 (E3SMv1). The simulator performs the computations at the same vertical resolution as the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP), making use of aerosol optics from the E3SMv1 model as inputs and assuming that aerosol is uniformly distributed horizontally within each model grid box. The simulator applies a cloud masking and an aerosol detection threshold to obtain the ATB and SR profiles that would be observed above clouds by CALIOP with its aerosol detection capability. Our analysis shows that the aerosol distribution simulated at a seasonal timescale is generally in good agreement with observations. Over the Southern Ocean, however, the model does not produce the SR maximum as observed in the real world. Comparison between clear-sky and all-sky SRs shows little differences, indicating that the cloud screening by potentially incorrect model clouds does not affect the mean aerosol signal averaged over a season. This indicates that the differences between observed and simulated SR values are due not to sampling errors, but to deficiencies in the representation of aerosol in models. Finally, we highlight the need for future applications of lidar observations at multiple wavelengths to provide insights into aerosol properties and distribution and their representation in Earth system models.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Machine Learning Screening of Metal-Ion Battery Electrode Materials

Here, in this work we present deep neural network regression machine learning models (ML) for predicting the average voltage and the percentage change in volume of battery electrodes upon charging and discharging with metal ions. Our models exhibit good performance as measured by the average mean absolute error obtained from a 10-fold cross-validation as well as on independent test sets. We further assess the robustness our ML models by investigating their screening potential beyond the training database. We produce novel Na-ion electrodes by systematically replacing Li-ions in the original database by Na-ions, and then selecting a set of 22 electrodes that exhibit a good performance in energy density as well as small volume variations upon charging and discharging, as predicted by the machine learning model. The ML predictions for these new materials are then compared to quantum-mechanics based calculations. Our results reaffirm the significant role of machine learning techniques in the exploration of materials for battery applications.

,electrode volume change↗
DOE OSTI2024

Simulating hindered grain boundary diffusion using the smoothed boundary method

Abstract Grain boundaries can greatly affect the transport properties of polycrystalline materials, particularly when the grain size approaches the nanoscale. While grain boundaries often enhance diffusion by providing a fast pathway for chemical transport, some material systems, such as those of solid oxide fuel cells and battery cathode particles, exhibit the opposite behavior, where grain boundaries act to hinder diffusion. To facilitate the study of systems with hindered grain boundary diffusion, we propose a model that utilizes the smoothed boundary method to simulate the dynamic concentration evolution in polycrystalline systems. The model employs domain parameters with diffuse interfaces to describe the grains, thereby enabling solutions with explicit consideration of their complex geometries. The intrinsic error arising from the diffuse interface approach employed in our proposed model is explored by comparing the results against a sharp interface model for a variety of parameter sets. Finally, two case studies are considered to demonstrate potential applications of the model. First, a nanocrystalline yttria-stabilized zirconia solid oxide fuel cell system is investigated, and the effective diffusivities are extracted from the simulation results and are compared to the values obtained through mean-field approximations. Second, the concentration evolution during lithiation of a polycrystalline battery cathode particle is simulated to demonstrate the method’s capability.

Materials Science↗
DOE OSTI2026

Neural Scaling Laws of Deep ReLU and Deep Operator Network: A Theoretical Study

Neural scaling laws play a pivotal role in the performance of deep neural networks and have been observed in a wide range of tasks. However, a complete theoretical framework for understanding these scaling laws remains underdeveloped. In this paper, we explore the neural scaling laws for deep operator networks, which involve learning mappings between function spaces, with a focus on the Chen and Chen style architecture. These approaches, which include the popular Deep Operator Network (DeepONet), approximate the output functions using a linear combination of learnable basis functions and coefficients that depend on the input functions. We establish a theoretical framework to quantify the neural scaling laws by analyzing its approximation and generalization errors. We articulate the relationship between the approximation and generalization errors of deep operator networks and key factors such as network model size and training data size. Moreover, we address cases where input functions exhibit low-dimensional structures, allowing us to derive tighter error bounds. These results also hold for deep ReLU networks and other similar structures. Our results offer a partial explanation of the neural scaling laws in operator learning and provide a theoretical foundation for their applications.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Vertical control of DIII-D discharges with strong negative triangularity

Through predictive modeling validated by a series of experiments on DIII-D, the vertical stability of low β diverted plasmas with strong negative triangularity (NT) ($\delta\sim-0.6$) is assessed. As a result of their unique magnetic geometry, NT plasmas feature larger Shafranov shifts and more elongated inner flux surfaces than positive triangularity counterparts, typically leading to enhanced vertical instability growth rates. However, coupling with the non-conformal vessel DIII-D wall reduces these growth rates to controllable values, providing a path forward for stabilizing strongly NT plasma with a diverted geometry. These discharges are used to validate GSdesign (part of the TokSys code suite) stability calculations in strong NT plasmas on DIII-D, with errors of no more than ~20% observed between modeled and experimentally measured growth rates. Additions of diagnostic noise and power supply tuning to the TokSys model are needed to accurately capture the time dependence of DIII-D NT discharges, assisting with the design of control schemes specific to the DIII-D poloidal field coils. Lastly, implications of these results on a future NT reactor are briefly described.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

SAXS Assistant: Automated SAXS analysis for structural discovery in biologics and polymeric nanoparticles

Small-angle x-ray scattering (SAXS) is a powerful technique for assessing macromolecular structure. High-throughput SAXS is limited by the time-consuming and, at times, subjective nature of SAXS data interpretation. Here, we present SAXS Assistant, a Python-based script that streamlines SAXS data analysis to extract features for machine learning (ML) and key structural parameters, including the Guinier radius of gyration (R g ), pair distance distribution function (PDDF)-derived R g , maximum particle dimension (D max ), and Kratky plots. The script builds upon BioXTAS RAW and validates reliability via Guinier/PDDF R g agreement, an important indicator of well-measured data sets. For assistance in D max estimation, a multilayer perceptron regressor was trained with 1940 data files from the Small Angle Scattering Biological Data Bank. The model achieved a test set performance R 2 = 0.90 and mean absolute error = 11.7 Å. Training exclusively with experimental data translates analyses from researchers, including experts in the field, to the ML model, which helps assess D max estimations from PDDF. Gaussian mixture model clustering was implemented to classify profiles into structural classes based on entries in the Small Angle Scattering Biological Data Bank. Users may therefore assess the similarity between experimental samples and known biomolecular shapes within the mapped repository entries. This probabilistic clustering aids in quantifying information from Kratky and generating shape-descriptive features. SAXS Assistant accelerates SAXS data analysis through enforced quality control, ML-ready outputs, and flags for low-confidence results. In addition to providing the ability to analyze large data sets at high throughput, this tool is versatile and may serve researchers in both biological and synthetic polymer research fields.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Standardizing reverberation-measured C iv time-lag quasars, and using them with standardized Mg ii quasars to constrain cosmological parameters

ABSTRACT We use 38 C iv quasar (QSO) reverberation-mapped observations, which span eight orders of magnitude in luminosity and the redshift range 0.001064 ≤ z ≤ 3.368, to simultaneously constrain cosmological-model and QSO radius–luminosity (R–L) relation parameters in six cosmological models, using an improved technique that more correctly accounts for the asymmetric errors bars of the time-lag measurements. We find that R–L relation parameters are independent of the cosmological models used in the analysis and so the R–L relation can be used to standardize the C iv QSOs. The C iv QSO cosmological constraints are consistent with those from Mg ii QSOs, allowing us to derive joint C iv + Mg ii QSO cosmological constraints which are consistent with currently accelerated cosmological expansion, as well as consistent with cosmological constraints derived using better-established baryon acoustic oscillation (BAO) and Hubble parameter [H(z)] measurements. When jointly analysed with H(z) + BAO data, current C iv + Mg ii QSO data mildly tighten current H(z) + BAO data cosmological constraints.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2021

Alignment tolerance analysis for divided-pulse nonlinear compression

We present an analytic model that describes the output pulse after Kerr-based spectral broadening with divided-pulse nonlinear compression that includes errors in unequal pulse division, birefringent plate retardance, and thermal drift. The model shows that alignment tolerances become impractically tight at high levels of nonlinearity and that the angle of incidence on the birefringent plates can be utilized as a compensator to loosen those tolerances. We present experimental verification of the model, which is expected to be a fast and flexible tool to design future divided-pulse nonlinear compression systems.

47 OTHER INSTRUMENTATION↗
DOE OSTI2024

A safe reinforcement learning algorithm for supervisory control of power plants

Traditional control theory-based methods require tailored engineering for each system and constant fine-tuning. In power plant control, one often needs to obtain a precise representation of the system dynamics and carefully design the control scheme accordingly. Model-free Reinforcement learning (RL) has emerged as a promising solution for control tasks due to its ability to learn from trial-and-error interactions with the environment. It eliminates the need for explicitly modeling the environment’s dynamics, which is potentially inaccurate. However, the direct imposition of state constraints in power plant control raises challenges for standard RL methods. To address this, we propose a chance-constrained RL algorithm based on Proximal Policy Optimization for supervisory control. Our method employs Lagrangian relaxation to convert the constrained optimization problem into an unconstrained objective, where trainable Lagrange multipliers enforce the state constraints. In conclusion, our approach achieves the smallest distance of violation and violation rate in a load-follow maneuver for an advanced Nuclear Power Plant design.

constrained optimization↗