Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “interpretable models”

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 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Use of Physics to Improve Solar Forecast: Part II, Machine Learning and Model Interpretability

Machine learning (ML) models have been applied to forecast solar energy; however, they often lack clarity of interpretability and underlying physics. This work addresses such challenges by developing a hierarchy of ML models that gradually introduce predictors to improve the forecast accuracy based on a physics-based framework. Three ML models (ARIMA, LSTM, and XGBoost) are examined and compared with four physics-informed persistence models reported in Part I and the simple persistence model to assess the improvement of different models. The 7-year measurements at the U.S. Department of Energy's Atmospheric Radiation Measurement's Southern Great Plains Central Facility site are used for forecasts and evaluations. The results reveal that the step-by-step introduction of predictors leads to different improvements for models at different hierarchical levels. Comparison of the ML models with persistence models shows that LSTM and XGBoost outperform all the persistence models, with LSTM having the overall best performance; however, ARIMA underperforms the four physics-informed persistence models. This study demonstrates the importance and utility of incorporating physics into ML models in improving forecast accuracy by introducing a hierarchy of physics-based predictors, distinguishing predictor contributions, and enhancing the ML interpretability. The combined use of Global Horizontal Irradiance (GHI) and Direct Normal Irradiance (DNI) significantly improves the forecast accuracy compared to using individual irradiances alone because the pair contains more information on cloud-radiation interactions.

interpretability↗
DOE OSTI2026

Interpretable Models for Workflow Differentiation in High-Performance Scientific Networks

Scientific workflows in high-performance networks spawn hundreds of interdependent flows that must be managed collectively—yet existing network classifiers treat each flow in isolation, leading to fragmented QoS decisions and missed interflow patterns. We present a novel traffic classification solution that operates at the workflow level, distinguishing entire filetransfer operations from streaming analytics by capturing how concurrent flows interact and burst together. We introduce a workflow identification window (WIW) that ingests raw packet headers from parallel flows into unified tensors, preserving the spatial-temporal patterns that differentiate scientific workflows. This approach achieves 98.7% accuracy using CNN, LSTM, and hybrid architectures, while maintaining 84% accuracy on production traffic collected a week later—demonstrating robustness to temporal drift. By integrating SHAP and GradCAM explainability, we reveal that early-packet timing patterns and cross-flow correlations drive classification decisions, providing operators with interpretable insights. Our system enables coherent workflow-level QoS enforcement and dynamic bandwidth allocation in scientific networks, eliminating manual per-flow configuration while maintaining classification latency at millisecond level.

Giannakou, Anna [LBL, Berkeley]↗
DOE OSTI2026

Knowledge-guided learning with curated prior genetic biomarkers for robust model interpretation

Abstract Motivation Knowledge-guided learning offers effective and robust model training strategies in data-scarce settings by incorporating established domain knowledge, thereby enhancing generalization, robustness, and interpretability. By contrast, conventional deep learning approaches rely purely on data-driven learning, which can limit robust model interpretability, particularly in high-dimensional settings with limited size samples. In computational biology, knowledge-guided learning has primarily leveraged network- and structural-based knowledge, leading to biologically interpretable representations and enhanced predictive performance compared to conventional approaches. However, curated biomarkers, one of the most accessible forms of biological knowledge, remain largely unexplored within knowledge-guided paradigms. Results In this study, we propose a model-agnostic training paradigm, Biomarker-driven Explainable Prior-guided Learning (BioExPL), that can be applied to any neural networks that incorporates curated prior knowledge. BioExPL enforces neural networks to reflect curated biomarker priors in their latent representations through a novel knowledge-alignment loss. BioExPL consistently demonstrated significantly improved predictive performance and enhanced model interpretability with minimized computational overhead in simulation studies and intensive experiments on multiple cancer datasets. BioExPL not only integrates prior curated knowledge into the model but also accurately identifies unknown associated signals additionally. BioExPL is model-agnostic and domain-independent, enabling its integration into diverse neural network architectures. Availability and implementation The open-source is publicly available at: https://github.com/datax-lab/BioExPL.

Baek, Beomsu [Department of Computer Science, Univ↗
DOE OSTI2023

Overview of interpretive modelling of fusion performance in JET DTE2 discharges with TRANSP

In the paper we present an overview of interpretive modelling of a database of JET-ILW 2021 D-T discharges using the TRANSP code. The main aim is to assess our capability of computationally reproducing the fusion performance of various D-T plasma scenarios using different external heating and D-T mixtures, and to understand the performance driving mechanisms. We find that interpretive simulations confirm a general power-law relationship between increasing external heating power and fusion output, which is supported by absolutely calibrated neutron yield measurements. A comparison of measured and computed D-T neutron rates shows that the calculations' discrepancy depends on the absolute neutron yield. The calculations are found to agree well with measurements for higher performing discharges with external heating power above ~20 MW, while low-neutron shots display an average discrepancy of around +40% compared to measured neutron yields. A similar trend is found for the ratio between thermal and beam-target fusion, where larger discrepancies are seen in shots with dominant beam-driven performance. We compare the observations to studies of JET-ILW D discharges, to find that on average the fusion performance is well modelled over a range of heating power, although an increased unsystematic deviation for lower-performing shots is observed. The ratio between thermal and beam-induced D-T fusion is found to be increasing weakly with growing external heating power, with a maximum value of ≳1 achieved in a baseline scenario experiment. An evaluation of the fusion power computational uncertainty shows a strong dependence on the plasma scenario type and fusion drive characteristics, varying between ±25% and 35%. D-T fusion alpha simulations show that the ratio between volume-integrated electron and ion heating from alphas is ≲10 for the majority of analysed discharges. Alphas are computed to contribute between ~15% and 40% to the total electron heating in the core of highest performing D-T discharges. An alternative workflow to TRANSP was employed to model JET D-T plasmas with the highest fusion yield and dominant non-thermal fusion component because of the use of fundamental radio-frequency heating of a large minority in the scenario, which is calculated to have provided ~10% to the total fusion power.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2026

Data from: "Towards CONUS-Wide ML-Augmented Conceptually-Interpretable Modeling of Catchment-Scale Precipitation-Storage-Runoff Dynamics"

This data package was generated to support the manuscript “Towards CONUS-Wide Machine Learning-Augmented Conceptually Interpretable Modeling of Catchment-Scale Precipitation-Storage-Runoff Dynamics.” It provides input files, model outputs, plotting data, scripts, notebooks, and documentation used to develop, evaluate, and reproduce Mass-Conserving Perceptron (MCP)-based hydrologic modeling experiments across 513 selected Catchment Attributes and Meteorology for Large-sample Studies in the United States (CAMELS-US) basins. The files are organized by modeling component and analysis purpose, including rainfall–runoff experiments, snow module experiments, coupled hydrologic-snow experiments, Long Short-Term Memory (LSTM) benchmark results, model skill metrics, initialization and epoch records, cell-state normalization files, Akaike Information Criterion (AIC)-based model comparison files, and data used to generate manuscript figures. Tabular files can be opened using standard spreadsheet software or Python/R data-analysis tools. Python scripts, Jupyter notebooks, and selected MATLAB scripts are included for model execution, postprocessing, plotting, and statistical analysis. Quality assurance and quality control were conducted through the source-data selection and modeling workflow. Meteorological forcing, streamflow, and static catchment attributes were derived from the CAMELS-US dataset, and snow water equivalent data were derived from the University of Arizona (UA) Snow Water Equivalent dataset. Selected basins and time periods were screened during the associated research workflow to avoid missing observations or poor-quality cases. Static geospatial features were processed primarily using Quantum Geographic Information System (QGIS) and Geospatial Data Abstraction Library (GDAL) workflows. Additional details are provided in the associated manuscript and documentation.

ESS-DIVE CSV File Formatting Guidelines Reporting ↗
DOE OSTI2025

Exploring the Whole Set of Accurate Sparse Interpretable Models

In data science applications, there are often many models that fit the data well. This phenomenon was called the Rashomon Effect by Leo Breiman. The set of good models is called the Rashomon Set, and the goal of this project is to locate, store, and study the Rashomon sets for classes of interpretable models, including decision trees and generalized additive models.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

An interpretable model of pre-mRNA splicing for animal and plant genes

Pre-mRNA splicing is a fundamental step in gene expression, conserved across eukaryotes, in which the spliceosome recognizes motifs at the 3' and 5' splice sites (SSs), excises introns, and ligates exons. SS recognition and pairing is often influenced by protein splicing factors (SFs) that bind to splicing regulatory elements (SREs). Here, we describe SMsplice, a fully interpretable model of pre-mRNA splicing that combines models of core SS motifs, SREs, and exonic and intronic length preferences. We learn models that predict SS locations with 83 to 86% accuracy in fish, insects, and plants and about 70% in mammals. Learned SRE motifs include both known SF binding motifs and unfamiliar motifs, and both motif classes are supported by genetic analyses. Our comparisons across species highlight similarities between non-mammals, increased reliance on intronic SREs in plant splicing, and a greater reliance on SREs in mammalian splicing.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Interpretive modeling of tungsten divertor leakage during experiments with neon gas seeding

Abstract Many existing and future tokamaks with tungsten divertors operate, or will operate, with low- Z impurity seeding, but the direct effect of these seeded impurities on tungsten Scrape-off-Layer (SOL) transport has not been explored in detail. This paper reports on a DIII-D experiment designed to test how tungsten divertor leakage from the Small-Angle Slot V-Shaped, tungsten-coated divertor is impacted by neon seeding at a variety of injection rates and poloidal injection locations. Measurements from the experiment show an inverse relationship between the neon injection rate and the tungsten core penetration factor. Interpretive modeling is performed with a combination of the SOLPS-ITER and DIVIMP codes to assess the underlying tungsten behavior. The modeling results show that the reduction in tungsten divertor leakage is driven by both an increase in the divertor collisionality as well as a reduction in the ion temperature gradient near the divertor target. Collisions between low- Z impurities and tungsten impurities are found to have a significant impact on the tungsten SOL transport, such that ignoring the low- Z impurity collisional effects on the tungsten transport can result in an overestimate of the divertor leakage by an order-of-magnitude. Given the importance of these localized interactions, neon seeding from the closed, slot-like divertor has a clear advantage in being able to reduce tungsten divertor leakage without the high levels of neon core contamination that occur when seeding from other poloidal locations.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

Super-X and conventional divertor configurations in MAST-U ohmic L-mode; a comparison facilitated by interpretative modelling

Measurements are presented, alongside corresponding interpretative SOLPS-ITER simulations, of the first MAST-U experiments comparing ohmically heated L-mode fuelling scans in Conventional divertor (CD) and Super-X divertor (SXD) configurations. In experiment, at comparable outer mid-plane separatrix electron density, $n_{e,\textrm{sep,OMP}}$, the maximum lower outer target heat load was found to be a factor 16 $\,\pm\,7$ lower in SXD compared to CD. In simulation, a factor 26.8 reduction was found (slightly higher than the experimental range), suggesting an additional reduction in SXD compared to the factor 9.3 expected from geometric considerations alone. According to the simulations, this additional reduction in the SXD is due to a net radial transport of the energy remaining downstream of the $T_e = 5$ eV location. This energy is carried out of the critical (highest heat load) flux tube by deuterium atoms, demonstrating the importance of a longer legged divertor which provides space for this to occur. Importantly, in both simulation and experiment, the SXD has minimal impact on the upstream n e and T e profiles. Spectral inferences of detachment front movement in SXD compare well between simulation and experiment. In regions of high magnetic field gradient, the parallel movement of the front towards the X-point becomes less sensitive to increasing $n_{e,\textrm{sep,OMP}}$, in qualitative agreement with simplified models and previous predictive simulations. Additional aspects, regarding the target ion flux rollover, upstream separatrix temperature and drift effects, are also presented and discussed.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2023

An accurate and interpretable model for antimicrobial resistance in pathogenic Escherichia coli from livestock and companion animal species

Understanding the microbial genomic contributors to antimicrobial resistance (AMR) is essential for early detection of emerging AMR infections, a pressing global health threat in human and veterinary medicine. Here we used whole genome sequencing and antibiotic susceptibility test data from 980 disease causing Escherichia coli isolated from companion and farm animals to model AMR genotypes and phenotypes for 24 antibiotics. We determined the strength of genotype-to-phenotype relationships for 197 AMR genes with elastic net logistic regression. Model predictors were designed to evaluate different potential modes of AMR genotype translation into resistance phenotypes. Our results show a model that considers the presence of individual AMR genes and total number of AMR genes present from a set of genes known to confer resistance was able to accurately predict isolate resistance on average (mean F 1 score = 98.0%, SD = 2.3%, mean accuracy = 98.2%, SD = 2.7%). However, fitted models sometimes varied for antibiotics in the same class and for the same antibiotic across animal hosts, suggesting heterogeneity in the genetic determinants of AMR resistance. We conclude that an interpretable AMR prediction model can be used to accurately predict resistance phenotypes across multiple host species and reveal testable hypotheses about how the mechanism of resistance may vary across antibiotics within the same class and across animal hosts for the same antibiotic.

Chung, Henri C.↗
DOE OSTI2024

Extreme sparsification of physics-augmented neural networks for interpretable model discovery in mechanics

Data-driven constitutive modeling with neural networks has received increased interest in recent years due to its ability to easily incorporate physical and mechanistic constraints and to overcome the challenging and time-consuming task of formulating phenomenological constitutive laws that can accurately capture the observed material response. However, even though neural network-based constitutive laws have been shown to generalize proficiently, the generated representations are not easily interpretable due to their high number of trainable parameters. Sparse regression approaches exist that allow for obtaining interpretable expressions, but the user is tasked with creating a library of model forms which by construction limits their expressiveness to the functional forms provided in the libraries. Here, in this work, we propose to train regularized physics-augmented neural network-based constitutive models utilizing a smoothed version of $L^0$-regularization. This aims to maintain the trustworthiness inherited by the physical constraints, but also enables interpretability which has not been possible thus far on any type of machine learning-based constitutive model where model forms were not assumed a priori but were actually discovered. During the training process, the network simultaneously fits the training data and penalizes the number of active parameters, while also ensuring constitutive constraints such as thermodynamic consistency. We show that the method can reliably obtain interpretable and trustworthy constitutive models for compressible and incompressible hyperelasticity, yield functions, and hardening models for elastoplasticity, using synthetic and experimental data. This work aims to set a new paradigm for interpretable machine learning models in the broad area of solid mechanics where low and limited data is available along with prior knowledge of physical constraints that the learned maps need to obey. This paradigm can potentially be extended to a broader spectrum of scientific exploration.

Data-driven constitutive models↗
DOE OSTI2022

Hypothesis testing via AI: Generating physically interpretable models of scientific data with machine learning (Full Technical Report)

Deep learning has demonstrated an exceptional ability to solve complex tasks (an engineering success); however, it has done so at the expense of the ability to generate new knowledge (a scientific failure). We propose an alternative framework—entitled Deep Symbolic Regression (DSR)—in which artificial neural networks (NNs) rapidly generate hypotheses about physical relationships among inputs. This framework bypasses the need to interpret an NN altogether, while still leveraging the representational power of deep learning. The resulting models are tractable mathematical expressions, which are inherently and readily human interpretable and can provide insights into underlying physical phenomena. Further, we fold this methodology into the scientific process by allowing the scientist to directly integrate a priori knowledge and beliefs to accelerate learning. We demonstrate this methodology on symbolic regression—the problem of rediscovering underlying expressions describing a dataset—and achieve state-of-the-art performance across a wide variety of symbolic regression problems. Further, we generalize our DSR framework to apply to the more general class of symbolic optimization problems, in which one seeks to optimize a sequence of symbols or “tokens” under a black-box reward function. Examples of other symbolic optimization problems include neural architecture search and computational antibody design. Our generalized tool, Deep Symbolic Optimization (DSO), has been demonstrated on the task of learning symbolic control policies for reinforcement learning environments, and has been adopted as an enabling capability for computational antibody design.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Interpretive modelling of boron transport in the boundary plasma of WEST experiments with the impurity powder dropper

Abstract Boron (B) powder injection is a potential alternative to glow discharge boronization as a wall conditioning method for tokamaks. This technique is currently being studied in WEST experiments, during which B powder is injected by an Impurity Powder Dropper developed by PPPL. In order to interpret and analyse experimental trends, and to help develop future experiments, a modelling workflow using a boundary plasma simulation (SOLEDGE-EIRENE) and powder ablation simulation (Dust Injection Simulator) was developed and tested. The effect of adding a B neutral source to simulated deuterium + oxygen (D + O) plasmas was compared to experimental data from the WEST C5 campaign, where B powder was injected in a dedicated experiment. While the impact of B injection on radiated power P rad measurements at the upper divertor was similar, there were significant differences in measurements of P rad , outer strike point electron temperature T e OSP and O-II line intensity at the lower divertor between experiment and simulation. This discrepancy suggests that those parameters were affected by phenomena not present in the simulations, with the most likely candidates being reduced D recycling and a reduced O sourcing from the divertor.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2023

Lifetime measurements in 206 Po with a shell-model interpretation

The lifetimes of the first excited 2⁺ and 4⁺ states in 206 Po were measured using the recoil-distance Dopplershift method. The experimental results were compared to large-scale shell-model calculations that describe the deduced transition probabilities well. Those calculations were extended to the neighboring 204,208 Po isotopes giving a good overall description of the yrast states. However, the calculations underpredict the energies of the 6+1 and 8+1 states, which suggests that further improvement of the proton-neutron interaction is required.

190 ≤ A ≤ 219↗
DOE OSTI2025

Challenges and approaches to interpretive modeling of boundary plasma and neutral transport in a closed, pumped divertor

An experimental discharge from the DIII-D tokamak is modeled using the SOLPS-ITER code suite and compared against measurements in the pumped and relatively closed upper divertor. Uncertainties of boundary plasma simulations are identified by attempting to match code inputs to experimental conditions, including iteratively solving transport coefficients to match upstream experimental profiles using varying quantities of core particle flux, different pumping models, and various assumptions of ion thermal transport. Simulated boundary conditions for particle injection at the core interface are shown to be relevant to the plasma solution at the divertor targets, even if upstream transport is modified so that plasma profiles are comparatively similar, although seperatrix density is not held constant. When upstream plasma profiles are matched to experimental measurements by varying diffusive transport coefficients, using either poloidally symmetric or ballooning structure, the model finds a majority of injected energy being transported radially off the computational domain, in conflict with experimental radiated power measurements and heat flux measurements at the divertor target. Imposing a maximum thermal diffusivity or radially shifting the experimental separatrix location of the fitted profiles to increase power conducted to the targets by increasing the upstream electron temperature does not significantly modify this result. Including a thermalizing plenum volume in the simulation domain is shown to maintain the experimental volumetric pumping rate without knowing the neutral energy distribution incident on the pump duct a priori. By modifying transport parameters to match different assumptions for ion temperature, downstream neutral pressure changes by more than a factor of two, suggesting that attention to ion thermal transport may be a critical parameter for simulations to accurately resolve recycling and neutral transport, particularly in a closed divertor geometry. In addition to quantifying various modeling uncertainties, this work motivates both further experimental study and modeling improvements to improve predictive capabilities.

divertor↗
DOE OSTI2023

Prediction of Long-Term Geochemical Change in Bentonite Based on the Interpretative THMC Model of the FEBEX In Situ Test

Since nuclear energy is crucial in the decarbonization of the energy supply, one hurdle to remove is the handling of high-level radioactive waste (HLW). Disposal of HLW in a deep geological repository has long been deemed a viable permanent option. In the design of a deep geological repository, compacted bentonite is the most commonly proposed buffer material. Predicting the long-term chemical evolution in bentonite, which is important for the safety assessment of a repository, has been challenging because of the complex coupled processes. Models for large-scale tests and predictions based on such models have been some of the best practices for such purposes. An 18-year-long in situ test with two dismantling events provided a unique set of chemical data that allowed for studying chemical changes in bentonite. In this paper, we first developed coupled thermal, hydrological, mechanical, and chemical (THMC) models to interpret the geochemical data collected in the in situ test and then extended the THMC model to 200 years to make long-term prediction of the geochemical evolution of bentonite. The interpretive coupled THMC model shows that the geochemical profiles were strongly affected by THM processes such as evaporation/condensation, porosity change caused by swelling, permeability change, and the shape of concentration profiles for major cations were largely controlled by transport processes, but concentration levels were regulated by chemical reactions, and the profiles of some species such as pH, bicarbonate, and sulfate were dominated by these reactions. The long-term THMC model showed that heating prolongs the time that bentonite becomes fully saturated in the area close to the heater/canister; however, once the bentonite becomes fully saturated, high concentrations of ions in bentonite near the heater, which was observed in the field test, will disappear; illitization continues for 50 years but will not proceed further.

Zheng, Liange↗
DOE OSTI2024

Integrating adaptive learning with post hoc model explanation and symbolic regression to build interpretable surrogate models

Abstract We develop a materials informatics workflow to build an interpretable surrogate model for micromagnetic simulations. Our goal is to predict the energy barrier of a moving isolated skyrmion in rare-earth-free $$\hbox {Mn}_4$$ Mn 4 N. Our approach integrates adaptive learning with post hoc model explanation and symbolic regression methods. We discuss an unexplored acquisition function (information condensing active learning) within the adaptive learning loop and compare it with the known standard deviation function for efficient navigation of the search space. Model-agnostic post hoc explanation techniques then uncover trends learned by the trained model, which we then leverage to constrain the expressions used for symbolic regression. Graphical abstract

Biswas, Ankita↗