Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “constrained learning”

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 361 records · Page 20
All sourcesNASA NTRSDOE OSTI
NASA NTRS2017

Lessons Learned from Preparing OSIRIS-REx Spectral Analog Samples for Bennu

NASA's OSIRIS-REx sample return mission launched on September 8th, 2016 to rendezvous with B-type asteroid (101955) Bennu in 2018. Type C and B asteroids have been linked to carbonaceous chondrites because of their similar visible - to - near infrared (VIS-NIR) spectral properties [e.g., 1,2]. The OSIRIS-REx Visible and Infrared Spectrometer (OVIRS) and the Thermal Emission Spectrometer (OTES) will make spectroscopic observations of Bennu during the encounter. Constraining the presence or absence of hydrous minerals (e.g., Ca-carbonate, phyllosilicates) and organic molecules will be key to characterizing Bennu [3] prior to sample site selection. The goal of this study was to develop a suite of analog and meteorite samples and obtain their spectral properties over the wavelength ranges of OVIRS (0.4- 4.3 micrometer) and OTES (5.0-50 micrometer). These spectral data were used to validate the mission science-data processing system. We discuss the reasoning behind the study and share lessons learned.

Schrader, D. L.↗
DOE OSTI2025

RLMolLM: Reinforcement Learning-Enhanced Language Model Framework for Inverse Molecular Design

Inverse molecular design faces significant challenges due to vast chemical space and complex property requirements. While language models show promise for molecular generation, they struggle with validity, multi-property optimization, and structural constraints. This work presents RLMolLM, a reinforcement learning framework combining Proximal Policy Optimization (PPO) with genetic algorithms to address these limitations. Our approach optimizes multiple user-specified properties including quantitative estimates of drug-likeness (QED), synthetic accessibility (SA), and ADMET (absorption, distribution, metabolism, excretion, and toxicity) endpoints without requiring complete model retraining, while maintaining capability for scaffold-constrained generation where specific substructures must be preserved. We outperform state-of-the-art methods for molecular optimization, achieving best QED scores across GDB13, Moses, and Zinc datasets with up to 31% improvement over previous methods while maintaining excellent validity, uniqueness, and novelty metrics. For simultaneous multi-property optimization, our framework achieves substantial improvements in ADMET properties including 4.5-fold reduction in hERG toxicity and enhanced Caco-2 permeability compared to Moses dataset. Under structural constraints, the framework significantly improves molecular validity while preserving scaffolds and effectively optimizing properties. In conclusion, this versatile solution advances pharmaceutical and materials molecular design through effective integration of reinforcement learning and genetic algorithms with multi-property optimization and scaffold preservation.

Genetic algorithms↗
DOE OSTI2023

Linear model decision trees as surrogates in optimization of engineering applications

Machine learning models are promising as surrogates in optimization when replacing difficult to solve equations or black-box type models. This work demonstrates the viability of linear model decision trees as piecewise-linear surrogates in decision-making problems. Linear model decision trees can be represented exactly in mixed-integer linear programming (MILP) and mixed-integer quadratic constrained programming (MIQCP) formulations. Furthermore, they can represent discontinuous functions, bringing advantages over neural networks in some cases. We present several formulations using transformations from Generalized Disjunctive Programming (GDP) formulations and modifications of MILP formulations for gradient boosted decision trees (GBDT). We then compare the computational performance of these different MILP and MIQCP representations in an optimization problem and illustrate their use on engineering applications. Importantly, we observe faster solution times for optimization problems with linear model decision tree surrogates when compared with GBDT surrogates using the Optimization and Machine Learning Toolkit (OMLT).

42 ENGINEERING↗
DOE OSTI2022

Rethinking the ill-posedness of the spectral function reconstruction — Why is it fundamentally hard and how Artificial Neural Networks can help

Reconstructing hadron spectral functions through Euclidean correlation functions are of the important missions in lattice QCD calculations. However, in a Källen–Lehmann (KL) spectral representation, the reconstruction is observed to be ill-posed in practice. It is usually ascribed to the fewer observation points compared to the number of points in the spectral function. In this paper, by solving the eigenvalue problem of continuous KL convolution, we show analytically that the ill-posedness of the inversion is fundamental and it exists even for continuous correlation functions. We discussed how to introduce regulators to alleviate the predicament, in which include the Artificial Neural Networks (ANNs) representations recently proposed by the Authors in another study. The uniqueness of solutions using ANNs representations is manifested analytically and validated numerically. Reconstructed spectral functions using different regularization schemes are also demonstrated, together with their eigen-mode decomposition. We observe that components with large eigenvalues can be reliably reconstructed by all methods, whereas those with low eigenvalues need to be constrained by regulators.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Search for nonresonant Higgs boson pair production in final state with two bottom quarks and two tau leptons in proton-proton collisions at $\sqrt{s}$ = 13 TeV

A search for the nonresonant production of Higgs boson pairs (HH ) via gluon-gluon and vector boson fusion processes in final states with two bottom quarks and two tau leptons is presented. The search uses data from proton-proton collisions at a center-of-mass energy of $\sqrt{s}$ = 13 TeV recorded with the CMS detector at the LHC, corresponding to an integrated luminosity of 138 fb -1 . Events in which at least one tau lepton decays hadronically are considered and multiple machine learning techniques are used to identify and extract the signal. The data are found to be consistent, within uncertainties, with the standard model (SM) predictions. Upper limits on the HH production cross section are set to constrain the parameter space for anomalous Higgs boson couplings. The observed (expected) upper limit at 95% confidence level corresponds to 3.3 (5.2) times the SM prediction for the inclusive HH cross section and to 124 (154) times the SM prediction for the vector boson fusion HH cross section. At 95% confidence level, the Higgs field self-coupling is constrained to be within -1.7 and 8.7 times the SM expectation, and the coupling of two Higgs bosons to two vector bosons is constrained to be within -0.4 and 2.6 times the SM expectation.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
NASA NTRS1997

Using APEX to Model Anticipated Human Error: Analysis of a GPS Navigational Aid

The interface development process can be dramatically improved by predicting design facilitated human error at an early stage in the design process. The approach we advocate is to SIMULATE the behavior of a human agent carrying out tasks with a well-specified user interface, ANALYZE the simulation for instances of human error, and then REFINE the interface or protocol to minimize predicted error. This approach, incorporated into the APEX modeling architecture, differs from past approaches to human simulation in Its emphasis on error rather than e.g. learning rate or speed of response. The APEX model consists of two major components: (1) a powerful action selection component capable of simulating behavior in complex, multiple-task environments; and (2) a resource architecture which constrains cognitive, perceptual, and motor capabilities to within empirically demonstrated limits. The model mimics human errors arising from interactions between limited human resources and elements of the computer interface whose design falls to anticipate those limits. We analyze the design of a hand-held Global Positioning System (GPS) device used for radical and navigational decisions in small yacht recalls. The analysis demonstrates how human system modeling can be an effective design aid, helping to accelerate the process of refining a product (or procedure).

VanSelst, Mark↗
DOE OSTI2026

Statistical and Machine Learning Approaches to Analyzing Pipeline Incidents in the United States (2010–2024)

This study applies machine learning methods to analyze natural gas pipeline incidents in the United States using the Pipeline and Hazardous Materials Safety Administration (PHMSA) Gas Distribution Incident Dataset (2010–2024). The dataset includes over 600 variables describing incident characteristics, infrastructure attributes, and contributing factors associated with unintentional gas releases. The objective is to assess whether these features can reliably predict the underlying cause of pipeline failures. Multinomial logistic regression and Random Forest models were developed to classify incident causes, including excavation damage, corrosion, equipment failure, and natural forces. Results show that excavation damage is both the most frequent and most predictable cause, with models achieving strong performance for this category. However, when excavation damage is excluded, model accuracy declines significantly, with some models performing near random levels. Across all approaches, severe class imbalance and limited variability in key predictors constrain predictive performance. Pipeline age and diameter emerge as the most influential variables, but they provide insufficient discriminatory power to distinguish among less frequent failure types. These findings indicate that non-excavation-related incidents are rare, heterogeneous, and weakly represented in the dataset, limiting the effectiveness of machine learning classification. Overall, this study highlights the structural limitations of the PHMSA dataset for predictive modeling and underscores the need for improved data balance and feature enrichment. The results reinforce excavation damage prevention as the most impactful strategy for reducing pipeline incidents.

03 NATURAL GAS↗
DOE OSTI2025

Combining Observations and Models: A Review of the CARDAMOM Framework for Data‐Constrained Terrestrial Ecosystem Modeling

The rapid increase in the volume and variety of terrestrial biosphere observations (i.e., remote sensing data and in situ measurements) offers a unique opportunity to derive ecological insights, refine process‐based models, and improve forecasting for decision support. However, despite their potential, ecological observations have primarily been used to benchmark process‐based models, as many past and current models lack the capability to directly integrate observations and their associated uncertainties for parameterization. In contrast, data assimilation frameworks such as the CARbon DAta MOdel fraMework (CARDAMOM) and its suite of process‐based models, known as the Data Assimilation Linked Ecosystem Carbon Model (DALEC), are specifically designed for model‐data fusion. This review, motivated by a recent CARDAMOM community workshop, examines the development and applications of CARDAMOM, with an emphasis on its role in advancing ecosystem process understanding. CARDAMOM employs a Bayesian approach, using a Markov Chain Monte Carlo algorithm to enable data‐driven calibration of DALEC parameters and initial states (i.e., carbon pool sizes) through observation operators. CARDAMOM's unique ability to retrieve localized model process parameters from diverse datasets—ranging from in situ measurements to global satellite observations—makes it a highly flexible tool for analyzing spatially variable ecosystem responses to environmental change. However, assimilating these data also presents challenges, including data quality issues that propagate into model skill, as well as trade‐offs between model complexity, parameter equifinality, and predictive performance. We discuss potential solutions to these challenges, such as reducing parameter equifinality by incorporating new observations. This review also offers community recommendations for incorporating emerging datasets, integrating machine learning techniques, strengthening collaboration with remote sensing, field, and modeling communities, and expanding CARDAMOM's relevance for localized ecosystem monitoring and decision‐making. CARDAMOM enables a deep, mechanistic understanding of terrestrial ecosystem dynamics that cannot be achieved through empirical analyses of observational datasets or weakly constrained models alone.

Bayesian inference↗
DOE OSTI2025

Using Machine Learning to Generate a GISS ModelE Calibrated Physics Ensemble (CPE)

A neural network (NN) surrogate of the NASA GISS ModelE atmosphere (version E3) is trained on a perturbed parameter ensemble (PPE) spanning 45 physics parameters and 36 outputs. The NN is leveraged in a Markov Chain Monte Carlo (MCMC) Bayesian parameter inference framework to generate a second posterior constrained ensemble coined a “calibrated physics ensemble,” or CPE. The CPE members are characterized by diverse parameter combinations and are, by definition, close to top-of-atmosphere radiative balance, and must broadly agree with numerous hydrologic, energy cycle and radiative forcing metrics simultaneously. Global observations of numerous cloud, environment, and radiation properties (provided by global satellite products) are crucial for CPE generation. The inference framework explicitly accounts for discrepancies (or biases) in satellite products during CPE generation. We demonstrate that product discrepancies strongly impact calibration of important model parameter settings (e.g., convective plume entrainment rates; fall speed for cloud ice). Structural improvements new to E3 are retained across CPE members (e.g., stratocumulus simulation). Notably, the framework improved the simulation of shallow cumulus and Amazon rainfall while not degrading radiation fields, an upgrade that neither default parameters nor Latin Hypercube parameter searching achieved. Analyses of the initial PPE suggested several parameters were unimportant for output variation. However, many “unimportant” parameters were needed for CPE generation, a result that brings to the forefront how parameter importance should be determined in PPEs. From the CPE, two diverse 45-dimensional parameter configurations are retained to generate radiatively-balanced, auto-tuned atmospheres that were used in two E3 submissions to CMIP6.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Peatland fires in Alaska will double by the end of the century

During recent summers, warm and dry conditions have increased the occurrence of wildfires and potentially peat-fires across Alaska. Limitations in resolving the fine-scale distribution of peatlands and climate observations have constrained our ability to accurately predict peat-fire dynamics. Using a new high-resolution peatland map of Alaska, we evaluated the climate and environmental controls of past and future peat-fire activity. Ensemble machine learning models identified reduced soil moisture, higher temperatures, and evapotranspiration as key predictors of annual total burned peatland area (tenfold CV R 2 = 0.62, RMSE = 221.1 km 2 ). By the end of the twenty-first century, models forced with climate datasets from representative concentration pathways (RCPs) 4.5, 6.0, and 8.5 emission scenarios project a statewide doubling of burned peatlands (increasing 61–121%), with regional increases ranging from 25–165% in polar, 61–95% in boreal, and 102–106% in maritime ecoregions. These projections indicate that wildfires will progressively encroach further into organic-rich moist and wet peaty soils, potentially amplifying soil carbon release across Alaska.

climate-change ecology↗
DOE OSTI2024

Search for long-lived particles using displaced vertices and missing transverse momentum in proton-proton collisions at s = 13 TeV

A search for the production of long-lived particles in proton-proton collisions at a center-of-mass energy of 13 TeV at the CERN LHC is presented. The search is based on data collected by the CMS experiment in 2016–2018, corresponding to a total integrated luminosity of 137 fb − 1 . This search is designed to be sensitive to long-lived particles with mean proper decay lengths between 0.1 and 1000 mm, whose decay products produce a final state with at least one displaced vertex and missing transverse momentum. A machine learning algorithm, which improves the background rejection power by more than an order of magnitude, is applied to improve the sensitivity. The observation is consistent with the standard model background prediction, and the results are used to constrain split supersymmetry (SUSY) and gauge-mediated SUSY breaking models with different gluino mean proper decay lengths and masses. This search is the first CMS search that shows sensitivity to hadronically decaying long-lived particles from signals with mass differences between the gluino and neutralino below 100 GeV. It sets the most stringent limits to date for split-SUSY models and gauge-mediated SUSY breaking models with gluino proper decay length less than 6 mm. © 2024 CERN, for the CMS Collaboration 2024 CERN

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2020

Predicting Volume of Distribution in Humans: Performance of In Silico Methods for a Large Set of Structurally Diverse Clinical Compounds

Volume of distribution at steady state (V D,ss ) is one of the key pharmacokinetic parameters estimated during the drug discovery process. Despite considerable efforts to predict V D,ss , accuracy and choice of prediction methods remain a challenge, with evaluations constrained to a small set (<150) of compounds. To address these issues, a series of in silico methods for predicting human V D,ss directly from structure were evaluated using a large set of clinical compounds. Machine learning (ML) models were built to predict V D,ss directly and to predict input parameters required for mechanistic and empirical V D,ss predictions. In addition, log D, fraction unbound in plasma (fup), and blood-to-plasma partition ratio (BPR) were measured on 254 compounds to estimate the impact of measured data on predictive performance of mechanistic models. Furthermore, the impact of novel methodologies such as measuring partition (Kp) in adipocytes and myocytes (n = 189) on V D,ss predictions was also investigated. In predicting V D,ss directly from chemical structures, both mechanistic and empirical scaling using a combination of predicted rat and dog V D,ss demonstrated comparable performance (62%–71% within 3-fold). The direct ML model outperformed other in silico methods (75% within 3-fold, r 2 = 0.5, AAFE = 2.2) when built from a larger data set. Scaling to human from predicted V D,ss of either rat or dog yielded poor results (<47% within 3-fold). Measured fup and BPR improved performance of mechanistic V D,ss predictions significantly (81% within 3-fold, r 2 = 0.6, AAFE = 2.0). Adipocyte intracellular Kp showed good correlation to the V D,ss but was limited in estimating the compounds with low V D,ss .

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

Learning Stochastic Parametric Differentiable Predictive Control Policies

We present a scalable unsupervised learning-based method for obtaining explicit control policies for model predictive control problems for stochastic linear systems with additive uncertainties subject to nonlinear chance constraints. We call the proposed method stochastic parametric differentiable predictive control (SP-DPC), which extends the recently proposed deterministic DPC policy optimization algorithm. We formulate the SP-DPC as a deterministic approximation to the stochastic parametric constrained optimal control problem via independent sampling of the problem's parameters and uncertainties. This formulation allows us to directly compute the policy gradients via automatic differentiation of the problem's value function, evaluated over sampled parameters and uncertainties. In particular, the computed expectation of the problem's value function is backpropagated through the finite-time closed-loop system rollouts parametrized by a known nominal system dynamics model and neural control policy. We also provide theoretical probabilistic guarantees on closed-loop stability and chance constraints satisfaction for systems controlled by learned neural policies. We demonstrate the computational efficiency and scalability of the proposed policy optimization algorithm in three numerical examples, including systems with a large number of states or subject to nonlinear constraints.

Drgona, Jan↗
DOE OSTI2026

Plasma confinement state classification via FPP relevant microwave diagnostics

We present a parsimonious and robust machine learning approach for identifying plasma confinement states in fusion power plants (FPPs) where reliable identification of the low-confinement and high-confinement regimes is critical for safe and efficient operation. Unlike research-oriented devices, FPPs must operate with a severely constrained set of diagnostics. To address this challenge, we demonstrate that a minimalist model, using only electron cyclotron emission (ECE) signals, can achieve accurate and reliable state classification. ECE provides electron temperature profiles without the engineering or survivability issues of in-vessel probes, making it a primary candidate for FPP-relevant diagnostics. Our framework employs ECE as input, extracts features using radial basis functions, and applies a gradient boosting classifier, achieving a test accuracy of 96% (correct predictions). Robustness analysis and feature importance analyzes confirm the approach’s reliability. These results demonstrate that state-of-the-art performance is attainable from a restricted diagnostic set, paving the way for minimalist yet resilient plasma control architectures for FPPs.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

Molecular Vision - Multimodal, multitask retrieval of molecular structure from measured signatures for reference-free compound identification

We are currently at risk of generating false conclusions based on limited methods to identify small molecules in biological systems and in chemical forensics. By definition, the chemical structures of novel small molecules have not been determined, let alone measured or synthesized. Currently, unambiguous structure determination of small molecules is constrained by the time and effort needed to isolate compounds and perform de novo structure elucidation using laboratory-based methods, significantly extending the time to inform mitigation strategies. To address this gap, we have developed a deep learning approach to directly map molecular structure to experimental signatures. We aim to unify measurement technologies employed in untargeted small molecule identification studies—such as infrared (IR) spectrometry, tandem mass spectrometry (MS/MS), ion mobility spectrometry-derived collision cross section (CCS)—through use of a multimodal, multitask deep learning architecture. Where existing methods require direct generation of information-rich spectra and/or properties, an inherently difficult task, we will simplify molecular signature-based identification by posing the problem as a recognition or retrieval task. The model is thus presented with relevant endpoints – structure and one or more molecular signatures – and need only determine whether they are semantically related. Thus, our approach offers the following advantages over existing techniques: (i) circumvents difficulties associated with direct generation of molecular signatures from structure and structure from signatures; (ii) incorporates multiple molecular signatures simultaneously, as available, to support identification; and (iii) enables rapid computation of structural embeddings toward broad coverage of known chemical space. Taken together, the approach removes the need to explicitly obtain or compute reference spectra, representing a powerful method for compound identification that requires only experimentally observed signatures.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Sim2Real Autonomous Robotic Exploration [Poster]

Autonomous robots offer promising solutions for exploration in environments that are inaccessible or hazardous to humans. Despite this, physical training of such robots is often constrained by safety risks, high cost or limited accessibility. This project presents an end-to-end simulation to reality pipeline leveraging Nvidia Isaac Sim and Boston Dynamics' Spot to enable autonomous navigation in indoor environments. A reinforcement learning policy is first trained using Nvidia Isaac Lab to establish Spot's locomotion pattern. Virtual LiDAR sensors are then integrated to perform SLAM-based navigation using simulated odometry. Finally, the simulated navigation scheme is transferred to a physical Spot robot to inspect and record images of a real-world room by repeating the learnt trajectory. The proposed framework highlights the potential of scalable training in simulation and reliable deployment in physical environments. Future directions include dynamic trajectory generation in unseen and challenging environments and integration of environmental sensing like temperature, radiation or humidity via sensor and material simulation.

97 - MATHEMATICS AND COMPUTING↗
DOE OSTI2026

AI/ML-Enhanced Wind Forecasts for Reducing Uncertainty in Prescribed Fire Planning

Prescribed fire is a vital tool for ecosystem management and wildfire risk reduction but its escalation is constrained by overly conservative burn windows because of uncertainties, for instance, in wind forecasts. This review describes the state of the art in weather product use by fire/smoke models and identifies three priority research gaps that artificial intelligence/machine learning (AI/ML) is well positioned to address: (1) spatial and temporal downscaling to meter-scale, sub-hourly wind fields; (2) bias correction for systematic model errors in complex terrain; and (3) robust uncertainty quantification to inform ensemble-based simulations. Emerging AI/ML techniques offer promising frameworks to address all three challenges. By providing high-resolution, bias-corrected, and probabilistic wind fields, AI/ML-enhanced forecasts will allow for expanded burn windows, improved ignition strategy design and a reduced reliance on expert intuition, especially when a prescribed fire is introduced into new areas.

54 ENVIRONMENTAL SCIENCES↗