SEARCH · Search NASA

Results for “probabilistic computing”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 613 records · Page 34

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Ensemble‐Based, Large‐Eddy Reconstruction of Wind Turbine Inflow in a Near‐Stationary Atmospheric Boundary Layer Through Generative Artificial Intelligence

ABSTRACT To validate the second‐by‐second dynamics of turbines in field experiments, it is necessary to accurately reconstruct the winds going into the turbine. Current time‐resolved inflow reconstruction techniques estimate wind behavior in unobserved regions using relatively simple spectral‐based models of the atmosphere. Here, we develop a technique for time‐resolved inflow reconstruction that is rooted in a large‐eddy simulation model of the atmosphere. Our “large‐eddy reconstruction” technique blends observations and atmospheric model information through a diffusion model machine learning algorithm, allowing us to generate probabilistic ensembles of reconstructions for a single 10‐min observational period. Our generated inflows can be used directly by aeroelastic codes or as inflow boundary conditions in a large‐eddy simulation. We verify the second‐by‐second reconstruction capability of our technique in three synthetic field campaigns, finding positive Pearson correlation coefficient values () between ground‐truth and reconstructed streamwise velocity, as well as smaller positive correlation coefficient values for unobserved fields (spanwise velocity, vertical velocity, and temperature). We validate our technique in three real‐world case studies by driving large‐eddy simulations with reconstructed inflows and comparing to independent inflow measurements. The reconstructions are visually similar to measurements, follow desired power spectra properties, and track second‐by‐second behavior ().

17 WIND ENERGY↗

DOE OSTI2025

LandScan mosaic enables high-resolution gridded population estimates with explicit uncertainty

Gridded population datasets represent high-resolution distributions of human occupancy, enabling informed decision-making across a broad range of fields. These data products are valuable for assessing environmental risk, urban development, disaster preparedness and resource allocation—areas where accurate population estimates directly enhance policy effectiveness and optimize resource distribution. Despite the importance of gridded population datasets, traditional population modeling approaches often overlook inherent uncertainties in the estimation process. This limitation can create a false sense of certainty in population estimates, potentially leading to flawed decisions by those who rely on the data. To address this methodological gap, we introduce a probabilistic machine learning modeling framework, LandScan Mosaic, that explicitly incorporates uncertainty into the population modeling process. Our approach systematically quantifies uncertainty in three key modeling parameters of the LandScan HD gridded population dataset: building use types, floor counts, and occupancy rates. By employing Monte Carlo simulations, we propagate these uncertainties through the modeling process, yielding probability distributions of population counts in place of deterministic point estimates. We demonstrate the practical application of this framework in Iloilo City, Philippines, using structured decision-making techniques and our probabilistic estimates to identify and prioritize areas most affected by projected flooding, supporting targeted interventions that address both economic and social risks. In doing so, we propose a population-specific approach for incorporating confidence into structured decision making processes. Through a comparative analysis with conventional deterministic approaches and point estimate approaches, including LandScan HD and WorldPop, we evaluate how the incorporation of machine learning and uncertainty influences decision rankings. This research advances population distribution modeling by offering a robust, quantitative approach that explicitly accounts for uncertainty in the underlying data, along with guidance for how users can apply uncertainty in their decision-making.

Environmental sciences↗

DOE OSTI2021

Machine learning methods for probabilistic locked-mode predictors in tokamak plasmas

A rotating tokamak plasma can interact resonantly with the external helical magnetic perturbations, also known as error fields. This can lead to locking and then to disruptions. We leverage machine learning (ML) methods to predict the locking events. We use a coupled third-order nonlinear ordinary differential equation model to represent the interaction of the magnetic perturbation and the plasma rotation with the error field. This model is sufficient to describe qualitatively the locking and unlocking bifurcations. Here, we explore using ML algorithms with the simulation data and experimental data, focusing on the methods that can be used with sparse datasets. These methods lead to the possibility of the avoidance of locking in real-time operations. We describe the operational space in terms of two control parameters: the magnitude of the error field and the rotation frequency associated with the momentum source that maintains the plasma rotation. The outcomes are quan- tified by order parameters that completely characterize the state, whether locked or unlocked. We use unsupervised ML methods to classify locked/unlocked states and note the usefulness of a certain normalization of the order parameters. Three supervised ML classifiers are used in suite to estimate the probability of locking in the region of control parameter space with hysteresis, i.e., the set of control parameters for which both locked and unlocked states can exist. The results show that a neural network gives the best estimate of the locking probability. An analogy of the present locking model with the van der Waals equation of state is also provided.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2024

Generative AI for Power Grid Operations

Generative artificial intelligence (AI) has captured into the mainstream, demonstrating capabilities that once belonged solely to the realm of human cognition. From defeating world champions in complex games to generating human-quality text and images, Generative AI has proven its potential to revolutionize countless industries. The electric power grid is no exception. Generative AI's ability to process vast amounts of data rapidly, assist decision support and identify patterns could significantly enhance power grid operations. For example, Generative AI could improve state estimation where measurements are not available or integrate renewable energy sources more efficiently with probabilistic forecasting. The key contributions of this whitepaper are outlined below: (1) Comprehensive overview of Generative AI's applications in power grid operations: It highlights the opportunities in areas such as forecasting, state estimation, and demonstrating the potential for enhancing efficiency, reliability, and resilience. (2) Expanding Generative AI's impact through synergies with emerging technologies: The paper introduce NREL developed eGridGPT and explores how AI orchestration, multi-agent systems, and Digital Twins can collaborate to optimize grid operations, addressing the complexities of a decarbonized and electrified future. (3) In-depth analysis of challenges in implementing Generative AI: This includes considerations like data availability and quality, model validation, certification, and ethical concerns, ensuring responsible AI deployment. (4) Emphasizing human-AI collaboration: The whitepaper underscores the importance of trustworthy, transparency, and explainability in AI systems to promote seamless interaction between human operators and AI, ultimately improving decision-making. (5) Exploring future research and development: It identifies critical areas for further advancement to fully realize Generative AI's potential in power grid operations. This whitepaper serves as a valuable resource for researchers, practitioners, and policymakers looking to harness Generative AI for a more reliable, stable, and cost-effective power grid.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2019

Economic Assessment of Nuclear Hybrid Energy Systems: Nuclear-Renewable-Water Integration in Arizona

One of the goals of the Nuclear-Renewable Hybrid Energy Systems (N-R HES) modeling and simulation (M&S) project is to assess the economic viability of N-R HES in a market that contains Variable Renewable Energy (VRE) sources like solar. The N-R HES M&S project is a multi-laboratory effort, led by Idaho National Laboratory (INL) and supported by Argonne National Laboratory (ANL) and Oak Ridge National Laboratory (ORNL). Development of the N-R HES software framework was initiated at INL in 2016 [1] and has reached some level of maturity such that it can be applied to more than simple demonstration cases, e.g. real industry problems. Nevertheless, more capabilities are constantly added to accommodate the special needs of challenging, real-life problems. The N-R HES framework is built on the Risk Analysis Virtual ENvironment (RAVEN) code [2], which it uses as a driver and workflow manager for all calculations. There are four main cornerstones of the N-R HES simulation framework: 1) The generation of stochastic synthetic time series. 2) A set of probabilistic analysis and optimization algorithms available in RAVEN. 3) A set of models for representing the physical behavior of N-R HES. These can be reduced-order models or high-fidelity Modelica [3] models. 4) The RAVEN plug-in called CashFlow [4] that maps physical performance into economic performance. Within this framework, a broad spectrum of questions related to N-R HES can be addressed. One of the challenges currently of high interest is the increasing penetration of VRE altering the profile of the net demand (demand after removing all non-curtailable renewable energy sources, e.g. rooftop solar). The N-R HES software framework is capable of analyzing the potential feasibility of mitigating the resultant volatility in the net electricity demand. One possible solution to manage net demand volatility is adding stabilizing loads to the grid. These loads can be external industrial processes that are able to ramp production up and down quickly, or for regulated markets, adding variable internal loads that will effectively increase the baseload of the power plant and therefore reduce the risk of internal demand that is lower than the capacity of the power plant. The latter is a possible solution currently being considered by Arizona Public Service (APS), the operating owner of the Palo Verde (PV) Nuclear Power Plant (NPP).

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

An Inventory of AI-ready Benchmark Data for US Fires, Heatwaves, and Droughts

Extreme weather events, including fires, heatwaves, and droughts, have significant impacts on earth, environmental, and energy systems. Mechanistic and predictive understanding, as well as probabilistic risk assessment of these extreme weather events, are crucial for detecting, planning for, and responding to these extremes. Records of extreme weather events provide an important data source for understanding present and future extremes, but the existing data needs preprocessing before it can be used for analysis. Moreover, there are many nonstandard metrics defining the levels of severity or impacts of extremes. In this study, we compile a comprehensive benchmark data inventory of extreme weather events, including fires, heatwaves, and droughts. The dataset covers the period from 2001 to 2020 with a daily temporal resolution and a spatial resolution of 0.5°×0.5° (~55km×55km) over the continental United States (CONUS), and a spatial resolution of 1km × 1km over the Pacific Northwest (PNW) region, together with the co-located and relevant meteorological variables. By exploring and summarizing the spatial and temporal patterns of these extremes in various forms of marginal, conditional, and joint probability distributions, we gain a better understanding of the characteristics of climate extremes. The resulting AI/ML-ready data products can be readily applied to ML-based research, fostering and encouraging AI/ML research in the field of extreme weather. This study can contribute significantly to the advancement of extreme weather research, aiding researchers, policymakers, and practitioners in developing improved preparedness and response strategies to protect communities and ecosystems from the adverse impacts of extreme weather events.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2018

Application of Multivariate Data Analysis Techniques for the Portable Isotopic Neutron Spectroscopy system

The Portable Isotopic Neutron Spectroscopy (PINS) is a commercialized system developed by Idaho National Laboratory (INL) to examine chemical compounds non-destructively, utilizing Prompt Gamma Neutron Activation Analysis (PGNAA) techniques. The PINS systems have been successfully deployed around the world to identify chemical munitions and containers. The PINS system takes advantage of a high-resolution gamma-ray spectrum from a high-purity germanium (HPGe) detector, and gamma-ray peak analysis provides input to its chemical identification logic with the probabilistic decision tree (PDT). Gamma-ray peak analyses, however, require knowledge of pre-selected gamma-ray peaks whose energies, intensities, and origins are well studied, and some peaks might be excluded from the decision making processes due to lack of current nuclear data. In contrast, Multivariate Analysis (MVA) treats a whole gamma-ray spectrum as a collection of multiple variables or a pattern. The effectiveness of chemical identification algorithm is determined by the availability of a wide range of data to train an algorithm to identify chemical fills with accuracy. INL has a collection of gamma-ray spectra of various chemical-fills from the field-deployed PINS systems over the years, and it was envisaged to combine such a database with the principle of multivariate statistics. Therefore, in parallel with the current decision tree algorithm, an MVA-based chemical identification algorithm was developed as an independent verification of the current results. The Principal Component Analysis (PCA) method was adopted to build an MVA-based identification algorithm, using the collected field data to refine and validate the algorithm. A benchmarking study of the MVA-based algorithm’s performance was conducted alongside the current algorithm, with the results presented and discussed in this study.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2019

A probabilistic perspective on thermodynamic parameter uncertainties: Understanding aqueous speciation of mercury

Speciation plays an important role in determining the fate and transport of metals in terrestrial surface and subsurface systems. Equilibrium speciation modeling in aqueous systems relies on thermodynamic constants (log K values) of complexes, which are subject to uncertainties. In this study, using Monte Carlo (MC) simulations with Latin hypercube sampling (LHS) we systematically analyze the propagation of thermodynamic constant uncertainty through speciation modeling of an inorganic mercury-sulfide-chloride-water system. We find that seemingly small variances of the input log K normal distributions can lead to output species concentrations spanning multiple orders of magnitude, with highly skewed probability distributions. When equilibrium with mineral metacinnabar (β-HgS(s)) is neglected, the relative uncertainty of each output species is strongly positively correlated with the skewness of its concentration probability distribution, i.e., the lowest uncertainty occurs when the species concentration probability distribution is the most negatively skewed as its concentration approaches the total element concentration limit. The highest uncertainty in the identity of dominant species is located around species equivalence points. For cases where the mineral equilibrium is included, we derive analytical probability density functions for the log concentrations of all major species in the system. The mineral log K uncertainty is found to be an important contributor to all output concentration uncertainties. The analysis of combined effects of both pH and total sulfide concentration on output concentration uncertainties shows that high concentration uncertainties occur under highly sulfidic alkaline conditions and low uncertainties at low pH and sulfide concentrations. Finally, an analysis as presented here can distinguish between conditions that require a full uncertainty analysis and those for which the classical deterministic speciation modeling suffices.

58 GEOSCIENCES↗

DOE OSTI2026

Throughput Estimation of Data Transport Networks From Digital Twin Measurements

Digital twins of networked infrastructures, known as Virtual Infrastructure Twins (VITs), are increasingly used for software development, pre-deployment testing, and design space exploration. While VITs avoid the costs and potential disruptions associated with experiments on operational networks, their throughput measurements are typically not sufficiently accurate for performance profiling of wide-area networks that they emulate. Here, machine learning (ML) methods are developed to transform these inaccurate VIT network throughput measurements to closely match in peak and overall profile of those from a physical testbed or production network. First, a micro kernel network reflecting a physical network is utilized to collect one-time measurements on a host to support this ML transformation. Then, a generic multi-modal ML method is developed to learn a map that transforms measurements from subsequent VITs on the same host to match past, current and follow-on testbed and cloud networks. ML generalization equations are derived to establish its correctness and probabilistically guarantee its generalization accuracy. Experimental results are presented for a variety of VIT hosts with target testbed and cloud networks; they include a case study of a four-site science ecosystem wherein inaccurate convex VIT measurement profiles are transformed into accurate concave profiles of target networks.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

Essential barrier height and a probabilistic approach in characterizing potential landscape

In this work we propose a probabilistic approach to investigate the shape of landscapes of multi-dimensional potential functions. Under a suitable coupling scheme, two copies of the overdamped Langevin dynamics associated with the potential function are coupled, and the coupling times are collected. Assuming a set of intuitive yet technically challenging conditions on the coupling scheme, it is shown that the tail distributions of the coupling times exhibit qualitatively different dependencies on the noise magnitude for single-well versus multi-well potential functions. More specifically, for convex single-well potentials, the negative tail exponent of the coupling time distribution is uniformly bounded away from zero by the convexity parameter and is independent of the noise magnitude. In contrast, for multi-well potentials, the negative tail exponent decreases exponentially as the noise vanishes, with the decay rate governed by the essential barrier height, a quantity introduced in this paper to characterize the non-convex nature of the potential function. Numerical investigations are conducted for a variety of examples, including the Rosenbrock function, interacting particle systems, and loss functions arising in artificial neural networks. These examples not only illustrate the theoretical results in various contexts but also provide crucial numerical validation of the conjectured assumptions, which are essential to the theoretical analysis yet lie beyond the reach of standard technical tools.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Hydrogen Plus Other Alternative Fuels Risk Assessment Models (HyRAM+) Version 4.1 Technical Reference Manual

The HyRAM+ software toolkit provides a basis for conducting quantitative risk assessment and consequence modeling for hydrogen, methane, and propane systems. HyRAM+ is designed to facilitate the use of state-of-the-art models to conduct robust, repeatable assessments of safety, hazards, and risk. HyRAM+ integrates deterministic and probabilistic models for quantifying accident scenarios, predicting physical effects, characterizing hazards (thermal effects from jet fires, overpressure effects from delayed ignition), and assessing impacts on people. HyRAM+ is developed at Sandia National Laboratories to support the development and revision of national and international codes and standards, and to provide developed models in a publicly-accessible toolkit usable by all stakeholders. This document provides a description of the methodology and models contained in HyRAM+ version 4.1. The two most significant changes for HyRAM+ version 4.1 from HyRAM+ version 4.0 are direct incorporation of unconfined overpressure into the QRA calculations and modification of the models for cryogenic liquid flow through an orifice. In QRA mode, the user no longer needs to input peak overpressure and impulse values that were calculated separately; rather, the unconfined overpressure is estimated for the given system inputs, leak size, and occupant location. The orifice flow model now solves for the maximum mass flux through the orifice at constant entropy while conserving energy, which does not require a direct speed of sound calculation. This does not affect the mass flow for all-gaseous releases; the method results in the same speed of sound for choked flow. However, this method does result in a higher (and more realistic) mass flow rate for a given leak size for liquid releases than was previously calculated.

08 HYDROGEN↗

DOE OSTI2026

VRN3P: Variational Recurrent Neural Network Based Net-Load Prediction under High Solar Penetration

This is the final technical report for the SETO-funded VRN3P project (PNNL# 76914). The goal of this project, led by Pacific Northwest National Laboratory (PNNL), in collaboration with Lawrence Livermore National Laboratory (LLNL) and Portland General Electric (PGE), was to develop and validate a deep variational recurrent neural network-based net-load prediction (VRN3P) framework for probabilistic time-series forecasting of day-ahead net-load under high solar penetration scenarios. The project team reports successful design of a novel probabilistic net-load forecasting architecture, comprising of a variational autoencoder and a recurrent neural network, which demonstrates 30% improvement in forecast performance, 60% improvement in training time, and consumes 44% less memory, when compared with conventional baseline models. The team tested the VRN3P model performance on GridLAB-D test-cases representing varying BTM solar penetration levels of 20%, 30%, and 50%, with integrated time-series net-load profiles provided by the utility partner (PGE). The VRN3P model demonstrate <2% hourly MAPE (averaged over the year) for day- ahead net-load forecast on the test scenario with 20% BTM solar. Transfer learning extension of the VRN3P model has demonstrated 8.33× speed-up in training, while still achieving acceptable forecast performance of 2.24% hourly MAPE on the 30% BTM solar penetration test-scenario. A preliminary version of the VRN3P GridAPPS-D™has been developed, along with a web-based interactive user-interface (named ‘Forte’) which has made available on GitHub for public use.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2025

As part of its commitment to advancing safety and reliability assessment methodologies, Argonne National Laboratory pioneered the use of an evaluation method called the Success Path Method (SPM) to improve risk management for offshore oil and gas operations. The development of the SPM at Argonne has been driven by the need to improve existing risk assessment methodologies by focusing on the steps necessary for success rather than failure modes alone. This is particularly important for industrial environments like offshore facilities that perform multiple functions under a continuously evolving set of operational conditions – such as water depth and temperature, currents, and weather conditions. In these dynamic environments, the traditional Probabilistic Risk Assessment (PRA) approach is far too complex as it focuses on what can go wrong – which comprises an infinite failure space that must be fully explored and understood. By shifting the focus to a finite space of success paths, the SPM enables operators and decision makers to prioritize a manageable number of steps that must go right to ensure success. Building on its five decades of experience in safety assessments for the nuclear industry, Argonne made major adaptations to existing risk assessment methods utilizing features similar to fault trees that are traditionally used in PRA to map all pathways in which the system can malfunction. In contrast, SPM identifies the components and processes that must function correctly to achieve specific outcomes – such as preventing the uncontrolled release of hydrocarbons during drilling operations. The SPM framework integrates equipment, procedures, software, processes, and human actions to ensure that physical barriers meet critical safety functions in dynamic operational conditions. This approach helps identify failure modes and improve operational risk management by narrowing the focus to key success elements, which in turn reduces uncertainty and helps users understand, manage, and respond to failures.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

Risk Importance Ranking of Fire Data Parameters to Enhance Fire PRA Model Realism

Fire is historically and analytically a significant contributor to nuclear power plant risk. The level of fire risk and the methods, tools and data for modeling this risk is highly debated by experts. One area of debate is the input data used in fire modeling and how to deal with this data’s high uncertainty. This report outlines initial work performed for determining the key parameters causing this uncertainty and how it propagates into nuclear power plant models. This research paves the way for the development of methods to reduce fire data uncertainty used in modeling. The Nuclear Regulatory Commission has mandated that nuclear power plants perform fire risk modeling. However, there are several issues with the current risk modeling implementation that affect the results. Approved modeling methods can be overly conservative and often do not match plant experience. Also, the data used in the modeling can have high uncertainties and is influenced by expert judgement. To evaluate input data uncertainty, researchers performed an initial review of several fire experiments done at Sandia National Laboratories. Uncertainties for fire data can come from many sources, such as experiment design constraints, environmental conditions, or other plant-specific aspects. There are many different significant and insignificant parameters driving the uncertainty. Additionally, the uncertainty of the different input data used in the fire modeling could have a significant or insignificant effect on the entire plant risk. A four-step methodology was developed to perform Integrated Probabilistic Risk Assessment Importance Ranking. A demonstration case using these steps was set up and three of the four steps were completed in fiscal year (FY) 2019 and the fourth step done FY 2020. These steps are: 1. The qualitative analysis of potential sources was conducted with the following items identified for the demonstration. • Maximum heat release rate • Time to maximum heat release rate • Duration of max heat release rate • Time to decay • Thermal conductivity of concrete • Specific heat of concrete • Density of concrete • Cable jacket thickness 2. A quantitative characterization of dominant sources of uncertainty was performed. A list of distributions and determined values of the dominant sources is shown in Appendix A. 3. A quantitative screening of the potential sources of uncertainty using Morris Elementary Effects Analysis was completed. An experimental model using the physics-based fire modeling tool Fire Dynamics Simulator was developed and coupled with the Risk Analysis Virtual Environment. The Morris analysis identified at least two parameters that can be eliminated as significant contributors (specific heat of concrete and cable jacket thickness). 4. Global importance measure (Global IM) analysis to generate a comprehensive ranking based on their influence on the plant risk. In this research, a moment-independent Global IM is used since it can address (a) uncertainty in the input parameters of the fire model, (b) uncertainty in the risk outputs, and (c) non-linearity and interactions among input parameters in the fire model, more accurately than the correlation-based and variance-based global methods. The observations from the research showed that, depending on the initial and boundary conditions of the fire scenarios, fire-induced damage could have a very small probability and could be dominated by the tail of the uncertainty distribution; hence, the accuracy of the correlation-based and variance-based methods is questionable. The moment-independent Global IM analysis in this research provides a better understanding of how experimental uncertainty data affects industry’s plant models and where improvements in that data will have the largest benefit for improving fire modeling accuracy in causing core damage. Among the five unscreened parameters obtained from the Morris EE analysis, the Global IM analysis results for the case study indicated that max heat release rate and fire location are the most important parameters. The report also outlines benefits of using a unified computational platform that integrates the underlying simulations (e.g., a fire progression model), quantitative screening (using the Morris EE method), and the Global IM analysis. A unified platform can (i) facilitate the ranking of input parameters considering multiple key fire scenarios simultaneously, rather than considering one scenario at a time, (ii) contribute to more explicit and accurate treatment of dependencies at multiple levels of Fire PRA, (iii) facilitate the sampling-based uncertainty quantification for Fire PRA, and (iv) help generating both “industry-wide” and "plant-specific" ranking of uncertainty sources in Fire PRA. Future research should be done to include additional parameters such as detection/suppression or cable fire spread. Adding a PRA software such as SAPHIRE to the RAVEN platform would help with plant model integration and improve treatment of fire-induced dependency. The I-PRA risk importance ranking methodology offered in this report can provide valuable information for efficiently (a) enhancing the realism of Fire PRA for existing plants and (b) supporting the development of Dynamic Fire PRA for advanced reactors and new plants.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Generative large language models for predictive maintenance planning

Maintenance planning and the generation of necessary components for tasks can prove time-consuming and complex. Automating the creation of recurring or similar tasks by leveraging previous planning packages and data, while uncovering insights to automate planning package generation, presents an opportunity to conserve valuable time and resources. This work aims to harness the textual and probabilistic capabilities of large language models (LLMs) to automate the generation of planning packages. Utilizing diverse data sources ranging from raw data to handwritten text, both singular and collaborative LLMs are trained and tested. Results demonstrate their capability to generate essential planning package components, effectively replicating the statistical patterns in the data. This demonstrates the use of these tools inside a digital asset for automated planning. This work outlines a methodology for constructing datasets, a training suite, and evaluation methods for LLM-based textual and conversational planning tools utilized in an asset digital twin. Results indicate that the fine-tuned models generate estimated planning information within the statistical ranges observed in real maintenance data. The models achieve high accuracy (>90%) in document question-answering and instruction generation tasks. Furthermore, the conversational retrieval-augmented generation (RAG) assistant system achieves 100% document retrieval accuracy, while conversational information capture exceeds 98% across the majority of work-package assistant modules.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

Frequency and time multiplexing for multiphoton state generation

One of the primary challenges with photonic quantum information processing is that single-photon states are created and heralded probabilistically, rather than being created on demand. One solution to this problem is multiplexing , which is attempting probabilistic generation of a single photon at multiple locations, times, or frequencies, and switching one successfully generated photon into an output mode. Motivated by the development of frequency-based photonic quantum information processing, we consider multiplexing using frequency and time as our options to use to get many photon generation opportunities. We devise an approach for generating multiphoton states, with photons populating multiple frequency modes in the same spatiotemporal mode. This method uses a variable-length optical delays to manipulate the temporal mode of the photons, and spaced fiber Bragg grating (FBG) reflectors to jointly manipulate the frequency and temporal modes of the photons. Experimental progress toward implementing this multiplexing scheme is proceeding along 2 fronts, each with different ways to achieve the variable-length optical delay. First, we will implement this scheme using a free-space optical quantum memory with multiple discretely adjustable free-space delays. For this implementation method, I calculate multiphoton generation rates, accounting for loss, that are realistically achievable with commercially available hardware. This work will appear in a theory paper, currently in preparation. Second, we will implement this multiplexing scheme in an integrated manner: the variable-length optical delay will happen on-chip using a long-lifetime Q-switchable Fabry-Perot cavity. The joint manipulation of the frequency and temporal modes of the photons will still happen off-chip in fiber with FBG reflectors. I report on experimental progress in constructing integrated Q-switchable Fabry-Perot cavities.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Added value of site load measurements in probabilistic lifetime extension: a Lillgrund case study

Site-specific fatigue estimation is an essential part of wind turbine lifetime extension, with various methods depending on data availability. The present study compares probabilistic lifetime extension assessment results for rotor blades with and without load measurements. It also addresses two key questions in such assessments: the applicability of the Frandsen model for estimating waked turbulence under complex and mixed wake conditions and the extrapolation of mid-term data over longer time periods. The case study wind turbine is SWT-2.3-93, located at the edge of the Lillgrund wind farm, situated in the Øresund Strait between Denmark and Sweden. The turbine is extensively instrumented, with 5 years of data available from its supervisory control and data acquisition (SCADA) system. Although the Frandsen turbulence estimates deviate in a different manner from measurements at below- and above-rated mean wind speeds, the model remains a conservative approach for fatigue load prediction and reliability. In the current case study, the site-specific assessment using strain gauge measurements yields a 33 % higher annual fatigue reliability index after 35 years compared to a scenario based on the Frandsen estimation combined with ambient environmental data and a generic aeroelastic model. The results also demonstrate that the sensitivity of fatigue reliability to load uncertainty is negligible when load measurements are used directly but relatively high when relying on the Frandsen model in combination with a generic aeroelastic model. Overall, the high variability of the lifetime extension in different scenarios of data availability and accuracy shows the importance and added value of high-quality measurements combined with wind-farm-level SCADA and a model updated in real time (digital twins).

17 WIND ENERGY↗

DOE OSTI2019

Hybridizing Classifiers and Collection Systems to Maximize Intelligence and Minimize Uncertainty in National Security Data Analytics Applications

There are numerous applications that combine data collected from sensors with machine-learning based classification models to predict the type of event or objects observed. Both the collection of the data itself and the classification models can be tuned for optimal performance, but we hypothesize that additional gains can be realized by jointly assessing both factors together. Through this research, we used a seismic event dataset and two neural network classification models that issued probabilistic predictions on each event to determine whether it was an earthquake or a quarry blast. Real world applications will have constraints on data collection, perhaps in terms of a budget for the number of sensors or on where, when, or how data can be collected. We mimicked such constraints by creating subnetworks of sensors with both size and locational constraints. We compare different methods of determining the set of sensors in each subnetwork in terms of their predictive accuracy and the number of events that they observe overall. Additionally, we take the classifiers into account, treating them both as black-box models and testing out various ways of combining predictions among models and among the set of sensors that observe any given event. We find that comparable overall performance can be seen with less than half the number of sensors in the full network. Additionally, a voting scheme that uses the average confidence across the sensors for a given event shows improved predictive accuracy across nearly all subnetworks. Lastly, locational constraints matter, but sometimes in unintuitive ways, as better-performing sensors may be chosen instead of the ones excluded based on location. This being a short-term research effort, we offer a lengthy discussion on interesting next-steps and ties to other ongoing research efforts that we did not have time to pursue. These include a detailed analysis of the subnetwork performance broken down by event type, specific location, and model confidence. This project also included a Campus Executive research partnership with Texas A&M University. Through this, we worked with a professor and student to study information gain for UAV routing. This was an alternative way of looking at the similar problem space that includes sensor operation for data collection and the resulting benefit to be gained from it. This work is described in an Appendix.

97 MATHEMATICS AND COMPUTING↗