Search NASA⌕ Search

SEARCH · Search NASA

Results for “data gaps”

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 505 records · Page 28
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Preliminary Results for Using Uncertainty and Out-of-distribution Detection to Identify Unreliable Predictions.

As machine learning (ML) models are deployed into an ever-diversifying set of application spaces, ranging from self-driving cars to cybersecurity to climate modeling, the need to carefully evaluate model credibility becomes increasingly important. Uncertainty quantification (UQ) provides important information about the ability of a learned model to make sound predictions, often with respect to individual test cases. However, most UQ methods for ML are themselves data-driven and therefore susceptible to the same knowledge gaps as the models themselves. Specifically, UQ helps to identify points near decision boundaries where the models fit the data poorly, yet predictions can score as certain for points that are under-represented by the training data and thus out-of-distribution (OOD). One method for evaluating the quality of both ML models and their associated uncertainty estimates is out-of-distribution detection (OODD). We combine OODD with UQ to provide insights into the reliability of the individual predictions made by an ML model.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

BEYONDPLANCK VI. Noise characterization and modeling

We present a Bayesian method for estimating instrumental noise parameters and propagating noise uncertainties within the global BeyondPlanck Gibbs sampling framework, and apply this to Planck Low Frequency Instrument (LFI) time-ordered data. Following previous literature, we initially adopt a 1/ f model for the noise power spectral density (PSD), but find the need for an additional lognormal component in the noise model for the 30 and 44 GHz bands. We implement an optimal Wiener-filter (or constrained realization) gap-filling procedure to account for masked data. We then use this procedure to both estimate the gapless correlated noise in the time-domain, n corr , and to sample the noise PSD parameters, ξ n = {σ 0 , f knee , α, A p }. In contrast to previous Planck analyses, we assume piecewise stationary noise only within each pointing period (PID), not throughout the full mission, but we adopt the LFI Data Processing Center (DPC) results as priors on α and f knee . On average, we find best-fit correlated noise parameters that are mostly consistent with previous results, with a few notable exceptions. However, a detailed inspection of the time-dependent results reveals many important findings. First and foremost, we find strong evidence for statistically significant temporal variations in all noise PSD parameters, many of which are directly correlated with satellite housekeeping data. Second, while the simple 1/ f model appears to be an excellent fit for the LFI 70 GHz channel, there is evidence for additional correlated noise not described by a 1/ f model in the 30 and 44 GHz channels, including within the primary science frequency range of 0.1–1 Hz. In general, most 30 and 44 GHz channels exhibit deviations from 1/ f at the 2–3σ level in each one hour pointing period, motivating the addition of the lognormal noise component for these bands. For some periods of time, we also find evidence of strong common mode noise fluctuations across the entire focal plane. Overall, we conclude that a simple 1/ f profile is not adequate to fully characterize the Planck LFI noise, even when fitted hour-by-hour, and a more general model is required. These findings have important implications for large-scale CMB polarization reconstruction with the Planck LFI data, and the current work is a first attempt at understanding and mitigating these issues.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2026

Iterative HOMER with uncertainties

We present iHOMER, an iterative version of the HOMER method to extract Lund fragmentation functions from experimental data. Through iterations, we address the information gap between latent and observable phase spaces and systematically remove bias. To quantify uncertainties on the inferred weights, we use a combination of Bayesian neural networks and uncertainty-aware regression. We find that the combination of iterations and uncertainty quantification produces well-calibrated weights that accurately reproduce the data distribution. A parametric closure test shows that the iteratively learned fragmentation function is compatible with the true fragmentation function.

Butter, Anja [Heidelberg Univ. (Germany); Sorbonne↗
DOE OSTI2025

Mapping Nitrogen Pathways in Organic Manure through GCAM

As the world has become more and more globalized, a need for a comprehensive and extensive modelling system that maps out the vast majority of processes and items around the world and can display how interdependent they are. Furthermore, such a model should be able to predict future processes and values such that it can guide scientific thinking and policy. This is where the importance of GCAM becomes so apparent. GCAM, the Global Change Analysis Model, that incorporates many different fields and can predict such a wide variety of processes. As the model expands and encompasses more fields and goes into greater detail, there is a need for previously incorporated processes to be further fleshed out. My project is dedicated to creating the organic nitrogen fertilizer pathway, such that there are two nitrogen fertilizer pathways, synthetic and organic (manure). Previously, only synthetic fertilizer was included, with all organic fertilizer (manure) being omitted from the model. Developing the synthetic path is relatively easy, as the process for creating synthetic fertilizer is an industrial process that has been perfected over decades, and thus has a very precise input/output coefficient of nitrogen, but determining that same coefficient for organic manure is much harder, because, well, digestion is not an industrial process, especially considering the many different types of animals have different processes, and a good amount of the manure they produce won’t be applied to the soil. I obtained data from the UN’s FAO on manure nitrogen applied to soil, and filled in the gaps and created many different visualizations to help comprehend the data. These findings will foster a greater understanding of nitrogen flows, which is crucial to fields like industrial agriculture, polluting runoff, and nitrifying bacteria and archaea in the soil.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Physics-Informed Recurrent Neural Networks to Predict Reactor Operations of the AGN-201 Nuclear Reactor

4 page paper submitted to ANS Student conference. Summary of paper similar to the following abstract: The ability to predict how a reactor will operate, understand when anomalous conditions arise, and ensure a reactor is being operated as expected is crucial for deploying new nuclear facilities. Digital twins serve as a unique solution to recognizing reactor behavior; however, they require data to be useful. For next-generation reactors, this data may not currently be available. To explore how synthetic physics-informed reactor data can be used to predict reactor operations, a recurrent neural network was implemented for the Idaho State University AGN-201 digital twin. The goal of this work is to determine how synthetic data can be used to train a recurrent neural network model for predicting the reactor power of the AGN-201. The recurrent neural network was validated using both synthetic and real operational data. We envision this approach will help bridge the gap between the virtual and physical sides of a digital twin, where reactor physics models based on as-built data can be corrected for actual operating parameters to ensure the virtual model mirrors reality.

98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL P↗
DOE OSTI2024

A new data-driven map predicts substantial undocumented peatland areas in Amazonia

Tropical peatlands are among the most carbon-dense terrestrial ecosystems yet recorded. Collectively, they comprise a large but highly uncertain reservoir of the global carbon cycle, with wide-ranging estimates of their global area (441 025–1700 000 km 2 ) and below-ground carbon storage (105–288 Pg C). Substantial gaps remain in our understanding of peatland distribution in some key regions, including most of tropical South America. Here we compile 2413 ground reference points in and around Amazonian peatlands and use them alongside a stack of remote sensing products in a random forest model to generate the first field-data-driven model of peatland distribution across the Amazon basin. Our model predicts a total Amazonian peatland extent of 251 015 km 2 (95th percentile confidence interval: 128 671–373 359), greater than that of the Congo basin, but around 30% smaller than a recent model-derived estimate of peatland area across Amazonia. The model performs relatively well against point observations but spatial gaps in the ground reference dataset mean that model uncertainty remains high, particularly in parts of Brazil and Bolivia. For example, we predict significant peatland areas in northern Peru with relatively high confidence, while peatland areas in the Rio Negro basin and adjacent south-western Orinoco basin which have previously been predicted to hold Campinarana or white sand forests, are predicted with greater uncertainty. Similarly, we predict large areas of peatlands in Bolivia, surprisingly given the strong climatic seasonality found over most of the country. Very little field data exists with which to quantitatively assess the accuracy of our map in these regions. Data gaps such as these should be a high priority for new field sampling. This new map can facilitate future research into the vulnerability of peatlands to climate change and anthropogenic impacts, which is likely to vary spatially across the Amazon basin.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Effects of Chlorine Capture and a Proposed Density Law on the Reactivity of Plutonium Solution Systems

During fissionable material processing, all normal and credible abnormal conditions must remain safely subcritical. Nuclear Criticality Safety (NCS) uses a number of methods to determine subcriticality, one of which is the use of neutron transport codes such as MCNP6. In order to create models for use with MCNP6, both the geometry and materials in fissionable material processes must be known, or assumptions must be made and quantified for the impact to bias. One of the systems with a significant amount of bias due to material modeling assumptions is in the area of aqueous plutonium processing. These solutions are typically plutonium nitrate solutions or plutonium chloride solutions, which are modeled as fictitious plutonium metal-water mixtures because little is known about the actual density of the solution and there is no current predictive capability approved for use at Los Alamos National Laboratory (LANL) for modeling them. This research is currently underway to fill the gap and develop an algorithm for use with MCNP6 to model the density of plutonium chloride solutions. The method is to be validated with experimental data for density, and also validated with critical experiments using MCNP6. Note that the Chlorine Worth Study (CWS) was performed in December 2021 to help bridge the gap in chlorine data for critical experiments, and is currently awaiting International Criticality Safety Benchmark Evaluation Project (ICSBEP) review. This study was performed by LANL at the National Criticality Experiments Research Center (NCERC) at the Nevada National Security Site (NNSS). Additional information regarding this experiment may currently be found in LA-UR- 22-29180. Additionally, the Chemistry-Actinide Analytical Chemistry (C-AAC) at LANL has performed a number of solution density measurements for PuCl 3 -HC 1 -H 2 O, allowing for such data be used to create a semi-empirical density via the Pitzer method. The published dataset for the measurements is documented in LA-UR-22-25454. This method has already been tested successfully for aqueous plutonium nitrate solutions in SCALE. Current solution density measurements exist of plutonium concentrations of 0-~142g/L, all at 2M HC1, at temperatures 20-40°C. Additional data was taken for HC1-corrected density values, which essentially mimics the data for a pure PuCl x -water solution. The calculations in this report aim to support the current research by demonstrating the difference in system reactivity for the current modeling method when compared to the new proposed modeling with a density law implementation, which is being written as a Python tool to be used with MCNP6.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗
DOE OSTI2026

Topsoil bulk geochemical compositions - An updated harmonized global dataset

Mineral weathering is a key biogeochemical process because of the capacity of minerals to stabilize organic matter. However, predicting soil weathering status across large spatial areas still isn’t possible due to a lack of global data and theoretical frameworks. To address this knowledge gap, multiple global datasets of bulk topsoil geochemical compositions have been harmonized using R. These datasets document topsoil bulk geochemical compositions across five continents (n = ~16,000 observations). Source data for these observations include the EuroGEOSurveys Geochemical Baseline Database (FOREGS), the US Geological Survey National Geochemical Database (NASGLP), the Geochemical Atlas of Australia (GAA), the US Geological Survey Alaska Geochemical Database (AGD84), the National Cooperative Soil Survey (NCSS), the European Geochemical Mapping of Agricultural Soil (GEMAS), Ecorespira-Amazon (ERA), the New Zealand Geochemical Baseline Survey (NZ_GBS), and the African Soil Information Service (AFSIS). Major elements observed include Aluminum (Al), Calcium (Ca), Iron (Fe), Potassium (K), Magnesium (Mg), Sodium (Na), Titanium (Ti), Manganese (Mn), Phosphorus (P), Carbon (C), and Sulfur (S). This data package includes the harmonized dataset itself, and the R scripts necessary to harmonize these datasets, in addition to metadata that describes all columns, files, and databases used in this project. Methods & Sampling Step 1 – Databases of geochemical data identified This study aimed to leverage existing measurements of topsoil geochemical data. Databases were first identified and deemed appropriate for inclusion if they were measuring soils and performed these measurements on the <2mm soil fraction. Databases such as NCSS and AGD84 needed more post processing to include in the database and this was done using the NCSS_datamerge_031626 R file and Alaska_USGSmerge_031626 R file, respectively. Step 2 – Database harmonization Once appropriate databases were identified, they were harmonized for ease of analysis using the R script Database_Harmonization_031826. This included removing columns from original datasets that would not be used in analysis (removed columns are noted in the code). Then, data cleaning procedures specific to each dataset were undertaken. This includes standardizing columns to include units and adding metadata columns regarding procedures for analyzing specific elements. Functions for standardizing measurements and units are outline in R files: calculate element_mg_kg_031626, calculate_oxide_wt_perc_031626, change_oxide_caps_031626, and conv_2_numeric_031626. This also included adding a unique identifier for each sample to identify it with its respective database (see CD_ID in data dictionary). Geographic information: Data reflect a compilation of datasets collected globally. Geographic areas covered by each of the datasets include: - EuroGEOSurveys Geochemical Baseline Database (FOREGS) - European continent - North American Soil Geochemical Landscapes (NASGLP) - continental United States and limited parts of Canada (see database key for more details) - National Geochemical Survey of Australia (GAA) - Australia - Alaska geochemical database (AGDB4) - Alaska - National Cooperative Soil Survey (NCSS) - Global measurements, but concentrated in the continental United States - Geochemical data for arable land and land under permanent grass cover in continental Europe (GEMAS) - continental Europe - Ecorespira-Amazon (ERA) - Geochemical data from the Amazon basin - Geochemical baseline data for New Zealand (NZGBS) - New Zealand - Geochemical data collected across continental Africa (AfSIS) - Measurements across Africa

EARTH SCIENCE > LAND SURFACE > SOILS↗
DOE OSTI2025

Ubiquity and Causes of Soil Water Preferential Flow Across 17 Ecoregions

Abstract Preferential flow (PF) in soil causes the rapid transport of water, nutrients, and contaminants into the subsurface, influencing groundwater recharge and streamflow. Data scarcity has hindered the quantification of PF occurrence and the identification of its drivers across diverse ecoregions. We address this gap by analyzing high‐frequency, multi‐depth soil moisture data across 17 ecoregions in the USA, using ∼1,500 sensors at 40 sites. We discovered that PF is widespread, with sites experiencing PF in up to 60% of rainfall events ≥2 mm. Multiple approaches consistently show that PF is more likely to occur with increased peak rainfall intensity, finer textured material, low soil moisture variability, humid climate, and higher net primary productivity. This suggests that PF patterns could shift with projected climate changes, increasing uncertainty in predictions of groundwater recharge, water quality, and streamflow generation. Plain Language Summary Water can bypass part of the soil's matrix through a process called preferential flow (PF). This quick transport of water through the soil brings with it nutrients and contaminants and eventually makes it to groundwater and streams. To ensure ample amounts of good quality groundwater and surface water we need to understand when and where PF occurs. We inferred when PF occurred at 40 different sites across 17 ecoregions in the USA using soil moisture and rainfall data. We found that PF happened at all sites and in up to 60% of rainfall events ≥2 mm. Preferential flow was most likely at sites with high rainfall intensities, high clay content in soils, low variability in soil moisture, and high vegetation productivity. As rainfall intensities are predicted to increase due to climate change and vegetation becomes more productive, PF occurrence becomes more important for predicting groundwater recharge, water quality, and streamflow generation. Key Points Preferential flow (PF) is ubiquitous across the USA and occurs in up to 60% of all rainfall events ≥2 mm Rainfall intensity, soil texture, and antecedent soil moisture emerge as critical in generating PF across diverse ecoregions Two different PF detection approaches show similar relationships between key drivers and occurrence of PF

Li, Bonan↗
DOE OSTI2021

A series of Rb 4 Ln 2 (P 2 S 6 )(PS 4 ) 2 (Ln = La, Ce, Pr, Nd, Sm, Gd) rare earth thiophosphates with two distinct thiophosphate units [P V S 4 ] 3- and [P IV 2 S 6 ] 4-

A series of rubidium rare earth thiophosphates with the formula Rb 4 Ln 2 (P 2 S 6 )(PS 4 ) 2 (Ln = La, Ce, Pr, Nd, Sm, and Gd) were synthesized using the high temperature molten flux crystal growth method utilizing a RbBr flux. Single crystals of all title compounds, as well as phase pure powders of the La-, Ce-, and Sm-containing compositions, were obtained. Single crystals of the title compounds were characterized by single crystal and powder X-ray diffraction for structure and phase identification. Rb 4 Ln 2 (P 2 S 6 )(PS 4 ) 2 crystallizes in the monoclinic crystal system adopting the P2 1 /n space group for the large rare earths (Ln = La, Ce, Pr) and the C2/c space group for the smaller rare earths (Ln = Nd, Sm, Gd). This Rb 4 Ln 2 (P 2 S 6 )(PS 4 ) 2 series is a rare example of thiophosphates containing both tetrahedral [P V S 4 ] 3– and dimeric [P IV 2 S 6 ] 4– thiophosphate units that, in this structural family, link corrugated rare earth sulfide chains into sheets. Here, the band gaps of the materials were determined from UV–Vis data and the fluorescence spectrum of Rb 4 Ce 2 (P 2 S 6 )(PS 4 ) 2 was collected. Optical band gaps were estimated to be 2.9 and 2.4 for the Nd and Sm analogues, respectively.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Modeling of the metal–insulator transition temperature in alio-valently doped VO 2 through symbolic regression

The correlated semiconductor vanadium dioxide (VO 2 ) exhibits an insulator–metal transition (IMT) near room temperature, which is of interest in various device applications. Precise IMT temperature control is crucial to determine the use cases across technologies such as thermochromic windows, actuators for robots or neuronal oscillators. Doping the cation or anion sites can modulate the IMT by several tens of degrees and control hysteresis. However, modeling the effects of control parameters (e.g., doping concentration, type of dopants) is challenging due to complex experimental procedures and limited data, hindering the use of traditional data-driven machine learning approaches. Symbolic regression (SR) can bridge this gap by identifying nonlinear expressions connecting key input parameters to target properties, even with small data sets. In this work, we develop SR models to capture the IMT trends in VO 2 influenced by different dopant parameters. Using experimental data from the literature, our study reveals a dual nature of the IMT temperature with varying tungsten (W) doping concentrations. The symbolic model captures data trends and accounts for experimental variability, providing a complementary approach to first-principles calculations. Our feature-driven analysis across a broader class of dopants informs selectivity and provides qualitative insights into tuning phase transition properties valuable for neuromorphic computing and thermochromic windows.

36 MATERIALS SCIENCE↗
DOE OSTI2023

A bi-level data-driven framework for fault-detection and diagnosis of HVAC systems

Long-term operation of heating, ventilation, and air conditioning (HVAC) systems will eventually lead to a range of HVAC system failures, resulting in excessive energy consumption and maintenance costs. Here, to avoid HVAC malfunctioning, fault detection diagnostic (FDD) is utilized as a common practice. Machine learning methods have lately received considerable interest for FDD analysis of HVAC systems due to their high detection accuracy. Meanwhile, HVAC malfunctions are regarded as rare occurrences, hence normal operating data samples are much more accessible than data samples in faulty and malfunctioning conditions. The dominating frequency of normal operation in HVAC datasets has also led to heavily biased classification algorithms within the literature. Moreover, the focus of previous literature has been on increasing the accuracy of the models which leads to a high number of false positives (misleading alarms) in the system. In order to enhance the performance of diagnostic procedures and fill the mentioned gaps, this study proposes a novel data-driven framework. A bi-level machine learning framework is developed for diagnosing faults in air handling units (AHUs) and rooftop units (RTUs) based on principal component analysis (PCA), time series anomaly detection, and random forest (RF). It is shown that PCA can reduce the dataset dimension with one principal component accounting for 95% of data variance. Also, the random forest could classify the faults with 89% precision for single-zone AHU, 85% precision for RTU, and 79% for multi-zone AHU. By proposing this framework, three persistent challenges are addressed: (I) minimizing false positives; (II) accounting for data imbalance; and (III) normal condition monitoring of equipment.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2025

AI in Astrophysics: Tackling Domain Shift, Model Robustness and Uncertainty

Artificial Intelligence (AI) is revolutionizing physics research from probing the large-scale structure of the Universe to modeling subatomic interactions and fundamental forces. Yet, a major challenge persists: AI models trained on simulations or old experiment / astronomical survey often perform poorly when applied to new data exposing issues of dataset (domain) shift, model robustness, and uncertainty in predictions. This talk will introduce common challenges in applying AI across domains and present solutions based on domain adaptation a set of techniques designed to improve model generalization under domain shift. We will cover foundational ideas, practical strategies, and current research frontiers in this area. Through examples in astrophysics, we'll explore how domain adaptation can help bridge the gap between synthetic and real-world data, improve trust in model outputs, and advance scientific discovery.

Ciprijanvoic, Aleksandra [Fermilab] (ORCID:0000000↗
DOE OSTI2025

Shift Happens: Building Robust AI Models with Domain Adaptation

Artificial Intelligence (AI) is revolutionizing physics research—from probing the large-scale structure of the Universe to modeling subatomic interactions and fundamental forces. Yet, a major challenge persists: AI models trained on simulations or old experiment / astronomical survey often perform poorly when applied to new data—exposing issues of dataset (domain) shift, model robustness, and uncertainty in predictions. This summer school session will introduce students to common challenges in applying AI across domains and present solutions based on domain adaptation—a set of techniques designed to improve model generalization under domain shift. We will cover foundational ideas, practical strategies, and current research frontiers in this area. Through examples in astrophysics, we'll explore how domain adaptation can help bridge the gap between synthetic and real-world data, improve trust in model outputs, and advance scientific discovery. The concepts discussed are broadly applicable across physics and other scientific disciplines, making this a valuable topic for anyone interested in building robust, transferable AI models for science.

Ciprijanovic, A. [Fermilab] (ORCID:000000031281719↗
DOE OSTI2024

A Survey on Privacy in Graph Neural Networks: Attacks, Preservation, and Applications

Graph Neural Networks (GNNs) have gained significant attention owing to their ability to handle graph-structured data and the improvement in practical applications. However, many of these models prioritize high utility performance, such as accuracy, with a lack of privacy consideration, which is a major concern in modern society where privacy attacks are rampant. To address this issue, researchers have started to develop privacy-preserving GNNs. Despite this progress, there is a lack of a comprehensive overview of the attacks and the techniques for preserving privacy in the graph domain. In this survey, we aim to address this gap by summarizing the attacks on graph data according to the targeted information, categorizing the privacy preservation techniques in GNNs, and reviewing the datasets and applications that could be used for analyzing/solving privacy issues in GNNs. We also outline potential directions for future research in order to build better privacy-preserving GNNs.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Scale‐Dependent Influence of Permafrost on Riverbank Erosion Rates

Abstract Whether permafrost systematically alters the rate of riverbank erosion is a fundamental geomorphic question with significant importance to infrastructure, water quality, and biogeochemistry of high‐latitude watersheds. For over four decades, this question has remained unanswered due to a lack of data. Using remotely sensed imagery, we addressed this knowledge gap by quantifying riverbank erosion rates across the Arctic and subarctic. To compare these rates to non‐permafrost rivers, we assembled a global data set of published riverbank erosion rates. We found that erosion rates in rivers influenced by permafrost are on average nine times lower than non‐permafrost systems; erosion rate differences increase up to 40 times for the largest rivers. To test alternative hypotheses for the observed erosion rate difference, we examined differences in total water yield and erosional efficiency between these rivers and non‐permafrost rivers. Neither of these factors nor differences in river sediment loads provided compelling alternative explanations, leading us to conclude that permafrost limits riverbank erosion rates. This conclusion was supported by field investigations of rates and patterns of erosion along three rivers flowing through discontinuous permafrost in Alaska. Our results show that permafrost limits maximum bank erosion rates on rivers with stream powers greater than 900 Wm −1 . On smaller rivers, however, hydrology rather than thaw rate may be the dominant control on bank erosion. Our findings suggest that Arctic warming and hydrological changes should increase bank erosion rates on large rivers but may reduce rates on rivers with drainage areas less than a few thousand km 2 .

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Reliable and Efficient Machine Learning (Final Technical Report)

Modern scientific experiments generate massive amounts of data at a pace much faster than humans can manually analyze. While machine learning has revolutionized commercial data analysis (such as recommending movies or recognizing faces), applying these tools to complex scientific discovery is challenging because scientific answers must be precise, interpretable, and adhere to physical laws. The research under this project aims to develop new mathematical tools and computer algorithms specifically designed for scientific applications. Major progress has been made in automatically cleaning and deconstructing messy experimental data, analyzing the visual information of physical phenomena, determining the underlying physical variables, and providing rig orous mathematical analysis of interesting algorithms and concepts widely used in machine learning. This project addressed the critical gap between our ability to generate massive scientific data and our ability to extract interpretable information from it. We established mathematical foundations for Scientific Machine Learning (SciML) aimed at effective data analytics and automated discovery. Our work focused on three core objectives: (1) developing reliable feature extraction methods for dynamic high-dimensional data, (2) establishing mathematical foundations for discovering dynamics via neural networks, and (3) creating rigorous optimization techniques for these models. Key outcomes come from two fronts. On the practical side, they include the development of algorithms that significantly enhance the extraction of signals from field data, as well as the capability to handle situations that exhibit smooth variations or physical stretching due to temperature changes. They also include the creation of an automated framework for discovering fundamental state variables from raw experimental data, demonstrating the ability to identify intrinsic physical dimensions without prior knowledge of the governing laws. On the theoretical front, the research results in theoretical advances in Optimal Transport, a widely used notion in SciML, specifically regarding functions with fixed-size nodal sets, provide sharp bounds relevant to uncertainty quantification. Meanwhile, the outcomes also include the establishment of convergence theories for nonlocal gradient descent methods, enabling robust optimization with noisy data in high-dimensional settings commonly encountered in scientific modeling. The project also helps creating opportunities to train the next generation of researchers, equipping them with the necessary technical skills for today’s workplace and preparing them for future advances.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Comparison of time-frequency-analysis techniques applied in building energy data noise cancellation for building load forecasting: A real-building case study

Time-frequency analysis that disaggregates a signal in both time and frequency domain is an important supporting technique for building energy analysis such as noise cancellation in data-driven building load forecasting. There is a gap in the literature related to comparing various time–frequency-analysis techniques, especially discrete wavelet transform (DWT) and empirical mode decomposition (EMD), to guide the selection and tuning of time–frequency-analysis techniques in data-driven building load forecasting. This article provides a framework to conduct a comprehensive comparison among thirteen DWT/EMD techniques with various parameters in a load forecasting modeling task. A real campus building is used as a case study for illustration. The DWT and EMD techniques are also compared under various data-driven modeling algorithms for building load forecasting. The results in the case study show that the load forecasting models trained with noise-cancelled energy data have increased their accuracy to 9.6% on average tested under unseen data. This study also shows that the effectiveness of DWT/EMD techniques depends on the data-driven algorithms used for load forecasting modeling and the training data. Hence, DWT/EMD-based noise cancellation needs customized selection and tuning to optimize their performance for data-driven building load forecasting modeling.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗