Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “generative machine 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 649 records · Page 36
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Coolant Pump Predictive Data Analytics from Signatures Generated by the Recursive Short Time Fast Fourier Transform

Although a nuclear reactor is a hostile environment for sensors and signal transmissions, the reactor core is amenable to acoustic communication. An acoustic measurement infrastructure installed at the Advanced Test Reactor (ATR) nozzle trench area records acoustic signals that can capture reactor operating states. The distinct states produce unique signatures that can be identified and tracked using data processing and data analytics. The infrastructure relies on acoustic transmission through ATR in-pile structural components, piping, and coolant that transmit acoustically modified signals generated by the coolant pumps. This paper will discuss results from using the Recursive Short Time Fast Fourier Transform (RSTFFT) technique used to process acoustic signals and provide signatures that are identified and monitored by analytics. The RSTFFT is applied to ATR data to understand the vibration levels and signatures for different operating regimes as displayed by the spectrogram. The combination of coolant pumps for normal and high-power operation generate unique signatures. These acoustic signatures are used to develop machine learning approaches to automatically classify operating regimes. Two machine-learning models, Support Vector Machines and Linear Discriminant Analysis, were developed to classify two event classes. Class 1 is a normal steady-state operation, and Class 2 is any event that is due to start up, shut down, or other actions. Both types of machine learning models had over a 96% prediction accuracy for the two classes. These results lay the foundation for predictive analytic frameworks that can be leveraged by ATR to optimize operations and maintenance.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2025

Data-driven Community-centered Resilient Assessment and Planning Toolkit for Nexus of Energy and Water (DCRAPT-NEW)

Urban areas, including Detroit and Pittsburgh, have suffered significant dual outages of the electrical and water infrastructure in the past decade due, in part, to the increasing number of extreme weather events. With increasing temperatures and rainfall intensity, these regions need to prepare for increasing extreme events through community-based energy and water resilience analysis, planning, and enhancement. This project developed a suite of open-source, open-access, community-centered, data-driven assessment and distributed energy resource (DER) and planning tools for energy and water resilience enhancement in urban areas. Through establishing a multi-level community awareness and engagement mechanism and a comprehensive collection of power outage and flooding data, an innovative group of community energy and water resilience assessment and planning tools have been developed for a wide range of users with differing and variable sets of data available to them. The developed tools include (1) DOE EAGLE-I data-driven, deep-learning assisted resilience assessment and DER planning tools at the county level with socioeconomic factors incorporated; (2) Utility annual power outage data-driven tools for long term resilience assessment and DER planning and 15-min power outage data-driven tools for short term resilience assessment and planning; (3) Detailed engineering tools for energy and water systems resilience assessment and planning when the system topology and component fragility curves are available; (4) Alternative Resiliency Metric Calculation that extracts and separates outage and restoration processes; and (5) Co-optimization tools that evaluate the resilience of the power and sewage system and allow users to conduct joint planning with energy and wastewater systems. The developed tools provide planners, decision-makers, and stakeholders with powerful capabilities to systematically evaluate system/community resilience and optimal and actionable guidance for enhancing resilience while prioritizing DER investments. The tools have been used and validated in Detroit and Pittsburgh and can be used in other areas of the nation. In addition, this project will (1) advance the knowledge and applications of machine-learning methods in analyzing and fusing different layers of information and generating meaningful data points such as generating rare weather events; (2) significantly improve the energy and water resilience of the identified communities in Detroit and Pittsburgh and prepare for more frequent and severe weather conditions; (3) help communities assess extreme weather event impacts and address short-term and long-term resilience-related issues The developed tools have been made public via GitHub and demonstrated to community stakeholders and utility companies via the two annual workshops and numerous community engagement meetings. The project outcomes are also disseminated through publications in various journals and conference proceedings, and presentations at top conferences.

13 HYDRO ENERGY↗
DOE OSTI2022

Revenue prediction for integrated renewable energy and energy storage system using machine learning techniques

Revenue estimation for integrated renewable energy and energy storage systems is important to support plant owners or operators’ decisions in battery sizing selection that leads to maximized financial performances. A common approach to optimizing revenues of a hybrid hydro and energy storage system is using mixed-integer linear programming (MILP). Although MILP models can provide accurate production cost estimations, they are typically very computationally expensive. To provide a fast yet accurate first-step information to hydropower plant owners or operators who consider integrating energy storage systems, we propose an innovative approach to predicting optimal revenues of an integrated energy generation and storage system. In this study, we examined the performance of two prediction techniques: Generalized Additive Models (GAMs) and machine learning (ML) models developed based on artificial neural networks (ANN). Predictive equations and models are generated based on optimized solutions from a market participation optimization model, the Conventional Hydropower Energy and Environmental Resource System (CHEERS) model. The two predicting techniques reduce the computational time to evaluate annual revenue for one set of battery configurations from 3 h to 1 to 4 min per run while also being implementable with significantly less data. The model validation prediction errors of developed GAMs and ML models are generally below 5%; for model testing predictions, the ML models consistently outperform the regression equations in terms of root mean square errors. This new approach allows plant owners, operators, or potential investors to quickly access multiple battery configurations under different energy generation and market scenarios. This new revenue prediction method will therefore help reduce the barriers, and thereby promoting the deployment of battery hybridization with existing renewable energy sources.

13 HYDRO ENERGY↗
DOE OSTI2021

A HPC Theory-Guided Machine Learning Cyberinfrastructure for Communicating Hydrometeorological Data Across Scales

High-resolution predictions of hydrometeorological variables are critical for supporting hydropower generation decisions and flood control at hydroelectric power plants. Traditional climate and hydrologic models rely on the numerical simulation of detailed physical processes. Therefore, running these simulations is time-, labor-, and computation-intensive. Improving the spatial and temporal resolution in these modeling outputs could lead to cubic increases in both the simulation time and computational demands, rendering high-resolution hydrometeorological predictions expensive and impractical. Many past studies apply the super resolution (SR) technique to downscale climate models using deep learners. However, deep learners are deemed “black-boxes,” as their derivation processes from low-resolution outputs to high-resolution outputs are often hidden. Their results are difficult for domain scientists to interpret and validate. Thus, there is a need for an exploratory machine learning approach that can partially integrate domain-specific theory and knowledge into the data-driven mapping process between simulation outputs of different spatial scales. The domain-specific theory and knowledge can be incorporated into the data model through an inductive approach in which process-related environmental variables are used and analyzed as key drivers (i.e., environmental surrogates) to reflect the complex physical processes. Many of these variables, such as land use land cover, soil types, topography, digital elevation, air temperature, and various watershed characteristics, can be directly measured through sensors or remote sensing techniques. Additionally, SR applications that can downscale hydrological and hydrodynamics models to efficiently produce high-resolution (1 m) flood depth grids are still rare. Since the flood depth grid can be used to support critical decisions for flood control operation at hydroelectric power plants, it is crucial to enable an SR-based capability for interpolating high-resolution flood inundation maps.

13 HYDRO ENERGY↗
DOE OSTI2025

Transformers and Long Short-Term Memory Transfer Learning for GenIV Reactor Temperature Time Series Forecasting

Automated monitoring of the coolant temperature can enable autonomous operation of generation IV reactors (GenIV), thus reducing their operating and maintenance costs. Automation can be accomplished with machine learning (ML) models trained on historical sensor data. However, the performance of ML usually depends on the availability of large amount of training data, which is difficult to obtain for GenIV, as this technology is still under development. We propose the use of transfer learning (TL), which involves utilizing knowledge across different domains, to compensate for this lack of training data. TL can be used to create pre-trained ML models with data from small-scale research facilities, which can then be fine-tuned to monitor GenIV reactors. In this work, we develop pre-trained Transformer and long short-term memory (LSTM) networks by training them on temperature measurements from thermal hydraulic flow loops operating with water and Galinstan fluids at room temperature at Argonne National Laboratory. The pre-trained models are then fine-tuned and re-trained with minimal additional data to perform predictions of the time series of high temperature measurements obtained from the Engineering Test Unit (ETU) at Kairos Power. The performance of the LSTM and Transformer networks is investigated by varying the size of the lookback window and forecast horizon. The results of this study show that LSTM networks have lower prediction errors than Transformers, but LSTM errors increase more rapidly with increasing lookback window size and forecast horizon compared to the Transformer errors.

LSTM↗
DOE OSTI2024

Quantification of the Crack Evolution Process by Extracting Relevant Signal Components from Wave Propagation and Diffusive Transport Front Measurements

Wave propagation and diffusive transport phenomena in a geological rock sample undergoing crack evolution process are expected to interact with the mechanical discontinuities in the medium. The measurements of the signals associated with these phenomena can be used to assess and monitor the crack-driven micromechanical alterations in the rock. Different wave/diffusion phenomena, such as sonic propagation, pressure diffusion, and acoustic emission (AE), are sensitive to different elements of the mechanical discontinuities generated during the evolution of the crack clusters from initiation to coalescence. Sonic propagation, AE, and pressure diffusion monitoring have the potential to map the crack evolution because the transmitter-receiver arrays can be designed, arranged and tuned to (1) achieve maximum recovery of the scattered waveforms and travel times, (2) capture the later arrivals and multiple reflections, and (3) illuminate large rock volume. However, the structural/topological complexities of the mechanical discontinuities, complex distribution of the stress fields, complex mechanical alterations in media, and fluid redistribution in the crack system pose serious challenges for the detection and modeling of the crack evolution process (from here on, we will use the term ‘crack evolution process’ to mean that the crack evolution occurred under shallow crustal conditions). For purposes of accurately accounting such complexities and heterogeneities in the absence of reliable physical laws, simulation methods, and signal processing techniques, my early-career research proposal will develop and apply novel data-driven machine learning methods to: (1) extract signal components relevant to the various phases of crack evolution and (2) generate a 2D visual map of the crack evolution process.

58 GEOSCIENCES↗
DOE OSTI2022

Development of a plug-and-play anti-noise module for fault diagnosis of rotating machines in nuclear power plants

The health of rotating machines is crucial to the stable operation and safety of nuclear power plants. However, research on machine learning-based fault diagnosis of rotating machines in the nuclear industry is still in its infancy. The signal noise generated in the plant may negatively affect the effectiveness of the analysis. In this paper, a plug-and-play anti-noise machine learning module is proposed to fill the knowledge and capability gap. The modules are loaded into a convolutional neural network called deep residual network (ResNet) to obtain a new model with noise reduction capability. The basic idea is that the module is able to identify noise features and include them in the subsequent analysis, effectively filtering noise at the feature level of the network. Nine variants of the new model are compared with the original ResNet as well as four classical machine learning models to test the effectiveness of the module and to examine the impact of the module's loading modes on the performance of the new model. Finally, this research helps facilitate the application of machine learning in the plant noise environment.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2024

Novel ceramic capacitors with ultrahigh energy density and efficiency (Final Technical Report)

Antiferroelectric ceramics are a special class of material that have shown great potential as the dielectric in electrical capacitors due to their high energy- and power-density. During each charge-discharge cycle, the ceramic undergoes transformation to a ferroelectric phase and resumes its antiferroelectric phase. The hysteresis associated with the transitions leads to a mediocre energy efficiency and service lifetime of antiferroelectric capacitors and, hence, their almost absence in commercial products. Under the support of this research project, we first formulated a universal lattice-compatibility theory that included electrostatic polarization energy along with elastic energy and thermal energy to understand the origin of the hysteresis in antiferroelectric oxides. Guided by this compatibility theory, we conducted high-throughput density functional theory (DFT) calculations to assess chemical modifiers and their effect on crystal structures of 400+ PbZrO 3 -based compositions. Down-selected compositions were experimentally validated for their suppressed hysteresis and higher energy efficiency. The verified low-hysteresis compositions were then expanded to an antiferroelectric ceramic library with nearly 500 new compositions (more than 1,500 samples) using high-throughput experiments involving ceramic synthesis and property screening. The large quantity of data generated (theory and experimental) in these tasks were processed by machine-learning techniques and identified trends were fed to the next iteration. In the end, we successfully discovered four compositions with near-zero hysteresis, yielding a world-record energy efficiency of 98.2% at an energy density of 3.0 J/cm 3 . Furthermore, our antiferroelectric ceramic capacitor reaches 79.5 million charge-discharge cycles lifetime, a factor of 80 enhancement over previous antiferroelectric ceramics with large hysteresis. These research accomplishments have not only met the milestones set in the SOPO, but also led to two patent filings, three journal publications (one of them was in Advanced Materials, impact factor 29.4), and nine oral presentations at various venues. Through the course of the project, three postdocs, four Ph.D. students, and one M.S. student were trained. In short, our project established a new methodology in searching next-generation functional ceramics on the fundamental side and discovered several high-efficiency antiferroelectric compositions for capacitors on the applied side. Once fabricated into the multilayer form for commercial applications, these ceramic capacitors can potentially enable the high temperature high power density DC-link capacitors that are critical for the next generation inverters in electric vehicles. The project also significantly contributed to the nation’s workforce development in the STEM fields.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Integration of machine learning with neutron scattering for the Hamiltonian tuning of spin ice under pressure

Quantum materials research requires co-design of theory with experiments and involves demanding simulations and the analysis of vast quantities of data, usually including pattern recognition and clustering. Artificial intelligence is a natural route to optimise these processes and bring theory and experiments together. Here, we propose a scheme that integrates machine learning with high-performance simulations and scattering measurements, covering the pipeline of typical neutron experiments. Our approach uses nonlinear autoencoders trained on realistic simulations along with a fast surrogate for the calculation of scattering in the form of a generative model. We demonstrate this approach in a highly frustrated magnet, Dy 2 Ti 2 O 7 , using machine learning predictions to guide the neutron scattering experiment under hydrostatic pressure, extract material parameters and construct a phase diagram. Our scheme provides a comprehensive set of capabilities that allows direct integration of theory along with automated data processing and provides on a rapid timescale direct insight into a challenging condensed matter system.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Optimization of microwave emission from laser filamentation with a machine learning algorithm

We demonstrate that is it possible to optimize the yield of microwave radiation from plasmas generated by laser filamentation in atmosphere through manipulation of the laser wavefront. A genetic algorithm controls a deformable mirror that reconfigures the wavefront using the microwave waveform amplitude as feedback. Optimization runs performed as a function of air pressure show that the genetic algorithm can double the microwave field strength relative to when the mirror surface is flat. Here, an increase in the volume and brightness of the plasma fluorescence accompanies the increase in microwave radiation, implying an improvement in the laser beam intensity profile through the filamentation region due to the optimized wavefront.

Englesbe, Alexander↗
DOE OSTI2023

Automatic point Cloud Building Envelope Segmentation (Auto-CuBES) using Machine Learning

Modern retrofit construction practices use 3D point cloud data of the building envelope to obtain the as-built dimensions. However, manual segmentation by a trained professional is required to identify and measure window openings, door openings, and other architectural features, making the use of 3D point clouds labor-intensive. In this study, the Automatic point Cloud Building Envelope Segmentation (Auto-CuBES) algorithm is described, which can significantly reduce the time spent during point cloud segmentation. The Auto-CuBES algorithm inputs a 3D point cloud generated by commonly available surveying equipment and outputs a wire-frame model of the building envelope. Unsupervised machine learning methods were used to identify facades, windows, and doors while minimizing the number of calibration parameters. Additionally, Auto-CuBES generates a heat map of each facade indicating non-planar characteristics that are crucial for the optimization of connections used in overclad envelope retrofits. With a scan resolution of 3 mm, the resulting window dimensions showed a mean absolute error of 4.2 mm compared to manual laser measurements.

Maldonado Puente, Bryan↗
DOE OSTI2021

Expansion of Machine-Learning Method for Classifying Neutron Resonances

The understanding of astrophysics processes and the performance of nuclear reactors and other nuclear systems depend on a precise description of the neutron interaction cross sections for materials and nuclei present in these environments. At low neutron energies, these cross sections exhibit resonance structure represented by sharp enhancements when the neutron energy is sufficiently close to excited levels in a compound nucleus. Such resonances can be characterized by their quantum numbers relative to angular momenta, which are often deduced in an ad hoc and irreproducible manner from the shape of the cross sections. The correct assignment of the quantum numbers of neutron resonances is therefore of paramount importance. To address this we have developed a machine-learning method to automate the identification and correction of these spin assignments. The algorithm is trained from simulated data, generated from statistical properties of resonance data for a given nucleus, to mimic the errors found in real data. In this project we describe five independent approaches to further develop and expand the applicability of the machine-learning spin classifier: i) Feature impact; ii) Integration with the Atlas; iii) Training optimization; iv) Spacings systematics; and v) Validation with polarized data. The premises, methods, results, and future perspectives are discussed.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2023

Theoretical Multinary Perovskite Oxides Dataset

Perovskite oxides (ternary chemical formula ABO3) are a diverse class of materials with applications including heterogeneous catalysis, solid-oxide fuel cells, thermochemical conversion, and oxygen transport membranes. However, their multicomponent chemical space is underexplored due to the immense number of possible compositions. To expand the number of computed compounds we report a dataset of 66,516 theoretical multinary oxides, 59,708 of which are perovskites. First, the AA'BB'O6 compositions were generated in the a-b+a- Glazer tilting mode using the computationally-inexpensive Structure Prediction and Diagnostic Software (SPuDS) program. Next, we optimized these structures with DFT using parameters compatible with the Materials Project (MP) database. Our dataset contains these optimized structures and their formation (ΔHf) and decomposition enthalpies (ΔHd) computed relative to MP tabulated elemental references and competing phases, respectively. This dataset can be mined, used to train machine learning models, and rapidly and systematically expanded by optimizing more SPuDS-generated AA'BB'O6 perovskite structures using MP-compatible DFT calculations

Bare, Zachary↗
DOE OSTI2023

Machine Learning-Based Regression Models for Ironmaking Blast Furnace Automation

Computational fluid dynamics (CFD)-based simulation has been the traditional way to model complex industrial systems and processes. One very large and complex industrial system that has benefited from CFD-based simulations is the steel blast furnace system. The problem with the CFD-based simulation approach is that it tends to be very slow for generating data. The CFD-only approach may not be fast enough for use in real-time decisionmaking. To address this issue, in this work, the authors propose the use of machine learning techniques to train and test models based on data generated via CFD simulation. Regression models based on neural networks are compared with tree-boosting models. In particular, several areas (tuyere, raceway, and shaft) of the blast furnace are modeled using these approaches. The results of the model training and testing are presented and discussed. The obtained R 2 metrics are, in general, very high. The results appear promising and may help to improve the efficiency of operator and process engineer decisionmaking when running a blast furnace.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

ppdx : Automated modeling of protein–protein interaction descriptors for use with machine learning

This paper describes ppdx, a python workflow tool that combines protein sequence alignment, homology modeling, and structural refinement, to compute a broad array of descriptors for characterizing protein–protein interactions. The descriptors can be used to predict various properties of interest, such as protein–protein binding affinities, or inhibitory concentrations (IC 50 ), using approaches that range from simple regression to more complex machine learning models. The software is highly modular. It supports different protocols for generating structures, and 95 descriptors can be currently computed. More protocols and descriptors can be easily added. The implementation is highly parallel and can fully exploit the available cores in a single workstation, or multiple nodes on a supercomputer, allowing many systems to be analyzed simultaneously. As an illustrative application, ppdx is used to parametrize a model that predicts the IC 50 of a set of antigens and a class of antibodies directed to the influenza hemagglutinin stalk.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

Dataset of theoretical multinary perovskite oxides

Perovskite oxides (ternary chemical formula ABO 3 ) are a diverse class of materials with applications including heterogeneous catalysis, solid-oxide fuel cells, thermochemical conversion, and oxygen transport membranes. However, their multicomponent (chemical formula $A_xA^{'}_{1-x}B_yB^{'}_{1-y}O_3$) chemical space is underexplored due to the immense number of possible compositions. To expand the number of computed $A_xA^{'}_{1-x}B_yB^{'}_{1-y}O_3$ compounds we report a dataset of 66,516 theoretical multinary oxides, 59,708 of which are perovskites. First, 69,407 $A_{0.5}A^{'}_{0.5}B_{0.5}B^{'}_{0.5}O_3$ compositions were generated in the a - b + a - Glazer tilting mode using the computationally-inexpensive Structure Prediction and Diagnostic Software (SPuDS) program. Next, we optimized these structures with density functional theory (DFT) using parameters compatible with the Materials Project (MP) database. Our dataset contains these optimized structures and their formation (ΔH f ) and decomposition enthalpies (ΔH d ) computed relative to MP tabulated elemental references and competing phases, respectively. This dataset can be mined, used to train machine learning models, and rapidly and systematically expanded by optimizing more SPuDS-generated $A_{0.5}A^{'}_{0.5}B_{0.5}B^{'}_{0.5}O_3$ perovskite structures using MP-compatible DFT calculations.

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

Development of a machine learning model for polyethylene pyrolysis using a detailed reaction mechanism

Waste plastics have recently received significant attention as the issue of waste generation continues to increase. Thermal conversion processes, such as pyrolysis and gasification, are attractive potential technologies for utilizing waste plastics and reducing overall waste generation. Efficient utilization of plastics requires a detailed understanding of the conversion process such as pyrolysis and gasification. However, a mechanistic understanding of these processes lead to large and complex kinetic schemes that are not suited for large-scale and long-time simulation methods. Currently, most modeling approaches for pyrolysis and gasification rely on globally lumped, simplified kinetic schemes that provide results that are classified by their product type and not individual species, which limit the level of fidelity achieved via modeling. A machine learning (ML) model has been developed for the primary reactions of high-density polyethylene (HDPE) in an attempt to increase computational efficiency while still maintaining a high level of detail and accuracy. The ML model is trained on a detailed reaction mechanism containing 42 total species and 737 chemical reactions. A DeepONet branch and trunk architecture was adopted to train the model using time-steps relevant to computational fluid dynamics simulations. The ML used physics-informed loss functions to ensure mass conservation. The surrogate model has been deployed in simple MFiX CFD simulations, single particle and an experimental drop tube reactor, and has shown promising performance compared to the original scheme.

Houston, Ross↗