Results for “factorization machine” — Page 13

DOE OSTI2025

Machine Learning for Automated Weld Quality Monitoring and Control

Resistance Spot Welding (RSW) is a critical process in the automotive industry, valued for its cost-effectiveness, short cycle time, and robustness. However, achieving consistent high-quality joints remains challenging due to the complex interplay of various factors, like materials, processes, and manufacturing uncertainties, etc. Under the collaborative project between Oak Ridge National Laboratory (ORNL) and General Motors (GM), we have developed a robust and expansible machine learning (ML) framework aimed at enhancing quality control in RSW. By harnessing the power of machine learning, we have developed the ability to ensure every aspect of the welding process, from the initial process design stage to the final weld joint quality. The framework operates by analyzing a variety of data streams, including in-line process signals, process parameters, materials, and postprocessed weld joint data. Through this analysis, the models have been trained to detect deviations from optimal quality standards, leveraging their ability to identify signature data patterns and anomalies within in-line signals and construct complex correlations between these signals and weld quality parameters. Meanwhile, the machine learning framework is designed to adapt to a variety of materials, including high strength steels and aluminum alloys, etc. Its flexible architecture facilitates the incorporation of diverse data sources and features, enabling precise modeling and prediction across a broad range of material properties and weld quality variables. The expansible ML frameworks represent a promising transformation in weld quality monitoring and control, empowering industry to achieve high levels of efficiency, consistency, and reliability in manufacturing.

42 ENGINEERING↗

NASA NTRS1960

Over-Water Aspects of Ground-Effect Vehicles

The large thrust augmentation obtainable with annular-jet configurations in ground proximity has led to the serious investigation of ground-effect machines. The basic theoretical work on these phenomena has been done by Chaplin and Boehler. Large thrust-augmentation factors, however, can be obtained only at very low heights, that is, of the order of a few percent of the diameter of the vehicle. To take advantage of this thrust augmentation therefore the vehicle must be either very large or must operate over very smooth terrain. Over-land uses of these vehicles then will probably be rather limited. The water, however, is inherently smooth and those irregularities that do exist, that is waves, are statistically known. It appears therefore that some practical application of ground-effect machines may be made in over-water application.

Kuhn, Richard E.↗

DOE OSTI2025

Q-Cluster: Quantum Error Mitigation Through Noise-Aware Unsupervised Learning

Quantum error mitigation (QEM) is critical in reducing the impact of noise in the pre-fault-tolerant era, and is expected to complement error correction in fault-tolerant quantum computing (FTQC). In this work, we propose a novel QEM approach, Q-Cluster, that uses unsupervised learning (clustering) to reshape the measured bit-string distribution. Our approach starts with a simplified bit-flip noise model. It first performs clustering on noisy measurement results, i.e., bit-strings, based on the Hamming distance. The centroid of each cluster is calculated using a qubit-wise majority vote. Next, the noisy distribution is adjusted with the clustering outcomes and the bitflip error rates using Bayesian inference. Our simulation results show that Q-Cluster can mitigate high noise rates (up to 40% per qubit) with the simple bit-flip noise model. However, real quantum computers do not fit such a simple noise model. To address the problem, we (a) apply Pauli twirling to tailor the complex noise channels to Pauli errors, and (b) employ a machine learning model, ExtraTrees regressor, to estimate an effective bit-flip error rate using a feature vector consisting of machine calibration data (gate & measurement error rates), circuit features (number of qubits, numbers of different types of gates, etc.) and the shape of the noisy distribution (entropy). Our experimental results show that our proposed Q-Cluster scheme improves the fidelity by a factor of 1.46x, on average, compared to the unmitigated output distribution, for a set of low-entropy benchmarks on five different IBM quantum machines. Our approach outperforms the state-of-art QEM approaches RZNE [28], M3 [24], Hammer [35], and QBEEP [33] by 1.26x,1.29x,1.47x, and 2.65 x, respectively.

42 ENGINEERING↗

DOE OSTI2024

Dark Energy Survey Deep Field photometric redshift performance and training incompleteness assessment

Context. The determination of accurate photometric redshifts (photo-zs) in large imaging galaxy surveys is key for cosmological studies. One of the most common approaches are machine learning techniques. These methods require a spectroscopic or reference sample to train the algorithms. Attention has to be paid to the quality and properties of these samples since they are key factors in the estimation of reliable photo-zs. Aims. The goal of this work is to calculate the photo-zs for the Y3 DES Deep Fields catalogue using the DNF machine learning algorithm. Moreover, we want to develop techniques to assess the incompleteness of the training sample and metrics to study how incompleteness affects the quality of photometric redshifts. Finally, we are interested in comparing the performance obtained with respect to the EAzY template fitting approach on Y3 DES Deep Fields catalogue. Methods. We have emulated -- at brighter magnitude -- the training incompleteness with a spectroscopic sample whose redshifts are known to have a measurable view of the problem. We have used a principal component analysis to graphically assess incompleteness and to relate it with the performance parameters provided by DNF. Finally, we have applied the results about the incompleteness to the photo-z computation on Y3 DES Deep Fields with DNF and estimated its performance. Results. The photo-zs for the galaxies on DES Deep Fields have been computed with the DNF algorithm and added to the Y3 DES Deep Fields catalogue. They are available at https://des.ncsa.illinois.edu/releases/y3a2/Y3deepfields. Some techniques have been developed to evaluate the performance in the absence of "true" redshift and to assess completeness. We have studied... (Partial abstract)

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2024

Improving North American Wildfire Prediction by Integrating a Machine-Learning Fire Model in a Land Surface Model

Wildfires have shown increasing trends in both frequency and severity across the Contiguous United States (CONUS). However, process-based fire models have difficulties in accurately simulating the burned area over the CONUS due to a simplification of the physical process and cannot capture the interplay among fire, ignition, climate, and human activities. The deficiency of burned area simulation deteriorates the description of fire impact on energy balance, water budget, and carbon fluxes in the Earth System Models (ESMs). Alternatively, machine learning (ML) based fire models, which capture statistical relationships between the burned area and environmental factors, have shown promising burned area predictions and corresponding fire impact simulation. We develop a hybrid framework (ML4Fire-XGB) that integrates a pretrained eXtreme Gradient Boosting (XGBoost) wildfire model with the Energy Exascale Earth System Model (E3SM) land model (ELM) version 2.1. A Fortran-C-Python deep learning bridge is adapted to support online communication between ELM and the ML fire model. Specifically, the burned area predicted by the ML-based wildfire model is directly passed to ELM to adjust the carbon pool and vegetation dynamics after disturbance, which are then used as predictors in the ML-based fire model in the next time step. Evaluated against the historical burned area from Global Fire Emissions Database 5 from 2001-2020, the ML4Fire-XGB model outperforms process-based fire models in terms of spatial distribution and seasonal variations. Sensitivity analysis confirms that the ML4Fire-XGB well captures the responses of the burned area to rising temperatures. The ML4Fire-XGB model has proved to be a new tool for studying vegetation-fire interactions, and more importantly, enables seamless exploration of climate-fire feedback, working as an active component in E3SM.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2024

Simulated wildfire burned area over the CONUS during 2001-2020

Wildfires have shown increasing trends in both frequency and severity across the Contiguous United States (CONUS). However, process-based fire models have difficulties in accurately simulating the burned area over the CONUS due to a simplification of the physical process and cannot capture the interplay among fire, ignition, climate, and human activities. The deficiency of burned area simulation deteriorates the description of fire impact on energy balance, water budget, and carbon fluxes in the Earth System Models (ESMs). Alternatively, machine learning (ML) based fire models, which capture statistical relationships between the burned area and environmental factors, have shown promising burned area predictions and corresponding fire impact simulation. We develop a hybrid framework (ML4Fire-XGB) that integrates a pretrained eXtreme Gradient Boosting (XGBoost) wildfire model with the Energy Exascale Earth System Model (E3SM) land model (ELM). A Fortran-C-Python deep learning bridge is adapted to support online communication between ELM and the ML fire model. Specifically, the burned area predicted by the ML-based wildfire model is directly passed to ELM to adjust the carbon pool and vegetation dynamics after disturbance, which are then used as predictors in the ML-based fire model in the next time step. Evaluated against the historical burned area from Global Fire Emissions Database 5 from 2001-2020, the ML4Fire-XGB model outperforms process-based fire models in terms of spatial distribution and seasonal variations. Sensitivity analysis confirms that the ML4Fire-XGB well captures the responses of the burned area to rising temperatures. The ML4Fire-XGB model has proved to be a new tool for studying vegetation-fire interactions, and more importantly, enables seamless exploration of climate-fire feedback, working as an active component in E3SM.

Liu, Ye↗

NASA NTRS1977

Evaluation of a low aspect ratio small axial compressor stage, volume 1

A program was conducted to evaluate the effects of scaling, tip clearance, and IGV reset on the performance of a low aspect ratio compressor stage. Stage design was obtained by scaling an existing single stage compressor by a linear factor of 0.304. The design objective was to maintain the meanline velocity field of the base machine in the smaller size. Adjustments were made to account for predicted blockage differences and to chord lengths and airfoil edge radii to obtain reasonable blade geometries. Meanline velocity diagrams of the base stage were not maintained at the scaled size. At design speed and flowrate the scaled stage achieved a pressure ratio of 1.423, adiabatic efficiency of 0.822, and surge margin of 18.5%. The corresponding performance parameters for the base stage were 1.480, 0.872, and 25.2%, respectively. The base stage demonstrated a peak efficiency at design speed of 0.872; the scaled stage achieved a level of 0.838. When the scaled stage rotor and stator tip clearances were doubled, the stage achieved a pressure ratio of 1.413, efficiency of 0.799, and surge margin of 16.0% at the design flowrate. The peak stage efficiency at design speed was 0.825 with the increased clearance. Increased prewhirl lowered the stage pressure ratio as expected. Stage efficiency was maintained with ten degrees of increased prewhirl and then decreased substantially with ten additional degrees of reset.

Sawyer, C. W., III↗

NASA NTRS1977

Experience with a vectorized general circulation weather model on Star-100

A version of an atmospheric general circulation model was vectorized to run on a CDC STAR 100. The numerical model was coded and run in two different vector languages, CDC and LRLTRAN. A factor of 10 speed improvement over an IBM 360/95 was realized. Efficient use of the STAR machine required some redesigning of algorithms and logic. This precludes the application of vectorizing compilers on the original scalar code to achieve the same results. Vector languages permit a more natural and efficient formulation for such numerical codes.

Soll, D. B.↗

NASA NTRS1984

The Strength of the Metal. Aluminum Oxide Interface

The strength of the interface between metals and aluminum oxide is an important factor in the successful operation of devices found throughout modern technology. One finds the interface in machine tools, jet engines, and microelectronic integrated circuits. The strength of the interface, however, should be strong or weak depending on the application. The diverse technological demands have led to some general ideas concerning the origin of the interfacial strength, and have stimulated fundamental research on the problem. Present status of our understanding of the source of the strength of the metal - aluminum oxide interface in terms of interatomic bonds are reviewed. Some future directions for research are suggested.

Pepper, S. V.↗

NASA NTRS1988

Management of human error by design

Design-induced errors and error prevention as well as the concept of lines of defense against human error are discussed. The concept of human error prevention, whose main focus has been on hardware, is extended to other features of the human-machine interface vulnerable to design-induced errors. In particular, it is pointed out that human factors and human error prevention should be part of the process of transport certification. Also, the concept of error tolerant systems is considered as a last line of defense against error.

Wiener, Earl↗

NASA NTRS2001

Astrobiology Sample Analysis as a Design Driver

This effort supports the Astrobiology Objective 8 the Search for LIFE ON MARS PAST AND PRESENT -(Astrobiology Program Office, 1998, p.7). The essential trade analysis is between returning very small samples to the Earth while protecting them versus in situ analysis on Mars. Developing these explicit parameters encompasses design, instrumentation, system integration, human factors and surface operations for both alternatives. This allocation of capability approach incorporates a "humans and machines in the loop" model that recognizes that every exploration system involves both humans and automated systems. The question is where in the loop they occur whether on Earth, in the Mars Base, in the rover or creeping over the Mars surface.

Cohen, Marc M.↗

NASA NTRSDate not supplied

Linear Regression Model for Predictive Service Provider Selection

The increasing number of satellites in orbit has led to a growing reliance on third-party service providers for data transfer between Earth and space. Traditional approaches to managing satellite communications require human intervention, which becomes more burdensome with the escalating number of satellites. This research addresses the need for an efficient and automated system to optimize service provider selection for NASA space communication. Previous research has utilized human-operated approaches for service provider management. Our study fills a gap by developing a cognitive algorithm that automates and optimizes the selection process based on various parameters, such as data volume, priority, quality of service and cost. This novel solution reduces user burden, facilitates service management, and contributes to the development of cognitive spaceflight missions, ultimately supporting NASA’s research into Cognitive Communications technology. The algorithm design consists of three major steps: modeling data, developing a Link Selection Algorithm (LSA) based on a grading system, and applying machine learning using linear regression. The LSA evaluates providers based on user-defined constraints, considering factors such as delivery time, cost, and quality of service. We define a suitability metric which allows our algorithm to make a recommendation to a user regarding which commercial service providers to select. The addition of Linear Regression predicts the future suitability value. Our main findings demonstrate that the resulting algorithm can autonomously manage connections between satellites and providers, maximizing communication channel efficiency. This research has significant implications, as it not only addresses a pressing issue in satellite communication management but also advances the field of cognitive spaceflight missions.

Linear regression↗

DOE OSTI2025

Explainable machine learning to quantify the value of proximal remote sensing in latent energy flux estimation

Proximal remote sensing has the potential to provide critical information on vegetation biophysical factors that can predict land-atmosphere exchange of water and energy. Latent energy (LE) flux is traditionally estimated using process-based models which rely on vegetation parameters that change during the growing season. Data-driven models have the potential to address these issues by offering flexible predictor selection and more efficient utilization of the information in predictor sets. These models require careful choice of predictors to avoid redundancy and allow robust cross-validation. In this study we present a systematic and comprehensive evaluation of machine learning (ML) models to assess the capability of meteorological and proximal sensing data for predicting LE at a half-hourly temporal resolution across multiple growing seasons for an agricultural system. The results presented here demonstrate that a model using four environmental predictors in combination with two proximal sensing variables can capture 88 % of the variability in LE. ML models using only three predictors (one meteorological and two proximal remote sensing) captured 81 % of LE variability, offering the best trade-off between performance and complexity. An ML model utilizing only two predictors, one proximal remote sensing variable and downwelling radiation, captured 77 % of LE variability. These results demonstrate the power of proximal remote sensing and meteorological observations to estimate land-atmosphere water vapor exchange, providing a solution where more direct methods such as eddy covariance are not available and for evaluations of agronomic management and genotypic variations.

60 APPLIED LIFE SCIENCES↗

NASA NTRS1978

Man-machine interface analysis of the flight design system

The objective of the current effort was to perform a broad analysis of the human factors issues involved in the design of the Flight Design System (FDS). The analysis was intended to include characteristics of the system itself, such as: (1) basic structure and functional capabilities of FDS; (2) user backgrounds, capabilities, and possible modes of use; (3) FDS interactive dialogue, problem solving aids; (4) system data management capabilities; and to include, as well, such system related matters as: (1) flight design team structure; (2) roles of technicians; (3) user training; and (4) methods of evaluating system performance. Wherever possible, specific recommendations are made. In other cases, the issues which seem most important are identified. In some cases, additional analyses or experiments which might provide resolution are suggested.

Ramsey, H. R.↗

NASA NTRSDate not supplied

Transient Optimization of an Electrified Gas Turbine Engine Using Machine Learning

Gas turbine engines are designed with sufficient margin to prevent stall under normal operating conditions throughout their life. This compromise ensures that during rapid accelerations, compressor operation remains stable, but at the cost of efficiency and thrust responsiveness. The design margin encompasses multiple sources of uncertainty and systematic deviances from the operating line, the largest of which is the transient allowance. This set-aside accounts for the temporary incoordination of the engine spools during an acceleration while still enabling it to meet the certification requirement to accelerate from low to high power within a specified time, and without experiencing overtemperature, surge, stall, or other detrimental factors. Electrification of the powertrain provides the opportunity to address this reserve and truly optimize the design. The addition of electric machines inherent in hybrid propulsion concepts offers a means to interact with the engine shafts such that the necessary margin can be reduced, which can positively impact the engine design. By adjusting the amount of power extracted from or injected to the engine spools by the electric machines during transient operation, excursions from the operating line can be minimized. Past work using a dynamic engine model has shown that optimization of the fuel flow schedule during acceleration can reduce the required margin while still meeting the time requirement, and results are further improved when combined with power injection and extraction. The current work uses machine learning through a genetic algorithm to address the problem holistically by concurrently optimizing the electric machine power command and fuel flow acceleration schedule using an updated, higher fidelity version of the original engine model.

Stall Margin↗

NASA NTRSDate not supplied

Transient Optimization of an Electrified Gas Turbine Engine Using Machine Learning

Gas turbine engines are designed with sufficient margin to prevent stall under normal operating conditions throughout their life. This compromise ensures that during rapid accelerations, compressor operation remains stable, but at the cost of efficiency and thrust responsiveness. The design margin encompasses multiple sources of uncertainty and systematic deviances from the operating line, the largest of which is the transient allowance. This set-aside accounts for the temporary incoordination of the engine spools during an acceleration while still enabling it to meet the certification requirement to accelerate from low to high power within a specified time, and without experiencing overtemperature, surge, stall, or other detrimental factors. Electrification of the powertrain provides the opportunity to address this reserve and truly optimize the design. The addition of electric machines inherent in hybrid propulsion concepts offers a means to interact with the engine shafts such that the necessary margin can be reduced, which can positively impact the engine design. By adjusting the amount of power extracted from or injected to the engine spools by the electric machines during transient operation, excursions from the operating line can be minimized. Past work using a dynamic engine model has shown that optimization of the fuel flow schedule during acceleration can reduce the required margin while still meeting the time requirement, and results are further improved when combined with power injection and extraction. The current work uses machine learning through a genetic algorithm to address the problem holistically by concurrently optimizing the electric machine power command and fuel flow acceleration schedule using an updated, higher fidelity version of the original engine model.

Stall Margin↗

NASA NTRS2024

Transient Optimization of an Electrified Gas Turbine Engine Using Machine Learning

Gas turbine engines are designed with sufficient margin to prevent stall under normal operating conditions throughout their life. This compromise ensures that during rapid accelerations, compressor operation remains stable, but at the cost of efficiency and thrust responsiveness. The design margin encompasses multiple sources of uncertainty and systematic deviances from the operating line, the largest of which is the transient allowance. This set-aside accounts for the temporary incoordination of the engine spools during an acceleration while still enabling it to meet the certification requirement to accelerate from low to high power within a specified time, and without experiencing overtemperature, surge, stall, or other detrimental factors. Electrification of the powertrain provides the opportunity to address this reserve and truly optimize the design. The addition of electric machines inherent in hybrid propulsion concepts offers a means to interact with the engine shafts such that the necessary margin can be reduced, which can positively impact the engine design. By adjusting the amount of power extracted from or injected to the engine spools by the electric machines during transient operation, excursions from the operating line can be minimized. Past work using a dynamic engine model has shown that optimization of the fuel flow schedule during acceleration can reduce the required margin while still meeting the time requirement, and results are further improved when combined with power injection and extraction. The current work uses machine learning through a genetic algorithm to address the problem holistically by concurrently optimizing the electric machine power command and fuel flow acceleration schedule using an updated, higher fidelity version of the original engine model.

Stall Margin↗

NASA NTRS2003

Gravitational Effects on Distortion in Sintering

During sintering a powder compact gains strength through low-temperature interparticle bonding, usually induced by solid-state surface diffusion, followed by further strength contributions from high-temperature densification. In cases where a liquid phase forms, sintering densification is accelerated and shape retention is sustained while open pores remain and contribute capillary forces. Unfortunately, sintering densification requires the compact become thermally softened to a point where creep strain rates reach levels near 10(exp -2)/s when the liquid forms. On the other hand, thermal softening of the powder compact substantially reduces the strength at high temperatures. Therefore, the in situ strength evolution during sintering is a primary focus to separate compact densification (as required for high performance) with minimized distortion (as required for net-shaping). With respect to gravitation effects on distortion during sintering there are two points of substantial weakness - prior to significant interparticle bonding and during final pore closure. This research is focused on understanding the competition among interparticle neck growth, densification, thermal softening, grain boundary wetting, capillary effects associated with liquid wetting and residual porosity, and gravity. Most surprising is the apparent role of gravity, where the deviatoric stress acting on the powder structure induces skeletal formation that reduces distortion. In contrast with theory, microgravity samples exhibit more distortion yet fail to fully densify. Results are presented on the experimental concepts supporting an emerging model of sintering strength evolution that enables understanding of both distortion and densification. The experiments have relied on tungsten heavy alloys, various combinations of dihedral angle, pore size, initial porosity, liquid:solid ratio, and heating rates. On Earth, the dominant factor with respect to distortion is the starting body heterogeneity. Current modeling efforts are seeking some means to uniformly predict the distortion based on a starting pore structure heterogeneity parameter. Densification is largely unaffected by the initial pore structure, but distortion is highly variable, suggesting that nonuniform pore closure might be a significant parameter during sintering. With respect to flight experiments, plans include removal of the solid body forces acting on the solid grains, allowing stabilization of the pore structure and examination of the buoyancy effects with regard to distortion. In microgravity there is the surprising result that compacts do not densify, yet distort a factor that is contrary to all current sintering models. Densification without distortion during liquid phase sintering was achieved by manipulating microstructure and its evolution during sintering. Microstructure parameters such as the solid volume fraction, dihedral angle, initial porosity, and pore size were varied to measure densification and distortion behavior during LPS using W-Ni-Cu alloys. Green compacts were formed using ethylene-bis-stearamide as a pore-forming agent with the amount of polymer controlling the initial porosity. Different initial pore sizes were generated by varying the polymer particle size. Dihedral angle was varied by changing the Ni:Cu ratio in the alloys. Finally, the solid volume fraction was adjusted via the tungsten content. Distortion was quantified using profiles determined with a coordinate measuring machine to calculate a distortion parameter. Sintering results showed that solid volume fraction and dihedral angle are the dominant factors on densification and distortion during liquid phase sintering. Distortion decreases with increasing solid volume fraction and dihedral angle, while initial porosity and pore size have no observable effect on distortion at nearly full densification. Various strategies emerge to improve distortion control in liquid phase sintering.

German, Randall M.↗