SEARCH · Engineering Papers

Results for “machine learning optimization”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

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

At least 1,009 records · Page 56

All sources NASA NTRS DOE OSTI

DOE OSTI2023

Adrastea: An Efficient FPGA Design Environment for Heterogeneous Scientific Computing and Machine Learning

We present Adrastea, an efficient FPGA design environment for developing scientific machine learning applications. FPGA development is challenging, from deployment, proper toolchain setup, programming methods, interfacing FPGA kernels, and more importantly, the need to explore design space choices to get the best performance and area usage from the FPGA kernel design. Adrastea provides an automated and scalable design flow to parameterize, implement, and optimize complex FPGA kernels and associated interfaces. We show how virtualization of the development environment via virtual machines is leveraged to simplify the setup of the FPGA toolchain while deploying the FPGA boards and while scaling up the automated design space exploration to leverage multiple machines concurrently. Adrastea provides an automated build and test environment of FPGA kernels. By exposing design space hyper-parameters, Adrastea can automatically search the design space in parallel to optimize the FPGA design for a given metric, usually performance or area. Adrastea simplifies the task of interfacing with the FPGA kernels with a simplified interface API. To demonstrate the capabilities of Adrastea, we implement a complex random forest machine learning kernel with 10,000 input features while achieving extremely low computing latency without loss of prediction accuracy, which is required by a scientific edge application at SNS. We also demonstrate Adrastea using an FFT kernel and show that for both applications Adrastea is able to systematically and efficiently evaluate different design options, which reduced the time and effort required to develop the kernel from months of manual work to days of automatic builds.

Young, Aaron↗

DOE OSTI2024

Human limits in machine learning: prediction of potato yield and disease using soil microbiome data

Abstract Background The preservation of soil health is a critical challenge in the 21st century due to its significant impact on agriculture, human health, and biodiversity. We provide one of the first comprehensive investigations into the predictive potential of machine learning models for understanding the connections between soil and biological phenotypes. We investigate an integrative framework performing accurate machine learning-based prediction of plant performance from biological, chemical, and physical properties of the soil via two models: random forest and Bayesian neural network. Results Prediction improves when we add environmental features, such as soil properties and microbial density, along with microbiome data. Different preprocessing strategies show that human decisions significantly impact predictive performance. We show that the naive total sum scaling normalization that is commonly used in microbiome research is one of the optimal strategies to maximize predictive power. Also, we find that accurately defined labels are more important than normalization, taxonomic level, or model characteristics. ML performance is limited when humans can’t classify samples accurately. Lastly, we provide domain scientists via a full model selection decision tree to identify the human choices that optimize model prediction power. Conclusions Our study highlights the importance of incorporating diverse environmental features and careful data preprocessing in enhancing the predictive power of machine learning models for soil and biological phenotype connections. This approach can significantly contribute to advancing agricultural practices and soil health management.

Aghdam, Rosa↗

DOE OSTI2024

Machine learning visualization tool for exploring parameterized hydrodynamics

We are interested in the computational study of shock hydrodynamics, i.e. problems involving compressible solids, liquids, and gases that undergo large deformation. These problems are dynamic and nonlinear and can exhibit complex instabilities. Due to advances in high performance computing it is possible to parameterize a hydrodynamic problem and perform a computational study yielding $\mathscr{O} (TB)$ of simulation state data. We present an interactive machine learning tool that can be used to compress, browse, and interpolate these large simulation datasets. This tool allows computational scientists and researchers to quickly visualize 'what-if' situations, perform sensitivity analyses, and optimize complex hydrodynamic experiments.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Investigation of acoustic waves under subsurface conditions to improve the predictions of rock mechanical properties and natural fracture characteristics

Mechanical properties and natural fracture characteristics are critical to investigate for subsurface engineering applications, including carbon storage, well drilling, and stimulation, as they govern rock stability, fluid flow, and mechanical behavior under stress. This dissertation integrates experimental and machine learning approaches to enhance the prediction and understanding of these properties by analyzing acoustic wave behavior under varied subsurface conditions. First, the influence of temperature, pore pressure, and supercritical CO2 (scCO2) saturation on poroelastic properties is examined using Gray Berea sandstone samples. The results show that temperature and pore pressure significantly affect the bulk modulus and Biot’s coefficient, while scCO2 saturation impacts rock compressibility, informing strategies for effective geological carbon storage. The study extends this understanding by experimentally evaluating the impact of reservoir depletion on the dynamic mechanical properties of the emerging Caney shale in South Oklahoma with the employment of unsupervised machine learning to predict static mechanical properties across the Caney shale. Integrating petrophysical data and chemostratigraphy, the workflow—featuring K-means clustering, principal component analysis (PCA), and inverse distance weighting (IDW)—improves stratigraphic characterization and the estimation of static-to-dynamic modulus ratios, which is vital for optimizing drilling and stimulation strategies. Finally, the work explores how natural fracture characteristics in shale influence acoustic waveforms and shear wave splitting (SWS) analysis. Experimental data on fractured samples under different stress and temperature conditions, combined with machine learning models such as K-nearest neighbors (KNN) and extreme gradient boosting (XGBoost), reveal key fracture properties impacting SWS and wave propagation. Together, these studies provide a comprehensive framework for linking acoustic wave behavior with rock properties, advancing the methods for monitoring and predicting geomechanical changes. The insights offered valuable implications for safer, more efficient CO2 injection, hydrocarbon extraction, and subsurface management.

Elkholy, Sherif↗

DOE OSTI2020

Optimizing prediction of reservoir properties with artificial intelligence, big data, and the Subsurface Trend Analysis method

Presented by LRST at the Machine Learning in Oil and Gas Conference, 11/9

Mark-Moser, Mackenzie K.↗

DOE OSTI2020

Stability boundary and optimal operating parameter identification in milling using Bayesian learning

Here, this paper describes a novel Bayesian learning approach for stability boundary and optimal parameter identification in milling without the knowledge of the underlying tool dynamics or material cutting force coefficients. The paper is divided into two parts. First, a Bayesian learning method for stability lobe identification using test results is described. Each axial depth and spindle speed combination is characterized by a probability of stability which is updated using Bayes’ rule when a test result (stable or unstable) is made available. A novel likelihood function is defined which incorporates knowledge of the stability behavior. Numerical results show convergence to the analytical stability lobe diagram. Second, an adaptive experimental strategy to identify optimal operating parameters that maximize material removal rate is described. Numerical evaluation shows convergence to the optimal operating point with error less than 15 % within ten tests on average. The approach is validated using experimental results. Results show that the proposed method is an efficient and robust learning method to identify the stability lobe diagram and optimal operating parameters with a limited number of tests/data points.

Bayesian machine learning↗

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.

99 GENERAL AND MISCELLANEOUS↗

DOE OSTI2025

Machine Learning for Automated Weld Quality Monitoring and Control

42 ENGINEERING↗

DOE OSTI2026

HydraGNN v5.0

HydraGNN v5.0 expands the code base into a more portable, scalable, and flexible framework for scientific graph learning, with particular strength in atomistic machine-learning interatomic potentials and large-scale distributed training. The release adds Fully Sharded Data Parallel (FSDP) support alongside existing DDP and DeepSpeed paths, including FSDP-aware checkpointing and optimizer integration, and introduces a configurable multi-precision training workflow supporting FP32, BF16, and FP64 across GPUs and Intel XPUs. For atomistic modeling, HydraGNN v5.0 strengthens its MLIP capabilities through dynamic graph construction at every forward pass, energy-conserving force prediction via automatic differentiation, and per-atom energy loss formulations, while extending EGNN models to properly handle periodic boundary conditions. The release also broadens model expressiveness through graph-level attribute conditioning, adds new multi-task and model-parallel extensions such as MACE support and encoder/decoder branch optimization, and expands application coverage with integrated examples for datasets including OC25, Nabla2-DFT, QCML, Open Polymers 2026, and OPF. In parallel, HydraGNN v5.0 improves production readiness through performance optimizations for large-scale runs, stratified sampling and linear-regression preprocessing utilities, and tested installation scripts for DOE supercomputers including Frontier, Aurora, Perlmutter, and Andes. Overall, the release advances HydraGNN as a robust software platform for scalable graph neural networks across materials science, chemistry, and scientific machine learning workflows

Lupo Pasini, Massimiliano [Oak Ridge National Labo↗

DOE OSTI2021

Quantum Divide and Compute: Exploring the Effect of Different Noise Sources

Abstract Our recent work (Ayral et al. in Proceedings of IEEE computer society annual symposium on VLSI, ISVLSI, pp 138–140, 2020. 10.1109/ISVLSI49217.2020.00034 ) showed the first implementation of the Quantum Divide and Compute (QDC) method, which allows to break quantum circuits into smaller fragments with fewer qubits and shallower depth. This accommodates the limited number of qubits and short coherence times of quantum processors. This article investigates the impact of different noise sources—readout error, gate error and decoherence—on the success probability of the QDC procedure. We perform detailed noise modeling on the Atos Quantum Learning Machine, allowing us to understand tradeoffs and formulate recommendations about which hardware noise sources should be preferentially optimized. We also describe in detail the noise models we used to reproduce experimental runs on IBM’s Johannesburg processor. This article also includes a detailed derivation of the equations used in the QDC procedure to compute the output distribution of the original quantum circuit from the output distribution of its fragments. Finally, we analyze the computational complexity of the QDC method for the circuit under study via tensor-network considerations, and elaborate on the relation the QDC method with tensor-network simulation methods.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

Operator learning for energy-efficient building ventilation control with computational fluid dynamics simulation of a real-world classroom

Energy-efficient ventilation control plays an important role in reducing building energy consumption while ensuring occupant health and comfort. While Computational Fluid Dynamics (CFD) simulations provide detailed and physically accurate representations of indoor airflow, their high computational cost limits their use in real-time building control. In this work, we present a neural operator learning framework that combines the physical accuracy of CFD with the computational efficiency of machine learning to enable building ventilation control with the high-fidelity fluid dynamics models. Our method jointly optimizes the airflow supply rates and vent angles to reduce energy use and adhere to air quality constraints. We train an ensemble of neural operator transformer models to learn the mapping from building control actions to airflow fields using high-resolution CFD data. This learned neural operator is then embedded in an optimization-based control framework for building ventilation control. Experimental results show that our approach achieves significant energy savings compared to maximum airflow rate control, rule-based control, as well as data-driven control methods using spatially averaged CO 2 prediction and deep learning–based reduced-order models, while consistently maintaining safe indoor air quality. These results highlight the practicality and scalability of our method in maintaining energy efficiency and indoor air quality in real-world buildings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

DOE OSTI2022

Hardware-accelerated inference for real-time gravitational-wave astronomy

The field of transient astronomy has seen a revolution with the first gravitational-wave detections and the arrival of multi-messenger observations they enabled. Transformed by the first detection of binary black hole and binary neutron star mergers, computational demands in gravitational-wave astronomy are expected to grow by at least a factor of two over the next five years as the global network of kilometer-scale interferometers are brought to design sensitivity. With the increase in detector sensitivity, real-time delivery of gravitational-wave alerts will become increasingly important as an enabler of multi-messenger followup. In this work, we report a novel implementation and deployment of deep learning inference for real-time gravitational-wave data denoising and astrophysical source identification. This is accomplished using a generic Inference-as-a-Service model that is capable of adapting to the future needs of gravitational-wave data analysis. Overall, our implementation allows seamless incorporation of hardware accelerators and also enables the use of commercial or private (dedicated) as-a-service computing. Based on our results, we propose a paradigm shift in low-latency and offline computing in gravitational-wave astronomy. Such a shift can address key challenges in peak-usage, scalability and reliability, and provide a data analysis platform particularly optimized for deep learning applications. The achieved sub-millisecond scale latency will also be relevant for any machine learning-based real-time control systems that may be invoked in the operation of near-future and next generation ground-based laser interferometers, as well as the front-end collection, distribution and processing of data from such instruments.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2021

A Multi-Analysis Approach for Estimating Regional Health Impacts from the 2017 Northern California Wildfires

In the evening of October 8 and early hours of October 9, 2017, high winds in Northern California downed trees and power lines, igniting some of the most devastating wildfires the state had seen, and within hours unhealthy air quality impacted millions of people. We simulated these air quality conditions using fire detection information from the MODIS, VIIRS, and GOES-16 ABI satellite instruments, and applying a set of three WRF–CMAQ simulations, one data fusion, and three machine learning methods. We investigated using the 5-min available GOES-16 fire detection data to simulate timing of fire activity to allocate emissions hourly for the WRF-CMAQ air quality modeling system. Interestingly, this approach did not necessarily improve results compared to the baseline case, which used a default time profile. However, this approach was key to simulating the initial 12-hr explosive fire activity and smoke impacts. The WRF-CMAQ simulations compared well with observational data for the October 8-15 time period and tended to overestimate concentrations October 16-20. To improve these results, we applied three machine learning algorithms. We also had a unique opportunity to evaluate results with temporary monitors deployed specifically for wildfires, and performance was markedly different. For example, at the permanent monitoring locations, the WRF-CMAQ simulations had a Pearson correlation of 0.65, and the data fusion approach improved this (Pearson correlation = 0.95), while at the temporary monitor locations across the WRF-CMAQ, data fusion, and machine learning datasets, the best Pearson correlation was 0.5. The data fusion and machine learning results were biased low and WRF-CMAQ results were biased high. Finally, we applied the optimized PM2.5 exposure estimate in a short-term exposure-response function. Total estimated mortality attributable to PM2.5 exposure during the smoke episode was 83 (95% confidence interval: 0, 196) with 47% of these deaths attributable to wildland fire smoke.

O'Neill, Susan↗

DOE OSTI2023

Adaptive sampling for accelerating neutron diffraction-based strain mapping *

Abstract Neutron diffraction is a useful technique for mapping residual strains in dense metal objects. The technique works by placing an object in the path of a neutron beam, measuring the diffracted signals and inferring the local lattice strain values from the measurement. In order to map the strains across the entire object, the object is stepped one position at a time in the path of the neutron beam, typically in raster order, and at each position a strain value is estimated. Typical dwell times at neutron diffraction instruments result in an overall measurement that can take several hours to map an object that is several tens of centimeters in each dimension at a resolution of a few millimeters, during which the end users do not have an estimate of the global strain features and are at risk of incomplete information in case of instruments outages. In this paper, we propose an object adaptive sampling strategy to measure the significant points first. We start with a small initial uniform set of measurement points across the object to be mapped, compute the strain in those positions and use a machine learning technique to predict the next position to measure in the object. Specifically, we use a Bayesian optimization based on a Gaussian process regression method to infer the underlying strain field from a sparse set of measurements and predict the next most informative positions to measure based on estimates of the mean and variance in the strain fields estimated from the previously measured points. We demonstrate our real-time measure-infer-predict workflow on additively manufactured steel parts—demonstrating that we can get an accurate strain estimate even with 30%–40% of the typical number of measurements—leading the path to faster strain mapping with useful real-time feedback. We emphasize that the proposed method is general and can be used for fast mapping of other material properties such as phase fractions from time-consuming point-wise neutron measurements.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2020

Inference offers a metric to constrain dynamical models of neutrino flavor transformation

The multimessenger astrophysics of compact objects presents a vast range of environments where neutrino flavor transformation may occur and may be important for nucleosynthesis, dynamics, and a detected neutrino signal. Development of efficient techniques for surveying flavor evolution solution spaces in these diverse environments, which augment and complement existing sophisticated computational tools, could leverage progress in this field. To this end we continue our exploration of statistical data assimilation (SDA) to identify solutions to a small-scale model of neutrino flavor transformation. SDA is a machine learning formula wherein a dynamical model is assumed to generate any measured quantities. Specifically, we use an optimization formulation of SDA wherein a cost function is extremized via the variational method. Regions of state space in which the extremization identifies the global minimum of the cost function will correspond to parameter regimes in which a model solution can exist. Our example study seeks to infer the flavor transformation histories of two monoenergetic neutrino beams coherently interacting with each other and with a matter background. We require that the solution be consistent with measured neutrino flavor fluxes at the point of detection, and with constraints placed upon the flavor content at various locations along their trajectories, such as the point of emission, and the locations of the Mikheyev-Smirnov-Wolfenstein resonances. We show how the procedure efficiently identifies solution regimes and rules out regimes where solutions are infeasible. Overall, results in this work intimate the promise of this “variational annealing” methodology to efficiently probe an array of fundamental questions that traditional numerical simulation codes render difficult to access.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2022

Embedding Learning in Hybrid Quantum-Classical Neural Networks

Quantum embedding learning is an important step in the application of quantum machine learning to classical data. In this paper we propose a quantum few-shot embedding learning paradigm, which learns embeddings useful for training downstream quantum machine learning tasks. Crucially, we identify the circuit bypass problem in hybrid neural networks, where learned classical parameters are optimized to represent the dataset without the quantum kernel. We observe that the few-shot learning embeddings generalize to unseen classes, and suffer less from the circuit bypass problem in terms of better occupation of the parameter space compared with embeddings learned from regression and classification.

Liu, Minzhao↗

DOE OSTI2020

Neural networks for modeling and control of particle accelerators

Charged particle accelerators support a wide variety of scientific, industrial, and medical applications. They range in scale and complexity from systems with just a few components for beam acceleration and manipulation, to large scientific user facilities that span many kilometers and have hundreds-to-thousands of individually-controllable components. Specific operational requirements must be met by adjusting the many controllable variables of the accelerator. Meeting these requirements can be challenging, both in terms of the ability to achieve specific beam quality metrics in a reliable fashion and in terms of the time needed to set up and maintain the optimal operating conditions. One avenue toward addressing this challenge is to incorporate techniques from the fields of machine learning (ML) and artificial intelligence (AI) into the way particle accelerators are modeled and controlled. While many promising approaches within AI/ML could be used for particle accelerators, this dissertation focuses on approaches based on neural networks. Neural networks are particularly well-suited to modeling, control, and diagnostic analysis of nonlinear systems, as well as systems with large parameter spaces. They are also very appealing for their ability to process high-dimensional data types, such as images and time series (both of which are ubiquitous in particle accelerators). In this work, key studies that demonstrated the potential utility of modern neural network-based approaches to modeling and control of particle accelerators are presented. The context for this work is important: at the start of this work in 2012, there was little interest in AI/ML in the particle accelerator community, and many of the advances in neural networks and deep learning that enabled its present success had not yet been made at that time. As such, this work was both an exploration of possible application areas and a generator of proof-of-concept demonstrations in these areas.

43 PARTICLE ACCELERATORS↗

DOE OSTI2025

Methods and Tools To Assess Robustness of Nuclear Plant Outages

Refueling outages are one of the most challenging phases in a nuclear power plant (NPP) operating cycle. Refueling outages are extremely costly for an NPP due to the large amount of required resources and because of lost revenue due to plant being off the grid. Outage durations have steadily decreased across the industry over that last few decades primarily due to improved planning and coordination, but there are still many plants that struggle to meet the performance metrics accomplished by other utilities. Schedule resilience is one of the issues. NPP outages require scheduling thousands of activities within 30 days on average. Despite detailed planning, once the outage starts, numerous emergent issues typically appear along with schedule delays requiring continuous replanning and adjustment. When schedule disruption occurs during an outage, plant staff make urgent efforts to recover but are often not able to maintain the planned outage duration. These outage delays can cost a utility several million dollars per day. Tools that could help outage schedulers create a more resilient schedule and allow them to optimally reschedule emergent work could significantly reduce outage delays. One key aspect of creating a resilient schedule is to have accurate estimates for activity duration. Another important outage scheduling capability is the ability to schedule emergent work with minimal disruption. This paper focuses on developing tools and methods to support NPPs with outage schedule optimization and it describes the initial development of tools to support outage management that leverage computational and machine learning methods.

97 - MATHEMATICS AND COMPUTING↗