Search NASA⌕ Search

SEARCH · Search NASA

Results for “machine architecture”

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 217 records · Page 12
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

A Machine Learning Approach to the Detection of Ghosting Artifacts in Dark Energy Survey Images

Unwanted artifacts that often plague astronomical images arise from a number of sources that include imperfect optics, faulty image sensors, cosmic ray hits, and even airplanes and satellites. One major source that is dicult to avoid is known as ghosting, and is caused by the scattering and internal reflections of light o of the telescope’s mechanical and optical components. Detecting ghosting artifacts eciently in the large cosmological surveys that will acquire petabytes of data can be a daunting task. In this paper, we use data from the Dark Energy Survey to develop, train, and validate a machine learning model to detect ghosts based on convolutional neural networks. The model architecture and training procedure is discussed in detail, and the performance on the training and validation set presented. Testing is performed on data and results are compared with those from a ray-tracing algorithm. A proof-of-principle is demonstrated that shows promise for Stage IV surveys and beyond.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2021

Deep Cellular Recurrent Network for Efficient Analysis of Time-Series Data With Spatial Information

Efficient processing of large-scale time series data is an intricate problem in machine learning. Conventional sensor signal processing pipelines with hand engineered feature extraction often involve huge computational cost with high dimensional data. Deep recurrent neural networks have shown promise in automated feature learning for improved time-series processing. However, generic deep recurrent models grow in scale and depth with increased complexity of the data. This is particularly challenging in presence of high dimensional data with temporal and spatial characteristics. Consequently, this work proposes a novel deep cellular recurrent neural network (DCRNN) architecture to efficiently process complex multi-dimensional time series data with spatial information. Here, the cellular recurrent architecture in the proposed model allows for location-aware synchronous processing of time series data from spatially distributed sensor signal sources. Extensive trainable parameter sharing due to cellularity in the proposed architecture ensures efficiency in the use of recurrent processing units with high-dimensional inputs. This study also investigates the versatility of the proposed DCRNN model for classification of multi-class time series data from different application domains. Consequently, the proposed DCRNN architecture is evaluated using two time-series datasets: a multichannel scalp EEG dataset for seizure detection, and a machine fault detection dataset obtained in-house. The results suggest that the proposed architecture achieves state-of-the-art performance while utilizing substantially less trainable parameters when compared to comparable methods in the literature.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2026

Future foundries: A convergent manufacturing platform

This article introduces the Future Foundries platform developed at Oak Ridge National Laboratory, a first-generation research system designed to demonstrate convergent manufacturing. Convergent manufacturing brings together additive, subtractive, and transformative processes in a digitally interconnected environment to enable end-to-end production workflows. By linking traditionally discrete steps, convergent platforms accelerate production, improve repeatability, and support high-mix, low-volume manufacturing. The Future Foundries platform exemplifies this vision in practice by combining four modular, vendor-agnostic process cells that include robotic WAAM, induction heating, optical metrology, and machining, coordinated through an automated pallet handler and a ROS 2-based digital thread. This architecture provides the flexibility and scalability needed for agile production in small and medium-sized manufacturing enterprises and for field deployable manufacturing. Two case studies illustrate the platform’s capabilities. The first presents an integrated workflow for fabricating, transforming, and repairing critical replacement components, showing how consolidated thermal, additive, inspection, and machining operations reduce manual part handling and streamline process flow. The second case study highlights coordinated multi-part production enabled by automated pallet logistics and multi-cell scheduling. Together, these examples showcase convergent manufacturing as a practical and scalable strategy for strengthening domestic casting and forging capacity, improving supply-chain resilience, and enabling rapid, adaptable production of mission-critical components.

Convergent manufacturing↗
DOE OSTI2024

Comparison of Real-Time Pressure Rail Selection Algorithms for the Hybrid Hydraulic Electric Architecture: Case Study on a Track Loader

Abstract The hybrid hydraulic electric architecture (HHEA) seeks to combine the high power/torque/force density of hydraulics with the efficiency of electric machines. A set of common pressure rails is used to provide a majority of the power and this power is modulated by small electric machines to provide precise control for the operator. The HHEA has been studied in previous work using off-line dynamic programming optimization to determine energy efficient pressure rail selections, but this approach requires drive cycle information apriori. A Lagrange multiplier method has also been investigated where a set of gains (Lagrange multipliers) are optimized off-line with the idea the these gains, once determined, could be used for real-time operation. In this work, three new real-time pressure rail selection algorithms that do not require future drive cycle information are investigated; greedy, torque minimizing, and thresholding. The greedy control is found to only use 1% more energy than the globally optimal dynamic programming solution; but a model of energy loss is required.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2025

DeepHyper: A Python Package for Massively Parallel Hyperparameter Optimization in Machine Learning

Machine learning models are increasingly applied across scientific disciplines, yet their effectiveness often hinges on heuristic decisions—such as data transformations, training strategies, and model architectures—that are not learned by the models themselves. Automating the selection of these heuristics and analyzing their sensitivity is crucial for building robust and efficient learning workflows. DeepHyper addresses this challenge by democratizing hyperparameter optimization, providing accessible tools to streamline and enhance machine learning workflows from a laptop to the largest supercomputer in the world. Building on top of hyperparameter optimization, it unlocks new capabilities around ensembles of models for improved accuracy and uncertainty quantification. All of these organized around efficient parallel computing.

ensemble↗
DOE OSTI2019

Porting a 3D seismic modeling code (SW4) to CORAL machines

Seismic waves fourth order (SW4) solves the seismic wave equations on Cartesian and curvilinear grids using large compute clusters with O (100,000) cores. In this article we discuss the porting of SW4 to run on the CORAL architecture using the RAJA performance portability abstraction layer. The performances of key kernels using RAJA and CUDA are compared to estimate the performance penalty of using the portability abstraction layer. Code changes required for efficiency on GPUs and minimizing time spent in Message Passing Interface (MPI) are discussed. This article describes a path for efficiently porting large code bases to GPU-based machines while avoiding the pitfalls of a new architecture in the early stages of its deployment. Current bottlenecks in the code are discussed along with possible architectural or software mitigations. SW4 runs 28× faster on one 4-GPU CORAL node than on a CTS-1 node (Dual Intel Xeon E5-2695 v4). SW4 is now in routine use on problems of unprecedented resolution (203 billion grid points) and scale on 1,200 nodes of Summit.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

CommAwareNet: Towards Communication‐Aware Smart Facilities: Designing an Energy‐Efficient High‐Data‐Rate and Reliable Hybrid THz/VLC Comm. Arch. Reinforced with Intelligent Surfaces for Future Network

This project will develop foundations for energy-efficient, high-data-rate, and reliable communication links for next-generation advanced wireless technologies. The Internet of Everything is emerging as a major technology that can establish a massively connected network among both human-type users and machine-type devices. Particularly, massive machine-type communication is becoming the dominant communication paradigm for numerous emerging areas, including data center networking and wireless backhaul, healthcare (e-health), manufacturing (industry 4.0), utilities, transportation (connected cars and public safety), and virtual or augmented reality for human-machine interaction applications. The rising demand for high data rates, especially in indoor scenarios and densely populated outdoor environments, will eventually overload conventional RF-based technologies. This project aims to conduct fundamental research to design, optimize, and validate a communication-aware, energy-efficient, and reliable network (CommAwareNet) architecture for future smart facilities and environments that require unprecedented high throughput needs for massive numbers of the machine- and human-type users. The scientific and technical merit of this work lies in the research required to bring the proposed transformative CommAwareNet solution into fully functional and practical architecture.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2026

Generalist multimodal AI: A review of architectures, challenges and opportunities

Multimodal models are expected to be a critical component to future advances in artificial intelligence. Here, this field is starting to grow rapidly with a surge of new design elements motivated by the success of foundation models in natural language processing (NLP) and vision. It is widely hoped that further extending the foundation models to multiple modalities (e.g., text, image, video, sensor, time series, graph, etc.) will ultimately lead to generalist multimodal models, i.e. one model across different data modalities and tasks. However, there is little research that systematically analyzes recent multimodal models (particularly the ones that work beyond text and vision) with respect to the underling architecture proposed. Therefore, this work provides a fresh perspective on generalist multimodal models (GMMs) via a novel architecture and training configuration specific taxonomy. This includes factors such as Unifiability, Modularity, and Adaptability that are pertinent and essential to the wide adoption and application of GMMs. The review further highlights key challenges and prospects for the field and guide the researchers into the new advancements.

Artificial intelligence (AI)↗
DOE OSTI2021

ARENA: Asynchronous Reconfigurable Accelerator Ring to Enable Data-Centric Parallel Computing

The next generation HPC and data centers are likely to be reconfigurable and data-centric due to the trend of hardware specialization and the emergence of data-driven applications. In this work, we propose ARENA – an asynchronous reconfigurable accelerator ring architecture as a potential scenario on how the future HPC and data centers will be like. Despite using the coarse-grained reconfigurable arrays (CGRAs) as the substrate platform, our key contribution is not only the CGRA-cluster design itself, but also the ensemble of a new architecture and programming model that enables asynchronous tasking across a cluster of reconfigurable nodes, so as to bring specialized computation to the data rather than the reverse. We presume distributed data storage without asserting any prior knowledge on the data distribution. Hardware specialization occurs at runtime when a task finds the majority of data it requires are available at the present node. In other words, we dynamically generate specialized CGRA accelerators where the data reside. The asynchronous tasking for bringing computation to data is achieved by circulating the task token, which describes the dataflow graphs to be executed for a task, among the CGRA cluster connected by a fast ring network. Evaluations on a set of HPC and data-driven applications across different domains show that ARENA can provide better parallel scalability with reduced data movement (53.9 percent). Compared with contemporary compute-centric parallel models, ARENA can bring on average 4.37× speedup. The synthesized CGRAs and their task-dispatchers only occupy 2.93mm 2 chip area under 45nm process technology and can run at 800MHz with on average 759.8mW power consumption. ARENA also supports the concurrent execution of multi-applications, offering ideal architectural support for future high-performance parallel computing and data analytics systems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

OPTIMAL CONTROL AND ENERGY-SAVING ANALYSIS OF COMMON PRESSURE RAIL ARCHITECTURES: HHEA & STEAM

In recent years several novel hydraulic architectures have been proposed with the intention of significantly increasing system efficiency. Two of these architectures, Steigerung der Energieeffizienz in der Arbeitshydraulik mobiler Arbeitsmaschinen (STEAM), and the Hybrid Hydraulic-Electric Architecture (HHEA), use a system of multiple common pressure rails (CPRs) to serve the multiple degrees-of-freedom of the machine. The key difference is that STEAM throttles hydraulic power from these rails while HHEA combines electric and hydraulic power to meet actuator demands. As a throttle-less architecture, HHEA is expected to save more energy than STEAM at the expense of added complexity. Therefore, it is useful to quantify this additional energy saving. Both systems have discrete operational choices corresponding to how the CPRs are utilized for each actuator. It is necessary to determine optimal operation for each of these architectures for analysis and fair comparison. Techniques for optimal operation of the HHEA have been developed previously from the Langrange multiplier method. Applying the same optimal control method to STEAM encountered some technical challenge leading to the optimal control algorithm not being able to satisfy certain constraints. The issue is analyzed and solved by adding noise to the optimization. Using this proposed algorithm, case studies are performed to compare the energy-saving potentials of STEAM and HHEA for two sizes of excavators and a wheel-loader performing representative duty cycles. The baseline is a standard load-sensing architecture. Results show that STEAM and HHEA can reduce energy consumption between 35-65% and 50-80% respectively

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2025

A methodology for decay heat characterization in molten salt reactors

Accurate decay heat prediction in molten salt reactors (MSRs) faces dual challenges: complex operational uncertainties and the need for interpretable models compatible with engineering workflows. This work presents a hybrid machine learning and segmented polynomial methodology that addresses both requirements through three key innovations. First, a modular data architecture encodes MSR-specific operational parameters (power density: 1-100 W cm -3 , humidity: 0-0.1 wt %, air ingress: 0-0.1 mol %) with uncertainty-aware temporal discretization spanning 15 orders of magnitude. Second, region-optimized machine learning models achieve 92.3 % root mean square error (RMSE) reduction over conventional polynomials while maintaining physical interpretability through automated piecewise equation generation. Third, dual front-end interfaces accelerate safety analyses — a Jupyter environment enables researchers to explore 10,000+ parameter combinations via interactive widgets, while a Streamlit web application reduces design iteration cycles through production-grade visualization tools. Operational deployment demonstrates prediction times of only a couple hundred milliseconds for 10 4 years decay profiles, enabling real-time optimization of spent fuel container designs.

42 - ENGINEERING↗
DOE OSTI2021

Automated Generation of Integrated Digital and Spiking Neuromorphic Machine Learning Accelerators

The growing numbers of application areas for artificial intelligence (AI) methods have led to an explosion of domain-specific accelerators that could support every new machine learning (ML) algorithm advancement, clearly highlighting the need for a capability to quickly and automatically transition from algorithm definition to hardware implementation and explore design space along a variety of SWaP (size, weight and Power). The software defined architectures (SODA) synthesizer implements a compiler-based modular infrastructure for the end-to-end generation of machine learning accelerators from high-level frameworks to hardware description language. At the same time, neuromorphic computing, by mimicking how the brain operates, promises to perform artificial intelligence tasks at efficiencies orders of magnitude higher than the current conventional tensor-processing based accelerators, as demonstrated by a variety of specialized designs leveraging Spiking Neural Networks (SNNs). Nevertheless, the mapping of an artificial neural network (ANN) to solutions supporting SNNs is still a non-trivial and very device-specific task, and completely lack the possibility to design hybrid systems that integrate conventional and spiking neural models. In this paper we discuss the support for such an integrated generation leveraging the SODA Synthesizer framework and its modular structure. In particular, we present a new MLIR dialect (part of the SODA frontend) that allows expressing spiking neural network features (e.g., available resources, spiking sequences, analog signal reading, etc.) and illustrate how it enables mapping to Spiking Neurons and deployment to the related specialized hardware (which, in the digital domain, could be generated through the other existing layers of the SODA Synthesizer). We then discuss the opportunities for even deeper integration afforded by the hardware compilation infrastructure, providing a path towards the generation of complex heterogeneous artificial intelligence systems.

Curzel, Serena↗
DOE OSTI2025

ReSpike: A Co-Design Framework for Evaluating SNNs on ReRAM-Based Neuromorphic Processors

With Moore’s law approaching its end, traditional von Neumann architectures are struggling to keep up with the exceeding performance and memory requirements of artificial intelligence and machine learning algorithms. Unconventional computing approaches such as neuromorphic computing that leverage spiking neural networks (SNNs) to perform computation are gaining traction and seek the paradigm shift necessary to sustain the increasing demands of modern applications. Novel memory technologies, such as resistive RAM (ReRAM), employ a crossbar architecture that possesses the inherent capability of efficiently computing vector-matrix multiplication—a dominant operation in SNNs. The prospect of naturally mapping SNNs to the crossbar structures provides a unique opportunity for achieving a high-performance, power-efficient neuromorphic system. In this work, we present ReSpike, which is a new framework, behavioral simulator, and architectural design based on ReRAM crossbar architectures, enabling modeling and co-design to achieve efficient execution of SNNs. We drive this co-design forward by quantifying the impact that ReRAM cell nonidealities have on the corresponding accuracy of an SNN application.

Asifuzzaman, Kazi [ORNL] (ORCID:0000000240044791)↗
DOE OSTI2022

H-GCN: A Graph Convolutional Network Accelerator on Versal ACAP Architecture

Recently Graph Neural Networks (GNNs) have drawn tremendous attentions due to their unique capability to extend the Machine Learning (ML) approaches to broadly defined applications with unstructured data, especially graphs. Comparing with other ML modalities, the acceleration of GNNs is as critical but even more challenging due to the irregularity and heterogeneity from graph typologies that together limit the performance. Existing efforts mainly focus on handling graphs’ irregularity, however, have not studied the heterogeneity. To this end, in this work, we propose H-GCN, a PL-AIE-based hybrid accelerator that leverages the emerging heterogeneity of Xilinx Versal ACAPs to achieve high-performance GNN inference. In particular, H-GCN partitions each graph into three subgraphs based on its inherent heterogeneity and processes them using PL and the newly emerged AIE respectively. To further improve the performance, we explore the sparsity support of AIE and develop an efficient density-aware method to map tiles of SpMM onto the systolic tensor array automatically. Compared with the current state-of-the-art GCN accelerator, HGCN achieves on average 1.5× speedups.

Zhang, Chengming↗
DOE OSTI2026

Machine Learning Digital Twin for Lithium Ion Battery State of Health Predictions

A digital twin system has been established to model the long term degradation of the state of health of lithium ion batteries. Two data streams result from the computational model of the system and from the physical experiment measurements. A machine learning pipeline has been developed at the nexus of these data streams. Leveraging the unique data sources in multiple transfer learning approaches has lead to the development of multiple cell specific machine learning digital twins. Our discussion on this first of its kind technology will cover challenges in data ingestion, scalability, architecture and orchestration, and research findings.

25 - ENERGY STORAGE↗
DOE OSTI2022

Time Sequence Machine Learning-Based Data Intrusion Detection for Smart Voltage Source Converter-Enabled Power Grid

Smart inverters of distributed energy resources can enable cloud computing, condition monitoring, result visualization, remote control, and peer-to-peer energy trading in advanced power systems. However, the advent of data injection attacks in the communication architecture can alter measurement characteristics of power grids and have devastating consequences. In this article, we propose a time sequence machine learning-based anomaly detection methodology for detecting cyber intrusion into control signal setpoints and dc voltage signal measurement bias of the voltage source converter (VSC) in wind generators. We first investigated the effects of four types of denial of service, tampering signal, and stealthy-type data intrusion attacks on smart VSCs and overall wind farms. We then proposed a novel time sequence machine learning-based intrusion detection framework that can be implemented to detect different cyberattacks in the VSCs. The performance of the proposed framework has been compared with that of autoencoder and clustering-based intrusion detection framework. The proposed framework was validated by using the IEEE 39 bus power system in the presence of four wind farms in different locations. Using several metrics for intrusion detection performance, we validated the effectiveness of the proposed framework.

42 ENGINEERING↗
DOE OSTI2025

Multi-head physics-informed neural networks for learning functional priors and uncertainty quantification

In numerous applications, the integration of prior knowledge and historical information is essential, particularly for tasks requiring the solution of ordinary or partial differential equations (ODEs/PDEs) in data-sparse or noisy environments. For instance, achieving accurate solutions to time-dependent PDEs with limited initial condition measurements necessitates an effective strategy for embedding prior knowledge. Hard-parameter sharing architectures in neural networks (NNs) have demonstrated success in both traditional and scientific machine learning domains, facilitating the learning of informative representations. Here, in this study, we introduce a novel, yet efficient, method to enhance physics-informed neural networks (PINNs) by incorporating a multi-head structure that enables the learning of functional priors from both empirical data and governing physical laws. This prior information can then be used to address data sparsity and high-level noise in solving ODE/PDE problems with uncertainty quantification (UQ). The approach, termed Multi-Head PINN (MH-PINN), consists of a shared body NN and multiple head NNs, each corresponding to an individual PINN instance. Our framework for functional prior learning is carried out in two stages: (1) training the MH-PINNs to develop a shared body NN alongside multiple head NNs, and (2) employing these trained head NNs to estimate a prior distribution through a normalizing flow-based density estimator. The learned functional prior can then be applied as a regularization mechanism in deterministic contexts or as an informative prior within a Bayesian inference framework, aiding in the resolution of subsequent ODE/PDE tasks. We evaluate the efficacy of MH-PINNs across five benchmark problems, including a high-dimensional parametric PDE, all characterized by data sparsity or substantial noise levels. Our findings reveal that MH-PINNs deliver accurate solutions and robust UQ, demonstrating adaptability across a range of complex and challenging scenarios.

Bayesian inference↗
DOE OSTI2020

Deep Generative Models that Solve PDEs: Distributed Computing for Training Large Data-Free Models

Recent progress in scientific machine learning (SciML) has opened up the possibility of training novel neural network architectures that solve complex partial differential equations (PDEs). Several (nearly data free) approaches have been recently reported that successfully solve PDEs, with examples including deep feed forward networks, generative networks, and deep encoder-decoder networks. However, practical adoption of these approaches is limited by the difficulty in training these models, especially to make predictions at large output resolutions (≥1024×1024). Here we report on a software framework for data parallel distributed deep learning that resolves the twin challenges of training these large SciML models - training in reasonable time as well as distributing the storage requirements. Our framework provides several out of the box functionality including (a) loss integrity independent of number of processes, (b) synchronized batch normalization, and (c) distributed higher-order optimization methods. We show excellent scalability of this framework on both cloud as well as HPC clusters, and report on the interplay between bandwidth, network topology and bare metal vs cloud. We deploy this approach to train generative models of sizes hitherto not possible, showing that neural PDE solvers can be viably trained for practical applications. We also demonstrate that distributed higher-order optimization methods are 2-3× faster than stochastic gradient-based methods and provide minimal convergence drift with higher batch-size.

PDEs↗