SEARCH · Search NASA
Results for “High Performance Networking”
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.
Soft Actor Critic Based Volt-VAR Co-optimization in Active Distribution Grids
Modern distribution networks are undergoing several technical challenges, such as voltage fluctuations, because of high penetration of distributed energy resources (DERs). This paper proposes a deep reinforcement learning (DRL)-based Volt VAR co-optimization technique for reducing voltage fluctuations as well as power loss under high penetration of DERs. In addition, the proposed approach minimizes the operational cost of the grid. A stochastic policy optimization based soft actor critic (SAC) agent is proposed to configure the optimal set-points of the reactive power outputs of the inverters. The performance of the proposed model is verified on the modified IEEE 34- and 123-bus systems and compared with a base case scenario with no reactive supply by inverters, and a local droop control approach. The results demonstrate that the proposed framework outperforms the conventional droop control method in improving the voltage profile, minimizing the network power loss, and reducing grid operational cost.
Implementing Spatio-Temporal 3D-Convolution Neural Networks and UAV Time Series Imagery to Better Predict Lodging Damage in Sorghum
Unmanned aerial vehicle (UAV)-based remote sensing is gaining momentum in a variety of agricultural and environmental applications. Very-high-resolution remote sensing image sets collected repeatedly throughout a crop growing season are becoming increasingly common. Analytical methods able to learn from both spatial and time dimensions of the data may allow for an improved estimation of crop traits, as well as the effects of genetics and the environment on these traits. Multispectral and geometric time series imagery was collected by UAV on 11 dates, along with ground-truth data, in a field trial of 866 genetically diverse biomass sorghum accessions. We compared the performance of Convolution Neural Network (CNN) architectures that used image data from single dates (two spatial dimensions, 2D) versus multiple dates (two spatial dimensions + temporal dimension, 3D) to estimate lodging detection and severity. Lodging was detected with 3D-CNN analysis of time series imagery with 0.88 accuracy, 0.92 Precision, and 0.83 Recall. This outperformed the best 2D-CNN on a single date with 0.85 accuracy, 0.84 Precision, and 0.76 Recall. The variation in lodging severity was estimated by the best 3D-CNN analysis with 9.4% mean absolute error (MAE), 11.9% root mean square error (RMSE), and goodness-of-fit (R2) of 0.76. This was a significant improvement over the best 2D-CNN analysis with 11.84% MAE, 14.91% RMSE, and 0.63 R2. The success of the improved 3D-CNN analysis approach depended on the inclusion of “before and after” data, i.e., images collected on dates before and after the lodging event. The integration of geometric and spectral features with 3D-CNN architecture was also key to the improved assessment of lodging severity, which is an important and difficult-to-assess phenomenon in bioenergy feedstocks such as biomass sorghum. This demonstrates that spatio-temporal CNN architectures based on UAV time series imagery have significant potential to enhance plant phenotyping capabilities in crop breeding and Precision agriculture applications.
Preliminary Design of Nuclear Reactor Heat Delivery Systems: Integration with Reference Oil Refinery, Methanol Synthesis, and Hydrogen Production
The Department of Energy’s (DOE) Integrated Energy Systems (IES) program is generating comprehensive analyses validating the opportunity for using nuclear energy in a variety of applications including future clean grids, providing heat for direct use, and providing heat to help reduce emissions in chemical commodities. This work focuses on the preliminary designs of thermal delivery systems that can integrate heat produced from a nuclear core to industrial processes. The key research question that needs to be answered is: what is the prospective method for integrating nuclear generated heat energy into non-electric applications that can facilitate combined heat and power operations by advanced nuclear reactor systems? This research is a composition of case studies showing preliminary conceptual designs for thermal delivery systems integrating advanced nuclear systems with a few industrial systems including high temperature steam electrolysis, a reference oil refinery, and potential future methanol systems that supplant some natural gas use with nuclear energy. Piping and instrumentation diagrams have been developed to show the conceptual integration of nuclear systems with representative industrial systems. Different features of the configurations are dependent on the specific integration requirements including energy source conditions, demand quantity, and require energy application conditions. Design concepts are validated using thermodynamic balance calculations to verify system performance including calculating system losses during transport. Key components: pumps and compressors, heat exchangers, and network piping are reported with key design information and sizing.
SMALE: Enhancing Scalability of Machine Learning Algorithms on Extreme-Scale Computing Platforms
Deployment and execution of machine learning tasks on extreme-scale computing platforms face several significant technical challenges: 1) High computing cost incurred by dense networks – The computing workload of deep networks with densely-connected topology increases rapidly with the network size, imposing a non-scalable computing model of extreme-scale computing platforms; 2) Non-optimized workload distribution – Many advanced deep learning algorithms, e.g., sparsification and irregular net-work topology, produce very unbalanced workload distribution on extreme-scale computing platforms. The computation efficiency is greatly hindered by the incurred data and computation redundancies as well as long tails of the node with extensive workload; 3) Constraints in data movement and I/O bottle-neck – Inter-node data movement in extreme-scale computing platforms are associated with high energy and latency costs, and subject to the constraints of I/O bandwidth; and 4) Generalization of algorithm realization and acceleration on computing platforms – The large varieties of machine learning algorithms and structures of extreme-scale computing platforms make the derivation of a generalized algorithm realization and acceleration method very challenging, which, however, is the requirement by domain scientists and interested users. We call the above challenges Smale’s Problems in Machine Learning and Understanding for High-Performance Computing Scientific Discovery. The objective of our three-year research project is to develop a holistic innovation set at structure, assembly, and acceleration layers of machine learning algorithms to address the above challenges in algorithm deployment and execution. Three tasks are particularly performed, including: At the algorithm structure level, we investigate the techniques that can structurally sparsify on the topology of deep networks for computing workload reduction. We also study clustering and pruning techniques that can optimize the workload distributions over the extreme-scale computing platforms; At the algorithm assembly level, we derive a unified learning framework for unsupervised transfer learning and dynamic growing capabilities. Novel training methods are also exploited to enhance the training efficiency of the proposed framework; At the algorithm acceleration level, we will develop a series of techniques that can accelerate the computation of sparse matrix operations, which are one of the core executions in deep learning and optimize memory access of the concerned platforms. Our proposed techniques attack the fundamental problems in machine learning algorithms running on extreme-scale computing platforms by vertically integrating the solutions at three closely entangled layers, paving the long-term scaling path of machine learning applications under DOE context. Three tasks corresponding to the above respective research orientations are performed during the three-year project period with our collaborators at ORNL. The outcome of the proposed project is anticipated to form a holistic solution set of novel algorithms and network topologies, efficient training techniques, and fast acceleration methods to promote the computing scalability of the machine learning applications of particular interest to DOE.
Supercritical, liquid, and gas CO 2 reactive transport and carbonate formation in portland cement mortar
In this paper, we investigate carbonate formation and reactive transport rate in variably saturated portland cement mortars when high concentrations of gas, liquid, or supercritical CO 2 flow through their pore network. Xray computed tomography completed during CO 2 flow is used to quantify the microstructural evolution as the mortar carbonates. After in situ tests, higher resolution scans, thermogravimetric analysis, and desorption isotherm analysis are performed to further quantify microstructural changes. We found that at dry conditions supercritical CO 2 moves more rapidly through the pore space and precipitates more carbonates than liquid or gas CO 2 . However, at 50% degree of saturation (DOS) the CO 2 state did not affect the rate of transport in that each specimen exposed to a different CO 2 state carbonated within the first hour of CO 2 exposure. When the pore space is at 50 or 100% DOS, supercritical CO 2 did not react with hydration products more rapidly nor did it result in more carbonate formation during exposure compared to gas or liquid CO 2 . The amount of Ca(OH) 2 that contributes to CaCO 3 formation is correlated to the DOS. For the mortar composition analyzed, Ca(OH) 2 contributes to approximately 40% of the carbonates formed in the 50% DOS specimens and 15% in the 100% DOS specimens. In other words, as the amount of moisture in the pore space increases, phases other than Ca(OH) 2 contribute to more than 50% of the total CaCO 3 formed.
Active learning of ternary alloy structures and energies
Abstract Machine learning models with uncertainty quantification have recently emerged as attractive tools to accelerate the navigation of catalyst design spaces in a data-efficient manner. Here, we combine active learning with a dropout graph convolutional network (dGCN) as a surrogate model to explore the complex materials space of high-entropy alloys (HEAs). We train the dGCN on the formation energies of disordered binary alloy structures in the Pd-Pt-Sn ternary alloy system and improve predictions on ternary structures by performing reduced optimization of the formation free energy, the target property that determines HEA stability, over ensembles of ternary structures constructed based on two coordinate systems: (a) a physics-informed ternary composition space, and (b) data-driven coordinates discovered by the Diffusion Maps manifold learning scheme. Both reduced optimization techniques improve predictions of the formation free energy in the ternary alloy space with a significantly reduced number of DFT calculations compared to a high-fidelity model. The physics-based scheme converges to the target property in a manner akin to a depth-first strategy, whereas the data-driven scheme appears more akin to a breadth-first approach. Both sampling schemes, coupled with our acquisition function, successfully exploit a database of DFT-calculated binary alloy structures and energies, augmented with a relatively small number of ternary alloy calculations, to identify stable ternary HEA compositions and structures. This generalized framework can be extended to incorporate more complex bulk and surface structural motifs, and the results demonstrate that significant dimensionality reduction is possible in thermodynamic sampling problems when suitable active learning schemes are employed.
Dynamical Sketching for Enhanced Communication Efficiency in Federated Learning
Federated learning (FL) has revolutionized distributed machine learning by enabling collaborative model training without sharing local data. However, communication efficiency and privacy guarantees remain significant challenges. This paper introduces a dynamic sketching mechanism in FL, optimizing the trade-off between communication efficiency and model accuracy. By dynamically selecting the sketch matrix size, our approach adapts to the evolving characteristics of the data and the model, ensuring optimal performance across diverse scenarios. We leverage Bayesian optimization to systematically tune the sketch parameters, achieving an effective balance between resource efficiency and model performance. Experimental results on the MNIST dataset using a convolutional neural network (CNN) architecture validate the proposed method's efficiency and scalability. Our dynamic sketching approach significantly outperforms fixed-size sketching techniques, achieving higher compression ratios (up to 62x) and providing better privacy guarantees while maintaining high model accuracy. These findings highlight the robustness and versatility of our approach and make it a valuable solution for privacy-preserving, communication-efficient federated learning.
Orchestrating Distributed Resource Ensembles for Petascale Science
Distributed, data-intensive computational science applications of interest to DOE scientific com- munities move large amounts of data for experiment data management, distributed analysis steps, remote visualization, and accessing scientific instruments. These applications need to orchestrate ensembles of resources from multiple resource pools and interconnect them with high-capacity multi- layered networks across multiple domains. It is highly desirable that mechanisms are designed that provide this type of resource provisioning capability to a broad class of applications. It is also important to have coherent monitoring capabilities for such complex distributed environments. In this project, we addressed these problems by designing an abstract API, enabled by novel semantic resource descriptions, for provisioning complex and heterogeneous resources from multiple providers using their native provisioning mechanisms and control planes: computational, storage, and multi-layered high-speed network domains. We used an extensible resource representation based on semantic web technologies to afford maximum flexibility to applications in specifying their needs. We evaluated the effectiveness of provisioning using representative data-intensive ap- plications. We also developed mechanisms for providing feedback about resource performance to the application, to enable closed-loop feedback control and dynamic adjustments to resource allo- cations (elasticity). This was enabled through development of a novel persistent query framework that consumes disparate sources of monitoring data, including perfSONAR, and provides scalable distribution of asynchronous notifications.
Aftershock Identification Using a Paired Neural Network Applied to Constructed Data
This report is intended to detail the findings of our investigation of the applicability of machine learning to the task of aftershock identification. The ability to automatically identify nuisance aftershock events to reduce analyst workload when searching for events of interest is an important step in improving nuclear monitoring capabilities and while waveform cross - correlation methods have proven successful, they have limitations (e.g., difficulties with spike artifacts, multiple aftershocks in the same window) that machine learning may be able to overcome. Here we apply a Paired Neural Network (PNN) to a dataset consisting of real, high quality signals added to real seismic noises in order to work with controlled, labeled data and establish a baseline of the PNN's capability to identify aftershocks. We compare to waveform cross - correlation and find that the PNN performs well, outperforming waveform cross - correlation when classifying similar waveform pairs, i.e., aftershocks.
An Ensemble Score Filter for Tracking High-Dimensional Nonlinear Dynamical Systems
We propose an ensemble score filter (EnSF) for solving high-dimensional nonlinear filtering problems with superior accuracy. A major drawback of existing filtering methods, e.g., particle filters or ensemble Kalman filters, is the low accuracy in handling high-dimensional and highly nonlinear problems. EnSF attacks this challenge by exploiting the score-based diffusion model, defined in a pseudo-temporal domain, to characterizing the evolution of the filtering density. EnSF stores the information of the recursively updated filtering density function in the score function, in stead of storing the information in a set of finite Monte Carlo samples (used in particle filters and ensemble Kalman filters). Unlike existing diffusion models that train neural networks to approximate the score function, we develop a training-free score estimation that uses mini-batch-based Monte Carlo estimator to directly approximate the score function at any pseudo-spatial-temporal location, which provides sufficient accuracy in solving high-dimensional nonlinear problems as well as saves tremendous amount of time spent on training neural networks. Another essential aspect of EnSF is its analytical update step, gradually incorporating data information into the score function, which is crucial in mitigating the degeneracy issue faced when dealing with very high-dimensional nonlinear filtering problems. High-dimensional Lorenz systems are used to demonstrate the performance of our method. EnSF provides surprisingly impressive performance in reliably tracking extremely high-dimensional Lorenz systems (up to 1,000,000 dimension) with highly nonlinear observation processes, which is a well-known challenging problem for existing filtering methods.
Gaussian Process Regression for Aggregate Baseline Load Forecasting
Demand response (DR) is one of the most effective ways to maintain the reliability and improve the flexibility of power systems. Accurate forecasts of baseline loads are essential for DR programs. In the era of big data, machine learning-based approaches present a unique opportunity for baseline load forecasting. Thus, this paper presents a machine learning-based approach using a relatively less explored algorithm, Gaussian process regression (GPR), to forecast aggregate baseline loads. As such, a dataset was generated using a set of EnergyPlus simulations. Using the generated dataset, a GPR-based forecasting model was developed. In addition, support vector regression (SVR)-, artificial neural network (ANN)-, and averaging-based models were developed as baseline models for comparison. These models were compared in terms of accuracy, simplicity, and integrity. The prediction performance of the models showed that the GPR-based model is more accurate and reliable than the others. Such high performance shows the potential of the GPR in baseline load forecasting. GPR, therefore, can be used for DR applications.
Privacy-Preserving Real-Time Action Detection in Intelligent Vehicles Using Federated Learning-Based Temporal Recurrent Network
This study introduces a privacy-preserving approach for the real-time action detection in intelligent vehicles using a federated learning (FL)-based temporal recurrent network (TRN). This approach enables edge devices to independently train models, enhancing data privacy and scalability by eliminating central data consolidation. Our FL-based TRN effectively captures temporal dependencies, anticipating future actions with high precision. Extensive testing on the Honda HDD and TVSeries datasets demonstrated robust performance in centralized and decentralized settings, with competitive mean average precision (mAP) scores. The experimental results highlighted that our FL-based TRN achieved an mAP of 40.0% in decentralized settings, closely matching the 40.1% in centralized configurations. Notably, the model excelled in detecting complex driving maneuvers, with mAPs of 80.7% for intersection passing and 78.1% for right turns. These outcomes affirm the model’s accuracy in action localization and identification. The system showed significant scalability and adaptability, maintaining robust performance across increased client device counts. The integration of a temporal decoder enabled predictions of future actions up to 2 s ahead, enhancing the responsiveness. Our research advances intelligent vehicle technology, promoting safety and efficiency while maintaining strict privacy standards.
Corroborative Evaluation of the Real-World Energy Saving Potentials of InfoRich Eco-Autonomous Driving (iREAD) System
There has been an increasing interest in exploring the potential to reduce energy consumption of future connected and automated vehicles. People have extensively studied various eco-driving implementations that leverage preview information provided by on-board sensors and connectivity, as well as the control authority enabled by automation. Quantitative real-world evaluation of eco-driving benefits is a challenging task. The standard regulatory driving cycles used for measuring exhaust emissions and fuel economy are not truly representative of real-world driving, nor for capturing how connectivity and automation might influence driving trajectories. To adequately consider real-world driving behavior and potential “off-cycle” impacts, this paper presents four collaborative evaluation methods: large-scale simulation, in-depth simulation, vehicle-in-the-loop testing, and vehicle road testing. These four approaches, spanning simulation and testing aspects, evaluate real-world fuel economy benefits with different ranges and resolutions. The large-scale simulations leverage an extensive real-world driving database to assess overall eco-driving benefits across a range of road network and driving scenarios. The real-world driving data are further leveraged to generate representative driving routes for deeper evaluation. Based on the representative routes, in-depth simulation relying on high-fidelity models investigates how different traffic scenarios can impact the eco-driving performance. The vehicle-in-the-loop setup reinforces the in-depth simulations by conducting tests with an actual vehicle operated on a chassis dynamometer; the measured energy savings were indeed found to agree with the in-depth simulation savings estimates. Finally, limited but representative road testing with the fully integrated vehicle will be conducted to demonstrate the eco-driving capability and conclude the overall evaluation regimen.
Cutting the Deployment Costs of Physics-Based MPC in Buildings by Simulation-Based Imitation Learning
It has been shown that model predictive control (MPC) is a promising solution for energy-efficient building operations. However, the deployment of MPC in a large portion of the building stock has not been possible partially because of high installation costs. Every building is unique and requires a tailored MPC solution. The best performing solutions are often based on physics-based modeling, which is, however, computationally expensive and requires dedicated software. A promising direction that tackles this problem is to train a neural network-based optimal control policy to imitate the behavior of physics-based MPC from the simulation data generated offline. The neural networks give control actions that closely approximate those produced by physics-based MPC, but with a fraction of the computational and memory requirements and without the need for licensed software. The main advantage of the proposed approach stems from simple evaluation at execution time, leading to low computational footprints and easy deployment on embedded HW platforms. In the case study, we present the energy savings potential of physics-based MPC applied to an office building in Belgium. We demonstrate how neural network approximators can be used to cut the implementation and maintenance costs of MPC deployment without compromising performance. We also critically assess the presented approach by pointing out the remaining challenges and open research questions.
Computational Approaches for Clean Energy Materials
Currently, 80% of the global final energy consumption occurs in form of fuels and only 20% as electricity. On the other hand, renewable energy additions come almost exclusively in the form of electricity (dominantly photovoltaics and wind). Thus, a successful energy transition will require enormous growth in renewables, sufficient to convert excess electricity into fuels, as well as the development of non-electricity based solar fuel technologies. As much as photovoltaic capacities have grown over the past 20 years, it is far from clear that current technologies and materials are up to the task to grow from here by yet another factor 100 until 2050. Therefore, sustained research efforts on emerging inorganic semiconductors for solar electricity and fuels are essential for facing the double challenge of climate change and energy security. Computational materials science can make important contributions, guiding and supporting research activities through both materials search and discovery and through detailed studies that help to develop a mechanistic understanding of materials performance and bottlenecks. This presentation will highlight three recent computational projects with relevance for photovoltaics and solar fuels (1) Defect graph neural networks (dGNN) for materials discovery in solar thermochemical hydrogen (STCH) [1]. The dGNN approach facilitates broad and fast materials screening for defect properties. (2) Modeling highly off-stoichiometric systems by evaluating the free energy of defect interaction [2]. This approach allows quantitative prediction of H2 production in complex STCH oxides. (3) First-principles atomic structure prediction for interfaces [3]. This work showed how an atomically thin CdCl2 interlayer phase enables in principle ideal electron transport across the incommensurate SnO2/CdTe interface. [1] M.D. Witman, A. Goyal, T. Ogitsu, A.H. McDaniel, S. Lany, Nat. Comput. Sci. (2023). https://doi.org/10.1038/s43588-023-00495-2. [2] A. Goyal, M.D. Sanders, R.P. O'Hayre, S. Lany, PRX Energy 3, 013008 (2024). https://doi.org/10.1103/PRXEnergy.3.013008. [3] A. Sharan, M. Nardone, D. Krasikov, N. Singh, S. Lany, Appl. Phys. Rev. 9, 041411 (2022). https://doi.org/10.1063/5.0104008.
ElectroCat: DOE's approach to PGM-free catalyst and electrode R&D
The successful development of high-performance, durable platinum group metal-free (PGM-free) electrocatalysts and electrodes for polymer electrolyte membrane fuel cells (PEMFCs) will ultimately improve the cost-competiveness of fuel cells in a wide range of applications. This is considered to be a critical development especially for automotive fuel cell applications in order to bring the system cost of an automotive fuel cell system down to the $30/kW cost target set by the U.S. Department of Energy (DOE). The platinum group metal (PGM) electrocatalysts are a major contributor to the system cost. Addressing the technical challenges to PGM-free electrocatalyst and electrode development, therefore, represents one of DOE's most pressing research and development (R&D) priorities. ElectroCat was formed by the DOE as part of the Energy Materials Network (EMN) in early 2016, and shares with other EMN consortia the goal of decreasing the time to market for advanced materials related to clean energy technologies, in the context of increasing U.S. fuel cell electric vehicle (FCEV) manufacturing competitiveness. To accomplish this, the consortium performs core research and development and provides universities and companies streamlined access to the unique, world-class set of tools and expertise relevant to early-stage applied PGM-free catalyst R&D of the member national laboratories. Moreover, ElectroCat fosters a systematic methodology by which prospective catalysts and electrodes are prepared and analyzed rapidly and comprehensively using high-throughput, combinatorial methods. Catalyst discovery is augmented by theory as well as foundational electrocatalysis and materials knowledge at the participating national laboratories. Furthermore, ElectroCat has developed a data sharing framework, requisite of all EMN consortia, for disseminating its findings to the public via a searchable database, to further expedite incorporation of PGM-free electrocatalysts into next-generation fuel cells by advancing the general understanding of the PGM-free electrocatalyst field.
Adaptive Interface-PINNs (AdaI-PINNs) for transient diffusion: Applications to forward and inverse problems in heterogeneous media
We model transient diffusion in heterogeneous materials using a novel physics-informed neural networks framework (PINNs) termed Adaptive interface physics-informed neural networks or AdaI-PINNs (Roy et al. arXiv preprint arXiv:2406.04626, 2024). AdaI-PINNs utilize different activation functions with trainable slopes tailored to each material region within the computational domain, allowing for a fully automated and adaptive PINNs approach to model interface problems with strongly and weakly discontinuous solutions. To enhance its performance in highly heterogeneous transient diffusion systems, we prescribe a suite of robust practices, including appropriate non-dimensionalization of equations, a biased sampling method, Glorot initialization, and the hard enforcement of boundary and initial conditions. Here we evaluate the efficacy of the proposed method on several benchmark forward and inverse problems. Comparative studies on one-dimensional and two-dimensional benchmark problems reveal that the modified AdaI-PINNs outperform its unmodified counterpart, achieving root-mean-square errors that are at least two orders of magnitude better in forward problems. For inverse problems, the maximum errors in the approximated diffusion coefficients by modified AdaI-PINNs are four orders of magnitude better than those of the unmodified version. Additionally, modified AdaI-PINNs demonstrate improved stability in problems with large material mismatches.