SEARCH · Engineering Papers

Results for “network regularization”

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 181 records · Page 10

All sources NASA NTRS DOE OSTI

DOE OSTI2022

Neural network–based pore flow field prediction in porous media using super resolution

Direct pore-scale simulations of fluid flow through porous media are computationally expensive to perform for realistic systems. Previous works have demonstrated using the geometry of the microstructure of porous media to predict the velocity fields therein based on neural networks. However, such trained neural networks do not perform well for unseen porous media with a large degree of heterogeneity. In this study we propose that incorporating a coarse velocity field in the input of neural networks is an effective way to improve the prediction performance. The coarse velocity field can be simulated with a low computational cost and provides global information to regularize the ill-posedness of the learning problem, which is usually caused by the use of local geometries due to the computational resource constraints. We show that incorporating the coarse-mesh velocity field significantly improves the prediction accuracy of the fine-mesh velocity field by comparison to the prediction that relies on geometric information alone, especially for the porous medium with a large interior vuggy pore space. We also show the flexibility of training the network in using coarse velocity fields with various resolutions. The results suggest that even using coarse velocity field with a very low resolution, the predictions are still enhanced and close to the ground truths. The feasibility of the method is further demonstrated by testing the trained network on real rocks. This study highlights the merits of incorporating a coarse-mesh velocity field into the input for neural networks, which provides global, physics-based information for the model, thereby improving the model's generalization capability.

42 ENGINEERING↗

DOE OSTI2021

Recommendations for Distributed Energy Resource Patching

While computer systems, software applications, and operational technology (OT)/Industrial Control System (ICS) devices are regularly updated through automated and manual processes, there are several unique challenges associated with distributed energy resource (DER) patching. Millions of DER devices from dozens of vendors have been deployed in home, corporate, and utility network environments that may or may not be internet-connected. These devices make up a growing portion of the electric power critical infrastructure system and are expected to operate for decades. During that operational period, it is anticipated that critical and noncritical firmware patches will be regularly created to improve DER functional capabilities or repair security deficiencies in the equipment. The SunSpec/Sandia DER Cybersecurity Workgroup created a Patching Subgroup to investigate appropriate recommendations for the DER patching, holding fortnightly meetings for more than nine months. The group focused on DER equipment, but the observations and recommendations contained in this report also apply to DERMS tools and other OT equipment used in the end-to-end DER communication environment. The group found there were many standards and guides that discuss firmware lifecycles, patch and asset management, and code-signing implementations, but did not singularly cover the needs of the DER industry. This report collates best practices from these standards organizations and establishes a set of best practices that may be used as a basis for future national or international patching guides or standards.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Optimizing Fast Charging and Wetting in Lithium-Ion Batteries with Optimal Microstructure Patterns Identified by Genetic Algorithm

To sustain the high-rate current required for fast charging electric vehicle batteries, electrodes must exhibit sufficiently high effective ionic diffusion. Additionally, to reduce battery manufacturing costs, wetting time must decrease. Both of these issues can be addressed by structuring the electrodes with mesoscale pore channels. However, their optimal spatial distribution, or patterns, is unknown. Herein, a genetic algorithm has been developed to identify these optimal patterns using a CPU-cheap proxy distance-based model to evaluate the impact of the added pore networks. Both coin-cell and pouch cell form factors have been considered for the wetting analysis, with their respective electrolyte infiltration mode. Regular hexagonal and mud-crack-like patterns, respectively, for fast charging and fast wetting were found to be optimal and have been compared with pre-determined, easier to manufacture, patterns. The model predicts that using cylindrical channels arranged in a regular hexagonal pattern is ∼6.25 times more efficient for fast charging as compared to grooved lines with both structuring strategies being restricted to a 5% electrode total volume loss. The model also shows that only a very limited electrode volume loss (1%–2%) is required to dramatically improve the wetting (5–20 times) compared to an unstructured electrode.

25 ENERGY STORAGE↗

NASA NTRS2002

Superlative TINs

TIN is short for 'triangulated irregular network,' which is a piecewise planar model of a surface. If properly constructed, a TIN can be more than 30 times as efficient as a regular triangulation. In our project (a ground combat simulation to support U.S. Army training exercises), the TIN is used to represent the Earth's surface and is used primarily to determine whether line of sight is blocked by terrain. High efficiency requires accurate identification of ridgelines with as few triangles as possible. The work currently in progress is the implementation of a TINning process that we hope will produce superlative TINs. This presentation describes that process.

Triangulated Irregular Networks TIN Delaunay trian↗

DOE OSTI2024

Shearless effective barriers to chaotic transport induced by even twin islands in nontwist systems

For several decades now it has been known that systems with shearless invariant tori, nontwist Hamiltonian systems, possess barriers to chaotic transport. These barriers are resilient to breakage under perturbation and therefore regions where they occur are natural places to look for barriers to transport. Here we describe a kind of effective barrier that persists after the shearless torus is broken. Because phenomena are generic, for convenience we study the standard nontwist map (SNM), an area-preserving map that violates the twist condition locally in the phase space. The barrier occurs in nontwist systems when twin even period islands are present, which happens for a broad range of parameter values in the SNM. With a phase space composed of regular and irregular orbits, the movement of chaotic trajectories is hampered by the existence of shearless curves, total barriers, and a network of partial barriers formed by the stable and unstable manifolds of the hyperbolic points. Being a degenerate system, the SNM has twin islands and, consequently, twin hyperbolic points. We show that the structures formed by the manifolds intrinsically depend on period parity of the twin islands. For this even scenario the structure that we call a torus free barrier occurs because the manifolds of different hyperbolic points form an intricate chain atop a dipole configuration and the transport of chaotic trajectories through the chain becomes a rare event. This structure impacts the emergence of transport, the escape basin for chaotic trajectories, the transport mechanism, and the chaotic saddle. The case of odd periodic orbits is different: we find for this case the emergence of transport immediately after the breakup of the last invariant curve, and this leads to a scenario of higher transport, with intricate escape basin boundary and a chaotic saddle with nonuniformly distributed points.

classical mechanics↗

DOE OSTI2022

Rethinking the ill-posedness of the spectral function reconstruction — Why is it fundamentally hard and how Artificial Neural Networks can help

Reconstructing hadron spectral functions through Euclidean correlation functions are of the important missions in lattice QCD calculations. However, in a Källen–Lehmann (KL) spectral representation, the reconstruction is observed to be ill-posed in practice. It is usually ascribed to the fewer observation points compared to the number of points in the spectral function. In this paper, by solving the eigenvalue problem of continuous KL convolution, we show analytically that the ill-posedness of the inversion is fundamental and it exists even for continuous correlation functions. We discussed how to introduce regulators to alleviate the predicament, in which include the Artificial Neural Networks (ANNs) representations recently proposed by the Authors in another study. The uniqueness of solutions using ANNs representations is manifested analytically and validated numerically. Reconstructed spectral functions using different regularization schemes are also demonstrated, together with their eigen-mode decomposition. We observe that components with large eigenvalues can be reliably reconstructed by all methods, whereas those with low eigenvalues need to be constrained by regulators.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

A physics-informed and hierarchically regularized data-driven model for predicting fluid flow through porous media

This paper presents a new deep learning data-driven model for predicting structure dependent pore-fluid velocity fields in rock. The model is based on a Convolutional Auto-Encoder (CAE) artificial neural network capable of learning from image data generated by direct numerical simulations of fluid flow through pore-structures, such as by Lattice Boltzmann or molecular dynamics methods. The main novelty of the model in comparison to previous CAE-based data-driven approaches consists of three parts. The first is a methodology for decomposing the full-domain of the porous media into sub-regions, or “sub-domains”, in order to reduce the overall size of the CAE, batch process the sub-domains in parallel, and enable the CAE to learn local and generalizable nonlinear mappings of pore-fluid velocities. The second consists of embedding the finite difference solutions of the incompressible Navier-Stokes and continuity equations into convolutional layers prior to the CAE in order to provide the CAE with knowledge of fluid dynamics physics (PhyFlow). The third main novelty is that the training of the CAE is regularized with a hierarchical loss function that encourages the learning of fluid flow patterns (in a way similar to ranked modes in principal component analysis), ranking from most to least important. This is shown to increase the stability in learning, reduce over-fitting, and promote interpretability of the CAE neural network layers (HierCAE). The comprehensive new data-driven model, which we call the PhyFlow-HierCAE model, is shown to exhibit improved accuracy and generalizability of flow field predictions over conventional CAE models, attributable to the embedded physical knowledge and the hierarchical regularization, as well as realize orders of magnitude speed-ups in computation times as a surrogate for the direct numerical simulations. Examples of training and forward predictions on unseen pore-structures are provided and evaluated for data from Lattice Boltzmann and molecular dynamics simulations of pore-fluid flow. The model is shown to be a fast and accurate emulator (or “surrogate”) for predicting effective permeability of unseen pore-structures based on learning from relatively small direct numerical simulation datasets.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

NASA NTRS2014

Scalable Asset Discovery, Vulnerability Scanning, and Penetration Testing for Remote Sites and Wireless Spectrums Utilizing an Embedded Linux Plug - PwniPlug and the Raspberry Pi B+ as a Sample Pen Test

All devices attached to the NASA KSC network are subject to security vulnerability scanning and/or penetration testing. In today's changing environment, vulnerable and/or unprotected systems can easily be overlooked. Systems that are not properly managed can become a potential threat to the operational integrity of our systems and networks. This includes all NASA (internal and external) information systems within NASA KSC Internet Protocol (IP) address space, and NASA KSC facilities. The Office of the Chief Information Officer (OCIO) recommends that all NASA Centers and information systems be subject to penetration testing on a regular interval in accordance with the guidelines identified by the National Institute of Standards and Technology (NIST). (ITS-HBK-2810.04-02A) Protecting information and equipment at NASA is an area of increasing concern. In addition to the CPU's on the network; Supervisory, Control and Data Acquisition (SCADA) systems are especially vulnerable because these systems have lacked standards, use embedded controllers with little computational power and informal software, are connected to physical processes, have few operators, and are increasingly also being connected to corporate networks. The scope of work is comprised of several individual components which together build upon previous work by Drew Branch, NASA KSC Intern. The Pwn Plug is the selected COTS (Commercial-Off-The-Shelf) device chosen to test simplification of mandatory IT Security tasks. The device will be utilized to provide services to NASA KSC and enable an assessment of infrastructure soundness and regulatory compliance in an efficient, economical, and business responsive manner. The Pwn Plug is designed as a pen testing appliance which provides a hardware platform that can support commercial penetration testing efforts at significantly reduced costs. The expected outcomes are: 1) External Penetration Testing, 2) Social Engineering, 3) Procedural Documentation, 4) Recommended Remediation Action Plan, 5) System Retest & Remediation Attestation and 6) Final Reports, out briefing and Presentation. Due to physical and material constraints beyond intern and mentor control, the project was redefined as a working pen-test scenario. Limitations of lab availability and tools dictated an academic exercise. This report was developed within the scenario guidelines suggested by the project mentor. The guidelines were to be creative in developing a Pen Test program for a client.

Penetration Testing↗

DOE OSTI2025

A Variational Autoencoder Model Toward Molecular Structure Representation Learning of Fuels

Here, in this work, a Variational Autoencoder (VAE)-based data-driven modeling framework is developed with the overarching goal of enabling fuel design. The VAE model is trained on a large dataset with several chemical species to learn a compressed latent space molecular representation. Chemical structure in the form of Simplified Molecular Input Line Entry System (SMILES) string is fed as input, encoded into the VAE latent space, and decoded back to the SMILES string using Long Short-Term Memory (LSTM) networks. Complexities of the VAE training loss function are thoroughly examined by varying the weightage (beta (𝜷) parameter) of the latent space regularization term, thereby assessing the balance between reconstruction accuracy and validity, and focusing on both accurate molecular structure reconstruction and latent space consistency. Two different strategies for 𝜷 variation are evaluated: linear annealing and cyclic annealing. In addition, the impact of total correlation adjustment and hierarchical priors is also studied with regard to the balance between reconstruction fidelity and latent space regularization, and potential issues such as posterior collapse, over-regularization, and poor disentanglement of latent variables. Overall, the best performance of the model is achieved with hierarchical priors and incrementally increasing 𝜷 from 0 to a threshold value of 0.25 over 75 epochs. The generative VAE model can be readily coupled with Quantitative Structure–Property Relationship (QSPR) analysis to develop an integrated end-to-end framework for fuel-property prediction and molecular design of novel promising fuels.

fuel design↗

NASA NTRS1992

Laser ranging data analysis

Center for Space Research efforts have focused on the near real-time analysis of Lageos laser ranging data and on the production of predictive ephemerides. The data are analyzed in terms of range bias, time bias, and internal precision, and estimates for the Earth orientation parameters X(sub p), Y(sub p) and UT1 are obtained. The results of these analyses are reported in a variety of formats. In addition several additional stations began sending not only quick-look observations but also normal points created on-site with new software. These normal points are transmitted in a new standard format different from either current quick-look or MERIT-II full-rate formats. Thus new preprocessing software was written and successfully tested on these data. Inspection of the Bendix produced Lageos full-rate normal points continued, with detailed analyses and filtering of all 1991 A and B release normal points for Lageos through the beginning of 1992. A summary of the combined full-rate and quick-look normal point data set created for 1991 is provided. New long-term ephemerides for Lageos satellite, as well as for Etalon-1 and Etalon-2 (the so-called high satellites used for laser ranging) were produced and distributed to the network stations in cooperation with the Crustal Dynamics Project and Eurolas. These predictions are used by essentially every laser ranging site obtaining regular returns from any of these three satellites.

Source record↗

NASA NTRS2023

DPOD2020: A DORIS Extension of the ITRF2020 for Precise Orbit Determination

As one of the tracking systems used to determine orbits of the altimeter mission satellites (such as TOPEX/Poseidon, Envisat, Jason-1/-2/-3, CryoSat-2, Saral/Altika, Sentinel-3A/-3B, HY-2A/C/D, Jason-CS/Sentinel-6A, SWOT), DORIS (Doppler Orbitography Radiopositionning Integrated by Satellite) allows to determine positions and velocities of tracking stations that define a stable reference for the estimation of the precise orbits and thus are fundamental for the quality of the altimeter data and derived mean sea level products. Due to the time evolution of the DORIS ground network, some stations included in the 2020 realization of the International Terrestrial Reference Frame (ITRF2020) have been decommissioned and since 2021.0 a few new stations were added to the tracking network. Therefore, to satisfy operational requirements for POD (Precise Orbit Determination) and routine delivery of geodetic products, the International DORIS Service (IDS) regularly updates the DPOD (DORIS terrestrial reference frame for Precise Orbit Determination). The DPOD solutions include mean positions and velocities of all the DORIS stations since 1993.0 derived from the stacking of the latest IDS weekly combined series aligned to the current ITRF. In this paper, we first present the stacking process of the DPOD2020 version 1.0. Then, we address the validation procedure of the DPO2020 including comparison with ITRF2020 and POD tests. For eighty percent of all the time segments of all the DORIS stations, the station position differences between DPOD2020 version 1.0 and ITRF2020 are smaller than ten millimeters. The major position differences between these two solutions are associated with the DORIS sites either localized in the South Atlantic Anomaly region or with time spans smaller than one year. Compared to DPOD2014, the DPOD2020 shows reduction of the main statistics of the DORIS-to-DORIS tie residuals (differences between the estimated and measured ties). The POD tests showed similar results for DPOD2020 and DPOD2014 for most of the altimetric satellites (TOPEX/Poseidon, Jason-1, CryoSat-2, Jason-3). In addition, we observed better POD results with DPOD2020 for the latest altimetric satellite Sentinel-3A as well as a slight degradation for Jason-2. That degradation was fully explained by slightly worst results for the stations localized in the South Atlantic Anomaly region.

DORIS↗

NASA NTRS1991

Communications infrastructure requirements for telemedicine/telehealth in the context of planning for and responding to natural disasters: Considering the need for shared regional networks

During the course of recent years the frequency and magnitude of major disasters - of natural, technological, or ecological origin - have made the world community dramatically aware of the immense losses of human life and economic resources that are caused regularly by such calamities. Particularly hard hit are developing countries, for whom the magnitude of disasters frequently outstrips the ability of the society to cope with them. In many cases this situation can be prevented, and the recent trend in disaster management has been to emphasize the importance of preparedness and mitigation as a means of prevention. In cases of disaster, a system is needed to respond to relief requirements, particularly the delivery of medical care. There is no generic telecommunications infrastructure appropriate for the variety of applications in medical care and disaster management. The need to integrate telemedicine/telehealth into shared regional disaster management telecommunications networks is discussed. Focus is on the development of infrastructure designed to serve the needs of disaster prone regions of the developing world.

Scott, John Carver↗

DOE OSTI2022

Kinetics-Informed Neural Networks

Chemical kinetics and reaction engineering consists of the phenomenological framework for the disentanglement of reaction mechanisms, optimization of reaction performance and the rational design of chemical processes. Here, we utilize feed-forward artificial neural networks as basis functions to solve ordinary differential equations (ODEs) constrained by differential algebraic equations (DAEs) that describe microkinetic models (MKMs). We present an algebraic framework for the mathematical description and classification of reaction networks, types of elementary reaction, and chemical species. Under this framework, we demonstrate that the simultaneous training of neural nets and kinetic model parameters in a regularized multi-objective optimization setting leads to the solution of the inverse problem through the estimation of kinetic parameters from synthetic experimental data. We analyze a set of scenarios to establish the extent to which kinetic parameters can be retrieved from transient kinetic data, and assess the robustness of the methodology with respect to statistical noise. Furthermore, this approach to inverse kinetic ODEs can assist in the elucidation of reaction mechanisms based on transient data.

36 MATERIALS SCIENCE↗

DOE OSTI2026

Quantum Ising model on (2+1)-dimensional anti–de Sitter space using tensor networks

We study the quantum Ising model on (2+1)-dimensional anti-de Sitter space using matrix product states (MPS) and matrix product operators (MPOs). We explore the bulk phase diagram of the theory on regular tessellations of hyperbolic space with coordination number seven and find disordered and ordered phases separated by a phase transition. We find that the boundary-boundary spin correlation function exhibits power law scaling deep in the disordered phase of the Ising model consistent with holography. At the critical point, we find the boundary entanglement entropy scales logarithmically with subsystem size but away from this, we see a linear scaling. In comparison, the full system exhibits a volume law scaling, which is expected in chaotic and/or highly connected systems. We also measure out of time ordered correlators (OTOCs) to explore the scrambling behavior of the theory.

Quantum spin models↗

DOE OSTI2024

TNet: A Model-Constrained Tikhonov Network Approach for Inverse Problems

Deep learning (DL), in particular deep neural networks, by default is purely data-driven and in general does not require physics. This is the strength of DL but also one of its key limitations when applied to science and engineering problems in which underlying physical properties—such as stability, conservation, and positivity—and accuracy are required. DL methods in their original forms are often not capable of respecting the underlying mathematical models or achieving desired accuracy even in big-data regimes. On the other hand, many data-driven science and engineering problems, such as inverse problems, typically have limited experimental or observational data, and DL would overfit the data in this case. Leveraging information encoded in the underlying mathematical models, we argue, not only compensates for missing information in low data regimes but also provides opportunities to equip DL methods with the underlying physics, hence promoting better generalization. This paper develops a model-constrained DL approach and its variant TNet—a Tikhonov neural network—which are capable of learning not only information hidden in the training data but also in the underlying mathematical models to solve inverse problems governed by partial differential equations in low data regimes. We provide the constructions and some theoretical results for the proposed approaches for both linear and nonlinear inverse problems. Since TNet is designed to learn inverse solutions with Tikhonov regularization, it is interpretable: in fact it recovers Tikhonov solutions for linear cases while potentially approximating Tikhonov solutions for nonlinear inverse problems. We also prove that data randomization can enhance not only the smoothness of the networks but also their generalizations. Comprehensive numerical results confirm the theoretical findings and show that with even as little as 1 training data sample for one-dimensional (1D) deconvolution, 5 for an inverse 2D heat conductivity problem, 100 for inverse initial conditions for a time-dependent 2D Burgers’s equation, and 50 for inverse initial conditions for 2D Navier–Stokes equations, TNet solutions can be as accurate as Tikhonov solutions while being several orders of magnitude faster. Furthermore, this is possible owing to the model-constrained term, replications, and randomization.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

A Real-Time Optimization with Warm-Start of Multiperiod AC Optimal Power Flows

We present a real-time optimization strategy based on warm-start for solving a moving horizon of multi period AC optimal power flow (ACOPF) problems. In each horizon, ACOPFs are temporally interlinked via generator ramp constraints, and we assume that each horizon needs to be solved every few seconds or minutes. We introduce two approximate tracking schemes that closely follow a solution path consisting of strongly regular points. We present theoretical results bounding the tracking error by the square of the parameter changes between time periods. Experimental results for networks of sizes up to 9K buses show a fast computation time while maintaining a good solution quality, thus making our approach well suited for real-time circumstances.

moving horizon↗

NASA NTRS1974

Dynamic behaviour of coastal sedimentation in the Lions Gulf

The author has identified the following significant results. A number of ERTS-1 images covering this geographical zone were studied and compared with cartographic maps, air photographs, and thermal-IR images. Old and recent sediments leave traces in the landscape which are decoded by interpreting the shapes of the clear zones forming a network against the black background representing water and humid zones. Current sedimentation and its mechanism were investigated. It had been hoped that a regular sequence of images would make it possible to follow the dynamics of the Rhone and the coastal rivers in relation to meteorological conditions. In any event only a small number of images spread over a wide period of time were obtained, and a complete study was therefore impossible. However, in comparing some of the ERTS-1 images certain thermal-IR images and information on the flow of the Rhone provided some clarification of mechanisms associated with river dynamics.

Guy, M.↗

DOE OSTI2023

Developing an AI-Powered Zero-Trust Cybersecurity Framework for Malware Prevention in Nuclear Power Plants

This study presents the development of an AI-powered Zero-Trust cybersecurity framework for malware prevention in nuclear power plants. The framework aims to enhance the security of critical systems within nuclear power plants by adopting the principles of Zero-Trust and leveraging artificial intelligence (AI) technologies. By assuming no implicit trust in any user or device and continuously authenticating and authorizing access, the framework ensures a robust defense against malware attacks. The integration of AI allows for the detection and prevention of malware through behavioral analytics, endpoint protection, network segmentation, and continuous monitoring. The paper discusses the key considerations, steps, and technologies involved in developing this framework, emphasizing the importance of regular updates, training, compliance, and auditing. The proposed framework serves as a comprehensive approach to safeguarding nuclear power plants from sophisticated malware threats and protecting the integrity and safety of critical infrastructure.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗