Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “inversion methods”

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 361 records · Page 20
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Underground hydrogen storage leakage detection and characterization based on machine learning of sparse seismic data

Underground hydrogen storage (UHS) is considered as a scalable approach for massive storage and seasonal extraction of hydrogen (H 2 ). Although conventional leakage detection and characterization methods based on time-lapse seismic imaging and inversion generally apply to H 2 leakage detection problem, a high-fidelity yet cost effective geophysics approach is still missing to reliably inform leakage location and properties based on very sparse data. In response, we develop a novel supervised machine learning method to detect and characterize H 2 leakage from UHS. The input to our neural network are sparse time-lapse seismic waveforms, while the output from the neural network includes the spatial location and physical properties of a H 2 leakage. Here, we generate high-quality time-lapse waveforms using the elastic-wave equations to train the neural network. We train and validate our machine learning model and find that it attains high accuracy in using extremely sparse time-lapse seismic data to detect and characterize H 2 leakage. Our investigation is the first systematic study that focuses on applying machine learning to subsurface H 2 leakage detection and characterization and could potentially serve as a cost-effective geophysical tool for underground hydrogen leakage detection and characterization with high fidelity.

08 HYDROGEN↗
DOE OSTI2025

ARM-IRL: Adaptive Resilience Metric Quantification Using Inverse Reinforcement Learning

The resilience of safety-critical systems is gaining importance due to the rise in cyber and physical threats, especially within critical infrastructure. Traditional static resilience metrics may not capture dynamic system states, leading to inaccurate assessments and ineffective responses to cyber threats. This work aims to develop a data-driven, adaptive method for resilience metric learning. We propose a data-driven approach using inverse reinforcement learning (IRL) to learn a single, adaptive resilience metric. The method infers a reward function from expert control actions. Unlike previous approaches using static weights or fuzzy logic, this work applies adversarial inverse reinforcement learning (AIRL), training a generator and discriminator in parallel to learn the reward structure and derive an optimal policy. The proposed approach is evaluated on multiple scenarios: optimal communication network rerouting, power distribution network reconfiguration, and cyber–physical restoration of critical loads using the IEEE 123-bus system. The adaptive, learned resilience metric enables faster critical load restoration in comparison to conventional RL approaches.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

AutoPhaseNN: unsupervised physics-aware deep learning of 3D nanoscale Bragg coherent diffraction imaging

Abstract The problem of phase retrieval underlies various imaging methods from astronomy to nanoscale imaging. Traditional phase retrieval methods are iterative and are therefore computationally expensive. Deep learning (DL) models have been developed to either provide learned priors or completely replace phase retrieval. However, such models require vast amounts of labeled data, which can only be obtained through simulation or performing computationally prohibitive phase retrieval on experimental datasets. Using 3D X-ray Bragg coherent diffraction imaging (BCDI) as a representative technique, we demonstrate AutoPhaseNN, a DL-based approach which learns to solve the phase problem without labeled data. By incorporating the imaging physics into the DL model during training, AutoPhaseNN learns to invert 3D BCDI data in a single shot without ever being shown real space images. Once trained, AutoPhaseNN can be effectively used in the 3D BCDI data inversion about 100× faster than iterative phase retrieval methods while providing comparable image quality.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Massively parallel modeling and inversion of electrical resistivity tomography data using PFLOTRAN

Abstract. Electrical resistivity tomography (ERT) is a broadly accepted geophysical method for subsurface investigations. Interpretation of field ERT data usually requires the application of computationally intensive forward modeling and inversion algorithms. For large-scale ERT data, the efficiency of these algorithms depends on the robustness, accuracy, and scalability on high-performance computing resources. In this regard, we present a robust and highly scalable implementation of forward modeling and inversion algorithms for ERT data. The implementation is publicly available and developed within the framework of PFLOTRAN, an open-source, state-of-the-art massively parallel subsurface flow and transport simulation code. The forward modeling is based on a finite-volume discretization of the governing differential equations, and the inversion uses a Gauss–Newton optimization scheme. To evaluate the accuracy of the forward modeling, two examples are first presented by considering layered (1D) and 3D earth conductivity models. The computed numerical results show good agreement with the analytical solutions for the layered earth model and results from a well-established code for the 3D model. Inversion of ERT data, simulated for a 3D model, is then performed to demonstrate the inversion capability by recovering the conductivity of the model. To demonstrate the parallel performance of PFLOTRAN's ERT process model and inversion capabilities, large-scale scalability tests are performed by using up to 131 072 processes on a leadership class supercomputer. These tests are performed for the two most computationally intensive steps of the ERT inversion: forward modeling and Jacobian computation. For the forward modeling, we consider models with up to 122 ×106 degrees of freedom (DOFs) in the resulting system of linear equations and demonstrate that the code exhibits almost linear scalability on up to 10 000 DOFs per process. On the other hand, the code shows superlinear scalability for the Jacobian computation, mainly because all computations are fairly evenly distributed over each process with no parallel communication.

58 GEOSCIENCES↗
DOE OSTI2023

Seismic Monitoring at the Farnsworth CO2-EOR Field Using Time-Lapse Elastic-Waveform Inversion of 3D-3C VSP Data

During the Development Phase of the U.S. Southwest Regional Partnership on Carbon Sequestration, supercritical CO2 was continuously injected into the deep oil-bearing Morrow B formation of the Farnsworth Unit in Texas for Enhanced Oil Recovery (EOR). The project injected approximately 94 kilotons of CO2 to study geologic carbon storage during CO2-EOR. A three-dimensional (3D) surface seismic dataset was acquired in 2013 to characterize the subsurface structures of the Farnsworth site. Following this data acquisition, the baseline and three time-lapse three-dimensional three-component (3D-3C) vertical seismic profiling (VSP) data were acquired at a narrower surface area surrounding the CO2 injection and oil/gas production wells between 2014 and 2017 for monitoring CO2 injection and migration. With these VSP datasets, we inverted for subsurface velocity models to quantitatively monitor the CO2 plume within the Morrow B formation. We first built 1D initial P-wave (Vp) and S-wave (Vs) velocity models by upscaling the sonic logs. We improved the deep region of the Vp and Vs models by incorporating the deep part of a migration velocity model derived from the 3D surface seismic data. We improved the shallow region of 3D Vp and Vs models using 3D traveltime tomography of first arrivals of VSP downgoing waves. We further improved the 3D baseline velocity models using elastic-waveform inversion (EWI) of the 3D baseline VSP upgoing data. Our advanced EWI method employs alternative tomographic and conventional gradients and total-variation-based regularization to ensure the high-fidelity updates of the 3D baseline Vp and Vs models. We then sequentially applied our 3D EWI method to the three time-lapse datasets to invert for spatiotemporal changes of Vp and Vs in the reservoir. Our inversion results reveal the volumetric changes of the time-lapse Vp and Vs models and show the evolution of the CO2 plume from the CO2 injection well to the oil/gas production wells.

42 ENGINEERING↗
DOE OSTI2025

Convergence Analysis of the Alternating Anderson–Picard Method for Nonlinear Fixed-Point Problems

Anderson acceleration (AA) has been widely used to solve nonlinear fixed-point problems due to its rapid convergence. This work focuses on a variant of AA in which multiple Picard iterations are performed between each AA step, referred to as the Alternating Anderson–Picard (AAP) method. Furthermore, despite introducing more “slow” Picard iterations, this method has been shown to be efficient and even more robust in both linear and nonlinear cases. However, there is a lack of theoretical analysis for AAP in the nonlinear case. In this paper, we address this gap by establishing the equivalence between AAP and a multisecant-GMRES method that uses GMRES to solve a multisecant linear system at each iteration. From this perspective, we show that AAP “converges” to the Newton-GMRES method. Specifically, as the residual approaches zero, the multisecant matrix, the approximate Jacobian inverse, the search direction, and the optimization gain of AAP converge to their counterparts in the Newton-GMRES method. These connections provide insights for analyzing the asymptotic convergence properties of AAP. Consequently, we show that AAP is locally 𝑞-linear convergent and provide an upper bound for the convergence factor of AAP. To validate the theoretical results, numerical examples are provided.

Anderson acceleration↗
DOE OSTI2024

Scientific machine learning for closure models in multiscale problems: A review

Here, closure problems are omnipresent when simulating multiscale systems, where some quantities and processes cannot be fully prescribed despite their effects on the simulation's accuracy. Recently, scientific machine learning approaches have been proposed as a way to tackle the closure problem, combining traditional (physics-based) modeling with data-driven (machine-learned) techniques, typically through enriching differential equations with neural networks. This paper reviews the different reduced model forms, distinguished by the degree to which they include known physics, and the different objectives of a priori and a posteriori learning. The importance of adhering to physical laws (such as symmetries and conservation laws) in choosing the reduced model form and choosing the learning method is discussed. The effect of spatial and temporal discretization and recent trends toward discretization-invariant models are reviewed. In addition, we make the connections between closure problems and several other research disciplines: inverse problems, Mori-Zwanzig theory, and multi-fidelity methods. In conclusion, much progress has been made with scientific machine learning approaches for solving closure problems, but many challenges remain. In particular, the generalizability and interpretability of learned models is a major issue that needs to be addressed further.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

An efficient method to propagate model uncertainty when inverting seismic data for time domain seismic moment tensors

SUMMARY We present a computationally efficient method to approximately propagate uncertainty when linearly inverting seismic data for point source, time variable moment tensor components. The method is based on the assumption that the data residual, given by the difference between the observed seismic data and the data predicated by a linear inversion, contains the effects of both data and model uncertainty. Our method uses a distribution of data residuals, added directly to the data, in a pseudo-Monte Carlo scheme. Using the assumption that the data residual is a stochastic process, we use the well-known Karhunen–Loève (KL) theorem to construct a distribution of data residuals, where the required basis functions are constructed using Fourier series. The Fourier series are scaled by a product of a random variable and the real-valued spectral amplitudes of the original data residual’s spectrum. Thus, the Fourier series and spectral amplitudes are eigenfunction-eigenvalue pairs used in the KL-based construction of data residual distribution. Using tests with synthetic data, we show that our method compares closely with a Finite Difference Monte Carlo (FDMC) method that we presented previously. More importantly, the method presented here is computationally several orders of magnitude faster than our previous FDMC method, and requires no a priori assumptions of model and/or data uncertainty.

Poppeliers, Christian (ORCID:0000000159526849)↗
DOE OSTI2022

A Bayesian Approach for Quantifying Data Scarcity when Modeling Human Behavior via Inverse Reinforcement Learning

Computational models that formalize complex human behaviors enable study and understanding of such behaviors. However, collecting behavior data required to estimate the parameters of such models is often tedious and resource intensive. Thus, estimating dataset size as part of data collection planning (also known as Sample Size Determination) is important to reduce the time and effort of behavior data collection while maintaining an accurate estimate of model parameters. In this paper, we present a sample size determination method based on Uncertainty Quantification (UQ) for a specific Inverse Reinforcement Learning (IRL) model of human behavior, in two cases: 1) pre-hoc experiment design—conducted in the planning stage before any data is collected, to guide the estimation of how many samples to collect; and 2) post-hoc dataset analysis—performed after data is collected, to decide if the existing dataset has sufficient samples and whether more data is needed. Here, we validate our approach in experiments with a realistic model of behaviors of people with Multiple Sclerosis (MS) and illustrate how to pick a reasonable sample size target. Our work enables model designers to perform a deeper, principled investigation of effects of dataset size on IRL.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Bayesian inference of heterogeneous epidemic models: Application to COVID-19 spread accounting for long-term care facilities

Here we propose a high dimensional Bayesian inference framework for learning heterogeneous dynamics of a COVID-19 model, with a specific application to the dynamics and severity of COVID-19 inside and outside long-term care (LTC) facilities. We develop a heterogeneous compartmental model that accounts for the heterogeneity of the time-varying spread and severity of COVID-19 inside and outside LTC facilities, which is characterized by time-dependent stochastic processes and time-independent parameters in ~ 1500 dimensions after discretization. To infer these parameters, we use reported data on the number of confirmed, hospitalized, and deceased cases with suitable post-processing in both a deterministic inversion approach with appropriate regularization as a first step, followed by Bayesian inversion with proper prior distributions. To address the curse of dimensionality and the ill-posedness of the high-dimensional inference problem, we propose use of a dimension-independent projected Stein variational gradient descent method, and demonstrate the intrinsic low-dimensionality of the inverse problem. We present inference results with quantified uncertainties for both New Jersey and Texas, which experienced different epidemic phases and patterns. Moreover, we also present forecasting and validation results based on the empirical posterior samples of our inference for the future trajectory of COVID-19.

42 ENGINEERING↗
DOE OSTI2021

Synthetic Evaluation of Infrasonic Multipole Waveform Inversion

Acoustic source inversions estimate the mass flow rate of volcanic explosions or yield of chemical explosions and provide insight into potential source directionality. However, the limitations of applying these methods to complex sources and their ability to resolve a stable solution have not been investigated in detail. We perform synthetic infrasound waveform inversions that use 3-D Green’s functions for a variety of idealized and realistic deployment scenarios using both a flat plane and Yasur volcano, Vanuatu as examples. We investigate the ability of various scenarios to retrieve the input source functions and relative amplitudes for monopole and multipole (monopole and dipole) inversions. Infrasound waveform inversions appear to be a robust method to quantify mass flow rates from simple sources (monopole) using deployments of infrasound sensors placed around a source, but care should be taken when analyzing and interpreting results from more complex acoustic sources (multipole) that have significant directional components. In the examples we consider the solution is stable for monopole inversions with a signal-to-noise ratio greater than five and the dipole component is small. For most scenarios investigated, the vertical dipole component of the multipole explosion source is poorly constrained and can impact the ability to recover the other source term components. Because multipole inversions are ill-posed for many deployments, a low residual does not necessarily mean the proper source vector has been recovered. Synthetic studies are able to help investigate the limitations and place bounds on information that may be missing using monopole and multipole inversions for potentially directional sources.

58 GEOSCIENCES↗
DOE OSTI2022

Finite Element Modeling of Diffusion in Fractured Porous Media by Using Hierarchical Material Properties

Abstract Fractured porous media challenge modeling approaches due to high computational costs and excessive mesh refinement imposed by the extreme scale variability of fractures and the heterogeneity of the surrounding porous rock. To overcome such difficulties, we utilize the hierarchical finite element method ( Hi ‐FEM) that has been developed previously to simulate the electrical potential distribution in complex geologic environments. The method employs the hierarchical basis functions in classical finite element analysis to enable representation of material properties on each dimensional component of a given 3D unstructured finite element, thereby inherently allowing for interactions at the boundary between fracture and a host rock. In this study, we extend its application to transient fluid flow and heat conduction in the Laplace domain. Time‐domain flow solutions are obtained by numerical inverse Laplace transform. We evaluate the accuracy of the method using different flow models and demonstrate its robustness for large‐scale, rock mass models featuring complex fracture networks. Moreover, for the computation of nodal Darcian velocity fields in fractured porous media where the fractures are represented as 2D features, a new approach that employs the Yeh's Galerkin model for both volume and facet elements is proposed. Results show that Hi ‐FEM can produce accurate flow solutions for fractured porous media without any need of coupling or transfer mechanism while still being computationally economical and numerically robust, even for large‐scale simulations.

Beskardes, G. D.↗
DOE OSTI2025

Resummation for lattice QCD calculation of generalized parton distributions at nonzero skewness

Large-momentum effective theory (LaMET) provides an approach to directly calculate the x-dependence of generalized parton distributions (GPDs) on a Euclidean lattice through power expansion and a perturbative matching. When a parton’s momentum becomes soft, the corresponding logarithms in the matching kernel become non-negligible at higher orders of perturbation theory, which requires a resummation. But the resummation for the off-forward matrix elements at nonzero skewness ξ is difficult due to their multi-scale nature. In this work, we demonstrate that these logarithms are important only in the threshold limit, and derive the threshold factorization formula for the quasi-GPDs in LaMET. We then propose an approach to resum all the large logarithms based on the threshold factorization, which is implemented on a GPD model. We demonstrate that the LaMET prediction is reliable for [−1 + x 0 , −ξ − x 0 ] ∪ [−ξ + x 0 , ξ − x 0 ] ∪ [ξ + x 0 , 1 − x 0 ], where x 0 is a cutoff depending on hard parton momenta. Through our numerical tests with the GPD model, we demonstrate that our method is self-consistent and that the inverse matching does not spread the nonperturbative effects or power corrections to the perturbatively calculable regions.

hadronic spectroscopy↗
DOE OSTI2025

Elucidating the factors controlling the methanol synthesis from CO 2 on ZnPd-ZnO/TiO 2 catalysts prepared by direct/inverse impregnation

The direct hydrogenation of CO 2 to methanol over supported ZnPd intermetallic catalysts remains a challenge due to the complex nature of the active sites, which are highly dependent on both the dispersion of ZnPd intermetallics and their proximity to ZnO. This study investigates the effect of the Pd and Zn incorporation order on methanol synthesis over TiO 2 -supported ZnPd catalysts. Catalysts were prepared via direct (Pd/Zn), inverse (Zn/Pd), and co-impregnation (Pd+Zn) methods using anatase TiO 2 as the support. Comprehensive characterization by XRD, HRTEM, CO-DRIFTS and XPS revealed that the formation of small and crystalline intermetallic β-ZnPd particles and their interaction with ZnO/ZnO x depends on the sequence of Pd and Zn incorporation. It was found that, under the reaction conditions, the ZnO x species plays a critical role in facilitating the formation of the intermetallic ZnPd through the interactions between Pd and ZnO x . In addition, ZnO x species have also been shown to modulate key catalytic properties, such as CO 2 activation and methanol selectivity. Specifically, the ZnPd/Pd surface ratio, determined by Pd dispersion and proximity to ZnO/ZnO x , was shown to determine the methanol production from CO 2 directly or via a CO intermediate. The Pd/Zn catalyst showed superior selectivity due to the in-situ formation of intermetallic species facilitated by Pd-ZnO x interactions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

A Monte-Carlo method for Coulomb collisions in moderately coupled plasmas

Here, we introduce a new approach for coupling cumulative and single-scattering distributions, extending Monte Carlo methods for Coulomb collisions in weakly coupled plasmas into the moderately coupled regime. Using time steps typical of cumulative-scattering methods, we show that this approach not only recovers leading-order corrections proportional to the inverse Coulomb logarithm but also solves the full Boltzmann collision integral. A key feature of the method is its simplicity: it can be integrated into cumulative-scattering methods for weakly coupled plasmas with only ~ 10 additional lines of code.

Boltzmann collision integral↗
DOE OSTI2022

Gamma-ray spectroscopy using angular distribution of Compton scattering

We present a novel method to perform gamma-ray spectroscopy suitable for extreme pileup situations, wherein traditional pulse-height spectroscopy is infeasible or impractical. One example of such application is spectral characterization of intense laser-based radiation sources that employ laser wakefield acceleration, in which a large number of photons can be produced in a well-collimated beam on a picosecond or shorter timescale. The method relies on the angularly-resolved calorimetric measurement of Compton scattering. The probability for a photon to undergo Compton scattering into a given solid angle is a function of its energy and is described by the Klein–Nishina formula. By first Compton scattering the incident beam and then measuring the total energies deposited in multiple detectors placed at various scattering angles, the incident beam energy spectrum can be reconstructed without resolving the individual events in detectors. Here, we compare the performance of an maximum likelihood expectation maximization algorithm with that of an artificial neural network to measure the energy spectrum of a beam of monoenergetic gamma rays from 137Cs decay. We demonstrate that the method can be used to reconstruct the characteristic inverse Compton scattering spectra.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2026

Regularization by denoising diffusion models for solving inverse PDE problems with application to full waveform inversion

Partial differential equation (PDE)-governed inverse problems are fundamental across various scientific and engineering applications; yet they face significant challenges due to nonlinearity, ill-posedness, and sensitivity to noise. Here, we introduce a computational framework, regularization by denoising using diffusion models for partial differential equations (RED-DiffEq), by integrating physics-driven inversion and data-driven learning. RED-DiffEq leverages pretrained diffusion models as a regularization mechanism for PDE-governed inverse problems. We apply RED-DiffEq to solve the full waveform inversion problem in geophysics, a challenging seismic imaging technique that seeks to reconstruct high-resolution subsurface velocity models from seismic measurement data. Our method shows enhanced accuracy and robustness compared to benchmark methods. Additionally, it exhibits strong generalization and domain decomposition capacity, enabling the inversion of more complex velocity models with larger domains than those used in training the diffusion model. Our framework can also be directly applied to diverse PDE-governed inverse problems.

Shan, Siming [Yale University, New Haven, CT (Unit↗
DOE OSTI2025

Summary of the 5th IAEA technical meeting on fusion data processing, validation and analysis (FDPVA)

The purpose of the 5th International Atomic Energy Agency technical meeting on fusion data processing, validation and analysis (FDPVA) (Ghent University, Ghent, Belgium, 12–15 June 2023) was to provide a platform during which a set of topics relevant to FDPVA were discussed with the view of meeting the needs of next step fusion devices such as ITER. The validation and analysis of experimental data obtained from diagnostics used to characterize fusion plasmas are crucial for a knowledge-based understanding of the physical processes governing the dynamics of these plasmas. This paper presents the recent progress and achievements in the domain of plasma diagnostics data analysis and synthetic diagnostics reported at the meeting, including concept description of new devices; fusion databases; integrated data analysis; inverse problems; uncertainty propagation, verification and validation; probabilistic methods and machine learning. The relevant results underline trends observed in the current major fusion confinement devices.

fusion databases↗