Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “few shot learning”

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 181 records · Page 10
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Neural networks for estimation of divertor conditions in DIII-D using C III imaging

Deep learning approaches have been applied to images of C III emission in the lower divertor of DIII-D to develop models for estimating the level of detachment and magnetic configuration (X-point location and strike point radial location). The poloidal distance from the target to the C III emission front is used to represent the level of detachment. The models perform well on a test dataset not used in training, achieving $F_1$ scores as high as 0.99 for detachment state classification and root mean squared error (RMSE) as low as 2cm for front location regression. Predictions for shots with intermittent reattachment are studied, with class activation mapping used to aid in interpretation of the model predictions. Based on the success of these models, a third model was trained to predict the X-point location and strike point radial position from C III images. Though the dataset covers only a small range of possible magnetic configurations, the model shows promising results, achieving RMSE around 1cm for the test data.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Improving the efficiency of learning-based error mitigation

Error mitigation will play an important role in practical applications of near-term noisy quantum computers. Current error mitigation methods typically concentrate on correction quality at the expense of frugality (as measured by the number of additional calls to quantum hardware). To fill the need for highly accurate, yet inexpensive techniques, we introduce an error mitigation scheme that builds on Clifford data regression (CDR). The scheme improves the frugality by carefully choosing the training data and exploiting the symmetries of the problem. We test our approach by correcting long range correlators of the ground state of XY Hamiltonian on IBM Toronto quantum computer. We find that our method is an order of magnitude cheaper while maintaining the same accuracy as the original CDR approach. The efficiency gain enables us to obtain a factor of 10 improvement on the unmitigated results with the total budget as small as 2 ⋅ 10 5 shots. Furthermore, we demonstrate orders of magnitude improvements in frugality for mitigation of energy of the LiH ground state simulated with IBM's Ourense-derived noise model.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2004

U.S. Centennial of Flight Commision: Born of Dreams - Inspired by Freedom

The U.S. Centennial of Flight Commission developed and maintained a public web site that included activities related to the centennial of flight celebration and the history of aviation. The web site, www.centennialofflight.gov, was continually updated with educational and historical information, events, sights and sounds, and Commission information from its inception to June 2004. This DVD contains a 'snap shot' of the web site as of April 2004. The Web site on this DVD can be enjoyed without an Internet connection although in some places, you will be given links to online content. DVD content includes: 1) About the Commission - Information on the legislation, the Commissioners and Advisory Board members, news, the National Plans, meeting minutes and status reports; 2) Calendar of Events - A comprehensive list of activities, symposiums, exhibits, air shows, educational activities and more that took place through March 2004; 3) Wright Brothers History - The Library of Congress bibliography of Wright-related resources as well as the Chronology and Flight Log; the Brunsman articles; interactive learning modules from The Wright Experience; short informative essays and a series of links to other Wright brothers information sources. 4) History of Flight - Essays and images on the history of flight; 5) Sights and Sounds - Images, movies and special collections that capture the accomplishments of the Wright brothers and others who made significant contributions throughout the history of aviation and aerospace. As part of the NASA Art Program, a centennial song, 'Way Up There,' was commissioned; 6) Licensed Products - View collections of souvenirs and gift items to commemorate the 100th anniversary of the first powered flight; 7) Education - Resources that will help educators and their students celebrate 100 years of flight. Teachers can download Wright brothers posters and a Centennial of Flight bookmark, view live Web casts, and access an Educational Resources Center Matrix representing more than 50 government, industry and labor organizations promoting aviation and aerospace education.

Source record↗
DOE OSTI2021

Turn-key constrained parameter space exploration for particle accelerators using Bayesian active learning

Abstract Particle accelerators are invaluable discovery engines in the chemical, biological and physical sciences. Characterization of the accelerated beam response to accelerator input parameters is often the first step when conducting accelerator-based experiments. Currently used techniques for characterization, such as grid-like parameter sampling scans, become impractical when extended to higher dimensional input spaces, when complicated measurement constraints are present, or prior information known about the beam response is scarce. Here in this work, we describe an adaptation of the popular Bayesian optimization algorithm, which enables a turn-key exploration of input parameter spaces. Our algorithm replaces the need for parameter scans while minimizing prior information needed about the measurement’s behavior and associated measurement constraints. We experimentally demonstrate that our algorithm autonomously conducts an adaptive, multi-parameter exploration of input parameter space, potentially orders of magnitude faster than conventional grid-like parameter scans, while making highly constrained, single-shot beam phase-space measurements and accounts for costs associated with changing input parameters. In addition to applications in accelerator-based scientific experiments, this algorithm addresses challenges shared by many scientific disciplines, and is thus applicable to autonomously conducting experiments over a broad range of research topics.

43 PARTICLE ACCELERATORS↗
DOE OSTI2023

Investigating boosted decision trees as a guide for inertial confinement fusion design

Inertial confined fusion experiments at the National Ignition Facility have recently entered a new regime approaching ignition. Improved modeling and exploration of the experimental parameter space were essential to deepening our understanding of the mechanisms that degrade and amplify the neutron yield. The growing prevalence of machine learning in fusion studies opens a new avenue for investigation. Here in this paper, we have applied the Gradient-Boosted Decision Tree machine-learning architecture to further explore the parameter space and find correlations with the neutron yield, a key performance indicator. We find reasonable agreement between the measured and predicted yield, with a mean absolute percentage error on a randomly assigned test set of 35.5%. This model finds the characteristics of the laser pulse to be the most influential in prediction, as well as the hohlraum laser entrance hole diameter and an enhanced capsule fabrication technique. We used the trained model to scan over the design space of experiments from three different campaigns to evaluate the potential of this technique to provide design changes that could improve the resulting neutron yield. While these data-driven model cannot predict ignition without examples of ignited shots in the training set, it can be used to indicate that an unseen shot design will at least be in the upper range of previously observed neutron yields.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Rapid wavefield forecasting for earthquake early warning via deep sequence to sequence learning

We propose a deep learning model, WaveCastNet, to forecast high-dimensional wavefields. WaveCastNet integrates a convolutional long expressive memory architecture into a sequence-to-sequence forecasting framework, enabling it to model long-term dependencies and multiscale patterns in both space and time. By sharing weights across spatial and temporal dimensions, WaveCastNet requires significantly fewer parameters than more resource-intensive models such as transformers, resulting in faster inference times. Crucially, WaveCastNet also generalizes better than transformers to rare and critical seismic scenarios, such as high-magnitude earthquakes. Here, we show the ability of the model to predict the intensity and timing of destructive ground motions in real time, using simulated data from the San Francisco Bay Area. Furthermore, we demonstrate its zero-shot capabilities by evaluating WaveCastNet on real earthquake data. Our approach does not require estimating earthquake magnitudes and epicenters, steps that are prone to error in conventional methods, nor does it rely on empirical ground-motion models, which often fail to capture strongly heterogeneous wave propagation effects.

Geophysics↗
DOE OSTI2022

Ambient-temperature liquid jet targets for high-repetition-rate HED discovery science

High-power lasers can generate energetic particle beams and astrophysically relevant pressure and temperature states in the high-energy-density (HED) regime. Recently-commissioned high-repetition-rate (HRR) laser drivers are capable of producing these conditions at rates exceeding 1 Hz. However, experimental output from these systems is often limited by the difficulty of designing targets that match these repetition rates. To overcome this challenge, we have developed tungsten microfluidic nozzles, which produce a continuously replenishing jet that operates at flow speeds of approximately 10 m/s and can sustain shot frequencies up to 1 kHz. The ambient-temperature planar liquid jets produced by these nozzles can have thicknesses ranging from hundreds of nanometers to tens of micrometers. In this work, we illustrate the operational principle of the microfluidic nozzle and describe its implementation in a vacuum environment. Further, we provide evidence of successful laser-driven ion acceleration using this target and discuss the prospect of optimizing the ion acceleration performance through an in situ jet thickness scan. Future applications for the jet throughout HED science include shock compression and studies of strongly heated nonequilibrium plasmas. When fielded in concert with HRR-compatible laser, diagnostic, and active feedback technology, this target will facilitate advanced automated studies in HRR HED science, including machine learning-based optimization and high-dimensional statistical analysis.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

Forming a database to study reversed magnetic shear from the National Spherical Torus eXperiment using machine learning

Achieving a long-lived reversed magnetic shear (RMS) target plasma in the National Spherical Torus eXperiment Upgrade will require developing various sustainment scenarios. To help with the ongoing plasma control efforts, the development of a new analysis for the motional Stark effect (MSE) diagnostic using a machine learning algorithm, namely, MSE-ML, is described. MSE-ML will be used to identify patterns during RMS discharges, some of which suffer magnetohydrodynamic (MHD) events resulting in current redistribution and monotonic q-profiles. A database consisting of q and magnetic shear profiles is being constructed primarily based on the existing National Spherical Torus eXperiment data with equilibrium reconstructions constrained by the magnetic field pitch angle profile measured using the multi-channel MSE diagnostic. An unsupervised k-means clustering of the data is developed to study the RMS formation as a function of time. The initial clustering from the q-profiles shows significant differences in both amplitude and the duration of the RMS period. As a goal, the clustering results that detect and distinguish shots with substantial and sustained RMS are to be used as a preprocessing step in a supervised algorithm to identify the underlying conditions that lead to long-lasting improved confinement with RMS. Another aim of the MSE-ML study is to identify precursors of RMS-destroying MHD events in either derived data such as the q-profile or directly measured data such as the magnetic field pitch angle profile.

Uzun-Kaymak, I. U. (ORCID:0000000276251493)↗
DOE OSTI2022

Optimism is not a strategy: A white paper on how to give IFE a fighting chance to be real

With NIF shot N210808, we now have an existence proof of ignition (i.e. Lawson-like criteria exceeded and capsule gain well exceeding unity) in the laboratory and it has generated renewed interest in IFE. However, it is important to recognize that ignition on the NIF has been much more difficult than what was originally envisioned. Moreover, the design for the target that actually obtained burning plasma (Kritcher, Young, Robey, et al., Nature Phys. 2022; Zylstra, Hurricane, Callahan, et al., Nature, 601, 542, 2022) and ignition conditions is much different than the high gain design originally planned in the National Ignition Campaign (NIC; e.g. Lindl, Phys. Plasmas, 2, 3933, 1995; Lindl, Amendt, Berger, et al., Phys. Plasmas, 11, 339, 2004). In order to avoid squandering time and resources, the IFE community must learn the lessons of what happened on the NIF over the past decade.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2023

Combined Coherent Manipulation and Single-Shot Measurement of an Electron Spin in a Quantum Dot (Final Technical Report)

Semiconductor quantum dots (QDs) are promising candidates to act as single-photon sources and/or quantum bits in future optical quantum information applications. Their excellent optical properties such as high brightness, single-photon purity, and narrow linewidth have potential utility in many areas. One challenge is to control the energy levels of the QD without using a magnetic field. The AC Stark effect offers the opportunity to do this. We have demonstrated record-large AC Stark shifts of the energy states of a single charged QD. We showed that the shifts can be applied in a spin-selective manner by controlling the polarization of the laser producing the AC Stark effect. In order to characterize the effect, we developed a novel spectral filtering scheme to discriminate the high-power AC Stark laser from the QD fluorescence. We also developed a compact, low-cost, homemade polarimeter to help control the polarization of the AC Stark laser, which enabled the spin selectivity of the AC Stark effect. With the practical capabilities thus developed, we learned that the spin-selective AC Stark effect causes electron spin pumping, which in turn causes nuclear spin pumping via the hyperfine interaction. The nuclear spin pumping causes a mean field Zeeman interaction between the nuclear spin ensemble and the electron trapped in the QD, resulting in the so-called Overhauser shift. The magnitude of the Overhauser shift and the measured linewidth of the QD’s optical transitions enabled characterization of the mean nuclear spin polarization and fluctuations. The potentially rapid (ns-scale) control of the QD energy levels and nuclear spin polarization will enable measurements of electron and nuclear spin polarization in the absence of any real magnetic field, which is a completely unique capability. The ability to rapidly apply a spin-selective AC Stark effect in the presence of a weak real magnetic field will allow control of the polarization selection rules of the transitions, enabling all single-qubit operations on the electron spin, i.e., initialization, coherent manipulation, and quantum non-demolition measurement.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2024

GOLEM: GOld standard for Learning and Evaluation of Motifs

Motifs are distinctive, recurring, widely used idiom-like words or phrases, often originating from folklore, whose meaning is anchored in a narrative and have a significance as communicative devices across a wide range of media, including news, literature, and propaganda. Many motifs concisely imply a large constellation of culturally relevant information, and their broad usage suggests their cognitive importance as touchstones of cultural knowledge. As such, their detection is a step towards culturally aware natural language processing. We present GOLEM (GOld standard for Learning and Evaluation of Motifs) a dataset of English news articles, opinion pieces, and broadcast transcripts annotated for motific information. The dataset identifies 25,737 motif candidates across 34 motif types drawn from three cultural or national groups: Jewish, Irish, and Puerto Rican. The dataset contains 2,024,141 words split into 25,737 text snippets drawn from 8,073 articles. Each motif candidate is labeled according to a scheme which identifies the type of usage (motific, referential, eponymic, or unrelated), resulting in 1,743 actual motific instances in the data. Annotation was performed by individuals identifying as members of each group and achieved a Fleiss’ kappa (?) of > 0.55. In addition to the data, we demonstrate that classification of the candidate type is a challenging task for Large Language Models (LLMs) using a few-shot approach; recent models such as T5, FLAN-T5, GPT-2, and Llama 2 (7B) achieved a performance of 41% accuracy at best, where the majority class accuracy is 41% and the average chance accuracy is 27%. These data will support development of new models and approaches for detecting (and reasoning about) motific information in text.

motif, culture, natural language, artificial intel↗
NASA NTRS2022

Moon to Mars (M2M) Habitation Considerations: A Snap Shot As of January 2022

The following NASA Technical Memorandum (TM) is intended to provide a snapshot in time of NASA’s current considerations (ground rules and assumptions, functional allocations, logistics) for habitation systems for the lunar surface (non-roving) and Mars transits. As NASA continues to refine the reference designs to meet the needs of an evolving architecture, it is expected that this information will also be updated as a result. Where appropriate, relevant publicly released documents will be referenced to provide further detail. NASA’s human lunar exploration plan under the Artemis program calls for achieving the goal of sending the first woman and first person of color to the surface of the Moon in the mid- 2020s and working toward sustainable exploration by the end of the decade. Working with both commercial and international partners, NASA will establish a permanent human presence on the Moon to uncover new scientific discoveries and lay the foundation for private companies to build a lunar economy. Longer duration missions on the lunar surface and in lunar orbit will also serve as a test bed for technologies to support future Mars exploration campaigns. The agency will use what we learn on the Moon to prepare for humanity's next giant leap – sending astronauts to Mars. NASA intends to establish a sustained lunar presence with the development of the Artemis Base Camp to prove technologies and capabilities that will one day enable humans to live and work on Mars, beginning with core elements including the Lunar Terrain Vehicle (LTV), the Pressurized Rover (PR), the lunar Surface Habitat (SH), power systems, and in-situ resource utilization (ISRU) systems. For in space operations and eventual transport of humans to Mars, NASA will utilize a Mars Transit Habitat (TH). Following deployment, the TH will complete a series of longer duration missions and shakedown testing while docked at Gateway, leveraging Gateway’s habitation redundancy for safety measures. Proposed Gateway-TH missions will far exceed the longest duration cislunar human missions to date. They will be the first operational readiness tests of our long-duration deep space systems, and of the split crew (two crew on the surface, two crew in space) operations that are vital to the approach for the first human Mars mission. Both the SH and TH are major architectural elements of NASA’s Moon to Mars (M2M) approach, each with very different concepts of operation. The SH is intended for use on the lunar surface as a home for astronauts, surface operations base, science facility, hub for communications, extravehicular activity (EVA) equipment repair site, waste processing facility, and supply hub. It serves as an enabler for a sustained surface presence and preparation for partial gravity operations during Mars missions. The SH will be designed to be self-sufficient for operations on the lunar surface. The SH will independently provide several functions, including its own power generation, energy storage capability, sleep quarters, hygiene areas, work areas, and dining areas. It will be capable of communicating with surface assets, orbital assets, and directly with Earth ground stations. It is planned to operate with two crew in the habitat for ~28- day stays with crew swap-outs in which the PR crew of two trades places with the habitat crew. During the swap-out, the habitat will nominally support four crew for a short period of time. For contingency scenarios, the habitat must also be capable of supporting four crew for up to 7 days. The TH will be designed to be capable of up to ~1,200-day Mars missions with the ability to carry all food and supplies needed to support a crew of four for this duration. An assumed Mars mission profile for the TH is to carry crew and supplies for ~850-day roundtrips between Earth and Mars orbit that allows 30-day stays on the Martian surface. To test the systems for this long journey, the TH will be used to extend the duration of missions at Gateway, enabling the orbiting outpost to be used as a Mars analog. These analog missions will be accomplished by attaching the TH to Gateway and conducting lunar surface operations from the TH. The TH may also need to perform free-flying shakedown missions to test out all systems prior to leaving for Mars. The habitat provides many critical functions including: a contingency airlock, crew quarters, galley, hygiene areas, safe haven capability, and science equipment. It can receive docked items from either axial end or on a radial port.

Habitat↗
DOE OSTI2020

Data Federation Challenges in Remote Near-Real-Time Fusion Experiment Data Processing

Fusion energy experiments and simulations provide critical information needed to plan future fusion reactors. As next-generation devices like ITER move toward long-pulse experiments, analyses, including AI and ML, should be performed in a wide range of time and computing constraints, from near-real-time constraints, between-shot analysis, and to campaign-wide long-term analysis. However, the data volume, velocity, and variety make it extremely challenging for analyses using only local computational resources. Researchers need the ability to compose and execute workflows spanning edge resources to large-scale high-performance computing facilities.We present Delta, a system to address data analysis challenges, including AI/ML, in fusion science, by leveraging the ADIOS I/O library and middleware, to support executing science workflows over the wide area network for near-real-time streaming. We discuss the data federation challenges in performing remote workflows, focusing on on-going research work in (1) managing, reducing, and streaming data to minimize I/O and data movement overheads, (2) decompressing and reorganizing data for analysis, and (3) executing workflows for automated data analysis. We introduce examples for deep-learning based data analysis for the fusion domain and demonstrate how we use Delta to construct end-to-end workflows for a fusion device in Korea, connecting a remote DOE facility in the USA. The capability demonstrated by this project is the basis for improving the state of the art for near-real-time data federation amongst remote facilities.

Choi, Jong Youl↗
DOE OSTI2024

Risk-Aware Framework Development for Disruption Prediction: Alcator C-Mod and DIII-D Survival Analysis

Abstract Survival regression models can achieve longer warning times at similar receiver operating characteristic performance than previously investigated models. Survival regression models are also shown to predict the time until a disruption will occur with lower error than other predictors. Time-to-event predictions from time-series data can be obtained with a survival analysis statistical framework, and there have been many tools developed for this task which we aim to apply to disruption prediction. Using the open-source Auton-Survival package we have implemented disruption predictors with the survival regression models Cox Proportional Hazards, Deep Cox Proportional Hazards, and Deep Survival Machines. To compare with previous work, we also include predictors using a Random Forest binary classifier, and a conditional Kaplan-Meier formalism. We benchmarked the performance of these five predictors using experimental data from the Alcator C-Mod and DIII-D tokamaks by simulating alarms on each individual shot. We find that developing machine-relevant metrics to evaluate models is an important area for future work. While this study finds cases where disruptive conditions are not predicted, there are instances where the desired outcome is produced. Giving the plasma control system the expected time-to-disruption will allow it to determine the optimal actuator response in real time to minimize risk of damage to the device.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2023

Advanced data analysis in inertial confinement fusion and high energy density physics

Bayesian analysis enables flexible and rigorous definition of statistical model assumptions with well-characterized propagation of uncertainties and resulting inferences for single-shot, repeated, or even cross-platform data. This approach has a strong history of application to a variety of problems in physical sciences ranging from inference of particle mass from multi-source high-energy particle data to analysis of black-hole characteristics from gravitational wave observations. The recent adoption of Bayesian statistics for analysis and design of high-energy density physics (HEDP) and inertial confinement fusion (ICF) experiments has provided invaluable gains in expert understanding and experiment performance. In this Review, we discuss the basic theory and practical application of the Bayesian statistics framework. We highlight a variety of studies from the HEDP and ICF literature, demonstrating the power of this technique. Due to the computational complexity of multi-physics models needed to analyze HEDP and ICF experiments, Bayesian inference is often not computationally tractable. Two sections are devoted to a review of statistical approximations, efficient inference algorithms, and data-driven methods, such as deep-learning and dimensionality reduction, which play a significant role in enabling use of the Bayesian framework. We provide additional discussion of various applications of Bayesian and machine learning methods that appear to be sparse in the HEDP and ICF literature constituting possible next steps for the community. We conclude by highlighting community needs, the resolution of which will improve trust in data-driven methods that have proven critical for accelerating the design and discovery cycle in many application areas.

47 OTHER INSTRUMENTATION↗
DOE OSTI2025

Crack Identification and Characterization in Deformed Nb3Sn Rutherford Cable Stacks Using Machine Learning

An investigation of instance segmentation of cracks in Nb3Sn 4-stack 40-strand Rutherford cables using machine learning is presented. Three samples were uniaxially and biaxially loaded before metallographic inspections were performed. The Mask R-CNN model was used in the Detectron2 framework with pre-trained weights but fine-tuned to detect and segment cracks. The model detected cracks with bounding box and mask average precisions (AP) of 42.8 and 27.9, respectively, and was used for instance segmentation of all cracks in the three samples. More cracks were found in the sample pre-loaded along the z-axis (i.e., along the cable length). Pre-loading along the x-axis (i.e., on the cables edges) reduced the number of cracks and changed the crack orientation distribution, away from being highly aligned with the y-axis (i.e., normal to the cables broad faces), i.e., the direction with the highest applied load. Fine-tuning of the Segment Anything Model (SAM) was also studied but performed poorly without human-provided prompts. However, the zero-shot capability of SAM showed high promises to accelerate the image annotation process for applications beyond this study.

Croteau, Jean-Francois↗
DOE OSTI2024

Using convolutional neural networks to detect edge localized modes in DIII-D from Doppler backscattering measurements

In H-mode tokamak plasmas, the plasma is sometimes ejected beyond the edge transport barrier. These events are known as edge localized modes (ELMs). ELMs cause a loss of energy and damage the vessel walls. Understanding the physics of ELMs, and by extension, how to detect and mitigate them, is an important challenge. In this paper, we focus on two diagnostic methods—deuterium-alpha (D α ) spectroscopy and Doppler backscattering (DBS). The former detects ELMs by measuring Balmer alpha emission, while the latter uses microwave radiation to probe the plasma. DBS has the advantages of having a higher temporal resolution and robustness to damage. These advantages of DBS diagnostic may be beneficial for future operational tokamaks, and thus, data processing techniques for DBS should be developed in preparation. In sight of this, we explore the training of neural networks to detect ELMs from DBS data, using D α data as the ground truth. With shots found in the DIII-D database, the model is trained to classify each time step based on the occurrence of an ELM event. The results are promising. When tested on shots similar to those used for training, the model is capable of consistently achieving a high f1-score of 0.93. Furthermore, this score is a performance metric for imbalanced datasets that ranges between 0 and 1. We evaluate the performance of our neural network on a variety of ELMs in different high confinement regimes (grassy ELM, RMP mitigated, and wide-pedestal), finding broad applicability. Beyond ELMs, our work demonstrates the wider feasibility of applying neural networks to data from DBS diagnostic.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2026

3D Deep Learning Joint Inversion of Active Seismic Full Waveform and Passive Seismic Traveltime Data for Reservoir Imaging and Uncertainty Quantification

Here, we present deep learning (DL) networks for three-dimensional (3D) joint inversion of active seismic full waveform and passive seismic traveltime data to image reservoirs and their properties and quantify imaging uncertainties. Active seismic full-waveform data can provide high-resolution monitoring images but are collected only intermittently because of their high acquisition cost. In contrast, passive seismic data can be gathered at relatively low cost between regular active surveys, although their imaging quality can be compromised by factors such as low signal-to-noise ratios and limited ray coverage of the target. Although these datasets are routinely acquired together at CO 2 storage sites, their combined inversion within a 3D DL framework has not been previously demonstrated. To our knowledge, this is the first study to address this gap, combining the strength of both data types. For efficient data storage and DL training with large 3D seismic datasets, we use a 3D data matrix in which a random number of passive seismic traveltime data are stored as parabolic envelopes using one-hot encoding and a 3D full-waveform data matrix in which multiple shot gathers are summed. Two network architectures are evaluated: a single-encoder U-Net for single-data type inversion and a dual-encoder U-Net for joint inversion of active and passive seismic data. We also evaluate the single-encoder U-Net for joint inversion by concatenating full-waveform data and traveltime data. We propose a systematic approach for selecting an optimal dropout rate that balances regularization during training and Monte Carlo dropout-based uncertainty quantification during prediction by examining the correlation coefficient between standard deviation and prediction error, along with the training misfit, across a range of dropout rates. 3D DL inversion experiments include five different network configurations, with evaluations under ideal, noisy and dropout-enabled conditions. Both model and data uncertainties are assessed, as well as their combined effects. Across all conditions, the networks consistently predict accurate CO 2 saturation models with low prediction errors, such as a structural similarity index of 0.993 and CO 2 difference of 1.1%. Uncertainty estimates show strong spatial correlation with prediction errors, confirming the effectiveness of the proposed dropout selection approach. The results demonstrate that our DL approach, utilizing compact data representations and appropriate uncertainty quantification, yields accurate subsurface images under various inversion conditions and provides valuable insights into the reliability of predictions.

Um, Evan Schankee [Lawrence Berkeley National Labo↗