Search NASA⌕ Search

SEARCH · Search NASA

Results for “Operator 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 1,171 records · Page 65
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Autoencoder Based Analysis of RF Parameters in the Fermilab Low Energy Linac

Machine learning (ML) has the potential for significant impact on the modeling, operation, and control of particle accelerators due to its ability to model nonlinear behavior, interpolate on complicated surfaces, and adapt to system changes over time. Anomaly detection in particular has been highlighted as an area where ML can significantly impact the operation of accelerators. These algorithms work by identifying subtle behaviors of key variables prior to negative events. Efforts to apply ML to anomaly detection have largely focused on subsystems such as RF cavities, superconducting magnets, and losses in rings. However, dedicated efforts to understand how to apply ML for anomaly detection in linear accelerators have been limited. In this paper the use of autoencoders is explored to identify anomalous behavior in measured data from the Fermilab low-energy linear accelerator.

Edelen, Jonathan P. (ORCID:0000000215180652)↗
DOE OSTI2024

Quantifying wildfire drivers and predictability in boreal peatlands using a two-step error-correcting machine learning framework in TeFire v1.0

Abstract. Wildfires are becoming an increasing challenge to the sustainability of boreal peatland (BP) ecosystems and can alter the stability of boreal carbon storage. However, predicting the occurrence of rare and extreme BP fires proves to be challenging, and gaining a quantitative understanding of the factors, both natural and anthropogenic, inducing BP fires remains elusive. Here, we quantified the predictability of BP fires and their primary controlling factors from 1997 to 2015 using a two-step correcting machine learning (ML) framework that combines multiple ML classifiers, regression models, and an error-correcting technique. We found that (1) the adopted oversampling algorithm effectively addressed the unbalanced data and improved the recall rate by 26.88 %–48.62 % when using multiple datasets, and the error-correcting technique tackled the overestimation of fire sizes during fire seasons; (2) nonparametric models outperformed parametric models in predicting fire occurrences, and the random forest machine learning model performed the best, with the area under the receiver operating characteristic curve ranging from 0.83 to 0.93 across multiple fire datasets; and (3) four sets of factor-control simulations consistently indicated the dominant role of temperature, air dryness, and climate extreme (i.e., frost) for boreal peatland fires, overriding the effects of precipitation, wind speed, and human activities. Our findings demonstrate the efficiency and accuracy of ML techniques in predicting rare and extreme fire events and disentangle the primary factors determining BP fires, which are critical for predicting future fire risks under climate change.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2019

Muon Monitor Data To Maintain The Quality Of The NuMI Neutrino Beam at Fermilab

The NuMI-AIP has been proposed to upgrade the NuMI target system for a 1-MW operation. The facility produces an intense muon neutrino beam by using a high power pion production target and a focusing magnetic horns for long baseline neutrino experiment. A muon monitor which measures muon beam profile, is a key beam element to maintain the quality of muon neutrino beam. The system will be upgraded for the high beam power operation. We studied the response of muon monitors with the proton beam power, proton beam position on the target, and focusing strength of the horn. The results were evaluated for future operations. The response function needs to be modeled by using Machine Learning algorithms which will be implemented into a muon monitor signal analyzer. The new analyzer will forecast the neutrino beam quality and control the beam position on the target if needed. Upgrading the helium gas handling system is the other key to improve the stability of the muon monitor signal. We will present the status of the upgrade plan.

43 PARTICLE ACCELERATORS↗
DOE OSTI2015

Recovery from feeding unconditioned feed to part of the solvent extraction process in Thorp - 14430

A recent event in the Thorp Reprocessing Plant involved unconditioned feed being sent to one of the main process solvent extraction vessels. As the feed was unconditioned the fissile material was lost to the aqueous raffinate stream. The plant was tripped and shutdown in line with the baseline criticality safety case. This paper describes the recovery option that was taken to remove the Pu from the raffinate tank and safely consign it to the downstream effluent vessels. Key assumptions that were fundamental to maintaining criticality safety throughout these operations were clearly highlighted in the supporting criticality safety assessment. Key learning points that are of relevance to the wider criticality safety assessment community are pulled out at the end of this paper. (authors)

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2025

AI-Enabled Operations at Fermi Complex: Multivariate Time Series Prediction for Outage Prediction and Diagnosis

The Main Control Room of the Fermilab accelerator complex continuously gathers extensive time-series data from thousands of sensors monitoring the beam. However, unplanned events such as trips or voltage fluctuations often result in beam outages, causing operational downtime. This downtime not only consumes operator effort in diagnosing and addressing the issue but also leads to unnecessary energy consumption by idle machines awaiting beam restoration. The current threshold-based alarm system is reactive and faces challenges including frequent false alarms and inconsistent outage-cause labeling. To address these limitations, we propose an AI-enabled framework that leverages predictive analytics and automated labeling. Using data from $2,703$ Linac devices and $80$ operator-labeled outages, we evaluate state-of-the-art deep learning architectures, including recurrent, attention-based, and linear models, for beam outage prediction. Additionally, we assess a Random Forest-based labeling system for providing consistent, confidence-scored outage annotations. Our findings highlight the strengths and weaknesses of these architectures for beam outage prediction and identify critical gaps that must be addressed to fully harness AI for transitioning downtime handling from reactive to predictive, ultimately reducing downtime and improving decision-making in accelerator management.

Jain, Milan [PNL, Richland] (ORCID:000000021676111↗
DOE OSTI2022

Data-Driven Distribution System Coordinated PV Inverter Control Using Deep Reinforcement Learning

The deployment of distributed solar photovoltaic (PV) systems has increased consistently over the past decades. High penetrations of PVs could cause a series of adverse grid impacts, such as voltage violations. The recent development of smart inverter technologies rises the incentives of developing PV control solutions that regulate the inverter output power and seeking the optimization on system operational objectives. This paper proposes a data-driven control solution based on deep reinforcement learning (DRL) to optimize PV inverters for voltage regulation. The proposed solution can minimize PV real power curtailment while maintaining network voltage at an acceptable range. Comparison results between the proposed DRL control algorithms with deep deterministic policy gradient (DDPG) and volt-var control on a real feeder in west Colorado highlight the advantage of the proposed framework in controlling the system voltage while minimizing the PV real power curtailment.

deep reinforcement learning↗
DOE OSTI2022

A Deep Reinforcement Learning-based Reserve Optimization in Active Distribution Systems for Tertiary Frequency Regulation

Federal Energy Regulatory Commission (FERC)Orders 841 and 2222 have recommended that distributed energy resources (DERs) should participate in energy and reserve markets; therefore, a mechanism needs to be developed to facilitate DERs’ participation at the distribution level. Although the available reserve from a single distribution system may not be sufficient for tertiary frequency regulation, stacked and coordinated contributions from several distribution systems can enable them participate in tertiary frequency regulation at scale. This paper proposes a deep reinforcement learning (DRL)-based approach for optimization of requested aggregated reserves by system operators among the clusters of DERs. The co-optimization of cost of reserve, distribution network loss, and voltage regulation of the feeders are considered while optimizing the reserves among participating DERs. The proposed framework adopts deep deterministic policy gradient (DDPG), which is an algorithm based on an actor-critic method. The effectiveness of the proposed method for allocating reserves among DERs is demonstrated through case studies on a modified IEEE 34-node distribution system.

deep reinforcement learning, distributed energy re↗
DOE OSTI2013

Hanford and Savannah River Site Programmatic and Technical Integration

Abstract only. The Hanford Site and the Savannah River Site (SRS) were the primary plutonium production facilities within the U.S. nuclear weapons complex. Radioactive wastes were generated as part of these missions and are stored in similar fashion. The majority of radioactivity maintained by the two sites is located in underground carbon steel tanks in the physical form of supernatant, saltcake, or sludge. Disposition of SRS tank waste is ongoing by converting it into glass (pathway for sludge and radionuclides separated from supernatant or dissolved saltcake) or cement (pathway for the decontaminated supernatant and dissolved saltcake). Tank closure activity has also begun at SRS and will continue for the duration of mission. The Hanford tank waste inventory is roughly 2/3rds larger than SRS's by volume- but nominally half the radioactivity. The baseline disposition path includes high-level and low-activity waste vitrification with separate disposition of contact-handled transuranic tank waste. Retrieval of tank waste from aging single­ shell tanks (SSTs) into double-shell tanks (DSTs) is currently ongoing. As vitrification commences later this decade, Hanford will be in a similar operations mode as SRS. Site integration is increasing as the missions align. The ongoing integration is centered on key issues that impact both sites- regardless of mission timeframe. Three recent workshop exchanges have been held to improve communication with the primary intent of improving operations and technical work organization. The topics of these workshops are as follows: DST space utilization, optimization, and closure; Waste Feed Qualification; and, Cementitious Waste Forms. Key goals for these and future exchanges include aligning research and technology, preparing for joint initiatives (to maximize budgetary value for the customer), and reviewing lessons learned. Each site has played a leading role in the development of technology and operational practices that can be used to meet current challenges and to minimize the impact of future challenges. This paper provides an overview of the exchanges held, but predominately focuses on the team development and actions leading from the workshops.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2018

The future of Sinbad: learning from the best practices of ICSBEP - 24622

Two projects sanctioned by the Organisation for Economic Co-operation and Development (OECD) Nuclear Energy Agency (NEA) have over two decades of experience established comprehensive data sets in handbooks supporting criticality safety and reactor physics. The International Criticality Safety Benchmark Evaluation Project (ICSBEP), and later the International Reactor Physics Experiment Evaluation Project (IRPhEP) that was patterned after the success of the ICSBEP, serves as an example of quality and excellence in preserving our nuclear data heritage and establishing benchmark standards upon which current and future modeling, validation, and safety efforts can be supported. Many of the best practices and lessons learned from these benchmark projects can be adopted and adapted for use with the Shielding Integral Benchmark Archive and Database (SINBAD) to prepare higher quality shielding benchmark data to effectively expand the availability of qualified experimental data available to the nuclear community. The unique experiments performed and aggregated within SINBAD represent characteristic measurements, techniques, and applications not captured with the specificity of nuclear criticality safety or traditional reactor physics applications. Application of evaluated SINBAD benchmark data would challenge and enhance our current nuclear data and modeling and simulation strategies. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2018

Development of 80- and 100- Mile Work Day Cycles Representative of Commercial Pickup and Delivery Operation

When developing and designing new technology for integrated vehicle systems deployment, standard cycles have long existed for chassis dynamometer testing and tuning of the powertrain. However, to this day with recent developments and advancements in plug-in hybrid and battery electric vehicle technology, no true 'work day' cycles exist with which to tune and measure energy storage control and thermal management systems. To address these issues and in support of development of a range-extended pickup and delivery Class 6 commercial vehicle, researchers at the National Renewable Energy Laboratory in collaboration with Cummins analyzed 78,000 days of operational data captured from more than 260 vehicles operating across the United States to characterize the typical daily performance requirements associated with Class 6 commercial pickup and delivery operation. In total, over 2.5 million miles of real-world vehicle operation were condensed into a pair of duty cycles, an 80-mile cycle and a 100-mile cycle representative of the daily operation of U.S. class 3-6 commercial pickup and delivery trucks. Using novel machine learning clustering methods combined with mileage-based weighting, these composite representative cycles correspond to 90th and 95th percentiles for daily vehicle miles traveled by the vehicles observed. In addition to including vehicle speed vs time drive cycles, in an effort to better represent the environmental factors encountered by pickup and delivery vehicles operating across the United States, a nationally representative grade profile and key status information were also appended to the speed vs. time profiles to produce a 'work day' cycle that captures the effects of vehicle dynamics, geography, and driver behavior which can be used for future design, development, and validation of technology.

100 mile↗
DOE OSTI2022

LANL Activities on Mechanistic Approach to Analyzing and Improving Unconventional Hydrocarbon Production

Hydrocarbon production from shale reservoirs is inherently inefficient and challenging since these are low permeability plays. In addition, there is a limited understanding of the fundamentals and the controlling mechanisms, further complicating how to optimize these plays. Herein, we summarize our experimental and computational efforts fully and partially supported by the fundamental shale portfolio to reveal unconventional shale fundamentals and devise development strategies to enhance extraction efficiency with a minimal environmental footprint. Integrating these fundamentals with machine learning, we outline a pathway to improve the predictive power of our models, which enhances the forecast quality of production, thereby improving the economics of operations in unconventional reservoirs. For instance, we have developed science informed workflows and platforms for optimizing pressure-drawdown at a site, which allow operators to make reservoir-management decisions that optimize recovery in consideration of future production. Recently, our work relies on the hybridization of physics-based prediction and machine learning, whereby accurate synthetic data (combined with available site data) can enable the application of machine learning methods for rapid forecasting and optimization. Consequently, the workflow and platform are readily extendable to operations at other sites, plays, and basins.

04 OIL SHALES AND TAR SANDS↗
DOE OSTI2023

Machine-learning-assisted automation of single-crystal neutron diffraction

Neutron scattering is a powerful but expensive technique to study materials and discover new matter. Advanced detector technology has significantly improved the efficiency of neutron experiments, increasing the complexity of neutron data reduction and analysis. Machine learning (ML) brings new directions for neutron diffraction data reduction and experiment operation. Here, this work presents an ML-assisted data reduction and analysis method for precise recognition of Bragg peaks and the corresponding regions of interest; it can then automatically screen and align a measured crystal using the recognized peaks, and subsequently plan and optimize the data collection with user-provided information and uncertainty quantification values of detected peaks. This method shows robust performance in different complex sample environments and enables automated single-crystal neutron diffraction.

47 OTHER INSTRUMENTATION↗
DOE OSTI2020

Multi-Agent Safe Policy Learning for Power Management of Networked Microgrids

This paper presents a supervised multi-agent safe policy learning (SMAS-PL) method for optimal power management of networked microgrids (MGs) in distribution systems. While unconstrained reinforcement learning (RL) algorithms are black-box decision models that could fail to satisfy grid operational constraints, our proposed method considers AC power flow equations and other operational limits. Accordingly, the training process employs the gradient information of operational constraints to ensure that the optimal control policy functions generate safe and feasible decisions. Furthermore, we have developed a distributed consensus-based optimization approach to train the agents’ policy functions while maintaining MGs’ privacy and data ownership boundaries. After training, the learned optimal policy functions can be safely used by the MGs to dispatch their local resources, without the need to solve a complex optimization problem from scratch. Lastly, numerical experiments have been devised to verify the performance of the proposed method.

, policy gradient↗
DOE OSTI2018

Texturizing Wind Turbine Towers to Reduce Bat Mortality

There is still a great need for the development of mitigation strategies that are economically viable and can be used alone or in combination with other options (e.g., operational minimization, acoustic deterrents) to ensure that bat fatalities are effectively reduced at wind facilities across the U.S. Our goal was to develop a wind turbine tower coating that 1) bats show little or no interest in approaching, 2) can be applied to currently deployed wind turbine towers and to towers as they are constructed, 3) is economically feasible to produce and apply, and 4) ultimately contributes to a reduction in bat mortality at utility-scale wind facilities. Wind turbine tower monopoles are large, smooth surfaces and numerous studies have reported bats making close investigative approaches at turbine tower surfaces that are suggestive of drinking and foraging behavior. We therefore hypothesized that such activity could be contributing to bat-wind turbine fatalities by increasing the amount of time that bats spend in or around the rotor-swept zone. We further hypothesized that a texturizing the tower might minimized bat activity at the towers and therefore reduce fatality risk. To explore this, we designed a series of experiments to characterize how bats interact with smooth and texture-treated surfaces in a controlled environment. We conducted 64 behavioral trials in a flight facility from 10 June to 25 September 2015 and an additional 76 behavioral trials from 9 June to 30 August 2016. The first set of trials showed that while bats did make contact with smooth surfaces, they did not make contact with textured surfaces. We ascertained that bat activity occurred at 0 to 1 m distances from surfaces, suggesting that the bats (used in this study) may need to be this close to a surface to accurately discern its texture. From the results were developed a texture coating. For the second set of trials, we tested the texture coating. Bats showed a >40% reduction in activity within 1 m of the textured surfaces in comparison to the smooth surfaces. Following these controlled experiments, we conducted field tests at Wolf Ridge Wind, LLC. We applied the texture coating to the midsection of turbine monopoles. The texture coating was manually applied to 2 turbine towers, although we noted during the application there was variation in texture coating between turbines. We considered that this variation could potentially impact bat behavior. We then conducted bat activity surveys at both texture-treated towers and 2 nearby control towers, over 55 nights from 24 June to 22 September, 2017. All species of bats known to be present at the site were detected in the acoustic recordings at these turbine towers. We found no significant differences in overall numbers of bats observed and levels of acoustic activity at the smooth and texture-treated towers. When we analyzed the experimental pairs separately, however, we found that a greater proportion of the close range bat activity within Pair 1 occurred at the texture-treated tower and that much of this activity could be attributed to foraging behavior. In contrast at turbine Pair 2, we observed no difference in the proportion of close range bat activity between the smooth and texture-treated tower, with no clear differences in behavior at the two tower types. When we examined acoustic bat activity at the tower pairs separately, we found species-specific differences in activity that differed between the smooth and texture-treated towers within the pairs. For example, hoary bat acoustic activity was significantly higher at the texture-treated tower in Pair 1, but significantly lower at the texture-treated tower in Pair 2. Given the observational nature of the field test, our limited sample size (2 turbine pairs), variation in the texture-coating between the 2 turbine pairs, and different patterns of bat activity observed at the 2 turbine pairs, inferences from this one field test are limited at best. In conclusion, the results from the flight facility experiments were promising (i.e., we observed a significant decrease in bat activity at smooth compared to texture-treated surfaces); however, due to coating installation and application challenges prior to our field test, we were unable to collect sufficient information to understand how these coatings affect bat activity at operational wind turbine towers. Although the results from the field test are inconclusive, the lessons learned will be informative for future research into turbine surface materials that can modify the surface of the turbine tower and therefore alter how bats perceive wind towers. Thus, the outcomes of this research project may have a significant impact on wind energy operators, wind developers, and OEMs by contributing to future impact mitigation strategies. In future research we will test new ideas to improve the development, application, and evaluation of this mitigation technique.

17 WIND ENERGY↗
DOE OSTI2025

RLGBS: Reinforcement Learning-Guided Beam Search for process optimization in a paper machine dryer section

Paper drying is responsible for over two-thirds of energy consumption in the U.S. pulp and paper industry, presenting significant potential for energy savings through optimization of process parameters. Current approaches often assume fixed operating conditions, neglecting dynamic ambient and process variations that limit achievable savings and real-world applicability. To this end, we develop a physics-based simulation environment for a paper machine dryer section and propose a reinforcement learning (RL) framework to minimize overall energy consumption by optimizing drying process parameters under diverse operating conditions. To mitigate overdrying and numerical instabilities caused by suboptimal local RL actions, we introduce Reinforcement Learning-Guided Beam Search (RLGBS), which explores multiple action sequences in parallel using beam search. Instead of making step-by-step decisions, RLGBS prioritizes solutions based on cumulative probability, reducing the impact of individual suboptimal actions. Experiments demonstrate that RLGBS achieves consistent energy savings under unseen operating conditions not encountered during training, outperforming conventional RL methods. While validated in drying optimization, this framework is broadly applicable to other RL-based industrial process control problems.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2021

Dynamic Process Intensification via Data-Driven Dynamic Optimization: Concept and Application to Ternary Distillation

Process intensification is a design philosophy aimed at making chemical processes safer and more efficient. Its implementation often results in significant modifications to the design and structure of the process, with several conventional unit operations occurring in the same physical device. Traditionally, process intensification has focused on steady-state operation. In our previous works, we introduced dynamic process intensification (DPI) as a new intensification paradigm based on operational changes for conventional or intensified units. DPI is predicated on switching operation between two auxiliary steady states selected via a steady-state optimization calculation that ensures that the system generates, on average and over time, the same products as in nominal steady-state operation, but with favorable economics. This paper extends the DPI concept and introduces a novel dynamic optimization-based DPI strategy (Do-DPI) that involves imposing a true cyclic operation rather than switching between two discrete states. We discuss its implementation using surrogate dynamic models learned via system identification. Here, an extensive case study concerning a ternary distillation column separating a canonical hydrocarbon mixture shows that Do-DPI can reduce energy use by more than 4% relative to steady-state operation, with no significant deviations in product quality and production rate.

42 ENGINEERING↗
DOE OSTI2023

Cycle-by-cycle Delay Estimation at Signalized Intersections by using Machine Learning and Simulated Video Detection Data

Accurate estimation of delay is crucial for efficient traffic signal operations. Estimation of delay in the real-time manner using traditional loop detectors requires advanced detectors (in addition to stop-bar detection). In cases when this detection layout is not in place, delay estimates are approximated with a lower accuracy. Video detection is one of the most frequently deployed detection systems at signalized intersections in recent years. In most cases video detection operates in the same way as traditional inductive loops. However, when coupled with computer vision algorithms, video detection systems could be used to retrieve additional information (e.g., vehicular arrivals and departures) that cannot be taken out from the conventional systems (e.g., long stop-bar loop detectors). Although present for several decades, video detection data were not frequently examined for delay estimation purposes. In this study, we proposed a novel delay estimation model which can be developed with only data from stop-bar video detectors. Relevant data were collected from a simulation model of 11 signalized intersections at downtown Chattanooga, TN and processed to create needed inputs for model development. With the use of multigene genetic programming the authors developed a delay model that outperforms accuracy of multi regression model. Furthermore, authors evaluated the developed model by comparison with the other benchmark delay models, such as HCM and approach delay model. It was found that the developed MGGP delay model outperforms benchmark models for a wide range of traffic and signal operation conditions.

Erdagi, Ismet Goksad↗
DOE OSTI2015

Can a radiological event impact a criticality safety program: Lessons learned at WIPP - 14350

Any accident at a nuclear facility impacts all departments involved in the operation of the facility. The radiological release that occurred on February 14, 2014 has halted all waste disposal operations at the Waste Isolation Pilot Plant (WIPP) and has resulted in detailed reviews of all safety management programs. This paper describes the initial reactions to the event, criticality issues with recovery operations, corrective actions for the criticality safety program from independent reviews, and the importance of configuration management to criticality safety. The nature, location, and timing of the event resulted in significant delays in determining the cause of the event. During this time multiple theories on the initiator of the event were postulated including and inadvertent criticality event. However, the timing of the event, average fissile gram equivalent (FGE) loading of waste containers compared to analyzed FGE loads, and isotopic analysis of fixed air samples taken during the event proved it was not an inadvertent criticality event. Recovery activities such as replacing contaminated high efficiency particulate (HEPA), filters and disposing of site generated waste have also been reviewed to prevent criticality safety issues. Post event corrective actions in all departments have impacted how future operations at WIPP will be conducted, including corrective actions for the criticality safety program. The event has also proven the importance of configuration management and the effect a change can have on a process. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗