Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “AUTOS”

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 1,387 records · Page 77

Front End Engineering Design Study on Gasification of Coal and Biomass to Generate Carbon- Free Electric Power and Hydrogen

A 2nd Phase Front End Engineering Design study of a gasification plant concept to co-produce electric power and hydrogen with net-negative CO2 emissions is being completed under the U.S. Department of Energy’s (DOE’s) 21st Century Power Plants initiative, whose goal is to advance innovative power plant concepts that are capable of flexible, net-zero carbon emission operations while producing cost-effective hydrogen to support economy-wide decarbonization goals. The proposed standalone plant would be constructed in Nebraska, USA. The specified design feedstock is a hybrid blend of Powder River Basin (PRB) subbituminous coal from Wyoming and local Nebraska biomass (corn stover), 50% each by weight (dry basis). Other potential feedstocks, including woody biomass (eastern red cedar) and waste plastic (auto shredder residue) were evaluated as alternates. The process block comprises a high-pressure, oxygen-blown fluidized bed gasifier coupled with water-gas shift, Selexol process for acid gas (H2S and CO2) removal, and pressure-swing adsorption (PSA) to yield 8,500 kg/h of high-purity hydrogen. Off-gas from the PSA unit is used in a gas turbine combined cycle plant (the power block) to supply 50 MWe net electric power to the grid. Overall thermal efficiency of the plant is 50% (HHV) with net atmospheric CO2 removal at a rate of 32 t/h. Design activities necessary to provide input to the current front-end engineering design (FEED) study, including, site selection, gasifier technology selection, investment case preparation, and the development of the Environmental Information Volume (EIV) for the host site, have been completed. These, as well as the current FEED activities, are described in this presentation.

gasification, biomass, coal, hydrogen, power gener↗

Safeguards Technology Development Program: FY 2024 Mid-Year Report

This project has 2 major thrusts and 2 minor tasks – One of the major ones is a scoping study for improving uranium isotopics by analyzing higher energy, lower yield U-235 gamma emissions. The second major one benefits past projects such as auto-enrichment, and peak-based analysis routines previously added to GADRAS-DRF [1]. The two minor tasks involve an in-person or remote GADRAS-DRF training for the IAEA, and characterizing detectors of interest.

07 ISOTOPE AND RADIATION SOURCES↗

Seismic Signal Detection on International Monitoring System 3-Component Stations using PhaseNet

In this report we discuss training a deep learning seismic signal detection model on 3-component stations from the International Monitoring System (IMS) using the PhaseNet architecture. Using 14 years of associated signals from the International Data Centre’s (IDC) Late Event Bulletin (LEB), we auto-curated training data consisting of signal windows containing associated arrivals, and noise windows that contain no LEB-associated signals. We trained several models using different waveform window durations (30 seconds and 100 seconds), with and without bandpass filtering. We evaluated the effectiveness of our models using associated signals from the Unconstrained Global Event Bulletin (UGEB) and found that several of our models outperformed the signal detections from the IDC’s Selected Event List 3 (SEL3) arrival table. The SEL3 bulletin evaluated on the UGEB dataset with 100-second waveform windows registered a precision and recall of .15 and .48, respectively, versus .19 and .59 for our filtered-data model. For the 30-second waveform window dataset, the SEL3 bulletin achieved a precision and recall of .31 and .47, respectively, versus .32 and .60 for our filtered-data model. Finally, our models detected signals from all source-to-receiver distances, suggesting it is feasible to use a single PhaseNet model for the IMS network.

58 GEOSCIENCES↗

Hardware-in-the-loop Testing of Network Protectors for Low-Voltage Networks with Distributed Energy Resources

In this report, we developed and validated a network protector relay digital twin model and interfaced a commonly used network protector relay hardware with our real-time simulation system. Hardware-in-the-loop protection studies are performed to assess the impact of distributed energy resources (DER) and benchmark a rate-of-change-based mitigation strategy. Simulation results suggest that the network protector reverse trip and auto-reclose functions are negatively impacted by the high distributed energy resource penetration. To accommodate DER backfeed while remaining secure and reliable for faults on primary feeders, we recommend options for a rate-of-change-based blocking scheme and a protection setting change. Finally, future mitigation ideas and standard revisions are discussed.

24 POWER TRANSMISSION AND DISTRIBUTION↗

GADRAS-DRF Enhancements for Safeguards – Custom Peak Fitting to Enhance Model Fitting and Isotopics

One major software update was accomplished within the Gamma Detector Response and Analysis Software-Detector Response Function (GADRAS-DRF) package. This update allows users to adjust individual peak fits for use in subsequent analysis processes within GADRAS-DRF. A graphical user interface (GUI) was implemented so users can see the effect of their fit adjustments on the spectrum. This new feature will enhance the capability of the previously funded auto-enrichment and peak-based 1D model fit feature that was implemented in FY23 (Fiscal Year). Isotopic analysis was performed using M400 data obtained from uranium standards and the isotopic assessment is given with and without manual peak adjustments.

07 ISOTOPE AND RADIATION SOURCES↗

Unifying Combinatorial and Graphical Methods in Artificial Intelligence

Recently, a new graph Laplacian, called the inner product Laplacian, was introduced which generalizes many existing Laplacians, including the normalized and combinatorial Laplacian and their weighted variants. The key observation behind the inner product Laplacian is that by defining appropriate inner product spaces on the vertices and edges, the standard Laplacians can be recovered as Hodge Laplacians over the simplicial complex formed by the edges and vertices. These inner product spaces form a natural way to incorporate non-combinatorial information into the definition of a domain-specific Laplacian. In particular, in contrast to current domain-specific weighting schemes which rely solely on edge weights, information regarding the similarity of non-adjacent vertices and arbitrary pairs of edges can be effectively incorporated into the Laplacian. In order to illustrate this approach we consider the problem of calculating the potential energy of an atomistic configuration using Graph Neural Networks. In comparison with start-of-the-art approaches, such as SchNet, our approach replaces a learned (via auto-encoder) representation of the atom types with an inner product space on atoms based on scientific knowledge (e.g., electronegativity). We will illustrate how this approach captures key chemical properties of the molecules and compare the energy calculations with state-of-the-art neural network approaches. However, to compute the resulting Laplacian involves a mixture of sparse and dense matrix computation and yields a dense matrix as the basis for the graph convolution. This dense convolutional kernel necessitates moving away from the standard message passing framework for graph neural networks and increases the computational cost of applying the kernel. In order to mitigate these costs we investigate means of leveraging the mixed sparse and dense computations to reduce the overall computational cost and how these approaches can be automatically transferred to energy efficient hardware (e.g., field programmable gate arrays (FPGAs)).

97 MATHEMATICS AND COMPUTING↗

Smart Pixels: In-pixel AI for on-sensor data filtering

We present a smart pixel prototype readout integrated circuit (ROIC) designed in CMOS 28 nm bulk process, with in-pixel implementation of an artificial intelligence (AI) / machine learning (ML) based data filtering algorithm designed as proof-of-principle for a Phase III upgrade at the Large Hadron Collider (LHC) pixel detector. The first version of the ROIC consists of two matrices of 256 smart pixels, each 25$\times$25 µm\textsuperscript{2} in size. Each pixel consists of a charge-sensitive preamplifier with leakage current compensation and three auto-zero comparators for a 2-bit flash-type ADC. The frontend is capable of synchronously digitizing the sensor charge within 25 ns. Measurement results show an equivalent noise charge (ENC) of $\sim$30e\textsuperscript{-} and a total dispersion of $\sim$100e\textsuperscript{-} The second version of the ROIC uses a fully connected two-layer neural network (NN) to process information from a cluster of 256 pixels to determine if the pattern corresponds to highly desirable high-momentum particle tracks for selection and readout. The digital NN is embedded in-between analog signal processing regions of the 256 pixels without increasing the pixel size and is implemented as fully combinatorial digital logic to minimize power consumption and eliminate clock distribution, and is active only in the presence of an input signal. The total power consumption of the neural network is $\sim$ 300 $\mu$W. The NN performs momentum classification based on the generated cluster patterns and even with a modest momentum threshold, it is capable of 54.4\% – 75.4\% total data rejection, opening the possibility of using the pixel information at 40MHz for the trigger. The total power consumption of analog and digital functions per pixel is $\sim$ 6 $\mu$W per pixel, which corresponds to $\sim$ 1 W/cm\textsuperscript{2} staying within the experimental constraints.

Parpillon, Benjamin↗

Online System ID for Predicting Power Plant Performance Throughout Cycling Operations

This presentation represents a review of the background research conducted by NETL to apply artificial intelligence, i.e. auto-recursive algorithms and data analytics to detect leaks in utility scale boilers and laboratory power systems. The new project being funded by the Advanced Sensors and Controls Program is part of the Field Work Proposal funded in EY21 as Task 53 to demonstrate the application of these techniques on a utility scale power system.

Shadle, Lawrence↗

Radiation-Hard Smart-Pixel Detector ASIC ReadOut with Digital AI in 28nm

Detectors at future high energy colliders will face enormous technical challenges. Disentangling the unprecedented numbers of particles expected in each event will require highly granular silicon pixel detectors with billions of readout channels. With event rates as high as 40 MHz, these detectors will generate petabytes of data per second. To enable discovery within strict bandwidth and latency constraints, future trackers must be capable of fast, power efficient, and radiation hard data-reduction at the source. This effort is pursuing the co-design development of high-performance readout smart pixel ASICs for a future Phase III High Luminosity upgrade of the Large Hadron Collider. A 1.6mm2 ASIC prototype was designed by Fermilab in CMOS 28 nm bulk process and submitted for manufacturing in February 2024. It leverages the analog front-end pixel design of a previous prototype fabricated and tested in 2023, which achieved a simulated detection level of ~400e- with 30fF input capacitance. The ROIC consists of two matrices of 16×16 smart pixels, each 25×25 μm2 in size. Each smart pixel contains a charge-sensitive preamplifier with leakage current compensation and three auto-zero comparators for a 2-bit flash-type ADC. There is digital space for the integration of our fully combinatorial AI that performs momentum classification at the bunch crossing rate. The total power consumption is ∼6μW per pixel, which corresponds to ~1mW/cm2. The ASIC incorporates programmable front-end charge injection circuitry to generate pixel cluster charges during characterization. The cluster profile will be generated to duplicate hit characteristics of various momentum (pT). We will present early results from chip testing.

Parpillon, Benjamin↗

Business Opportunities in Clean Energy Supply Chains: Guidebook for Small and Medium-Sized Automotive Suppliers

This is the final report for the referenced TCF award which ran from Oct. 2019 through Jan. 2024 in conjunction with the industry partner Swift Solar. The long duration of the program was due to several factors including the pandemic and a technical work stoppage due to changes in the scope. Additionally, Swift’s business focus shifted during the course of the program. As a result of these factors, this program underwent two distinct modifications that dramatically changed the work scope. The following report is organized into three separate sections, each of which addresses a distinct work scope. Those scope changes, what necessitated them, and any new tasks/milestones are given, briefly, at the top of each section. The work performed was research on the hybrid halide perovskites (HHP), both the absorber material itself as well as other material layers that are required in a full device stack. At the start of the period of performance, there were two distinct challenges that we sought to address: (1) a wide band gap absorber (>1.6 eV), and (2) a recombination layer for tandem devices that provided good sequestration of the individual cell absorber materials in their respective device layers. NREL had previously published, and patented, advancements in both of these areas and they were the focus of much of the work throughout the program. The auto manufacturing space is evolving quickly. Sales of various electrified vehicles (xEVs) continue to grow steadily, which has medium to long-term ramifications for the thousands of U.S. manufacturers and hundreds of thousands of workers that contribute to the domestic automotive supply chain. In tandem with this growth in xEV sales, domestic clean energy manufacturing is surging (both for the xEV supply chain and other technologies), with public incentives spurring billions in private sector investment. Moreover, the next 2–3 years will see new and expanded industrial facilities come online to make products like hydrogen electrolyzers, solar panels, batteries, advanced electronics—and will create new supply chain needs and business opportunities as they do. DOE’s Office of Manufacturing and Energy Supply Chains (MESC) and Argonne National Laboratory developed this report to help automotive manufacturers—alongside various business support partners— understand these trends and their response options, and to make the most of current federal assistance programs for manufacturers. These market changes may create opportunities for small- and medium-sized manufacturers (SMMs) to expand production, grow profit margins, and diversify their businesses. For instance, SMMs could take advantage of engineering and process design similarities with new end-use industries that might enable greater business growth and stability as markets continue to change.

14 SOLAR ENERGY↗

Wholly Sustainable, Cost-Effective Carbon Fiber-Nylon Compounds CRADA 592 (Final Report)

Carbon fiber composites have attracted considerable attention due to the potential for substantial mass savings, with many examples now implemented in the low-volume luxury car market. However, migration to higher volume applications has been hindered by: (a) high material cost, (b) high processing times, and (c) perception of low Sustainability. This project will address all three of these barriers: (a) carbon fiber material to replace aluminum in structural components at a cost penalty of no more than $5/Kg-saved (aka weight buy), (b) fitting into high-rate processes for automotive production like injection molding, and (c) end-to-end Sustainable material – based on post industrial waste carbon fiber and nylon 66 and ability to recycle end-of-life auto parts. The opportunity lies in combining DowAksa capabilities in carbon fiber manufacturing, resin chemistry intermediate production with the unique testing capabilities inherent within PNNL. The teams from PNNL and DowAksa held several meetings virtually and in-person in Michigan and at PNNL, including a lab tour at PNNL. Throughout, the teams discussed DowAksa material sources, commercially available recycled base materials, and preliminary material properties. The teams also engaged in multiple discussions and evaluations of potential automotive applications based on the ideas suggested by PNNL. The teams discussed several potential automotive applications in which recycled carbon fiber and recycled PA resin can be used. The PNNL team identified 38 cast aluminum components that can potentially be assessed for redesign using the DowAksa materials system. The PNNL team also identified 27 polyamide components. The teams discussed the lists and narrowed it down to a handful of applications that are exterior and interior to common vehicle architectures. The team also considered semi-structural and structural components and short-listed the highest potential candidates, such as cross-car-beam. The cross-car beam was considered to be highly suitable and potentially viable demonstration applications based on the properties of the materials as well as the weight savings potentials and the reduction in embodied energy by utilizing wholly sustainable materials, since both materials, carbon fiber and resin, were derived from recycled materials. The next step was to reach out to potential OEMs and/or Tier1s who were interested in exploring such technology for future applications. However, the project was terminated, and no further discussions or exchange of information took place. No new data were generated, including no IP and no publications.

36 MATERIALS SCIENCE↗

Zero-Power Wireless Infrared Digitizing Sensors for Large Scale Energy-Smart Farm

This project, funded by ARPA-E and led by Northeastern University, developed zeropower infrared digitizing sensors to optimize irrigation and enhance crop yields. Traditional water stress detection methods are costly and require frequent maintenance, limiting their effectiveness. Our research identified shortwave infrared (SWIR) transmittance as the most reliable indicator of plant water stress and developed plasmonically enhanced micromechanical photoswitches (PMPs) that operate with minimal power. The sensors offer low-cost, large-scale deployment, auto-calibration across different crops, and a 10-year battery life, significantly reducing maintenance costs. The system achieved 4x greater accuracy than conventional soil moisture sensors while ensuring economic feasibility. By enabling precision irrigation, this technology conserves water, enhances crop productivity, and lowers operational costs, making it a scalable solution for sustainable agriculture and global food security.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

20 kV Gallium Nitride pn Diode Electro-Magnetic Pulse Arrestor for Grid Reliability (Final Scientific/Technical Report)

This project developed vertical Gallium Nitride (GaN) pn diodes under two main thrusts: (1) A focus on relatively higher-voltage devices for use as fast EMP arrestors to protect the electric grid; and (2) A focus on a Foundry effort to establish the manufacturability of relatively lower voltage devices. For the first thrust, the aim was to develop devices that go into avalanche breakdown to clamp the voltage across sensitive grid equipment subject to voltage transients induced by electromagnetic pulses (EMPs). Devices with breakdown voltages exceeding 6.5 kV were achieved, and breakdown times shorter than 1 ns were demonstrated, which is sufficiently fast to protect against the fast component of an EMP-induced signal. Key challenges included the epitaxial growth of thick (50 µm or more), low-doped (low 10 15 cm -3 range) GaN layers comprising the drift regions of the diodes, as well as the design and fabrication of edge termination structures (step-etched junction termination extensions) to prevent premature breakdown. Midway through the project, an additional emphasis was put on large-area, high-current devices, and forward currents of approximately 400 A were achieved in composite devices towards the end of the project. Experimental and theoretical studies of impact ionization and avalanche ruggedness were also conducted. For the second thrust (the Foundry), the focus was primarily on 1.2-kV-class devices, although towards the end of the project outstanding results on 3.3-kV class devices were also achieved. The aim of the Foundry was to develop a high-yield, reliable, and economic vertical GaN pn diode process. The Foundry conducted characterization of incoming epitaxial material and correlated this information with the yield and performance of fully processed devices. Wafer maps of diode characteristics such as forward and reverse current as well as capacitance-voltage curves were measured using auto-probing. Several mask sets comprising different-area devices as well as multiple edge termination designs (implanted junction termination extensions, guard rings, and combinations thereof) were studied, and machine-learning-based approaches were utilized to analyze the data. Packaging and reliability efforts were also undertaken for the Foundry diodes, which are necessary for a viable commercial process.

20KV↗

20 kV Gallium Nitride pn Diode Electro-Magnetic Pulse Arrestor for Grid Reliability (Final Report)

This project developed vertical Gallium Nitride (GaN) pn diodes under two main thrusts: (1) A focus on relatively higher-voltage devices for use as fast EMP arrestors to protect the electric grid; and (2) A focus on a Foundry effort to establish the manufacturability of relatively lower-voltage devices. For the first thrust, the aim was to develop devices that go into avalanche breakdown to clamp the voltage across sensitive grid equipment subject to voltage transients induced by electromagnetic pulses (EMPs). Devices with breakdown voltages exceeding 6.5 kV were achieved, and breakdown times shorter than 1 ns were demonstrated, which is sufficiently fast to protect against the fast component of an EMP-induced signal. Key challenges included the epitaxial growth of thick (50 um or more), low-doped (low 10 15 cm -3 range) GaN layers comprising the drift regions of the diodes, as well as the design and fabrication of edge termination structures (step-etched junction termination extensions) to prevent premature breakdown. Midway through the project, an additional emphasis was put on large-area, high-current devices, and forward currents of approximately 400 A were achieved in composite devices towards the end of the project. Experimental and theoretical studies of impact ionization and avalanche ruggedness were also conducted. For the second thrust (the Foundry), the focus was primarily on 1.2-kV-class devices, although towards the end of the project outstanding results on 3.3-kV-class devices were also achieved. The aim of the Foundry was to develop a high-yield, reliable, and economic vertical GaN pn diode process. The Foundry conducted characterization of incoming epitaxial material and correlated this information with the yield and performance of fully processed devices. Wafer maps of diode characteristics such as forward and reverse current as well as capacitance-voltage curves were measured using auto-probing. Several mask sets comprising different-area devices as well as multiple edge termination designs (implanted junction termination extensions, guard rings, and combinations thereof) were studied, and machine-learning-based approaches were utilized to analyze the data. Packaging and reliability efforts were also undertaken for the Foundry diodes, which are necessary for a viable commercial process.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Model-based Hierarchical Reinforcement Learning for Improved Physical Security Design: A Prototype

Prior work in FY24 developed an adversarial AI agent aid in path analysis of physical protection systems. This agent, trained using a model-based reinforcement learning algorithm, was able to successfully learn the most vulnerable path in facilities. It was able to extend the current state of practice for physical protection design by exhibiting dynamic behavior based on current environmental conditions. Whereas PathTrace largely performs a static, graph-based analysis, the AI agent was able to make decisions based on relative position in the facility, current conditions (was the adversarial agnet discovered?), and proximity to secondary targets. The agent demonstrated some novel capabilities, but had limitations that need to be resolved before it can be used for production purposes. For example, the adversarial agent generalizes poorly and takes a relatively long time to train. Nonetheless, there is still considerable promise for developing the adversarial agent further in order to explore even richer, more dynamic behaviors (e.g., adversary motivations, environmental debris, and more). This work considers a complementary idea; development of a planning agent. The planning agent is envisioned as an auto-complete-like tool that can help accelerate security system design by human experts. The agent would respect existing barriers and sensors placed by a human expert while offering cost-effective suggestions (i.e., implicitly balancing effectiveness with cost) to improve the design. The goal is for this agent to be part of an expert’s toolbox, not to totally upend the current state-of-practice, or to displace human experts. The ultimate goal would be concurrent training of both the adversarial and planning agent together, to learn entirely through self-play. This would represent an entirely new way of performing system deign. We selected a hierarchical, model-based reinforcement learning algorithm to serve as the planning agent. This is an extension of concepts used in the prior FY24 adversarial agent work. There, we had a single agent acting an environment. Here, we have two different sub-agents (policies), working together, to form a complete agent. There is a manager policy, which can select abstract goals on slower time scales, and a worker, which performs primitive actions to reach goals selected by the manager. It is worth noting that this class of algorithm is challenging to work with. From our understanding, our work is one of the first successful uses of model-based reinforcement learning (MBRL) in nuclear energy1 , and likely the first hierarchical model-based reinforcement learning application in nuclear energy. Further, this work is one of the first known attempts to apply AI to perform a design tasks in nuclear energy. Consequently, there were significant implementation challenges and the bulk of the work was focused on successful implementation and algorithm design. The results presented here are very low technology readiness level as a consequence of the lack of related literature, but still represent a significant step forward in the pursuit of applied AI for design.

42 ENGINEERING↗

PyARC Status Report: New Integrations and Upgrades to the Fast Reactor Analysis Workflow Management Tool

PyARC was initially developed as an open source tool to support fast reactor analyses using the Argonne Reactor Computation (ARC) code suite as a part of the Nuclear Energy Advanced Modeling and Simulation (NEAMS) Workbench initiative in FY17. The goal of this initiative is to provide a common user interface for model generation, real-time validation, execution, output processing, and visualization for all integrated codes. This is accomplished through the reliance on tools available in the Workbench framework and runtime environment. While initially developed to support the ARC codes, PyARC was extended in FY22 to wrap other NEAMS and non-ARC codes, including Griffin and OpenMC, in the supported other neutronics workflows, and support users in the adoption of NEAMS-supported high fidelity analysis codes. Most recently, NUBOW-3D, a recently adopted ARC code, was integrated to support reactor bowing calculations as well. Integration of these codes into the NEAMS Workbench directly benefits the advanced reactor modeling community by: • Providing a set of controlled, maintained, documented and validated scripts to generate inputs, which promotes best practices, reduces the learning curve, and facilitates project collaboration. • Improving the user experience: the Workbench interface provides assistance for building an input through auto-completion, real-time validation, document navigation, and geometry and results visualization. • Automating complex calculations and workflows for reactor analysis. • Helping users transition to using high-fidelity NEAMS codes along-side the ARC codes. In FY22, a progress report was published that described the state of each of the tools integrated into PyARC. Since then, there have been many enhancements and upgrades to the existing integrations as well as entirely new code integrations as well. This report details all new integrations and major developments in PyARC since the version 2.0.0 release highlighted in the FY22 report.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Electron-Proton Scattering Event Generation using Structured Tokenization

Recent work such as Omnijet-$\alpha$ has demonstrated that effective tokenization combined with transformer-based architectures can produce effective foundation models for jet physics. While tokenization may help models capture generalizable event characteristics, it also introduces discretization errors that may compromise the precision required for downstream physics analyses. As the number and complexity of the particle features grow, these errors are likely to grow proportionally. In this study, we investigate new tokenization strategies to improve the application of generative transformer models to \textsc{Pythia8} simulations of electron-proton scattering at the Electron-Ion Collider. Specifically, we propose a feature-based structured tokenization approach that utilizes multiple tokens per particle, improving expressivity, while reducing the total number of unique tokens needed. We evaluate this method against grid-based binning, K-means clustering, and vector-quantized variational auto-encoders on the event simulations. Our results show that feature-based structured tokenization reduces discretization error, leading to more accurate generative modeling of particle-level events.

Goldenberg, Steven [Thomas Jefferson National Acce↗

Continual Learning for Particle Accelerators

Particle accelerators operate under dynamically changing conditions, which often lead to data distribution drifts. These drifts pose significant challenges for Machine Learning (ML) models, which typically fail to maintain performance when faced with such non-stationary data. In particle accelerators, the primary sources of these data drifts include changes in accelerator settings and non-measured parameters such as machine degradation and environmental factors. Previous research has proposed conditional models to handle multiple beam configurations effectively; however, it is challenging to train the ML models on all possible configuration settings. Additionally, conditional models alone can not address performance degradation caused by drifts due to non-measured factors. These limitations contribute to a significant gap between ML development and its deployment in real-world operational settings. To bridge this gap, in this paper, we identify some of the key areas within particle accelerators where continual learning can help mitigate drift-induced performance degradation. In addition, we present a practical use case where a conditional Auto-Encoder model coupled with memory-based continual learning has been employed to demonstrate stable performance even when underlying data drifts.

Schram, Malachi [Thomas Jefferson National Acceler↗