Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Real Time Optimization”

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 883 records · Page 49

Data Curation for Machine Learning Applied to Geothermal Power Plant Operational Data for GOOML: Geothermal Operational Optimization with Machine Learning: Preprint

Geothermal Operational Optimization with Machine Learning (GOOML) is a transferable and extensible component-based geothermal asset modeling framework that considers complex steamfield relationships and identifies optimization prospects using a data-driven approach to physics-guided, data-centric machine learning. This framework has been used to develop digital twins that provide steamfield operators with operational environments to analyze and understand historical and forecasted power production, explore new steamfield configuration possibilities, and seek optimal asset management in real world applications. To create, test, and apply the GOOML framework, diverse time-series datasets spanning multiple years were sourced from various geothermal power plant components within several complex real-world geothermal operations. These operations are based in the United States and New Zealand and include a variety of technologies, end-uses and configurations, collectively covering nearly all relevant operating conditions for modern geothermal fields. Datasets were acquired from multiple sources to ensure that machine learning experiments generalized properly to various operating conditions. It was found that the data varied in quality, format, and completeness. To ensure consistency between the various datasets, a standardized data curation process was developed to reliably streamline data preparation. This paper will discuss best practices as learned from the GOOML data curation process which takes the following steps: 1) acquisition of large quantities of data from power plant operators, 2) digestion of data to gain an initial understanding of what is included, 3) data transformation, which includes converting the data into a standardized machine-readable format so that they can be visualized, quality checked, and cleaned, 4) quality assurance and quality control, involving identification of significant data gaps and apparent anomalies through mapping of data features to real world componentry via the GOOML historical model, followed by discussion with modelers and power plant operators to identify additional data needs and to resolve issues, 5) use in machine learning algorithms, and 6) repetition of steps one through five until all data needs are met and data are deemed suitable for producing trustworthy modeling results which may be disseminated, ideally along with the curated dataset. This iterative process is focused on improving the quality of the data rather than tuning machine learning model parameters and supports a shift towards data-centric AI as a means to improving real-world applicability of geothermal machine learning projects.

access↗

AI in Science Communication

Generative AI has brought great innovations across multiple fields, offering great tools for enhanced communication and efficiency. This project focused on developing a custom AI chatbot using OpenAI's Chat GPT (GPT-4o) to support the Fermilab communications team. An analysis identified Chat GPT as the optimal choice, leading to the adoption of its team version and the implementation of a real-time JSON schema for website scanning. Four distinct personas were created to tailor responses to specific audiences, and Fermilab's published content was uploaded to ensure tone consistency. The training involved iterative prompt trials, resulting in a responsive and effective communication assistant. Initial evaluations indicate that the custom GPT shows promise.

Valle, Diego↗

STUDIES IN ASTRONOMICAL TIME SERIES ANALYSIS. VI. BAYESIAN BLOCK REPRESENTATIONS

This paper addresses the problem of detecting and characterizing local variability in time series and other forms of sequential data. The goal is to identify and characterize statistically significant variations, at the same time suppressing the inevitable corrupting observational errors. We present a simple nonparametric modeling technique and an algorithm implementing it-an improved and generalized version of Bayesian Blocks-that finds the optimal segmentation of the data in the observation interval. The structure of the algorithm allows it to be used in either a real-time trigger mode, or a retrospective mode. Maximum likelihood or marginal posterior functions to measure model fitness are presented for events, binned counts, and measurements at arbitrary times with known error distributions. Problems addressed include those connected with data gaps, variable exposure, extension to piecewise linear and piecewise exponential representations, multivariate time series data, analysis of variance, data on the circle, other data modes, and dispersed data. Simulations provide evidence that the detection efficiency for weak signals is close to a theoretical asymptotic limit derived by Arias-Castro et al. In the spirit of Reproducible Research all of the code and data necessary to reproduce all of the figures in this paper are included as supplementary material.

79 ASTRONOMY AND ASTROPHYSICS↗

HGQ: High Granularity Quantization for Real-time Neural Networks on FPGAs

Neural networks with sub-microsecond inference latency are required by many critical applications. Targeting such applications deployed on FPGAs, we present High Granularity Quantization (HGQ), a quantization-aware training framework that optimizes parameter bit-widths through gradient descent. Unlike conventional methods, HGQ determines the optimal bit-width for each parameter independently, making it suitable for hardware platforms supporting heterogeneous arbitrary precision arithmetic. In our experiments, HGQ shows superior performance compared to existing network compression methods, achieving orders of magnitude reduction in resource consumption and latency while maintaining the accuracy on several benchmark tasks. These improvements enable the deployment of complex models previously infeasible due to resource or latency constraints. HGQ is open-source and is used for developing next-generation trigger systems at the CERN ATLAS and CMS experiments for particle physics, enabling the use of advanced machine learning models for real-time data selection with sub-microsecond latency.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A Multi-Objective Bayesian Optimized Human Assessed Multi-Target Generated Spectral Recommender System for Rapid Pareto Discoveries of Material Properties

Optimization for different tasks like material characterization, synthesis, and functional properties for desired applications over multi-dimensional control parameter and function spaces need a rapid strategic search through active learning. However, in all cases prior to optimization, the target material properties are assumed known and fixed, which mostly deviates from real-world scenarios in material synthesis. This can be critical for running expensive experiments on new materials, when the experimental results are fuzzy for any scientific outcomes due to improper target setting, ultimately wasting time and cost. The failure rate and cost are even higher over exploring on multi-target space, where we want to learn the pareto among multiple properties, to jointly optimize during material synthesis for desired applications. To address the challenge, here we introduce the human-operator attempt flexibility in the active learning based automated experiment framework, with generating multiple human assessed targets through a voting-based recommender system during real-time microscope measurements over the large material image space, sequentially learn/update multiple desired targets through a weighting system, and adaptively search in multiple material properties functional space for non-dominated pareto discoveries to maximize the custom structural similarity based acquisition function. We term this a multi-objective Bayesian optimized human assessed multi-target generated spectral recommender systems (MOBO-HAM-SRS). The approach has been demonstrated to peizoresponse force spectroscopy of a ferroelectric thin film, exploring with different kernels and acquisition functions. This work shows an advancement towards human-AI collaborated automated experiments, steering optimization trajectories through human overpowering AI at the early stage when uncertainty is high and AI overpowering human at the later stage with rapid exploration towards optimal goal, following human-assessed multiple targets properties.

Biswas, Arpan↗

Hardware-accelerated inference for real-time gravitational-wave astronomy

The field of transient astronomy has seen a revolution with the first gravitational-wave detections and the arrival of multi-messenger observations they enabled. Transformed by the first detection of binary black hole and binary neutron star mergers, computational demands in gravitational-wave astronomy are expected to grow by at least a factor of two over the next five years as the global network of kilometer-scale interferometers are brought to design sensitivity. With the increase in detector sensitivity, real-time delivery of gravitational-wave alerts will become increasingly important as an enabler of multi-messenger followup. In this work, we report a novel implementation and deployment of deep learning inference for real-time gravitational-wave data denoising and astrophysical source identification. This is accomplished using a generic Inference-as-a-Service model that is capable of adapting to the future needs of gravitational-wave data analysis. Overall, our implementation allows seamless incorporation of hardware accelerators and also enables the use of commercial or private (dedicated) as-a-service computing. Based on our results, we propose a paradigm shift in low-latency and offline computing in gravitational-wave astronomy. Such a shift can address key challenges in peak-usage, scalability and reliability, and provide a data analysis platform particularly optimized for deep learning applications. The achieved sub-millisecond scale latency will also be relevant for any machine learning-based real-time control systems that may be invoked in the operation of near-future and next generation ground-based laser interferometers, as well as the front-end collection, distribution and processing of data from such instruments.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Genetically encoded sensors enable real-time observation of metabolite production

Advances in biotechnology are enabling engineers to harness natural processes for the production of valuable new chemicals and materials. Cells engineered for chemical production act as renewable factories and redefine what is possible in industries as diverse as manufacturing, pharmaceuticals, and energy. Despite this potential, long and uncertain timelines for biobased product development hinder progress. Engineering cells for chemical production is challenging because the complexity of biology necessitates many attempted designs before an optimal combination of genetic elements or culture conditions is discovered. In this work, we develop a framework that allows design evaluation to be performed with fluorescence such that microbial chemical production can be monitored in real time without the need for sample preparation or laborious analytical techniques.

59 BASIC BIOLOGICAL SCIENCES↗

Overview of the KSTAR experiments and future plan

The Korea Superconducting Tokamak Advanced Research (KSTAR) device has recently entered a new phase of operation following the installation of a tungsten mono-block divertor, providing a unique platform to investigate high-performance, long-pulse plasma scenarios for future reactors. The 2023–2024 experimental campaigns demonstrated significant progress in commissioning the tungsten divertor, developing plasma operation scenarios, and achieving real-time control. Comparative studies revealed increased core radiation losses and associated performance degradation relative to the carbon divertor environment. Alleviation strategies, such as optimized neutral beam injection timing, boron powder seeding, and impurity control through gas puffing, were shown to reduce tungsten accumulation and extend high-performance phases, including record H-mode operation exceeding 100 s. International collaboration enabled the first demonstration of high poloidal beta scenarios with an internal transport barrier on KSTAR. Advances in control included upgraded real-time resonant magnetic perturbation schemes and disruption forecasting. Furthermore, divertor detachment control using a new surrogate-model-based approach with real-time radiation imaging demonstrated active impurity and heat flux management. Supported by enhanced heating, current drive, and diagnostic systems, these achievements offer valuable insights into tungsten impurity behavior, transport physics, and control strategies for future reactors including ITER. Planned upgrades toward a full-tungsten wall and ITER-aligned real-time plasma control will further establish KSTAR as a leading experimental platform for developing operational scenarios required to achieve steady-state, burning plasma conditions.

KSTAR↗

Bridging Cloud and Edge Computing at NREL Using CONNECT: Cloud Optimized Networking for Next-Gen Edge Computing Technologies [Slides]

CONNECT is an innovative on-premise hardware and software solution that integrates edge and cloud computing infrastructure at NREL. Built on the AWS Greengrass middleware and leveraging the MQTT protocol, CONNECT enables real-time data streaming from IoT devices and gateways to both cloud and local services, empowering researchers to rapidly capture, analyze, and act upon edge-generated data while leveraging cloud capabilities. The platform addresses research infrastructure challenges by providing a pre-approved platform which is already configured with the correct networking and cybersecurity baselines thus eliminating procurement delays and enabling on-demand availability. CONNECT's hybrid architecture efficiently manages burstable workloads, allowing research teams to dynamically scale computational capacity, handle peak data loads, and reduce operational bottlenecks. Advanced capabilities include built-in GPU support for executing machine learning models which enables low-latency inference at the edge from models trained in the cloud. This architecture supports real-time analytics and filtering, providing a mechanism to allow only transmitting and processing high-value data. Cloud-based configuration management permits engineers to manage on-premise systems remotely, optimizing operational efficiency. By bridging edge and cloud computing, CONNECT provides NREL researchers with a flexible, scalable platform that accelerates scientific discovery while maintaining robust security and performance standards.

97 MATHEMATICS AND COMPUTING↗

Improvements in Hanford TRU Program Utilizing Systems Modeling and Analyses

Hanford's Transuranic (TRU) Program is responsible for certifying contact-handled (CH) TRU waste and shipping the certified waste to the Waste Isolation Pilot Plant (WIPP). Hanford's CH TRU waste includes material that is in retrievable storage as well as above ground storage, and newly generated waste. Certifying a typical container entails retrieving and then characterizing it (Non-Destructive Examination [NDE], Non-Destructive Assay [NDA], and Head Space Gas Sampling [HSG]), validating records (data review and reconciliation), and designating the container for a payload. The certified payload is then shipped to WIPP. Systems modeling and analysis techniques were applied to Hanford's TRU Program to help streamline the certification process and increase shipping rates. The modeling and analysis yields several benefits: - Maintains visibility on system performance and predicts downstream consequences of production issues. - Predicts future system performance with higher confidence, based on tracking past performance. - Applies speculation analyses to determine the impact of proposed changes (e.g., apparent shortage of feed should not be used as basis to reassign personnel if more feed is coming in the queue). - Positively identifies the appropriate queue for all containers (e.g., discovered several containers that were not actively being worked because they were in the wrong 'physical' location - method used previously for queuing up containers). - Identifies anomalies with the various data systems used to track inventory (e.g., dimensional differences for Standard Waste Boxes). A model of the TRU Program certification process was created using custom queries of the multiple databases for managing waste containers. The model was developed using a simplified process chart based on the expected path for a typical container. The process chart was augmented with the remediation path for containers that do not meet acceptance criteria for WIPP. Containers are sorted into queues based on their current status in the process. A container can be in only one queue at any given time. Existing data systems are queried to establish the quantity of containers in each queue on any given day. This sets the amount of feed available that is then modeled to be processed according to the daily production plans. The daily production plans were created by identifying the equipment necessary and the staff that performs each process step, and determining the expected production rate for each step. Production performance is monitored on a weekly basis with Project senior staff to establish a total operating efficiency (TOE) for each step (comparing actual performance to production capacity). The unit operations were modeled to be constrained by each day's feed queue plus the performance of the preceding step. The TOE for each unit operation was applied to an integrated model to determine bottlenecks and identify areas for improvement. All of the steps were linked to predict future system performance based on available feed and integrated system-level TOE. It has been determined that at times sub-optimization of a particular unit operation is necessary to ensure the system remains balanced (e.g., having excess capacity in assay does no good if there is no feed available because the real-time radiography [RTR] is working at half capacity). Several recommendations have been provided to the Project management team resulting in improvements in the performance of TRU certification activities by Hanford's TRU Program. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Automatic Classification of Biological Targets in a Tidal Channel Using a Multibeam Sonar

Multibeam sonars are widely used for environmental monitoring of fauna at marine renewable energy sites. However, they can rapidly accrue vast volumes of data, which poses a challenge for data processing. Here, using data from a deployment in a tidal channel with peak currents of 1–2 m s –1 , we demonstrate the data-reduction benefits of real-time automatic classification of targets detected and tracked in multibeam sonar data. First, we evaluate classification capabilities for three machine learning algorithms: random forests, support vector machines, and k-nearest neighbors. For each algorithm, a hill-climbing search optimizes a set of hand-engineered attributes that describe tracked targets. Here, the random forest algorithm is found to be most effective—in postprocessing, discriminating between biological and nonbiological targets with a recall rate of 0.97 and a precision of 0.60. In addition, 89% of biological targets are correctly classified as either seals, diving birds, fish schools, or small targets. Model dependence on the volume of training data is evaluated. Second, a real-time implementation of the model is shown to distinguish between biological targets and nonbiological targets with nearly the same performance as in postprocessing. From this, we make general recommendations for implementing real-time classification of biological targets in multibeam sonar data and the transferability of trained models.

16 TIDAL AND WAVE POWER↗

A Cross-Domain Optimization Framework of PMU and Communication Placement for Multidomain Resiliency and Cost Reduction

Phasor measurement units (PMUs) play a crucial role in real-time monitoring and control of power grids. They rely on a communication network to transfer measurement data to the phasor data concentrator (PDC) for further processing and analysis. In this paper, a resilient cross-domain PMU and communication link placement method for minimizing the overall installation cost of the wide-area measurement system (WAMS) is proposed. Here, the main idea is to break down the barrier between the power grid domain and the communication domain, and consider the impact of one when design the other. The PMU placement in the power grid domain takes into account the cost of communication links by generating multiple solutions with equally minimum PMU costs for communication link placement evaluation. On the other hand, the communication link placement problem reduces the cost by customizing the routing policies based on the different roles of PMUs in grid observability. The proposed WAMS design is capable of withstanding any single component failure in the power domain (PMU failure or power branch failure) or in the communication domain (communication link failure or PDC failure). Numerical study on the IEEE 57-bus system reveals that the developed cross-domain optimization framework can significantly reduce the overall installation cost of WAMS while attaining multi-domain resiliency.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Quality Assurance of Radiation Therapy Planning Systems: Current Status and Remaining Challenges

Computerized radiation therapy planning systems (RTPSs) are pivotal for treatment planning. The acceptance, commissioning, and quality control of RTPSs are uniquely complex and are described in the American Association of Physicists in Medicine Task Group Report 53 (1998) and International Atomic Energy Agency Technical Report Series No. 430 (2004). The International Atomic Energy Agency also developed a document and data package for use by vendors and purchasers to aid with acceptance testing of RTPSs. This document is based on International Electrotechnical Commission standard 62083 (2000) and describes both 'type' tests to be performed in the factory and 'site' tests to be performed in the clinic. The American Association of Physicists Task Group Report 67 described benchmark tests for the validation of dose calculation algorithms. Test data are being produced with the backing of the U.S. National Cancer Institute. However, significant challenges remain. Technology keeps evolving rapidly, thus requiring new quality assurance (QA) procedures. Intensity-modulated radiation therapy with its use of inverse optimization has added a new dimension to QA, because the results are not intuitively obvious. New technologies such as real-time ultrasound guidance for brachytherapy, TomoTherapy, and Cyberknife, require their own specialized RTPSs with unique QA requirements. On-line imaging allows for the generation of dose reconstructions using image warping techniques to determine the daily dose delivered to the patient. With increasing computer speeds, real-time reoptimization of treatment plans will become a reality. Gating technologies will require four-dimensional dose calculations to determine the actual dose delivered to tissue voxels. With these rapidly changing technologies, it is essential that a strong QA culture is invoked in every institution implementing these procedures and that new protocols are developed as a part of the clinical implementation process.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

A Multi-Contact, Low Capacitance HPGe Detector for High Rate Gamma Spectroscopy

The detection, identification and non-destructive assay of special nuclear materials and nuclear fission by-products are critically important activities in support of nuclear non-proliferation programs. Both national and international nuclear safeguard agencies recognize that current accounting methods for spent nuclear fuel are inadequate from a safeguards perspective. Radiation detection and analysis by gamma-ray spectroscopy is a key tool in this field, but no instrument exists that can deliver the required performance (energy resolution and detection sensitivity) in the presence of very high background count rates encountered in the nuclear safeguards arena. The work of this project addresses this critical need by developing a unique gamma-ray detector based on high purity germanium that has the previously unachievable property of operating in the 1 million counts-per-second range while achieving state-of-the-art energy resolution necessary to identify and analyze the isotopes of interest. The technical approach was to design and fabricate a germanium detector with multiple segmented electrodes coupled to multi-channel high rate spectroscopy electronics. Dividing the germanium detector’s signal electrode into smaller sections offers two advantages; firstly, the energy resolution of the detector is potentially improved, and secondly, the detector is able to operate at higher count rates. The design challenges included the following; determining the optimum electrode configuration to meet the stringent energy resolution and count rate requirements; determining the electronic noise (and therefore energy resolution) of the completed system after multiple signals are recombined; designing the germanium crystal housing and vacuum cryostat; and customizing electronics to perform the signal recombination function in real time. In this phase I work, commercial off-the-shelf electrostatic modeling software was used to develop the segmented germanium crystal geometry, which underwent several iterations before an optimal electrode configuration was found. The model was tested and validated against real-world measurements with existing germanium detectors. Extensive modeling of electronic noise was conducted using established formulae, and real-world measurements were performed on candidate front-end electronic components. This initial work proved the feasibility of the design with respect to expected high count rate and energy resolution performance. Phase I also delivered the mechanical design of the detector housing and vacuum cryostat to be built in Phase II. Finally, a Monte Carlo simulation was created to show the response of the complete design to a Cs-137 source. This development presents a significant advance for nuclear safeguards instrumentation with increased speed and accuracy of detection and identification of special nuclear materials. Other significant applications are foreseen for a gamma-ray detector that delivers high energy resolution (1keV FWHM noise) at high count rate (1 Mcps), especially in the areas of physics research and materials analysis.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Optimizing first- and last-mile public transit services leveraging transportation network companies (TNC)

First-mile last-mile (FMLM) mobility services that connect riders to public transit can lead to improved transit accessibility and network efficiency if such services are convenient and reliable. However, many current FMLM services are inefficient and costly because they are inflexible (e.g., fixed supply of shuttles) and do not leverage collected data for optimized decision making. At the same time, new forms of shared mobility can provide added flexibility and real-time analytics to FMLM systems when carefully integrated. This study evaluates performance and cost implications of public/private coordination between transit shuttles and transportation network companies (TNC) in the FMLM context. A real-time operations model was developed to simulate daily operations for an existing FMLM system using real-world demand data. Three supply strategies were tested with varying levels of flexibility: (1) Status Quo (two 23-passenger on-demand shuttles), (2) Hybrid (one 23-passenger on-demand shuttle + TNC), and (3) TNC Only (exclusively use TNC services). Results indicated that the added flexibility of the Hybrid service design (using shuttles and TNCs) improved service performance (a 7.7% improvement), reduced daily operating costs (– 6.0%), and improved service reliability (95th percentile travel times decreased by up to 40% during peak periods). In addition, the Hybrid service design was more robust to variations in demand. Here, the Hybrid service was significantly cheaper to operate (– 31.6%) at reduced demand levels (50% of normal), and improved service performance (a 10.2% improvement) when demand levels were increased (150% of normal). These findings emphasize the importance of flexibility in FMLM service designs, especially when demand is sparse and variable.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

A Machine Learning Model for Predicting Composition of Catalytic Coprocessing Products from Molecular Beam Mass Spectra

Demand for the development of an automated and integrated refining process for biofuels has increased in recent years due to the lack of generalized process inspection tools. In bio-oil upgrading processes, all process variables are maintained based on the offline specification of intermediates and products. A lack of real-time product specifications in batch-wise monitoring can cause process failure and wasted resources. Therefore, there is a need for a fast and accurate intermediates/product specification tool that can be used for real-time specification to reduce waste and mitigate the risk of process failure. Here, to address this gap, we developed a machine learning (ML) model for predicting speciated bio-oil composition, including paraffin, iso-paraffins, olefins, naphthene, and aromatics. The model is trained using the mass spectra from upgraded products collected in the vapor phase before condensation and predicts the composition of the condensed product. Training ML models using raw mass spectra is challenging due to numerous overlapped peaks originating from different parent compounds. With this in mind, we propose a protocol that (i) transforms raw mass spectra to chemistry-inspired predefined features and (ii) trains decision tree-based models using these features. Our results show that the random forest model was robust against overfitting and had the highest accuracy compared to other models. Moreover, a stochastic ablation method determined the eight most significant features while maximizing the accuracy. Our protocol facilitates real-time compositional analysis of upgraded bio-oils and thus real-time process monitoring. Additionally, this protocol enables the rational design of efficient catalysts and the determination of optimal process conditions.

09 BIOMASS FUELS↗

Individual Sawtooth Pacing by Synchronized ECCD in TCV

Previous real-time sawtooth control scenarios using EC actuators have attempted to shorten or lengthen the sawtooth period by optimally positioning the EC absorption near the q = 1 surface. In new experiments we demonstrate for the first time that individual sawtooth crashes can be repetitively induced at predictable times by reducing the stabilizing ECCD power after a predetermined time from the preceding crash. Other stabilizing actuators (e.g. ICRF, NBI) are expected to produce similar effects. Armed with these results, we present a new sawtooth / NTM control paradigm for improved performance in burning plasmas. The potential appearance of neo-classical tearing modes, triggered by long period sawtooth crashes even at low beta, becomes predictable and therefore amenable to preemptive ECCD. The ITER Electron Cyclotron Upper Launcher (EC-UL) design incorporates the needed functionalities for this method to be applied. The methodology and associated TCV experiments will be presented.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A hybrid neural architecture: Online attosecond x-ray characterization

The emergence of high-repetition-rate x-ray free-electron lasers (XFELs), such as SLAC’s LCLS-II, serves as our canonical example for autonomous controls that necessitate high-throughput diagnostics paired with streaming computational pipelines capable of single-shot analysis with extremely low latency. We present the deterministic characterization with an integrated parallelizable hybrid resolver architecture, a hybrid machine learning framework designed for fast, accurate analysis of XFEL diagnostics using angular streaking-based sinogram images. This architecture integrates convolutional neural networks and bidirectional long short-term memory models to denoise input, identify x-ray sub-spike features, and extract sub-spike relative delays with sub-30 attosecond temporal resolution. Deployed on low-latency hardware, it achieves over 10 kHz throughput with 168.3 μs inference latency, indicating scalability to 14 kHz with field-programmable gate array integration. By transforming regression tasks into classification problems and leveraging optimized error encoding, we achieve high precision with low-latency performance that is critical for real-time streaming event selection and experimental control feedback signals. This represents a key development in real-time control pipelines for next-generation autonomous science, generally, and high repetition-rate x-ray experiments in particular.

Accelerator Physics (physics.acc-ph)↗