Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “trigger algorithms”

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 163 records · Page 9
All sourcesNASA NTRSDOE OSTI
NASA NTRS2013

Distributed Prognostics and Health Management with a Wireless Network Architecture

A heterogeneous set of system components monitored by a varied suite of sensors and a particle-filtering (PF) framework, with the power and the flexibility to adapt to the different diagnostic and prognostic needs, has been developed. Both the diagnostic and prognostic tasks are formulated as a particle-filtering problem in order to explicitly represent and manage uncertainties in state estimation and remaining life estimation. Current state-of-the-art prognostic health management (PHM) systems are mostly centralized in nature, where all the processing is reliant on a single processor. This can lead to a loss in functionality in case of a crash of the central processor or monitor. Furthermore, with increases in the volume of sensor data as well as the complexity of algorithms, traditional centralized systems become for a number of reasons somewhat ungainly for successful deployment, and efficient distributed architectures can be more beneficial. The distributed health management architecture is comprised of a network of smart sensor devices. These devices monitor the health of various subsystems or modules. They perform diagnostics operations and trigger prognostics operations based on user-defined thresholds and rules. The sensor devices, called computing elements (CEs), consist of a sensor, or set of sensors, and a communication device (i.e., a wireless transceiver beside an embedded processing element). The CE runs in either a diagnostic or prognostic operating mode. The diagnostic mode is the default mode where a CE monitors a given subsystem or component through a low-weight diagnostic algorithm. If a CE detects a critical condition during monitoring, it raises a flag. Depending on availability of resources, a networked local cluster of CEs is formed that then carries out prognostics and fault mitigation by efficient distribution of the tasks. It should be noted that the CEs are expected not to suspend their previous tasks in the prognostic mode. When the prognostics task is over, and after appropriate actions have been taken, all CEs return to their original default configuration. Wireless technology-based implementation would ensure more flexibility in terms of sensor placement. It would also allow more sensors to be deployed because the overhead related to weights of wired systems is not present. Distributed architectures are furthermore generally robust with regard to recovery from node failures.

Goebel, Kai↗
NASA NTRS2010

Online Planning Algorithm

AVA v2 software selects goals for execution from a set of goals that oversubscribe shared resources. The term goal refers to a science or engineering request to execute a possibly complex command sequence, such as image targets or ground-station downlinks. Developed as an extension to the Virtual Machine Language (VML) execution system, the software enables onboard and remote goal triggering through the use of an embedded, dynamic goal set that can oversubscribe resources. From the set of conflicting goals, a subset must be chosen that maximizes a given quality metric, which in this case is strict priority selection. A goal can never be pre-empted by a lower priority goal, and high-level goals can be added, removed, or updated at any time, and the "best" goals will be selected for execution. The software addresses the issue of re-planning that must be performed in a short time frame by the embedded system where computational resources are constrained. In particular, the algorithm addresses problems with well-defined goal requests without temporal flexibility that oversubscribes available resources. By using a fast, incremental algorithm, goal selection can be postponed in a "just-in-time" fashion allowing requests to be changed or added at the last minute. Thereby enabling shorter response times and greater autonomy for the system under control.

Rabideau, Gregg R.↗
DOE OSTI2025

ATTNChecker: Highly-Optimized Fault Tolerant Attention for Large Language Model Training

Large Language Models (LLMs) have demonstrated remarkable performance in various natural language processing tasks. However, the training of these models is computationally intensive and susceptible to faults, particularly in the attention mechanism, which is a critical component of transformer-based LLMs. In this paper, we investigate the impact of faults on LLM training, focusing on INF, NaN, and near-INF values in the computation results with systematic fault injection experiments. We observe the propagation patterns of these errors, which can trigger non-trainable states in the model and disrupt training, forcing the procedure to load from checkpoints. To mitigate the impact of these faults, we propose ATTNChecker, the first Algorithm-Based Fault Tolerance (ABFT) technique tailored for the attention mechanism in LLMs. ATTNChecker is designed based on fault propagation patterns of LLM and incorporates performance optimization to adapt to both system reliability and model vulnerability while providing lightweight protection for fast LLM training. Evaluations on four LLMs show that ATTNChecker on average incurs on average 7% overhead on training while detecting and correcting all extreme errors. Compared with the state-of-the-art checkpoint/restore approach, ATTNChecker reduces recovery overhead by up to 49×.

Liang, Yuhang [University of Alabama - Birmingham]↗
DOE OSTI2023

First Impressions: Early-time Classification of Supernovae Using Host-galaxy Information and Shallow Learning

Substantial effort has been devoted to the characterization of transient phenomena from photometric information. Automated approaches to this problem have taken advantage of complete phase coverage of an event, limiting their use for triggering rapid follow-up of ongoing phenomena. In this work, we introduce a neural network with a single recurrent layer designed explicitly for early photometric classification of supernovae (SNe). Our algorithm leverages transfer learning to account for model misspecification, host-galaxy photometry to solve the data-scarcity problem soon after discovery, and a custom weighted loss to prioritize accurate early classification. We first train our algorithm using state-of-the-art transient and host-galaxy simulations, then adapt its weights and validate it on the spectroscopically confirmed SNe Ia, SNe II, and SNe Ib/c from the Zwicky Transient Facility Bright Transient Survey. On observed data, our method achieves an overall accuracy of 82% ± 2% within 3 days of an event’s discovery, and an accuracy of 87% ± 5% within 30 days of discovery. At both early and late phases, our method achieves comparable or superior results to the leading classification algorithms with a simpler network architecture. These results help pave the way for rapid photometric and spectroscopic follow-up of scientifically valuable transients discovered in massive synoptic surveys.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS2015

Detection and Modeling of High-Dimensional Thresholds for Fault Detection and Diagnosis

Many Fault Detection and Diagnosis (FDD) systems use discrete models for detection and reasoning. To obtain categorical values like oil pressure too high, analog sensor values need to be discretized using a suitablethreshold. Time series of analog and discrete sensor readings are processed and discretized as they come in. This task isusually performed by the wrapper code'' of the FDD system, together with signal preprocessing and filtering. In practice,selecting the right threshold is very difficult, because it heavily influences the quality of diagnosis. If a threshold causesthe alarm trigger even in nominal situations, false alarms will be the consequence. On the other hand, if threshold settingdoes not trigger in case of an off-nominal condition, important alarms might be missed, potentially causing hazardoussituations. In this paper, we will in detail describe the underlying statistical modeling techniques and algorithm as well as the Bayesian method for selecting the most likely shape and its parameters. Our approach will be illustrated by several examples from the Aerospace domain.

Computer Systems↗
NASA NTRS2016

Machine-z: Rapid Machine-Learned Redshift Indicator for Swift Gamma-Ray Bursts

Studies of high-redshift gamma-ray bursts (GRBs) provide important information about the early Universe such as the rates of stellar collapsars and mergers, the metallicity content, constraints on the re-ionization period, and probes of the Hubble expansion. Rapid selection of high-z candidates from GRB samples reported in real time by dedicated space missions such as Swift is the key to identifying the most distant bursts before the optical afterglow becomes too dim to warrant a good spectrum. Here, we introduce 'machine-z', a redshift prediction algorithm and a 'high-z' classifier for Swift GRBs based on machine learning. Our method relies exclusively on canonical data commonly available within the first few hours after the GRB trigger. Using a sample of 284 bursts with measured redshifts, we trained a randomized ensemble of decision trees (random forest) to perform both regression and classification. Cross-validated performance studies show that the correlation coefficient between machine-z predictions and the true redshift is nearly 0.6. At the same time, our high-z classifier can achieve 80 per cent recall of true high-redshift bursts, while incurring a false positive rate of 20 per cent. With 40 per cent false positive rate the classifier can achieve approximately 100 per cent recall. The most reliable selection of high-redshift GRBs is obtained by combining predictions from both the high-z classifier and the machine-z regressor.

gamma-ray burst: general↗
DOE OSTI2023

Reconstruction of interactions in the ProtoDUNE-SP detector with Pandora

The Pandora Software Development Kit and algorithm libraries provide pattern-recognition logic essential to the reconstruction of particle interactions in liquid argon time projection chamber detectors. Pandora is the primary event reconstruction software used at ProtoDUNE-SP, a prototype for the Deep Underground Neutrino Experiment far detector. ProtoDUNE-SP, located at CERN, is exposed to a charged-particle test beam. This paper gives an overview of the Pandora reconstruction algorithms and how they have been tailored for use at ProtoDUNE-SP. In complex events with numerous cosmic-ray and beam background particles, the simulated reconstruction and identification efficiency for triggered test-beam particles is above 80% for the majority of particle type and beam momentum combinations. Specifically, simulated 1 GeV/c charged pions and protons are correctly reconstructed and identified with efficiencies of 86.1$\pm 0.6$% and 84.1$\pm 0.6$%, respectively. The efficiencies measured for test-beam data are shown to be within 5% of those predicted by the simulation.

43 PARTICLE ACCELERATORS↗
DOE OSTI2022

Idaho National Laboratory Data Acquisition And Processing System

The INLDAS data acquisition and processing system is designed to develop prototype measurements and real-time processing techniques. The INLDAS hardware and firmware currently consists of National Instruments • NI-DAQmx 14.1 • cDAQ-9184 • 9205 • 9211 INLDAS can perform of standard data acquisition functions as well as novel functions and real-time processing algorithms. There are 3 acquisition modes to choose from: • Continuous mode runs when the user hits start, processing data and logging it to file • SWTrigger mode waits for one or more predefined triggers before acquiring data. It will buffer data as well, so it can record data that happened shortly before the trigger • Wakeup mode waits on predefined timers. When a time goes off, it acquires a preset amount of data There are also 3 Data Processing Modes • Normal mode does no processing besides the rolling average • FFT Mode Performs an FFT on incoming data every time the time window has elapsed • STFFT mode

Smith, JamesA↗
NASA NTRS2015

Abort Trigger False Positive and False Negative Analysis Methodology for Threshold-Based Abort Detection

This paper describes a quantitative methodology for bounding the false positive (FP) and false negative (FN) probabilities associated with a human-rated launch vehicle abort trigger (AT) that includes sensor data qualification (SDQ). In this context, an AT is a hardware and software mechanism designed to detect the existence of a specific abort condition. Also, SDQ is an algorithmic approach used to identify sensor data suspected of being corrupt so that suspect data does not adversely affect an AT's detection capability. The FP and FN methodologies presented here were developed to support estimation of the probabilities of loss of crew and loss of mission for the Space Launch System (SLS) which is being developed by the National Aeronautics and Space Administration (NASA). The paper provides a brief overview of system health management as being an extension of control theory; and describes how ATs and the calculation of FP and FN probabilities relate to this theory. The discussion leads to a detailed presentation of the FP and FN methodology and an example showing how the FP and FN calculations are performed. This detailed presentation includes a methodology for calculating the change in FP and FN probabilities that result from including SDQ in the AT architecture. To avoid proprietary and sensitive data issues, the example incorporates a mixture of open literature and fictitious reliability data. Results presented in the paper demonstrate the effectiveness of the approach in providing quantitative estimates that bound the probability of a FP or FN abort determination.

risk assessment↗
NASA NTRS1985

Study of muons near shower cores at sea level using the E594 neutrino detector

The E594 neutrino detector has been used to study the lateral distribution of muons of energy 3 GeV near shower cores. The detector consists of a 340 ton fine grain calorimeter with 400,000 cells of flash chamber and dimensions of 3.7 m x 20 m x 3.7 m (height). The average density in the calorimeter is 1.4 gm/sq cm, and the average Z is 21. The detector was triggered by four 0.6 sq m scintillators placed immediately on the top of the calorimeter. The trigger required at least two of these four counters. The accompanying extensive air showers (EAS) was sampled by 14 scintillation counters located up to 15 m from the calorimeter. Several off line cuts have been applied to the data. Demanding five particles in at least two of the trigger detectors, a total of 20 particles in all of them together, and an arrival angle for the shower 450 deg reduced the data sample to 11053 events. Of these in 4869 cases, a computer algorithm found at least three muons in the calorimeter.

Goodman, J. A.↗
DOE OSTI2024

Performance of the CMS high-level trigger during LHC Run 2

The CERN LHC provided proton and heavy ion collisions during its Run 2 operation period from 2015 to 2018. Proton-proton collisions reached a peak instantaneous luminosity of 2.1× 10 34 cm -2 s -1 , twice the initial design value, at √(s)=13 TeV. The CMS experiment records a subset of the collisions for further processing as part of its online selection of data for physics analyses, using a two-level trigger system: the Level-1 trigger, implemented in custom-designed electronics, and the high-level trigger, a streamlined version of the offline reconstruction software running on a large computer farm. This paper presents the performance of the CMS high-level trigger system during LHC Run 2 for physics objects, such as leptons, jets, and missing transverse momentum, which meet the broad needs of the CMS physics program and the challenge of the evolving LHC and detector conditions. Sophisticated algorithms that were originally used in offline reconstruction were deployed online. Highlights include a machine-learning b tagging algorithm and a reconstruction algorithm for tau leptons that decay hadronically.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
NASA NTRS2008

Development and Testing of the VAHIRR Radar Product

Lightning Launch Commit Criteria (LLCC) and Flight Rules (FR) are used for launches and landings at government and commercial spaceports. They are designed to avoid natural and triggered lightning strikes to space vehicles, which can endanger the vehicle, payload, and general public. The previous LLCC and FR were shown to be overly restrictive, potentially leading to costly launch delays and scrubs. A radar algorithm called Volume Averaged Height Integrated Radar Reflectivity (VAHIRR), along with new LLCC and FR for anvil clouds, were developed using data collected by the Airborne Field Mill II research program. VAHIRR is calculated at every horizontal position in the coverage area of the radar and can be displayed similar to a two-dimensional derived reflectivity product, such as composite reflectivity or echo tops. It is the arithmetic product of two quantities not currently generated by the Weather Surveillance Radar 1988 Doppler (WSR-88D): a volume average of the reflectivity measured in dBZ and the average cloud thickness based on the average echo top height and base height. This presentation will describe the VAHIRR algorithm, and then explain how the VAHIRR radar product was implemented and tested on a clone of the National Weather Service's (NWS) Open Radar Product Generator (ORPG-clone). The VAHIRR radar product was then incorporated into the Advanced Weather Interactive Processing System (AWIPS), to make it more convenient for weather forecasters to utilize. Finally, the reliability of the VAHIRR radar product was tested with real-time level II radar data from the WSR-88D NWS Melbourne radar.

Barrett, Joe III↗
DOE OSTI2025

Semi-inclusive direct photon + jet and 𝜋 0 + jet correlations measured in 𝑝 + 𝑝 and central Au + Au collisions at $\sqrt{s_{NN}}$ = 200GeV

The STAR experiment at RHIC reports new measurements of jet quenching based on the semi- inclusive distribution of charged-particle jets recoiling from direct photon (γ dir ) and neutral pion (π 0 ) triggers in pp and central Au + Au collisions at $\sqrt{s_{NN}}$ = 200 GeV, for triggers in the range 9 < $E$$^{trig}_{T}$ < 20 GeV. The datasets have integrated luminosities of 3.9nb −1 for Au + Au and 23pb −1 for 𝑝𝑝 collisions. Jets are reconstructed using the anti-𝑘 𝑇 algorithm with resolution parameters 𝑅 = 0.2 and 0.5. The large uncorrelated jet background in central Au + Au collisions is corrected using a mixed-event approach, which enables precise charged-particle jet measurements at low transverse momentum 𝑝$^{ch}_{𝑇,jet}$ and large 𝑅. Recoil-jet distributions are reported in the range 𝑝$^{ch}_{𝑇,jet}$ < 25 GeV/𝑐. Comparison of the distributions measured in 𝑝𝑝 and Au + Au collisions reveals strong medium-induced jet yield suppression for 𝑅 = 0.2 with markedly less suppression for 𝑅 = 0.5. Comparison is also made to theoretical models incorporating jet quenching. Furthermore, these data provide new insight into the mechanisms underlying jet quenching and the angular dependence of medium-induced jet-energy transport and provide new constraints on modeling such effects.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2023

Development of ML FPGA Filter for Particle Identification and Tracking in Real Time

Real-time data processing is a frontier field in experimental particle physics. Machine Learning methods are widely used and have proven to be very powerful in particle physics. The growing computational power of modern FPGA boards allows us to add more sophisticated algorithms for real time data processing. Many tasks could be solved using modern Machine Learning (ML) algorithms which are naturally suited for FPGA architectures. The FPGA-based machine learning algorithm provides an extremely low, sub-microsecond, latency decision and makes information-rich data sets for event selection. We report work has started to evaluate an FPGA based Machine Learning (ML) algorithm for a real-time particle identification and tracking with Transition Radiation Detector (TRD) and e/m calorimeter. The first target is the GlueX experiment, with a plan to build a TRD based on GEM technology. GlueX trigger latency is 3.3 μs.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
NASA NTRS1989

Improved effectiveness of GPS RAIM through ridge regression signal processing

A measurement processing method has been developed which markedly improves the GPS Receiver Autonomous Integrity Monitoring (RAIM) software-based algorithm system's effectiveness in detecting satellite signal failures. Detection is via the consistency of a redundant set of pseudorange measurements. When five satellites are in view, five different subsolutions can be calculated; the integrity alarm is triggered on the basis of subsolution comparisons. Because a poor distribution of the satellites also causes RAIM subsolution scattering, a methodology for selecting the covariance matrix is presented which incorporates ridge regression into a Kalman filter.

Kelly, Robert J.↗
NASA NTRS2011

Bridging Empirical and Physical Approaches for Landslide Monitoring and Early Warning

Rainfall-triggered landslides typically occur and are evaluated at local scales, using slope-stability models to calculate coincident changes in driving and resisting forces at the hillslope level in order to anticipate slope failures. Over larger areas, detailed high resolution landslide modeling is often infeasible due to difficulties in quantifying the complex interaction between rainfall infiltration and surface materials as well as the dearth of available in situ soil and rainfall estimates and accurate landslide validation data. This presentation will discuss how satellite precipitation and surface information can be applied within a landslide hazard assessment framework to improve landslide monitoring and early warning by considering two disparate approaches to landslide hazard assessment: an empirical landslide forecasting algorithm and a physical slope-stability model. The goal of this research is to advance near real-time landslide hazard assessment and early warning at larger spatial scales. This is done by employing high resolution surface and precipitation information within a probabilistic framework to provide more physically-based grounding to empirical landslide triggering thresholds. The empirical landslide forecasting tool, running in near real-time at http://trmm.nasa.gov, considers potential landslide activity at the global scale and relies on Tropical Rainfall Measuring Mission (TRMM) precipitation data and surface products to provide a near real-time picture of where landslides may be triggered. The physical approach considers how rainfall infiltration on a hillslope affects the in situ hydro-mechanical processes that may lead to slope failure. Evaluation of these empirical and physical approaches are performed within the Land Information System (LIS), a high performance land surface model processing and data assimilation system developed within the Hydrological Sciences Branch at NASA's Goddard Space Flight Center. LIS provides the capabilities to quantify uncertainty from model inputs and calculate probabilistic estimates for slope failures. Results indicate that remote sensing data can provide many of the spatiotemporal requirements for accurate landslide monitoring and early warning; however, higher resolution precipitation inputs will help to better identify small-scale precipitation forcings that contribute to significant landslide triggering. Future missions, such as the Global Precipitation Measurement (GPM) mission will provide more frequent and extensive estimates of precipitation at the global scale, which will serve as key inputs to significantly advance the accuracy of landslide hazard assessment, particularly over larger spatial scales.

Kirschbaum, Dalia↗
NASA NTRS2009

Processing Ocean Images to Detect Large Drift Nets

A computer program processes the digitized outputs of a set of downward-looking video cameras aboard an aircraft flying over the ocean. The purpose served by this software is to facilitate the detection of large drift nets that have been lost, abandoned, or jettisoned. The development of this software and of the associated imaging hardware is part of a larger effort to develop means of detecting and removing large drift nets before they cause further environmental damage to the ocean and to shores on which they sometimes impinge. The software is capable of near-realtime processing of as many as three video feeds at a rate of 30 frames per second. After a user sets the parameters of an adjustable algorithm, the software analyzes each video stream, detects any anomaly, issues a command to point a high-resolution camera toward the location of the anomaly, and, once the camera has been so aimed, issues a command to trigger the camera shutter. The resulting high-resolution image is digitized, and the resulting data are automatically uploaded to the operator s computer for analysis.

Veenstra, Tim↗
DOE OSTI2023

Devices and methods for increasing the speed or power efficiency of a computer when performing machine learning using spiking neural networks

A method for increasing a speed and efficiency of a computer when performing machine learning using spiking neural networks. The method includes computer-implemented operations; that is, operations that are solely executed on a computer. The method includes receiving, in a spiking neural network, a plurality of input values upon which a machine learning algorithm is based. The method also includes correlating, for each input value, a corresponding response speed of a corresponding neuron to a corresponding equivalence relationship between the input value to a corresponding latency of the corresponding neuron. Neurons that trigger faster than other neurons represent close relationships between input values and neuron latencies. Latencies of the neurons represent data points used in performing the machine learning. A plurality of equivalence relationships are formed as a result of correlating. The method includes performing the machine learning using the plurality of equivalence relationships.

Vineyard, Craig Michael↗