Search NASA⌕ Search

SEARCH · Search NASA

Results for “Filtering 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 1,747 records · Page 97
All sourcesNASA NTRSDOE OSTI
NASA NTRS2002

Spectral Re-Growth Reduction for CCSDS 8-D 8-PSK TCM

This report presents a study on the CCSDS recommended 8-dimensional 8 PSK Trellis Coded Modulation (TCM) scheme. The important steps of the CCSDS scheme include: conversion of serial data into parallel form, differential encoding, convolutional encoding, constellation mapping, and filtering the 8-PSK symbols using the square root raised cosine (SRRC) pulses. The last step, namely the filtering of the 8 PSK symbols using SRRC pulses, significantly affects the bandwidth of the signal. If a nonlinear power amplifier is used, the SRRC filtered signal creates spectral regrowth. The purpose of this report is to investigate a technique, called the smooth phase interpolated keying (SPIK), that can provide an alternative to SRRC filtering so that good spectral as well as power efficiencies can be obtained with the CCSDS encoder. The results of this study show that the CCSDS encoder does not affect the spectral shape of the SRRC filtered signal or the SPIK signal. When a nonlinear traveling wave tube amplifier (TWTA) is used, the spectral performance of the SRRC signal degrades significantly while the spectral performance of SPIK remains unaffected. The degrading effect of a nonlinear solid state power amplifier (SSPA) on SRRC is found to be less than that due to a nonlinear TWTA. However, in both cases, the spectral performance of the SRRC modulated signal is worse than that of the SPIK signal. The bit error rate (BER) performance of the SRRC signal in a linear amplifier environment is about 2.5 dB better than that of the SPIK signal when both the receivers use algorithms of similar complexity. In a nonlinear TWTA environment, the SRRC signal requires accurate phase tracking since the TWTA introduces additional phase distortion. This problem does not arise with SPIK signal due to its constant envelope property. When a nonlinear amplifier is used, the SRRC method loses nearly 1 dB in the bit error rate performance. The SPIK signal does not lose any performance. Thus the performance gap between SRRC and SPIK reduces. The BER performance of SPIK can be improved even further by using a more optimal receiver. A similar optimal receiver for SRRC is quite complex since the amplifier distorts the pulse shape. However, this requires further investigation and is not covered in this report.

Borah, Deva K.↗
NASA NTRS2003

EVA Robotic Assistant Project: Platform Attitude Prediction

The Robotic Systems Technology Branch is currently working on the development of an EVA Robotic Assistant under the sponsorship of the Surface Systems Thrust of the NASA Cross Enterprise Technology Development Program (CETDP). This will be a mobile robot that can follow a field geologist during planetary surface exploration, carry his tools and the samples that he collects, and provide video coverage of his activity. Prior experiments have shown that for such a robot to be useful it must be able to follow the geologist at walking speed over any terrain of interest. Geologically interesting terrain tends to be rough rather than smooth. The commercial mobile robot that was recently purchased as an initial testbed for the EVA Robotic Assistant Project, an ATRV Jr., is capable of faster than walking speed outside but it has no suspension. Its wheels with inflated rubber tires are attached to axles that are connected directly to the robot body. Any angular motion of the robot produced by driving over rough terrain will directly affect the pointing of the on-board stereo cameras. The resulting image motion is expected to make tracking of the geologist more difficult. This will either require the tracker to search a larger part of the image to find the target from frame to frame or to search mechanically in pan and tilt whenever the image motion is large enough to put the target outside the image in the next frame. This project consists of the design and implementation of a Kalman filter that combines the output of the angular rate sensors and linear accelerometers on the robot to estimate the motion of the robot base. The motion of the stereo camera pair mounted on the robot that results from this motion as the robot drives over rough terrain is then straightforward to compute. The estimates may then be used, for example, to command the robot s on-board pan-tilt unit to compensate for the camera motion induced by the base movement. This has been accomplished in two ways: first, a standalone head stabilizer has been implemented and second, the estimates have been used to influence the search algorithm of the stereo tracking algorithm. Studies of the image motion of a tracked object indicate that the image motion of objects is suppressed while the robot crossing rough terrain. This work expands the range of speed and surface roughness over which the robot should be able to track and follow a field geologist and accept arm gesture commands from the geologist.

Nickels, Kevin M.↗
NASA NTRS2012

Progress Toward Affordable High Fidelity Combustion Simulations Using Filtered Density Functions for Hypersonic Flows in Complex Geometries

Significant progress has been made in the development of subgrid scale (SGS) closures based on a filtered density function (FDF) for large eddy simulations (LES) of turbulent reacting flows. The FDF is the counterpart of the probability density function (PDF) method, which has proven effective in Reynolds averaged simulations (RAS). However, while systematic progress is being made advancing the FDF models for relatively simple flows and lab-scale flames, the application of these methods in complex geometries and high speed, wall-bounded flows with shocks remains a challenge. The key difficulties are the significant computational cost associated with solving the FDF transport equation and numerically stiff finite rate chemistry. For LES/FDF methods to make a more significant impact in practical applications a pragmatic approach must be taken that significantly reduces the computational cost while maintaining high modeling fidelity. An example of one such ongoing effort is at the NASA Langley Research Center, where the first generation FDF models, namely the scalar filtered mass density function (SFMDF) are being implemented into VULCAN, a production-quality RAS and LES solver widely used for design of high speed propulsion flowpaths. This effort leverages internal and external collaborations to reduce the overall computational cost of high fidelity simulations in VULCAN by: implementing high order methods that allow reduction in the total number of computational cells without loss in accuracy; implementing first generation of high fidelity scalar PDF/FDF models applicable to high-speed compressible flows; coupling RAS/PDF and LES/FDF into a hybrid framework to efficiently and accurately model the effects of combustion in the vicinity of the walls; developing efficient Lagrangian particle tracking algorithms to support robust solutions of the FDF equations for high speed flows; and utilizing finite rate chemistry parametrization, such as flamelet models, to reduce the number of transported reactive species and remove numerical stiffness. This paper briefly introduces the SFMDF model (highlighting key benefits and challenges), and discusses particle tracking for flows with shocks, the hybrid coupled RAS/PDF and LES/FDF model, flamelet generated manifolds (FGM) model, and the Irregularly Portioned Lagrangian Monte Carlo Finite Difference (IPLMCFD) methodology for scalable simulation of high-speed reacting compressible flows.

Drozda, Tomasz G.↗
NASA NTRS2020

Precision Assessment of the HPLC Phytoplankton Pigment Dataset Analyzed by NASA to Quantify Global Variability in Support of Ocean Color Remote Sensing

The ability to generate chlorophyll a (Chl a) assessments from ocean color orbital sensors, such as VIIRS and MODIS, that satisfy the requirements to be climate-quality data record (CDR) quality is contingent in part on the quality of the in situ ground or sea truth observations that serve as datasets for vicarious calibration and algorithm validation activities. NASA has a mandate to collect, analyze, and distribute in situ data of the highest possible quality with documented uncertainties and in keeping with established performance metrics. Using a dataset of over 18,000 HPLC phytoplankton pigment samples representing water collected in all major ocean basins analyzed a central laboratory (Field Support Group (FSG) of the Ocean Ecology Laboratory (OEL) at NASA Goddard Space Flight Center (GSFC)), we performed an assessment of the global precision among sample replicates of Chl a as well as major accessory pigments. We investigated the impacts of filtration volume, water basin, collection technique, pigment concentration, and different filtration volumes for replicate filters on replicate filter precision, as well as investigating any pigment-specific differences. Our results quantify sample variability with the goal of understanding any systemic biases or biogeographic influences.

Thomas, Crystal S.↗
DOE OSTI2020

Reinforcement-based Program Induction in a Neural Virtual Machine

We present a neural virtual machine that can be trained to perform algorithmic tasks. Rather than combining a neural controller with non-neural memory storage as has been done in the past, this architecture is purely neural and emulates tape-based memory via fast associative weights (onestep learning). Here we formally define the architecture, and then extend the system to learn programs using recurrent policy gradient reinforcement learning based on examples of program inputs labeled with corresponding output targets, which are compared against actual output to generate a sparse reward signal. We describe the policy gradient training procedure used, and report its empirical performance on a number of smallscale list processing tasks, such as finding the maximum list element, filtering out certain elements, and reversing the order of the elements. These results show that program induction via reinforcement learning is possible using sparse rewards and solely neural computations.

Katz, Garrett E.↗
NASA NTRS1987

Signal averaging technique for noninvasive recording of late potentials in patients with coronary artery disease

An advanced non invasive signal averaging technique was used to detect late potentials in two groups of patients: Group A (24 patients) with coronary artery disease (CAD) and without sustained ventricular tachycardia (VT) and Group B (8 patients) with CAD and sustained VT. Recorded analog data were digitized and aligned using a cross correlation function with fast Fourier transform schema, averaged and band pass filtered between 60 and 200 Hz with a non-recursive digital filter. Averaged filtered waveforms were analyzed by computer program for 3 parameters: (1) filtered QRS (fQRS) duration (2) interval between the peak of the R wave peak and the end of fQRS (R-LP) (3) RMS value of last 40 msec of fQRS (RMS). Significant change was found between Groups A and B in fQRS (101 -/+ 13 msec vs 123 -/+ 15 msec; p < .0005) and in R-LP vs 52 -/+ 11 msec vs 71-/+18 msec, p <.002). We conclude that (1) the use of a cross correlation triggering method and non-recursive digital filter enables a reliable recording of late potentials from the body surface; (2) fQRS and R-LP durations are sensitive indicators of CAD patients susceptible to VT.

NASA Discipline Cardiopulmonary↗
DOE OSTI2019

Ultrafast processing of pixel detector data with machine learning frameworks

Modern photon science performed at high repetition rate free-electron laser (FEL) facilities and beyond relies on 2D pixel detectors operating at increasing frequencies (towards 100 kHz at LCLS-II) and producing rapidly increasing amounts of data (towards TB/s). This data must be rapidly stored for offline analysis and summarized in real time for online feedback to the scientists. While at LCLS all raw data has been stored, at LCLS-II this would lead to a prohibitive cost; instead, enabling real time processing of pixel detector data (dark, gain, common mode, background, charge summing, subpixel position, photon counting, data summarization) allows reducing the size and cost of online processing, offline processing and storage by orders of magnitude while preserving full photon information. This could be achieved by taking advantage of the compressibility of sparse data typical for LCLS-II applications. Faced with a similar big data challenge a decade ago, computer vision stimulated revolutionary advances in machine learning hardware and software. We investigated if these developments are useful in the field of data processing for high speed pixel detectors and found that typical deep learning models and autoencoder architectures failed to yield useful noise reduction while preserving full photon information, presumably because of the very different statistics and feature sets in computer vision and radiation imaging. However, the raw performance of modern frameworks like Tensorflow inspired us to redesign in Tensorflow mathematically equivalent versions of the state-of-the-art, “classical” algorithms used at LCLS. The novel Tensorflow models resulted in elegant, compact and hardware agnostic code, gaining 1 to 2 orders of magnitude faster processing on an inexpensive consumer GPU, reducing by 3 orders of magnitude the projected cost of online analysis and compression without photon loss at LCLS-II. The novel Tensorflow models also enabled ongoing development of a pipelined hardware system expected to yield an additional 3 to 4 orders of magnitude speedup, necessary for meeting the data acquisition and storage requirements at LCLS-II, potentially enabling acquiring every single FEL pulse at full speed. Computer vision a decade ago was dominated by hand-crafted filters; their structure inspired the deep learning revolution resulting in modern deep convolutional networks; similarly, our novel Tensorflow filters provide inspiration for designing future deep learning models for ultrafast and efficient processing and classification of pixel detector images at FEL facilities.

47 OTHER INSTRUMENTATION↗
DOE OSTI2022

The miniJPAS survey quasar selection – I. Mock catalogues for classification

In this series of papers, we employ several machine learning (ML) methods to classify the point-like sources from the miniJPAS catalogue, and identify quasar candidates. Since no representative sample of spectroscopically confirmed sources exists at present to train these ML algorithms, we rely on mock catalogues. In this first paper, we develop a pipeline to compute synthetic photometry of quasars, galaxies, and stars using spectra of objects targeted as quasars in the Sloan Digital Sky Survey . To match the same depths and signal-to-noise ratio distributions in all bands expected for miniJPAS point sources in the range 17.5 ≤ r < 24, we augment our sample of available spectra by shifting the original r-band magnitude distributions towards the faint end, ensure that the relative incidence rates of the different objects are distributed according to their respective luminosity functions, and perform a thorough modelling of the noise distribution in each filter, by sampling the flux variance either from Gaussian realizations with given widths, or from combinations of Gaussian functions. Finally, we also add in the mocks the patterns of non-detections which are present in all real observations. Although the mock catalogues presented in this work are a first step towards simulated data sets that match the properties of the miniJPAS observations, these mocks can be adapted to serve the purposes of other photometric surveys.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Exploiting voxel-sparsity for bone imaging with sparse-view cone-beam computed tomography

An optimization-based image reconstruction frame work is developed specifically for bone imaging. This framework exploits voxel-sparsity by use of ℓ 1 -norm image regularization and it enables image reconstruction from sparse-view cone-beam computed tomography (CBCT) acquisition. The effectiveness of the voxel-sparsity regularization is enhanced by using a blurred image representation. Ramp-filtering is included in the data discrepancy term and it has the effect of acting as a preconditioner, reducing the necessary number of iterations. The bone image reconstruction framework is demonstrated on CBCT data taken from an equine metacarpal condyle specimen.

Bone imaging↗
NASA NTRS2002

Algorithms and Sensors for Small Robot Path Following

Tracked mobile robots in the 20 kg size class are under development for applications in urban reconnaissance. For efficient deployment, it is desirable for teams of robots to be able to automatically execute path following behaviors, with one or more followers tracking the path taken by a leader. The key challenges to enabling such a capability are (l) to develop sensor packages for such small robots that can accurately determine the path of the leader and (2) to develop path following algorithms for the subsequent robots. To date, we have integrated gyros, accelerometers, compass/inclinometers, odometry, and differential GPS into an effective sensing package. This paper describes the sensor package, sensor processing algorithm, and path tracking algorithm we have developed for the leader/follower problem in small robots and shows the result of performance characterization of the system. We also document pragmatic lessons learned about design, construction, and electromagnetic interference issues particular to the performance of state sensors on small robots.

path following↗
DOE OSTI2022

tapir: A tool for topologies, amplitudes, partial fraction decomposition and input for reductions

The demand for precision predictions in the field of high energy physics has dramatically increased over recent years. Experiments conducted at the LHC, as well as precision measurements at the intensity frontier such as Belle II require equally precise theoretical predictions to make full use of the acquired data. To match the experimental precision, second-, third- and, for certain quantities, even higher-order calculations in perturbative quantum field theory are required. To facilitate such calculations, computer software automating as many steps as possible is required. Yet, each calculation poses different challenges and thus, a high level of configurability is required. In this context we present tapir: a tool for identification, manipulation and minimization of Feynman integral families. It is designed to integrate in toolchains based on the computer algebra system FORM, the use of which is common practice in the field. tapir can be used to reduce the complexity of multi-loop problems with cut-filters, topology mapping, partial fraction decomposition and alike. Program Title:tapir CPC Library link to program files:https://doi.org/10.17632/ptc9t46xyn.1 Developer's repository link:https://gitlab.com/tapir-devs/tapir Licensing provisions: GPLv3 Programming language:python 3, C++ Nature of problem: Multi-loop computations require the automatization of a large number of different tasks related to Feynman integral topologies. Among them are the identification and minimization of integral topologies, partial fraction decomposition of topologies in the case of linearly dependent propagators as well as mapping scalar products of loop momenta to scalar functions. Solution method: The minimization of topologies is performed by comparison of their respective Nickel indices [1], even further minimization utilizes Pak's algorithm [2]. To efficiently map scalar products of loop momenta to scalar functions FORM [3] code is generated. Additional comments including restrictions and unusual features: Minimization based on Pak's algorithm slows down for many lines and scales. A coarser minimization using the Nickel indices, however, is still possible. [1]B. Nickel, D. Meiron, G.A.J. Baker, Compilation of 2-pt and 4-pt graphs for continuous spin model, Report, University of Guelph, 1977.[2]A. Pak, J. Phys. Conf. Ser. 368 (2012) 012049, https://doi.org/10.1088/1742-6596/368/1/012049, arXiv:1111.0868.[3]B. Ruijl, T. Ueda, J. Vermaseren, FORM version 4.2, arXiv:1707.06453, 7 2017.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
NASA NTRS2018

Evaluation of SMAP Downscaled Brightness Temperature Using SMAPEx-4/5 Airborne Observations

The Soil Moisture Active and Passive (SMAP) mission, launched by the National Aeronautics and Space Administration (NASA) on 31st January 2015, was designed to provide global soil moisture every 2 to 3 days at 9 km resolution by downscaling SMAP passive microwave observations obtained at 36 km resolution using active microwave observations at 3 km resolution, and then retrieving soil moisture from the resulting 9 km brightness temperature product. This study evaluated the SMAP Active/Passive (AP) downscaling algorithm together with other resolution enhancement techniques. Airborne passive microwave observations acquired at 1 km resolution over the Murrumbidgee River catchment in south-eastern Australia during the fourth and fifth Soil Moisture Active Passive Experiments (SMAPEx-4/5) were used as reference data. The SMAPEx-4/5 data were collected in May and September 2015, respectively, and aggregated to 9 km for direct comparison with a number of available resolution-enhanced brightness temperature estimates. The results show that the SMAP AP downscaled brightness temperature had a correlation coefficient (R) of 0.84 and Root-Mean-Squared Error (RMSE) of ~10 K, while SMAP Enhanced, Nearest Neighbour, Weighted Average, and the Smoothing Filter-based Modulation (SFIM) brightness temperature estimates had somewhat better performance (RMSEs of ~7 K and an R exceeding 0.9). Although the SFIM had the lowest unbiased RMSE of ~6 K, the effect of cloud cover on Ka-band observations limits data availability.

downscaled brightness temperature↗
DOE OSTI2019

Portable COTS RGB Wavefront System for Real-time Turbulence Mitigation

Atmospheric effects reduce the ability to detect or identify objects of interest due to spatial and spectral scintillation. Algorithms alone cannot correct for turbulence over long-range horizontal paths. While adaptive optics methods work well for astronomical viewing, they are not readily adapted for complex imagery over long-range horizontal paths. At MSS Parallel in 2016, we presented our work for the red/green/blue (RGB) wavefront sensor. Since then, we have completed our system to combine an off-the-shelf high-frame-rate USB3 color camera with a compact 90 mm Maksutov-Cassegrain telescope and an OKO deformable mirror (DM) to create a portable, real-time RGB wavefront sensor and turbulence mitigation system. The telescope’s aperture is covered with an aluminum plate holding three separate narrowband color filters. These filters, specifically matched to the relative spectral response of the sensor’s Bayer mask, provide nearly complete separation of the three color images. This device enables portable imaging through atmospheric turbulence and analysis thereof, where relative motion of the three-color Bayer-pixel video provides real-time wavefront information on the second-order statistics of Kolmogorov turbulence. We measured relative RGB pixel shifts and controlled this mirror using an ODROID single-board computer. We developed a bench setup to calibrate the relationship between pixel shifts and the deformable mirror actuators. We demonstrated the feasibility of this system on a benchtop and on scenes at 700 meters showing real time correction.

47 OTHER INSTRUMENTATION↗
NASA NTRS2021

Model-Independent Time-Delay Interferometry Based on Principal Component Analysis

With a laser interferometric gravitational-wave detector in separate free flying spacecraft, the only way to achieve detection is to mitigate the dominant noise arising from the frequency fluctuations of the lasers via postprocessing. The noise can be effectively filtered out on the ground through a specific technique called time-delay interferometry (TDI), which relies on the measurements of time-delays between spacecraft and careful modeling of how laser noise enters the interferometric data. Recently, this technique has been recast into a matrix-based formalism by several authors, offering a different perspective on TDI, particularly by relating it to principal component analysis (PCA). In this work, we demonstrate that we can cancel laser frequency noise by directly applying PCA to a set of shifted data samples, without any prior knowledge of the relationship between single-link measurements and noise, nor time-delays. We show that this fully data-driven algorithm achieves a gravitational-wave sensitivity similar to classic TDI.

Quentin Baghi↗
DOE OSTI2025

Comparisons of the v11.1 Orbiting Carbon Observatory‐2 (OCO‐2) X CO2 Measurements With GGG2020 TCCON

The Orbiting Carbon Observatory 2 (OCO-2) is NASA's first Earth observation satellite mission dedicated to studying the sources and sinks of carbon dioxide (CO 2 ) on a global scale. The observations of reflected sunlight are inverted in a retrieval algorithm to produce estimates of the dry air mole-fractions of CO 2 (X CO2 ). The OCO-2 Level 2 data release, version 11.1 (v11.1) retrievals from the Atmospheric Carbon Observations from Space (ACOS) algorithm, includes significant improvements in the X CO2 data product compared to older OCO-2 data versions. This work compares the v11.1 X CO2 from OCO-2 against X CO2 estimates collected from a global ground-based network known as the Total Carbon Column Observing Network (TCCON), OCO-2's primary validation source. The OCO-2 project provides a version of the Level 2 data product, called “lite” files that include calibrated and bias-corrected XCO2 values, accessible together with all OCO-2 data products through the NASA Goddard Earth Sciences Data and Information Services Center (GES DISC). This work shows that OCO-2 X CO2 observations made between September 2014 and December 2023, after quality filtering and the application of an averaging kernel correction, agree well with coincident TCCON data for all OCO-2 observational modes of land (nadir, glint, target) and ocean (glint). The aggregated, bias-corrected, and quality-filtered absolute average bias values are less than or equal to 0.20 parts per million (ppm) globally for all OCO-2 observation modes, where the biases do not indicate a statistically significant time dependence. The land nadir/glint mode has the lowest bias value of −0.03 ± 0.85 ppm.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS2011

4 Vesta in Color: High Resolution Mapping from Dawn Framing Camera Images

Rotational surface variations on asteroid 4 Vesta have been known from ground-based and HST observations, and they have been interpreted as evidence of compositional diversity. NASA s Dawn mission entered orbit around Vesta on July 16, 2011 for a year-long global characterization. The framing cameras (FC) onboard the Dawn spacecraft will image the asteroid in one clear (broad) and seven narrow band filters covering the wavelength range between 0.4-1.0 microns. We present color mapping results from the Dawn FC observations of Vesta obtained during Survey orbit (approx.3000 km) and High-Altitude Mapping Orbit (HAMO) (approx.950 km). Our aim is to create global color maps of Vesta using multi spectral FC images to identify the spatial extent of compositional units and link them with other available data sets to extract the basic mineralogy. While the VIR spectrometer onboard Dawn has higher spectral resolution (864 channels) allowing precise mineralogical assessment of Vesta s surface, the FC has three times higher spatial resolution in any given orbital phase. In an effort to extract maximum information from FC data we have developed algorithms using laboratory spectra of pyroxenes and HED meteorites to derive parameters associated with the 1-micron absorption band wing. These parameters will help map the global distribution of compositionally related units on Vesta s surface. Interpretation of these units will involve the integration of FC and VIR data.

Reddy, V.↗
NASA NTRS1994

Optimal addition of images for detection and photometry

In this paper we describe weighting techniques used for the optimal coaddition of charge coupled devices (CCD) frames with differing characteristics. Optimal means maximum signal to noise (S/N) for stellar objects. We derive formulas for four applications: (1) object detection via matched filter, (2) object detection identical to DAOFIND, (3) aperture photometry, and (4) ALLSTAR profile-fitting photometry. We have included examples involving 21 frames for which either the sky brightness or image resolution varied by a factor of 3. The gains in S/N were modest for most of the examples, except for DAOFIND detection with varying image resolution which exhibited a substantial S/N increase. Even though the only consideration was maximizing S/N, the image resolution was seen to improve for most of the variable resolution examples. Also discussed are empirical fits for the weighting and the availability of the program, WEIGHT, used to generate the weighting for the individual frames. Finally, we include appendices describing the effects of clipping algorithms and a scheme for star/galaxy and cosmic-ray/star discrimination. scheme for star/galaxy and cosmic-ray/star discrimination.

Fischer, Philippe↗
NASA NTRS2007

POD Model Reconstruction for Gray-Box Fault Detection

Proper orthogonal decomposition (POD) is the mathematical basis of a method of constructing low-order mathematical models for the "gray-box" fault-detection algorithm that is a component of a diagnostic system known as beacon-based exception analysis for multi-missions (BEAM). POD has been successfully applied in reducing computational complexity by generating simple models that can be used for control and simulation for complex systems such as fluid flows. In the present application to BEAM, POD brings the same benefits to automated diagnosis. BEAM is a method of real-time or offline, automated diagnosis of a complex dynamic system.The gray-box approach makes it possible to utilize incomplete or approximate knowledge of the dynamics of the system that one seeks to diagnose. In the gray-box approach, a deterministic model of the system is used to filter a time series of system sensor data to remove the deterministic components of the time series from further examination. What is left after the filtering operation is a time series of residual quantities that represent the unknown (or at least unmodeled) aspects of the behavior of the system. Stochastic modeling techniques are then applied to the residual time series. The procedure for detecting abnormal behavior of the system then becomes one of looking for statistical differences between the residual time series and the predictions of the stochastic model.

Park, Han↗