Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Misclassification”

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 91 records · Page 5
All sourcesNASA NTRSDOE OSTI
NASA NTRS1991

Multi-frequency and polarimetric radar backscatter signatures for discrimination between agricultural crops at the Flevoland experimental test site

We describe the calibration and analysis of multi-frequency, multi-polarization radar backscatter signatures over an agriculture test site in the Netherlands. The calibration procedure involved two stages: in the first stage, polarimetric and radiometric calibrations (ignoring noise) were carried out using square-base trihedral corner reflector signatures and some properties of the clutter background. In the second stage, a novel algorithm was used to estimate the noise level in the polarimetric data channels by using the measured signature of an idealized rough surface with Bragg scattering (the ocean in this case). This estimated noise level was then used to correct the measured backscatter signatures from the agriculture fields. We examine the significance of several key parameters extracted from the calibrated and noise-corrected backscatter signatures. The significance is assessed in terms of the ability to uniquely separate among classes from 13 different backscatter types selected from the test site data, including eleven different crops, one forest and one ocean area. Using the parameters with the highest separation for a given class, we use a hierarchical algorithm to classify the entire image. We find that many classes, including ocean, forest, potato, and beet, can be identified with high reliability, while the classes for which no single parameter exhibits sufficient separation have higher rates of misclassification. We expect that modified decision criteria involving simultaneous consideration of several parameters increase performance for these classes.

Freeman, A.↗
NASA NTRS1991

Mathematical morphology for automated analysis of remotely sensed objects in radar images

A symbiosis of pyramidal segmentation and morphological transmission is described. The pyramidal segmentation portion of the symbiosis has resulted in low (2.6 percent) misclassification error rate for a one-look simulation. Other simulations indicate lower error rates (1.8 percent for a four-look image). The morphological transformation portion has resulted in meaningful partitions with a minimal loss of fractal boundary information. An unpublished version of Thicken, suitable for watersheds transformations of fractal objects, is also presented. It is demonstrated that the proposed symbiosis works with SAR (synthetic aperture radar) images: in this case, a four-look Seasat image of sea ice. It is concluded that the symbiotic forms of both segmentation and morphological transformation seem well suited for unsupervised geophysical analysis.

Daida, Jason M.↗
NASA NTRS1993

Classifying multispectral data by neural networks

Several energy functions for synthesizing neural networks are tested on 2-D synthetic data and on Landsat-4 Thematic Mapper data. These new energy functions, designed specifically for minimizing misclassification error, in some cases yield significant improvements in classification accuracy over the standard least mean squares energy function. In addition to operating on networks with one output unit per class, a new energy function is tested for binary encoded outputs, which result in smaller network sizes. The Thematic Mapper data (four bands were used) is classified on a single pixel basis, to provide a starting benchmark against which further improvements will be measured. Improvements are underway to make use of both subpixel and superpixel (i.e. contextual or neighborhood) information in tile processing. For single pixel classification, the best neural network result is 78.7 percent, compared with 71.7 percent for a classical nearest neighbor classifier. The 78.7 percent result also improves on several earlier neural network results on this data.

Telfer, Brian A.↗
NASA NTRS1993

Satellite inventory of Minnesota forest resources

The methods and results of using Landsat Thematic Mapper (TM) data to classify and estimate the acreage of forest covertypes in northeastern Minnesota are described. Portions of six TM scenes covering five counties with a total area of 14,679 square miles were classified into six forest and five nonforest classes. The approach involved the integration of cluster sampling, image processing, and estimation. Using cluster sampling, 343 plots, each 88 acres in size, were photo interpreted and field mapped as a source of reference data for classifier training and calibration of the TM data classifications. Classification accuracies of up to 75 percent were achieved; most misclassification was between similar or related classes. An inverse method of calibration, based on the error rates obtained from the classifications of the cluster plots, was used to adjust the classification class proportions for classification errors. The resulting area estimates for total forest land in the five-county area were within 3 percent of the estimate made independently by the USDA Forest Service. Area estimates for conifer and hardwood forest types were within 0.8 and 6.0 percent respectively, of the Forest Service estimates. A trial of a second method of estimating the same classes as the Forest Service resulted in standard errors of 0.002 to 0.015. A study of the use of multidate TM data for change detection showed that forest canopy depletion, canopy increment, and no change could be identified with greater than 90 percent accuracy. The project results have been the basis for the Minnesota Department of Natural Resources and the Forest Service to define and begin to implement an annual system of forest inventory which utilizes Landsat TM data to detect changes in forest cover.

Bauer, Marvin E.↗
NASA NTRS1993

Optimal band selection for dimensionality reduction of hyperspectral imagery

Hyperspectral images have many bands requiring significant computational power for machine interpretation. During image pre-processing, regions of interest that warrant full examination need to be identified quickly. One technique for speeding up the processing is to use only a small subset of bands to determine the 'interesting' regions. The problem addressed here is how to determine the fewest bands required to achieve a specified performance goal for pixel classification. The band selection problem has been addressed previously Chen et al., Ghassemian et al., Henderson et al., and Kim et al.. Some popular techniques for reducing the dimensionality of a feature space, such as principal components analysis, reduce dimensionality by computing new features that are linear combinations of the original features. However, such approaches require measuring and processing all the available bands before the dimensionality is reduced. Our approach, adapted from previous multidimensional signal analysis research, is simpler and achieves dimensionality reduction by selecting bands. Feature selection algorithms are used to determine which combination of bands has the lowest probability of pixel misclassification. Two elements required by this approach are a choice of objective function and a choice of search strategy.

Stearns, Stephen D.↗
NASA NTRS1995

The distribution of early- and late-type galaxies in the Coma cluster

The spatial distribution and the morohology-density relation of Coma cluster galaxies are studied using a new homogeneous photmetric sample of 450 galaxies down to B = 16.0 mag with quantitative morphology classification. The sample covers a wide area (10 deg X 10 deg), extending well beyond the Coma cluster. Morphological classifications into early- (E+SO) and late-(S) type galaxies are made by an automated algorithm using simple photometric parameters, with which the misclassification rate is expected to be approximately 10% with respect to early and late types given in the Third Reference Catalogue of Bright Galaxies. The flattened distribution of Coma cluster galaxies, as noted in previous studies, is most conspicuously seen if the early-type galaxies are selected. Early-type galaxies are distributed in a thick filament extended from the NE to the WSW direction that delineates a part of large-scale structure. Spiral galaxies show a distribution with a modest density gradient toward the cluster center; at least bright spiral galaxies are present close to the center of the Coma cluster. We also examine the morphology-density relation for the Coma cluster including its surrounding regions.

Doi, M.↗
NASA NTRS1994

Design of Ternary Correlation Filters to Reduce Probability of Error

The problem of designing ternary phase and amplitude filters (TPAF's) that reduce the probability of image misclassification for a two-class image set is studied. The Fisher ratio is used as a measure of the correct classification rate, and an attempt is made to maximize this quantity in the filter designs. Given the nonanalytical nature of the design problem, a simulated annealing optimization technique is employed. Computer simulation results are presented for several cases including single in-class and out-of-class image sets and multiple image sets corresponding to the design of synthetic discriminant function filters. Significant improvements are found in expected rates of correct classification in comparison to binary phase-only filters and other TPAF designs. Approaches to accelerate the filter design process are also discussed.

Downie, John D.↗
NASA NTRS2001

Did the Eruption of the Mt. Pinatubo Volcano Affect Cirrus Properties?

Some observations suggest that the volcanic aerosols produced by the Mt. Pinatubo eruption may have altered cirrus properties. We look for evidence that such modification of cirrus is extensive enough to be climatically significant by comparing three satellite-based cirrus datasets produced by the ISCCP (International Satellite Cloud Climatology Project) analysis, the 'split-window' analysis, and 3I analysis. Since the former two have not been compared in detail before, we conduct such a comparison here. When applied to AVHRR (Advanced Very High Resolution Radiometer) data, both the ISCCP and split-window analyses identify about 0.2 - 0.3 cirrus cloud amount in tropical latitudes; however, there are detailed differences of classification for about half of these clouds. The discrepancies are attributed to the simplified assumptions made by both methods. The latter two datasets are derived from infrared radiances, so they are much less sensitive to volcanic aerosols than the ISCCP analysis. After the Mt. Pinatubo eruption, the ISCCP results indicate a dramatic decrease of thin cirrus (cloud top pressure less than 440 mb and visible optical thickness less than 1.3) over ocean, accompanied by a comparable increase of altocumulus and cumulus clouds; over land, there are no significant changes. In contrast, results from the split-window and 3I analyses show little change in thin cirrus amount over either ocean or land that is associated with the volcanic eruption. The ISCCP results can, therefore, be understood as a misclassification of thin cirrus because the additional reflected sunlight by the volcanic aerosol makes the cirrus clouds appear to be optically thicker. Examination of the split-window signature and the infrared emissivities from 3I show no significant change in infrared emissivity (or optical thickness). These results indicate that the Mt. Pinatubo volcanic aerosol did not have a significant systematic effect on tropical cirrus properties, but rather produced only temporary, local effects. Hence, these results indicate that there was no significant climate feedback produced by aerosol-cirrus-radiative interactions.

Luo, Zhengzhao↗
NASA NTRS2004

Use of Probability Distribution Functions for Discriminating Between Cloud and Aerosol in Lidar Backscatter Data

In this paper we describe the algorithm hat will be used during the upcoming Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) mission for discriminating between clouds and aerosols detected in two wavelength backscatter lidar profiles. We first analyze single-test and multiple-test classification approaches based on one-dimensional and multiple-dimensional probability density functions (PDFs) in the context of a two-class feature identification scheme. From these studies we derive an operational algorithm based on a set of 3-dimensional probability distribution functions characteristic of clouds and aerosols. A dataset acquired by the Cloud Physics Lidar (CPL) is used to test the algorithm. Comparisons are conducted between the CALIPSO algorithm results and the CPL data product. The results obtained show generally good agreement between the two methods. However, of a total of 228,264 layers analyzed, approximately 5.7% are classified as different types by the CALIPSO and CPL algorithm. This disparity is shown to be due largely to the misclassification of clouds as aerosols by the CPL algorithm. The use of 3-dimensional PDFs in the CALIPSO algorithm is found to significantly reduce this type of error. Dust presents a special case. Because the intrinsic scattering properties of dust layers can be very similar to those of clouds, additional algorithm testing was performed using an optically dense layer of Saharan dust measured during the Lidar In-space Technology Experiment (LITE). In general, the method is shown to distinguish reliably between dust layers and clouds. The relatively few erroneous classifications occurred most often in the LITE data, in those regions of the Saharan dust layer where the optical thickness was the highest.

Liu, Zhaoyan↗
NASA NTRS2004

Evaluation of an Enhanced Bank of Kalman Filters for In-Flight Aircraft Engine Sensor Fault Diagnostics

In this paper, an approach for in-flight fault detection and isolation (FDI) of aircraft engine sensors based on a bank of Kalman filters is developed. This approach utilizes multiple Kalman filters, each of which is designed based on a specific fault hypothesis. When the propulsion system experiences a fault, only one Kalman filter with the correct hypothesis is able to maintain the nominal estimation performance. Based on this knowledge, the isolation of faults is achieved. Since the propulsion system may experience component and actuator faults as well, a sensor FDI system must be robust in terms of avoiding misclassifications of any anomalies. The proposed approach utilizes a bank of (m+1) Kalman filters where m is the number of sensors being monitored. One Kalman filter is used for the detection of component and actuator faults while each of the other m filters detects a fault in a specific sensor. With this setup, the overall robustness of the sensor FDI system to anomalies is enhanced. Moreover, numerous component fault events can be accounted for by the FDI system. The sensor FDI system is applied to a commercial aircraft engine simulation, and its performance is evaluated at multiple power settings at a cruise operating point using various fault scenarios.

Kobayashi, Takahisa↗
NASA NTRS2001

Absence of diurnal variation of C-reactive protein concentrations in healthy human subjects

BACKGROUND: The concentration of C-reactive protein (CRP) in otherwise healthy subjects has been shown to predict future risk of myocardial infarction and stroke. CRP is synthesized by the liver in response to interleukin-6, the serum concentration of which is subject to diurnal variation. METHODS: To examine the existence of a time-of-day effect for baseline CRP values, we determined CRP concentrations in hourly blood samples drawn from healthy subjects (10 males, 3 females; age range, 21-35 years) during a baseline day in a controlled environment (8 h of nighttime sleep). RESULTS: Overall CRP concentrations were low, with only three subjects having CRP concentrations >2 mg/L. Comparison of raw data showed stability of CRP concentrations throughout the 24 h studied. When compared with cutoff values of CRP quintile derived from population-based studies, misclassification of greater than one quintile did not occur as a result of diurnal variation in any of the subjects studied. Nonparametric ANOVA comparing different time points showed no significant differences for both raw and z-transformed data. Analysis for rhythmic diurnal variation using a method fitting a cosine curve to the group data was negative. CONCLUSIONS: Our data show that baseline CRP concentrations are not subject to time-of-day variation and thus help to explain why CRP concentrations are a better predictor of vascular risk than interleukin-6. Determination of CRP for cardiovascular risk prediction may be performed without concern for diurnal variation.

Non-NASA Center↗
NASA NTRS2005

Enhanced Bank of Kalman Filters Developed and Demonstrated for In-Flight Aircraft Engine Sensor Fault Diagnostics

In-flight sensor fault detection and isolation (FDI) is critical to maintaining reliable engine operation during flight. The aircraft engine control system, which computes control commands on the basis of sensor measurements, operates the propulsion systems at the demanded conditions. Any undetected sensor faults, therefore, may cause the control system to drive the engine into an undesirable operating condition. It is critical to detect and isolate failed sensors as soon as possible so that such scenarios can be avoided. A challenging issue in developing reliable sensor FDI systems is to make them robust to changes in engine operating characteristics due to degradation with usage and other faults that can occur during flight. A sensor FDI system that cannot appropriately account for such scenarios may result in false alarms, missed detections, or misclassifications when such faults do occur. To address this issue, an enhanced bank of Kalman filters was developed, and its performance and robustness were demonstrated in a simulation environment. The bank of filters is composed of m + 1 Kalman filters, where m is the number of sensors being used by the control system and, thus, in need of monitoring. Each Kalman filter is designed on the basis of a unique fault hypothesis so that it will be able to maintain its performance if a particular fault scenario, hypothesized by that particular filter, takes place.

Kobayashi, Takahisa↗
NASA NTRS2005

Aircraft Fault Detection and Classification Using Multi-Level Immune Learning Detection

This work is an extension of a recently developed software tool called MILD (Multi-level Immune Learning Detection), which implements a negative selection algorithm for anomaly and fault detection that is inspired by the human immune system. The immunity-based approach can detect a broad spectrum of known and unforeseen faults. We extend MILD by applying a neural network classifier to identify the pattern of fault detectors that are activated during fault detection. Consequently, MILD now performs fault detection and identification of the system under investigation. This paper describes the application of MILD to detect and classify faults of a generic transport aircraft augmented with an intelligent flight controller. The intelligent control architecture is designed to accommodate faults without the need to explicitly identify them. Adding knowledge about the existence and type of a fault will improve the handling qualities of a degraded aircraft and impact tactical and strategic maneuvering decisions. In addition, providing fault information to the pilot is important for maintaining situational awareness so that he can avoid performing an action that might lead to unexpected behavior - e.g., an action that exceeds the remaining control authority of the damaged aircraft. We discuss the detection and classification results of simulated failures of the aircraft's control system and show that MILD is effective at determining the problem with low false alarm and misclassification rates.

Wong, Derek↗
NASA NTRS2005

Data Format Classification for Autonomous Software Defined Radios

We present maximum-likelihood (ML) coherent and noncoherent classifiers for discriminating between NRZ and Manchester coded (biphase-L) data formats for binary phase-shift-keying (BPSK) modulation. Such classification of the data format is an essential element of so-called autonomous software defined radio (SDR) receivers (similar to so-called cognitive SDR receivers in the military application) where it is desired that the receiver perform each of its functions by extracting the appropriate knowledge from the received signal and, if possible, with as little information of the other signal parameters as possible. Small and large SNR approximations to the ML classifiers are also proposed that lead to simpler implementation with comparable performance in their respective SNR regions. Numerical performance results obtained by a combination of computer simulation and, wherever possible, theoretical analyses, are presented and comparisons are made among the various configurations based on the probability of misclassification as a performance criterion. Extensions to other modulations such as QPSK are readily accomplished using the same methods described in the paper.

classification↗
NASA NTRS2011

Feature Acquisition with Imbalanced Training Data

This work considers cost-sensitive feature acquisition that attempts to classify a candidate datapoint from incomplete information. In this task, an agent acquires features of the datapoint using one or more costly diagnostic tests, and eventually ascribes a classification label. A cost function describes both the penalties for feature acquisition, as well as misclassification errors. A common solution is a Cost Sensitive Decision Tree (CSDT), a branching sequence of tests with features acquired at interior decision points and class assignment at the leaves. CSDT's can incorporate a wide range of diagnostic tests and can reflect arbitrary cost structures. They are particularly useful for online applications due to their low computational overhead. In this innovation, CSDT's are applied to cost-sensitive feature acquisition where the goal is to recognize very rare or unique phenomena in real time. Example applications from this domain include four areas. In stream processing, one seeks unique events in a real time data stream that is too large to store. In fault protection, a system must adapt quickly to react to anticipated errors by triggering repair activities or follow- up diagnostics. With real-time sensor networks, one seeks to classify unique, new events as they occur. With observational sciences, a new generation of instrumentation seeks unique events through online analysis of large observational datasets. This work presents a solution based on transfer learning principles that permits principled CSDT learning while exploiting any prior knowledge of the designer to correct both between-class and withinclass imbalance. Training examples are adaptively reweighted based on a decomposition of the data attributes. The result is a new, nonparametric representation that matches the anticipated attribute distribution for the target events.

Thompson, David R.↗
NASA NTRS2011

An Accuracy Assessment of the CALIOP/CALIPSO Version 2/Version 3 Daytime Aerosol Extinction Product Based on a Detailed Multi-Sensor, Multi-Platform Case Study

The Cloud Aerosol LIdar with Orthogonal Polarization (CALIOP), on board the CALIPSO platform, has measured profiles of total attenuated backscatter coefficient (level 1 products) since June 2006. CALIOP s level 2 products, such as the aerosol backscatter and extinction coefficient profiles, are retrieved using a complex succession of automated algorithms. The goal of this study is to help identify potential shortcomings in the CALIOP version 2 level 2 aerosol extinction product and to illustrate some of the motivation for the changes that have been introduced in the next version of CALIOP data (version 3, released in June 2010). To help illustrate the potential factors contributing to the uncertainty of the CALIOP aerosol extinction retrieval, we focus on a one-day, multi-instrument, multiplatform comparison study during the CALIPSO and Twilight Zone (CATZ) validation campaign on 4 August 2007. On that day, we observe a consistency in the Aerosol Optical Depth (AOD) values recorded by four different instruments (i.e. spaceborne MODerate Imaging Spectroradiometer, MODIS: 0.67 and POLarization and Directionality of Earth s Reflectances, POLDER: 0.58, airborne High Spectral Resolution Lidar, HSRL: 0.52 and ground-based AErosol RObotic NETwork, AERONET: 0.48 to 0.73) while CALIOP AOD is a factor of two lower (0.32 at 532 nm). This case study illustrates the following potential sources of uncertainty in the CALIOP AOD: (i) CALIOP s low signal-to-noise ratio (SNR) leading to the misclassification and/or lack of aerosol layer identification, especially close to the Earth s surface; (ii) the cloud contamination of CALIOP version 2 aerosol backscatter and extinction profiles; (iii) potentially erroneous assumptions of the aerosol extinction-to-backscatter ratio (Sa) used in CALIOP s extinction retrievals; and (iv) calibration coefficient biases in the CALIOP daytime attenuated backscatter coefficient profiles. The use of version 3 CALIOP extinction retrieval for our case study seems to partially fix factor (i) although the aerosol retrieved by CALIOP is still somewhat lower than the profile measured by HSRL; the cloud contamination (ii) appears to be corrected; no particular change is apparent in the observation-based CALIOP Sa value (iii). Our case study also showed very little difference in version 2 and version 3 CALIOP attenuated backscatter coefficient profiles, illustrating a minor change in the calibration scheme (iv).

Kacenelenbogen, M.↗
NASA NTRS2012

Exploring the Connection Between Star Formation and AGN Activity in the Local Universe

We study a combined sample of 264 star-forming, 51 composite, and 73 active galaxies using optical spectra from SDSS and mid-infrared (mid-IR) spectra from the Spitzer Infrared Spectrograph. We examine optical and mid-IR spectroscopic diagnostics that probe the amount of star formation and relative energetic con- tributions from star formation and an active galactic nucleus (AGN). Overall we find good agreement between optical and mid-IR diagnostics. Misclassifications of galaxies based on the SDSS spectra are rare despite the presence of dust obscuration. The luminosity of the [NeII] 12.8 micron emission-line is well correlated with the star formation rate (SFR) measured from the SDSS spectra, and this holds for the star forming, composite, and AGN-dominated systems. AGN show a clear excess of [NeIII] 15.6 micron emission relative to star forming and composite systems. We find good qualitative agreement between various parameters that probe the relative contributions of the AGN and star formation, including: the mid-IR spectral slope, the ratio of the [NeV] 14.3 micron to [NeII] micron 12.8 fluxes, the equivalent widths of the 7.7, 11.3, and 17 micron PAH features, and the optical "D" parameter which measures the distance a source lies from the locus of star forming galaxies in the optical BPT emission-line diagnostic diagram. We also consider the behavior of the three individual PAH features by examining how their flux ratios depend upon the degree of AGN-dominance. We find that the PAH 11.3 micron feature is significantly suppressed in the most AGN-dominated systems.

LaMassa, Stephanie M.↗
NASA NTRS2014

An Integrated Architecture for Aircraft Engine Performance Monitoring and Fault Diagnostics: Engine Test Results

This paper presents a model-based architecture for performance trend monitoring and gas path fault diagnostics designed for analyzing streaming transient aircraft engine measurement data. The technique analyzes residuals between sensed engine outputs and model predicted outputs for fault detection and isolation purposes. Diagnostic results from the application of the approach to test data acquired from an aircraft turbofan engine are presented. The approach is found to avoid false alarms when presented nominal fault-free data. Additionally, the approach is found to successfully detect and isolate gas path seeded-faults under steady-state operating scenarios although some fault misclassifications are noted during engine transients. Recommendations for follow-on maturation and evaluation of the technique are also presented.

propulsion system performance↗