Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “intervals”

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 685 records · Page 38

The recurrence behaviour of X-ray bursts from 4U/MXB 1735-44

During about 80 h of uninterrupted observations of the X-ray burst source 4 U/MXB 1735-44, made in August 1985 with Exosat, five bursts were observed. The bursts occur at irregular intervals, ranging between 30 min and 36 h. The burst sizes do not scale with the intervals preceding the bursts; in particular the energy in a burst which occurred after an interval of 50 min was only 30 percent less than that of a burst which occurred after more than 20 h. Due to the long uninterrupted observations possible with Exosat, this is the first time for 1735-44 that the burst intervals and the burst characteristics (e.g., fluence) can be compared. The observations show that the model recently proposed by Fujimoto et al. (1987) to explain irregular burst intervals, is not applicable to 1735-44. The amplitude of periodic X-ray flux variations at the 4.65 h orbital period is less than 4 percent.

Van Paradijs, J.↗

Evaluation of the concept of pressure proof testing fuselage structures

The FAA and NASA have recently completed independent technical evaluations of the concept of pressure proof testing the fuselage of commercial transport airplanes. The results of these evaluations are summarized. The objectives of the evaluations were to establish the potential benefit of the pressure proof test, to quantify the most desirable proof test pressure, and to quantify the required proof test interval. The focus of the evaluations was on multiple-site cracks extending from adjacent rivet holes of a typical fuselage longitudinal lap splice joint. The FAA and NASA do not support pressure proof testing the fuselage of aging commercial transport aircraft. The argument against proof testing is as follows: (1) a single proof test does not insure an indefinite life; therefore, the proof test must be repeated at regular intervals; (2) for a proof factor of 1.33, the required proof test interval must be below 300 flights to account for uncertainties in the evaluation; (3) conducting the proof test at a proof factor of 1.5 would considerably exceed the fuselage design limit load; therefore, it is not consistent with accepted safe practices; and (4) better safety can be assured by implementing enhanced nondestructive inspection requirements, and adequate reliability can be achieved by an inspection interval several times longer than the proof test interval.

Harris, Charles E.↗

Inference in infinite-dimensional inverse problems - Discretization and duality

Many techniques for solving inverse problems involve approximating the unknown model, a function, by a finite-dimensional 'discretization' or parametric representation. The uncertainty in the computed solution is sometimes taken to be the uncertainty within the parametrization; this can result in unwarranted confidence. The theory of conjugate duality can overcome the limitations of discretization within the 'strict bounds' formalism, a technique for constructing confidence intervals for functionals of the unknown model incorporating certain types of prior information. The usual computational approach to strict bounds approximates the 'primal' problem in a way that the resulting confidence intervals are at most long enough to have the nominal coverage probability. There is another approach based on 'dual' optimization problems that gives confidence intervals with at least the nominal coverage probability. The pair of intervals derived by the two approaches bracket a correct confidence interval. The theory is illustrated with gravimetric, seismic, geomagnetic, and helioseismic problems and a numerical example in seismology.

Stark, Philip B.↗

EGRET observations of the Vela pulsar PSR 0833-45

During the all sky survey (May 1991 - Nov. 1992) of the Compton Gamma Ray Observatory the Vela pulsar PSR0833-45 was in the field of view of the Energetic Gamma Ray Experiment Telescope (EGRET) in ten separate viewing periods. The pulsar was detected in each one. The average intensity from 100 MeV to 2 GeV was (7.8 +/- 1.0) x 10 (exp -6) photons cm(exp -2) s(exp -1), which indicates that the pulsar in the years 1991/92 was in a state comparable to the low fluxes observed in 1977-1980. No significant changes in intensity were detected during the EGRET observations. The total spectrum of PSR0833-45 measured by EGRET can be described by a power law with index -(1.70 +/- 0.02) over the range 30 MeV to 2 GeV. The extrapolation of this spectrum into the 3 to 30 MeV range agrees with the observations by COMPTEL. Above 2 to 4 GeV EGRET detects a strong spectral break. The lightcurves obtained show a familar structure in the phase histogram: two peaks separated by 0.424 +/- 0.002 in phase with considerable emission in the phase interval between the peaks. The first gamma ray peak maximum trails the single radio peak maximum by 10.54 +/- 0.09 ms (= 0.118 +/- 0.001 in phase). The widths of the emission peaks (FWHM) are 2.7 ms for the first peak (0.03 phase) and 4.1 ms for the second peak (0.05 phase). The widths are approximately constant below a GeV, but show a tendency to become narrower at higher energies. On Jul. 20 1991 a glitch of the Vela period was registered in monitor radio observations. No significant differences between the pre- and post-glitch gamma ray lightcurves were found. The statistics available for the Vela observations allow for a division of the lightcurve into eight phase intervals. The emission peak cores (central FWHM) with leading and trailing wings and two interval regions were defined and spectra were derived for all parts of the lightcurve. The energy spectra for the eight phase intervals show significant differences: the first peak (approximately E(exp -1.81 +/- 0.04)) is somewhat softer than the second peak (approximately E(exp -1.71 +/- 0.03)); the wings attached to each peak show softer spectra than the code of the peaks; the interval emission has the hardest spectrum (approximately E(exp -1.52 +/- 0.03)).

Kanbach, G.↗

Strain accumulation in the Shumagin Islands: Results of initial GPS measurements

Deformation in the Shumagin seismic gap has been monitored with repeated trilateration (EDM) in the 1980-1987 interval and with the Global Positioning System (GPS) in the 1987-1991 interval. The geodetic network extends for 100-km across the Shumagin Islands to the Alaska Peninsula. Results from the GPS surveys are consistent with those previously reported for the EDM surveys: we failed to detect significant strain accumulation in the N30 deg W direction of plate convergence. Using the method of simultaneous reduction for position and strain rates, we found the average rate of extension in the direction of plate convergence to be -25 +/- 25 nanostrain/yr (nstrain/yr) during the 1987-1991 interval of GPS surveys compared with -20 +/- 15 nstrain/yr during the 1981-1987 interval of complete EDM surveys. We found a marginally significant -26 +/- 12 nstrain/yr extension rate in the 1981-1991 interval covered by the combined EDM and GPS surveys. Strain rates are higher, but not significantly so, in the part of the network closest to the trench. Spatial variation in the deformation is observed in the 1980-1991 average station velocities, where three of the four stations closest to the trench have an arcward velocity of a few mm/yr. The observed strain rates are an order of magnitude lower than the -200 nstrain/yr rate predicted by dislocation models.

Larson, Kristine M.↗

EGRET observations of the Vela pulsar, PSR0833-45

During the all sky survey (May 1991 to November 1992) of the Compton Gamma Ray Observatory, the Vela pulsar PSR0833-45 was in the field of view of the Energetic Gamma Ray Experiment Telescope (EGRET) in ten separate viewing periods. The pulsar was detected in each one. The average intensity from 100 MeV to 2GeV was (7.8 +/- 1.0) x 10(exp -6) photons /sq cm/s, which indicates that the pulsar in the years 1991/92 was in a state comparable to the low fluxes observed in 1977-1980. No significant changes in intensity were detected during the EGRET observations. The total spectrum of PSR0833-45 measured by EGRET can be described by a power-law with index -1.70 +/- 0.02 over the range 30 MeV to 2 GeV. The extrapolation of this spectrum into the 3-30 MeV range agrees with the observations by COMPTEL. Above 2-4 GeV EGRET detects a strong spectral break. The lightcurves obtained show a familiar structure in the phase histogram: two peaks separated by 0.424 +/- 0.002 in phase with considerable emission in the phase interval between the peaks. The first gamma ray peak maximum trails the single radio peak maximum by 10.54 +/- 0.09 ms (= 0.118 +/- 0.001 in phase). The widths of the emission peaks (FWHM) are 2.7 ms for the first peak (0.03 phase) and 4.1 ms for the second peak (0.05 phase). The widths are approximately constant below a GeV, but show a tendency to become narrower at higher energies. The statistics available for the Vela observations allow for a division of the lightcurve into eight phase intervals. The emission peak cores (central FWHM) with leading and trailing wings and two interval regions were defined and spectra were derived for all parts of the lightcurve. The energy spectra for the eight phase intervals show significant differences: the first peak (approximately E(exp -1.81 +/- 0.04)) is somewhat softer than the second peak (approximately E(exp -1.71 +/- 0.03)); the wings attached to each peak show softer spectra than the core of the peaks; the interval emission has the hardest spectrum (approximately E(exp -1.52 +/- 0.03)).

Kanbach, G.↗

Reducing Formation-Keeping Maneuver Costs for Formation Flying Satellites in Low-Earth Orbit

Several techniques are used to synthesize the formation-keeping control law for a three-satellite formation in low-earth orbit. The objective is to minimize maneuver cost and position tracking error. Initial reductions are found for a one-satellite case by tuning the state-weighting matrix within the linear-quadratic-Gaussian framework. Further savings come from adjusting the maneuver interval. Scenarios examined include cases with and without process noise. These results are then applied to a three-satellite formation. For both the one-satellite and three-satellite cases, increasing the maneuver interval yields a decrease in maneuver cost and an increase in position tracking error. A maneuver interval of 8-10 minutes provides a good trade-off between maneuver cost and position tracking error. An analysis of the closed-loop poles with respect to varying maneuver intervals explains the effectiveness of the chosen maneuver interval.

Hamilton, Nicholas↗

Vagal and sympathetic mechanisms in patients with orthostatic vasovagal syncope

BACKGROUND: Autonomic and particularly sympathetic mechanisms play a central role in the pathophysiology of vasovagal syncope. We report direct measurements of muscle sympathetic nerve activity in patients with orthostatic vasovagal syncope. METHODS AND RESULTS: We studied 53 otherwise healthy patients with orthostatic syncope. We measured RR intervals and finger arterial pressures and in 15 patients, peroneal nerve muscle sympathetic activity before and during passive 60 degree head-up tilt, with low-dose intravenous isoproterenol if presyncope did not develop by 15 minutes. We measured baroreflex gain before tilt with regression of RR intervals or sympathetic bursts on systolic or diastolic pressures after sequential injections of nitroprusside and phenylephrine. Orthostatic vasovagal reactions occurred in 21 patients, including 7 microneurography patients. Presyncopal and nonsyncopal patients had similar baseline RR intervals, arterial pressure, and muscle sympathetic nerve activity. Vagal baroreflex responses were significantly impaired at arterial pressures below (but not above) baseline levels in presyncopal patients. Initial responses to tilt were comparable; however, during the final 200 seconds of tilt, presyncopal patients had lower RR intervals and diastolic pressures than nonsyncopal patients and gradual reduction of arterial pressure and sympathetic activity. Frank presyncope began abruptly with precipitous reduction of arterial pressure, disappearance of muscle sympathetic nerve activity, and RR interval lengthening. CONCLUSIONS: Patients with orthostatic vasovagal reactions have impaired vagal baroreflex responses to arterial pressure changes below resting levels but normal initial responses to upright tilt. Subtle vasovagal physiology begins before overt presyncope. The final trigger of human orthostatic vasovagal reactions appears to be the abrupt disappearance of muscle sympathetic nerve activity.

Non-NASA Center↗

Degree Of Diminution In Vagal-Cardiac Activity Predicts Sudden Death In Familial Dysautonomia When Resting Tachycardia Is Absent

Patients with familial dysautonomia (FD) have an increased risk of sudden death, but sensitive and specific predictors of sudden death in FD are lacking. Methods. We recorded 10-min resting high-fidelity 12-lead ECGs in 14 FD patients and in 14 age/gender-matched healthy subjects and studied 25+ different heart rate variability (HRV) indices for their ability to predict sudden death in the FD patients. Indices studied included those from 4 "nonlinear" HRV techniques (detrended fluctuation analysis, approximate entropy, correlation dimension, and PoincarC analyses). The predictive value of PR, QRS, QTc and JTc intervals, QT dispersion (QTd), beat-to-beat QT and PR interval variability indices (QTVI and PRVI) and 12- lead high frequency QRS ECG (150-250 Hz) were also studied. FD patients and controls (C) differed (Pless than 0.0l) with respect to 20+ of the HRV indices (FD less than C) and with respect to QTVI and PRVI (FDBC) and HF QRS- related root mean squared voltages (FDBC) and reduced amplitude zone counts (FD less than C). They differed less with respect to PR intervals (FD less than C) and JTc intervals (FD greater than C) (P less than 0.05 for both) and did not differ at all with respect to QRS and QTc intervals and to QTd. Within 12 months after study, 2 of the 14 patients succumbed to sudden cardiac arrest. The best predictor of sudden death was the degree of diminution in HRV vagal-cardiac (parasympathetic) parameters such as RMSSD, the SDl of Poincare plots, and HF spectral power. Excluding the two FD patients who had resting tachycardia (HR greater than 100, which confounds traditional HRV analyses), the following criteria were independently 100% sensitive and 100% specific for predicting sudden death in the remaining 12 FD patients during spontaneous breathing: RMSSD less than 13 ms and/or PoincarC SD1 less than 9 ms. In FD patients without supine tachycardia, the degree of diminution in parasympathetic HRV parameters (by high-fidelity ECG) predicts incipient death.

Schlegel, T. T.↗

Method and apparatus for assessing cardiovascular risk

The method for assessing risk of an adverse clinical event includes detecting a physiologic signal in the subject and determining from the physiologic signal a sequence of intervals corresponding to time intervals between heart beats. The long-time structure of fluctuations in the intervals over a time period of more than fifteen minutes is analyzed to assess risk of an adverse clinical event. In a preferred embodiment, the physiologic signal is an electrocardiogram and the time period is at least fifteen minutes. A preferred method for analyzing the long-time structure variability in the intervals includes computing the power spectrum and fitting the power spectrum to a power law dependence on frequency over a selected frequency range such as 10.sup.-4 to 10.sup.-2 Hz. Characteristics of the long-time structure fluctuations in the intervals is used to assess risk of an adverse clinical event.

Albrecht, Paul↗

An Extended Forecast of the Frequencies of North Atlantic Basin Tropical Cyclone Activity for 2009

An extended forecast of the frequencies for the 2009 North Atlantic basin hurricane season is presented. Continued increased activity during the 2009 season with numbers of tropical cyclones, hurricanes, and major hurricanes exceeding long-term averages are indicated. Poisson statistics for the combined high-activity intervals (1950-1965 and 1995-2008) give the central 50% intervals to be 9-14, 5-8, and 2-4, respectively, for the number of tropical cyclones, hurricanes, and major hurricanes, with a 23.4% chance of exceeding 14 tropical cyclones, a 28% chance of exceeding 8 hurricanes, and a 31.9% chance of exceeding 4 major hurricanes. Based strictly on the statistics of the current high-activity interval (1995-2008), the central 50% intervals for the numbers of tropical cyclones, hurricanes, and major hurricanes are 12-18, 6-10, and 3-5, respectively, with only a 5% chance of exceeding 23, 13, or 7 storms, respectively. Also examined are the first differences in 10-yr moving averages and the effects of global warming and decadal-length oscillations on the frequencies of occurrence for North Atlantic basin tropical cyclones. In particular, temperature now appears to be the principal driver of increased activity and storm strength during the current high-activity interval, with near-record values possible during the 2009 season.

Wilson, Robert M.↗

A Formally-Verified Decision Procedure for Univariate Polynomial Computation Based on Sturm's Theorem

Sturm's Theorem is a well-known result in real algebraic geometry that provides a function that computes the number of roots of a univariate polynomial in a semiopen interval. This paper presents a formalization of this theorem in the PVS theorem prover, as well as a decision procedure that checks whether a polynomial is always positive, nonnegative, nonzero, negative, or nonpositive on any input interval. The soundness and completeness of the decision procedure is proven in PVS. The procedure and its correctness properties enable the implementation of a PVS strategy for automatically proving existential and universal univariate polynomial inequalities. Since the decision procedure is formally verified in PVS, the soundness of the strategy depends solely on the internal logic of PVS rather than on an external oracle. The procedure itself uses a combination of Sturm's Theorem, an interval bisection procedure, and the fact that a polynomial with exactly one root in a bounded interval is always nonnegative on that interval if and only if it is nonnegative at both endpoints.

Narkawicz, Anthony J.↗

Discovery of Activities via Statistical Clustering of Fixation Patterns

Human behavior often consists of a series of distinct activities, each characterized by a unique pattern of interaction with the visual environment. This is true even in a restricted domain, such as a piloting an aircraft, where activities with distinct visual signatures might be things like communicating, navigating, and monitoring. We propose a novel analysis method for gaze-tracking data, to perform blind discovery of these hypothetical activities. The method is in some respects similar to recurrence analysis, but here we compare not individual fixations, but groups of fixations aggregated over a fixed time interval. The duration of this interval is a parameter that we will refer to as delta. We assume that the environment has been divided into a set of N different areas-of-interest (AOIs). For a given interval of time of duration delta, we compute the proportion of time spent fixating each AOI, resulting in an N-dimensional vector. These proportions can be converted to integer counts by multiplying by delta divided by the average fixation duration (another parameter that we fix at 280 milliseconds). We compare different intervals by computing the chi-square statistic. The p-value associated with the statistic is the likelihood of observing the data under the hypothesis that the data in the two intervals were generated by a single process with a single set of probabilities governing the fixation of each AOI. The method has been applied to approximately 100 hours of eye movement data collected from pilots in a high-fidelity B747 flight simulator, and the results have been compared to synthetic data in which the each activity is represented as first-order Markov process with random probabilities assigned to the AOIs. Randomly-generated synthetic activities can require thousands of fixations to be discriminated with statistical significance, while the human data can be clustered using averaging windows of some 10's of seconds, suggesting that the actual activities are much more narrowly focused than random Markov models.

activity analysis↗

TPSAS-NF1676L-12168-DND

The Federal Aviation Administration’s Surveillance and Broadcast Services Program Office considers Interval Management (IM) to be one of the three key, near-term applications to make use of ADS-B-In, the receiving and processing of Automatic Dependent Surveillance – Broadcast data on-board an aircraft. Interval Management is used to describe a wide range of operational applications and uses of airborne spacing technology. Interval Management makes use of improved scheduling and sequencing capabilities for the air traffic controllers (ATC) and precise, relative spacing of aircraft, to improve airport throughput, increase the use of Optimized Profile Descents (OPD), and reduce tactical controller workload. The IM applications are subdivided into two categories: one where the flight crew is authorized to manage their speed to achieve the IM goal while the controller retains separation responsibility and a second where the flight crew takes responsibility for both managing speed but also for separation from the specified target aircraft. The former is referred to as Interval Management-Spacing (IM-S) and the latter as Interval Management-Delegated Separation (IM-DS). For the past 10 years NASA has been working on a specific IM application called Airborne Precision Spacing (APS). This concept has focused on increasing runway throughput while enabling OPDs. It is seen as a specific application within the IM class of applications. This talk will present the current state of the APS concept including support for dependent parallel runway operations and how NASA is working to have APS support the broader IM activities. Planned development and testing of APS, including extensions to support IM-DS and FAA-planned flight trials will be presented along with several up-coming simulations.

Bryan E Barmore↗

TPSAS-NF1676L-12329-DND

The Federal Aviation Administration’s Surveillance and Broadcast Services Program Office considers Interval Management (IM) to be one of the three key, near-term applications to make use of ADS-B-In, the receiving and processing of Automatic Dependent Surveillance – Broadcast data on-board an aircraft. Interval Management is used to describe a wide range of operational applications and uses of airborne spacing technology. Interval Management makes use of improved scheduling and sequencing capabilities for the air traffic controllers (ATC) and precise, relative spacing of aircraft, to improve airport throughput, increase the use of Optimized Profile Descents (OPD), and reduce tactical controller workload. The IM applications are subdivided into two categories: one where the flight crew is authorized to manage their speed to achieve the IM goal while the controller retains separation responsibility and a second where the flight crew takes responsibility for both managing speed but also for separation from the specified target aircraft. The former is referred to as Interval Management-Spacing (IM-S) and the latter as Interval Management-Delegated Separation (IM-DS). For the past 10 years NASA has been working on a specific IM application called Airborne Precision Spacing (APS). This concept has focused on increasing runway throughput while enabling OPDs. It is seen as a specific application within the IM class of applications. This talk will present the current state of the APS concept including support for dependent parallel runway operations and how NASA is working to have APS support the broader IM activities. Planned development and testing of APS, including extensions to support IM-DS and FAA-planned flight trials will be presented along with several up-coming simulations.

Brian T Baxley↗

Summer temperatures during the last glaciation (MIS 5c to MIS 3) inferred from a 50,000-year chironomid record from Füramoos, southern Germany

There is a sparsity of long, continuous palaeotemperature records for the last glacial period in central Europe, particularly for the interval corresponding to Marine Isotope Stages (MIS) 4 and 3. Here we present a new, ca. 50-thousand year (ka)-long chironomid record from Füramoos, southern Germany, covering the interval from MIS 5a to MIS 3 that we use to examine lake development and then to quantitatively reconstruct mean July air temperatures. Chironomid assemblages with high abundances of taxa such as Polypedilum nubeculosum-type, Microtendipes pedellus-type, Cladopelma lateralis-type and Dicrotendipes nervosus-type imply a shallow-lake setting for the majority of the examined interval, which is corroborated by other aquatic remains such as oribatid mites, Sialidae and Ceratopogonidae. Assemblages from the interval ca. 99 to 80 ka (in the region corresponding to the Brörup Interstadial, Stadial B and early Odderade Interstadial) are dominated by taxa such as Tanytarsus glabrescens-type and Tanytarsus mendax-type and indicate relatively warm temperatures. Assemblages from the interval covering ca. 80 to 54 ka (corresponding to the late Odderade, Stadial C, Dürnten Interstadial and Stadial D) are dominated by taxa such as Sergentia coracina-type and Tanytarsus lugens-type and are typical for cooler conditions. Reconstructed July temperatures for the early Würmian (Brörup to early Odderade; ca. 99–80 ka) are 13–14 °C. Values decline to <10 °C during the late Odderade and Stadial C (ca. 80–77 ka) around the MIS 5a/4 transition. This decrease is coeval with a pronounced decrease in Northern Hemisphere summer insolation. Values stay in the range of 9–11 °C during the Dürnten and Stadial D (ca. 54–74.5 ka) and increase again to 12.5 °C during the Bellamont 1 interstadial (ca. 54–46 ka). Reconstructed July temperatures track changes in arboreal pollen percentages at Füramoos and agree with a summer-temperature decrease during the early to mid-Würmian as reported by other palaeotemperature records from Europe and the North Atlantic. Our chironomid record from Füramoos provides valuable new insights into Würmian climate dynamics in Central Europe, and corroborates other temperature reconstructions from the early to mid-Würmian glacial period.

Chironomids↗

Joint Polar Satellite System (JPSS)-3 VIIRS Prelaunch Geometric Calibration and Characterization Status

The NASA/NOAA Visible Infrared Imaging Radiometer Suite (VIIRS) is a key instrument in the JPSS missions (SNPP, JPSS-1-4). Being part of the calibration and validation process, JPSS-3 VIIRS prelaunch geometric performance assessment focuses on the sensor’s spatial response, band-to-band co-registration (BBR), and pointing knowledge. In general, JPSS-3 VIIRS’ prelaunch geometric performance is very good, and consistent with SNPP and JPSS-1-2 VIIRS sensors. This paper highlights some specific key findings from the JPSS-3 VIIRS’ prelaunch tests. We first show that with timing adjustments, JPSS-3 VIIRS scan BBR error between VisNIR and S/M/LWIR bands has been reduced from 0.12/0.03 M sampling intervals to 0.005/0.002 M sampling intervals, respectively. Focal Plane Assembly (FPA) rotations have been changed from 0.1 degree to 0.04 degree in VisVIR, from 0.18 degree to 0.09 degree in S/MWIR, and 0.16 degree to 0.08 degree in LWIR. Another finding from the prelaunch test is that JPSS-3 VIIRS has relatively large band-to-band co-registration errors between VisNIR and S/M/LWIR bands in the track direction. This BBR mis-registration increases when aft‐optics assembly (AOA) temperature decreases. The BBR mis-registration is about 0.10 M sampling intervals between VisNIR and S/MWIR bands, and 0.08 M sampling intervals between VisNIR and LWIR bands. By comparison, the track BBR offset in JPSS-2 is less than 0.02 M sampling intervals between VisNIR and S/M/LWIR bands. We also notice that the unsymmetrical Day Night Band (DNB) scan-direction Line Spread Function (LSF) of JPSS-2 VIIRS at different gain stages and aggregation modes have been corrected in J3 VIIRS.

VIIRS↗

Comparison of Entry Descent and Landing Aerodynamic Databases with Uncertainty Quantification Developed Using Machine Learning Techniques

When developing the aerodynamic databases for use in trajectory simulations, it is important to develop a system of metrics to qualify which aerodynamic models are best to use. Since aerodynamics are just one input into trajectory simulations, the results of these simulations do not reflect on the quality of the aerodynamic database used. This means that aerodynamic database comparisons must be done offline. While traditional metrics that focus on mean/nominal predictions are a good first step, more robust estimates of the prediction interval become important as more focused uncertainty models are developed. We explore the limitations of evaluating aerodynamic models based purely on nominal-centered response surfaces. Before elaborating and evaluating metrics based on distributed models, the value of evaluating prediction interval and confidence interval are discussed to conclude that prediction intervals are more relevant to the use of trajectory analysis. Several metrics to evaluate the prediction interval are introduced with a focus on the standard calibration metric. Finally, we compare candidate models using both mean and distributed metrics. A finalized candidate model developed using state of the art machine learning methods is compared to a baseline model developed using traditional aerodynamic database modeling techniques.

Aerodynamic Database↗