Results for “Model selection” — Page 45

NASA NTRS1993

Nonlinear aerodynamic modeling using multivariate orthogonal functions

A technique was developed for global modeling of nonlinear aerodynamic coefficients using multivariate orthogonal functions based on the data. Each orthogonal function retained in the model was decomposed into an expansion of ordinary polynomials in the independent variables, so that the final model could be interpreted as selectively retained terms from a multivariable power series expansion. A predicted squared-error metric was used to determine the orthogonal functions to be retained in the model; analytical derivatives were easily computed. The approach was demonstrated on the Z-body axis aerodynamic force coefficient (Cz) wind tunnel data for an F-18 research vehicle which came from a tabular wind tunnel and covered the entire subsonic flight envelope. For a realistic case, the analytical model predicted experimental values of Cz very well. The modeling technique is shown to be capable of generating a compact, global analytical representation of nonlinear aerodynamics. The polynomial model has good predictive capability, global validity, and analytical differentiability.

Morelli, Eugene A.↗

NASA NTRS1989

Constructor selection research

The process of space construction is defined by the construction goal and by the construction system selected to accomplish that goal. The former provides the tasks that must be executed, while the latter provides the capabilities to execute them. This presentation outlines a high-level model of space construction and illustrates the relationship between constructor and task. The construction model is then utilized to develop a preliminary theory of constructor selection. The need for broad models of construction, constructor selection, and constructor management is identified. From this preliminary work it can be concluded that task decomposition is dependent upon both task and constructor; the selection process is separable from the optimization process; and that logical, defensible selection is possible.

Helleckson, Brent↗

NASA NTRS1989

The intrinsic luminosity and initial period of pulsars

The observed properties of pulsars are modeled on the assumption that they evolve by emitting magnetic dipole radiation and have exponentially decaying magnetic fields. The models are constructed by using Monte Carlo techniques to generate a Galactic sample of pulsars having particular distributions of initial periods, magnetic dipole moments, and positions. By choosing a relationship between the radio luminosity L(r), the period (P), and the P derivative, it is determined which pulsars of the generated sample could be detected by one of the major pulsar surveys. Detailed account is taken of known selection effects in these surveys. Models in which the size of the pulsar emission cone is dependent on the period are considered. The properties of the detectable pulsars are compared with the properties of the observed pulsars, and the initial distributions and luminosity relation of the model are adjusted until a satisfactory agreement is obtained. A crucial point is that the observed relation between L(r) and P, P derivative is affected by selection effects and does not apply to the actual pulsar population. It is found that the best fits are obtained with models in which the pulsars are injected with relatively long periods, about 0.5 sec.

Emmering, Robert T.↗

NASA NTRSDate not supplied

Alternate Methods of Model Reduction to Avoid Dynamic Modal Truncation Error

Loads analysis is traditionally performed using dynamically reduced models, which provides the benefit to reduce run time. If the reduced model frequency cutoff is not chosen appropriately, the model will lack the dynamic content required to fully represent the response of the non-reduced model. Standard guidance for reduced model frequency content, provided in NASA-STD-5002, is to solve fixed base modes up to a minimum of 1.5x the model frequency content of interest and to employ static modal truncation methods such as residual vectors, the mode acceleration method, and the residual flexibility method to account for the truncated flexibility of the missing modes. Fixed base modes require the selection of a set of degrees of freedom to be constrained which, if not properly selected, may affect the accuracy of the reduced model by excluding some of the dynamic characteristic of the full model. In this case, the standard NASA guidance would be insufficient, but it may not be readily apparent that a portion of the reduced model response is missing. This error was encountered during an independent verification and validation (IV&V) effort, where it was observed that the resulting dynamic response was lower than the inline analysis. In this specific case, despite following the standard NASA model reduction guidelines in the selection of the frequency cutoff, the inline model still did not fully capture the necessary dynamic content. As part of the IV&V, an alternate reduction methodology was employed using an unconstrained mode acceleration method. The original model initially performed a constrained reduction to twice the frequency content of interest before doing a free-free run, employing the mode acceleration method to account for the truncated modes. In contrast in the IV&V, the unconstrained model reduced directly to the needed frequency content of the free-free run, avoiding any interactions between constraints and dynamic content. To verify the model, the original reduction methodology was used to generate a series of Hurty-Craig-Bampton reductions, each with a higher frequency cutoff than the previous. The results were shown to converge once the frequency cutoff increased past eight times the frequency content of interest. At the same frequency cutoff, the results of the Hurty-Craig-Bampton model converged with the results of the unconstrained mode acceleration model. This comparative study provided confidence that the results of the unconstrained modal acceleration reduced model were correct and that the Hurty-Craig-Bampton needed to increase its frequency cutoff to fully capture the dynamic response.

Erin Simmons↗

NASA NTRSDate not supplied

Alternate Methods of Model Reduction to Avoid Dynamic Modal Truncation Error

Loads analysis is traditionally performed using dynamically reduced models, which provides the benefit to reduce run time. If the reduced model frequency cutoff is not chosen appropriately, the model will lack the dynamic content required to fully represent the response of the non-reduced model. Standard guidance for reduced model frequency content, provided in NASA-STD-5002, is to solve fixed base modes up to a minimum of 1.5x the model frequency content of interest and to employ static modal truncation methods such as residual vectors, the mode acceleration method, and the residual flexibility method to account for the truncated flexibility of the missing modes. Fixed base modes require the selection of a set of degrees of freedom to be constrained which, if not properly selected, may affect the accuracy of the reduced model by excluding some of the dynamic characteristic of the full model. In this case, the standard NASA guidance would be insufficient, but it may not be readily apparent that a portion of the reduced model response is missing. This error was encountered during an independent verification and validation (IV&V) effort, where it was observed that the resulting dynamic response was lower than the inline analysis. In this specific case, despite following the standard NASA model reduction guidelines in the selection of the frequency cutoff, the inline model still did not fully capture the necessary dynamic content. As part of the IV&V, an alternate reduction methodology was employed using an unconstrained mode acceleration method. The original model initially performed a constrained reduction to twice the frequency content of interest before doing a free-free run, employing the mode acceleration method to account for the truncated modes. In contrast in the IV&V, the unconstrained model reduced directly to the needed frequency content of the free-free run, avoiding any interactions between constraints and dynamic content. To verify the model, the original reduction methodology was used to generate a series of Hurty-Craig-Bampton reductions, each with a higher frequency cutoff than the previous. The results were shown to converge once the frequency cutoff increased past eight times the frequency content of interest. At the same frequency cutoff, the results of the Hurty-Craig-Bampton model converged with the results of the unconstrained mode acceleration model. This comparative study provided confidence that the results of the unconstrained modal acceleration reduced model were correct and that the Hurty-Craig-Bampton needed to increase its frequency cutoff to fully capture the dynamic response.

Erin Simmons↗

NASA NTRS1986

Solvability conditions for dendritic growth in the boundary-layer model with capillary anisotropy

This paper is concerned primarily with the development of an analytic approach to the theory of steady-state velocity selection in the boundary-layer model of dendritic solidification. The two-dimensional version of this model with a fourfold crystalline anisotropy alpha in the surface tension is considered. By extending a WKB method introduced in an earlier paper, the alpha dependence of the selected growth rate is determined in the limit of small alpha; and this rate is studied for large alphas in the limit in which the dimensionless undercooling approaches unity. Portions of the paper are devoted to a reinterpretation of the mathematical structure of the solvability condition in problems of this kind.

Langer, J. S.↗

NASA NTRS2009

Configuring the HYSPLIT Model for National Weather Service Forecast Office and Spaceflight Meteorology Group Applications

For expedience in delivering dispersion guidance in the diversity of operational situations, National Weather Service Melbourne (MLB) and Spaceflight Meteorology Group (SMG) are becoming increasingly reliant on the PC-based version of the HYSPLIT model run through a graphical user interface (GUI). While the GUI offers unique advantages when compared to traditional methods, it is difficult for forecasters to run and manage in an operational environment. To alleviate the difficulty in providing scheduled real-time trajectory and concentration guidance, the Applied Meteorology Unit (AMU) configured a Linux version of the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) (HYSPLIT) model that ingests the National Centers for Environmental Prediction (NCEP) guidance, such as the North American Mesoscale (NAM) and the Rapid Update Cycle (RUC) models. The AMU configured the HYSPLIT system to automatically download the NCEP model products, convert the meteorological grids into HYSPLIT binary format, run the model from several pre-selected latitude/longitude sites, and post-process the data to create output graphics. In addition, the AMU configured several software programs to convert local Weather Research and Forecast (WRF) model output into HYSPLIT format.

Dreher, Joseph G.↗

NASA NTRS2018

Marginal protein stability drives subcellular proteome isoelectric point

There exists a positive correlation between the pH of subcellular compartments and the median isoelectric point (pI) for the associated proteomes. Proteins in the human lysosome—a highly acidic compartment in the cell—have a median pI of ∼6.5, whereas proteins in the more basic mitochondria have a median pI of ∼8.0. Proposed mechanisms reflect potential adaptations to pH. For example, enzyme active site general acid/base residue pKs are likely evolved to match environmental pH. However, such effects would be limited to a few residues on specific proteins, and might not affect the proteome at large. A protein model that considers residue burial upon folding recapitulates the correlation between proteome pI and environmental pH. This correlation can be fully described by a neutral evolution process; no functional selection is included in the model. Proteins in acidic environments incur a lower energetic penalty for burying acidic residues than basic residues, resulting in a net accumulation of acidic residues in the protein core. The inverse is true under alkaline conditions. The pI distributions of subcellular proteomes are likely not a direct result of functional adaptations to pH, but a molecular spandrel stemming from marginal stability.

Kaiser Loell↗

NASA NTRS1984

Shuttle derived atmospheric density model. Part 2: STS atmospheric implications for AOTV trajectory analysis, a proposed GRAM perturbation density model

A perturbation model to the Marshall Space Flight Center (MSFC) Global Reference Atmosphere Model (GRAM) was developed for use in the Aeroassist Orbital Transfer Vehicle (AOTV) trajectory and analysis. The model reflects NASA Space Shuttle experience over the first twelve entry flights. The GRAM was selected over the Air Force 1978 Reference Model because of its more general formulation and wider use throughout NASA. The add-on model, a simple scaling with altitude to reflect density structure encountered by the Shuttle Orbiter was selected principally to simplify implementation. Perturbations, by season, can be utilized to minimize the number of required simulations, however, exact Shuttle flight history can be exercised using the same model if desired. Such a perturbation model, though not meteorologically motivated, enables inclusion of High Resolution Accelerometer Package (HiRAP) results in the thermosphere. Provision is made to incorporate differing perturbations during the AOTV entry and exit phases of the aero-asist maneuver to account for trajectory displacement (geographic) along the ground track.

Findlay, J. T.↗

NASA NTRS1974

Some after-dinner reflections on system identification

The growth of the field of system identification is discussed along with changes in methodology which have taken place in recent years. The similarity between pattern recognition and system identification is pointed out, involving the modelling in the latter and the feature selection problem in the former. It is stated that once a model is formulated, including the disturbances and measurement errors, the parameter finding can be formulated as a statistical estimation problem. The various techniques and their application are discussed.

Balakrishnan, A. V.↗

NASA NTRS1987

Non-linear identification of a squeeze-film damper

Described is an experimental study to identify the damping laws associated with a squeeze-film vibration damper. This is achieved by using a non-linear filtering algorithm to process displacement responses of the damper ring to synchronous excitation and thus to estimate the parameters in an nth-power velocity model. The experimental facility is described in detail and a representative selection of results is included. The identified models are validated through the prediction of damper-ring orbits and comparison with observed responses.

Stanway, Roger↗

NASA NTRS1987

Earth observing system instrument pointing control modeling for polar orbiting platforms

An approach to instrument pointing control performance assessment for large multi-instrument platforms is described. First, instrument pointing requirements and reference platform control systems for the Eos Polar Platforms are reviewed. Performance modeling tools including NASTRAN models of two large platforms, a modal selection procedure utilizing a balanced realization method, and reduced order platform models with core and instrument pointing control loops added are then described. Time history simulations of instrument pointing and stability performance in response to commanded slewing of adjacent instruments demonstrates the limits of tolerable slew activity. Simplified models of rigid body responses are also developed for comparison. Instrument pointing control methods required in addition to the core platform control system to meet instrument pointing requirements are considered.

Briggs, H. C.↗

NASA NTRS1990

Projection and assembly method for multibody component model reduction

The problem addressed is that of obtaining reduced-order component models for use in simulating the dynamics of a multibody system. In certain cases, nonlinear system models may be constructed using linear dynamic models for each component, but allowing large angle motion between components. Without some form of model reduction, system models constructed in this manner may be too large for use in control system design and simulation trades. This paper analyzes one method of component model reduction that allows systems level requirements (e.g., capturing the effect of body 1 reaction wheel noise on body 2 camera pointing) to aid in the selection of the reduced-order component models. Briefly stated, important modes are selected at the system level and projected onto the components, and reduced-order components are then assembled into a reduced-order system model that retains the projected modes.

Bernard, Douglas E.↗

NASA NTRS1982

Comparison of the CEAS and Williams-type barley yield models for North Dakota and Minnesota

The CEAS and Williams type models were compared based on specified selection criteria which includes a ten year bootstrap test (1970-1979). Based on this, the models were quite comparable; however, the CEAS model was slightly better overall. The Williams type model seemed better for the 1974 estimates. Because that year spring wheat yield was particularly low, the Williams type model should not be excluded from further consideration.

Leduc, S.↗

NASA NTRS1992

Workshop on Engineering Turbulence Modeling

Discussed here is the future direction of various levels of engineering turbulence modeling related to computational fluid dynamics (CFD) computations for propulsion. For each level of computation, there are a few turbulence models which represent the state-of-the-art for that level. However, it is important to know their capabilities as well as their deficiencies in order to help engineers select and implement the appropriate models in their real world engineering calculations. This will also help turbulence modelers perceive the future directions for improving turbulence models. The focus is on one-point closure models (i.e., from algebraic models to higher order moment closure schemes and partial differential equation methods) which can be applied to CFD computations. However, other schemes helpful in developing one-point closure models, are also discussed.

Povinelli, Louis A.↗

NASA NTRS1999

Backus Effect on a Perpendicular Errors in Harmonic Models of Real vs. Synthetic Data

Measurements of geomagnetic scalar intensity on a thin spherical shell alone are not enough to separate internal from external source fields; moreover, such scalar data are not enough for accurate modeling of the vector field from internal sources because of unmodeled fields and small data errors. Spherical harmonic models of the geomagnetic potential fitted to scalar data alone therefore suffer from well-understood Backus effect and perpendicular errors. Curiously, errors in some models of simulated 'data' are very much less than those in models of real data. We analyze select Magsat vector and scalar measurements separately to illustrate Backus effect and perpendicular errors in models of real scalar data. By using a model to synthesize 'data' at the observation points, and by adding various types of 'noise', we illustrate such errors in models of synthetic 'data'. Perpendicular errors prove quite sensitive to the maximum degree in the spherical harmonic expansion of the potential field model fitted to the scalar data. Small errors in models of synthetic 'data' are found to be an artifact of matched truncation levels. For example, consider scalar synthetic 'data' computed from a degree 14 model. A degree 14 model fitted to such synthetic 'data' yields negligible error, but amplifies 4 nT (rmss) added noise into a 60 nT error (rmss); however, a degree 12 model fitted to the noisy 'data' suffers a 492 nT error (rmms through degree 12). Geomagnetic measurements remain unaware of model truncation, so the small errors indicated by some simulations cannot be realized in practice. Errors in models fitted to scalar data alone approach 1000 nT (rmss) and several thousand nT (maximum).

Voorhies, C. V.↗

NASA NTRS2000

Trainable Gene Regulation Networks with Applications to Drosophila Pattern Formation

This chapter will very briefly introduce and review some computational experiments in using trainable gene regulation network models to simulate and understand selected episodes in the development of the fruit fly, Drosophila melanogaster. For details the reader is referred to the papers introduced below. It will then introduce a new gene regulation network model which can describe promoter-level substructure in gene regulation. As described in chapter 2, gene regulation may be thought of as a combination of cis-acting regulation by the extended promoter of a gene (including all regulatory sequences) by way of the transcription complex, and of trans-acting regulation by the transcription factor products of other genes. If we simplify the cis-action by using a phenomenological model which can be tuned to data, such as a unit or other small portion of an artificial neural network, then the full transacting interaction between multiple genes during development can be modelled as a larger network which can again be tuned or trained to data. The larger network will in general need to have recurrent (feedback) connections since at least some real gene regulation networks do. This is the basic modeling approach taken, which describes how a set of recurrent neural networks can be used as a modeling language for multiple developmental processes including gene regulation within a single cell, cell-cell communication, and cell division. Such network models have been called "gene circuits", "gene regulation networks", or "genetic regulatory networks", sometimes without distinguishing the models from the actual modeled systems.

Mjolsness, Eric↗

NASA NTRS2017

Design of an Axisymmetric Afterbody Test Case for CFD Validation

As identified in the CFD Vision 2030 Study commissioned by NASA, validation of advanced RANS models and scale-resolving methods for computing turbulent flow fields must be supported by continuous improvements in fundamental, high-fidelity experiments designed specifically for CFD implementation. In accordance with this effort, the underpinnings of a new test platform referred to herein as the NASA Axisymmetric Afterbody are presented. The devised body-of-revolution is a modular platform consisting of a forebody section and afterbody section, allowing for a range of flow behaviors to be studied on interchangeable afterbody geometries. A body-of-revolution offers advantages in shape definition and fabrication, in avoiding direct contact with wind tunnel sidewalls, and in tail-sting integration to facilitate access to higher Reynolds number tunnels. The current work is focused on validation of smooth-body turbulent flow separation, for which a six-parameter body has been developed. A priori RANS computations are reported for a risk-reduction test configuration in order to demonstrate critical variation among turbulence model results for a given afterbody, ranging from barely-attached to mild separated flow. RANS studies of the effects of forebody nose (with/without) and wind tunnel boundary (slip/no-slip) on the selected afterbody are presented. Representative modeling issues that can be explored with this configuration are the effect of higher Reynolds number on separation behavior, flow physics of the progression from attached to increasingly-separated afterbody flows, and the effect of embedded longitudinal vortices on turbulence structure.

Disotell, Kevin J.↗