Engineering Papers⌕ Search

Engineering topics

Peters, C.

Publications and source records attributed to Peters, C..

15 records
All sourcesNASA NTRSDOE OSTI
NASA NTRS1984

Laser thermal propulsion

Laser thermal propulsion (LTP) is studied for the case in which laser power is absorbed by a small very high-temperature plasma (about 20,000 K) and transferred to the remainder of the pure hydrogen propellant by radiation and mixing. This concept could lead to the realization of a lightweight orbital transfer vehicle propulsion system having a specific impulse in the range 1000-2000 s. Approximately 12 percent of the input power may be radiated to the thruster walls, and 15 percent of the total propellant flow must be heated to 20,000 K to provide a bulk temperature of 5000 K prior to expansion. Three principal research issues identified are: (1) conditions for hydrogen plasma ignition, (2) control of the plasma position within the laser beam, plasma stability, and plasma absorption efficiency, and (3) characterization of the mixing of the plasma and buffer flows.

Keefer, D.↗
NASA NTRS1984

Bayesian estimation of normal mixture parameters

A Bayesian, or penalized maximum likelihood, approach to the problem of estimating the parameters of a mixture of multivariate normal distributions is proposed. The Bayesian formulation eliminates the problem of singularities in the likelihood function and results in an attractive EM-like procedure. Although the question of consistency is not settled, it is suggested that the proposed method has certain advantages over both the constrained and unconstrained maximum likelihood procedures.

Peters, C.↗
NASA NTRS1983

A reexamination of the laser supported combustion wave

Kantrowitz (1972) and Minovitch (1972) have proposed the use of laser sustained plasmas as a means to heat a rocket propellant. Recent studies of laser-powered propulsion have been directed toward the application of high-specific-impulse space propulsion systems for orbital transfer missions. Analyses of rocket performance relied heavily on the concept of the laser-supported combustion (LSC) wave. Raizer (1971) first drew the analogy between laser-sustained plasmas and combustion waves in an analysis. The Raizer model was later applied to hydrogen by Kemp and Root (1979). In connection with certain problems arising with the approach considered by Kemp and Root, the present investigation is concerned with a reexamination of the Raizer model. Attention is given to a numerical approach for the entire LSC wave in hydrogen, taking into account the incorporation of the proper boundary conditions far downstream of the wave.

Keefer, D.↗
NASA NTRS1983

Covariance hypotheses for LANDSAT data

Two covariance hypotheses are considered for LANDSAT data acquired by sampling fields, one an autoregressive covariance structure and the other the hypothesis of exchangeability. A minimum entropy approximation of the first structure by the second is derived and shown to have desirable properties for incorporation into a mixture density estimation procedure. Results of a rough test of the exchangeability hypothesis are presented.

Decell, H. P.↗
NASA NTRS1983

Mixture Models for Dependent Observations

Parametric mixture models appropriate for data presented in homogeneous blocks of varying sizes from several unidentified source populations are considered. For most applications, the data elements within each block are dependent. Models are proposed for multivariate normal data incorporating two types of dependence, exchangeability of elements within blocks, and a Markov structure for blocks. The consequences of assuming exchangeability, when in fact the Markov structure holds, are explored. Computational problems for each model are considered, and results of a simple test of the exchangeability hypothesis for LANDSAT data are presented.

Peters, C.↗
NASA NTRS1982

The possible modifications of the Hisse model for pure LANDSAT agricultural data

An idea, due to A. Feiveson, is presented for relaxing the assumption of class conditional independence of LANDSAT spectral measurements within the same patch (field). Theoretical arguments are given which show that any significant refinement of the model beyond Feiveson's proposal will not allow the reduction, essential to HISSE, of the pure data to patch summary statistics. A slight alteration of the new model is shown to be a reasonable approximation to the model which describes pure data elements from the same patch as jointly Guassian with a covariance function which exhibits exponential decay with respect to spatial separation.

Peters, C.↗
NASA NTRS1981

The possible modifications of the HISSE model for pure LANDSAT agricultural data

A method for relaxing the assumption of class conditional independence of LANDSAT spectral measurements within the same patch (field) is discussed. Theoretical arguments are given which show that any significant refinement of the model beyond this proposal will not allow the reduction, essential to HISSE, of the pure data to patch summary statistics. A slight alteration of the new model is shown to be a reasonable approximation to the model which describes pure data elements from the same patch as jointly Gaussian with a covariance function which exhibits exponential decay with respect to spatial separation.

Peters, C.↗
NASA NTRS1980

On the existence, uniqueness, and asymptotic normality of a consistent solution of the likelihood equations for nonidentically distributed observations: Applications to missing data problems

A general theorem is given which establishes the existence and uniqueness of a consistent solution of the likelihood equations given a sequence of independent random vectors whose distributions are not identical but have the same parameter set. In addition, it is shown that the consistent solution is a MLE and that it is asymptotically normal and efficient. Two applications are discussed: one in which independent observations of a normal random vector have missing components, and the other in which the parameters in a mixture from an exponential family are estimated using independent homogeneous sample blocks of different sizes.

Peters, C.↗
NASA NTRS1980

Numerical trials of HISSE

The mathematical description and implementation of the statistical estimation procedure known as the Houston integrated spatial/spectral estimator (HISSE) is discussed. HISSE is based on a normal mixture model and is designed to take advantage of spectral and spatial information of LANDSAT data pixels, utilizing the initial classification and clustering information provided by the AMOEBA algorithm. The HISSE calculates parametric estimates of class proportions which reduce the error inherent in estimates derived from typical classify and count procedures common to nonparametric clustering algorithms. It also singles out spatial groupings of pixels which are most suitable for labeling classes. These calculations are designed to aid the analyst/interpreter in labeling patches with a crop class label. Finally, HISSE's initial performance on an actual LANDSAT agricultural ground truth data set is reported.

Peters, C.↗
NASA NTRS1978

Feature selection for best mean square approximation of class densities

A criterion for linear feature selection is proposed which is based on mean square apporximation of class density functions. It is shown that for the widest possible class of approximants, the criterion reduces to Devijver's Bayesian distance. For linear approximants the criterion is equivalent to well known generalized Fisher criteria.

Peters, C.↗