SEARCH · Search NASA
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.
Variance relationships among certain count estimators and posterior probability estimators of probability of correct classification are investigated. An estimator using posterior probabilities is presented for use in stratified sampling designs. A test case involving three normal classes is examined.
A new approach is taken to the problem of tracking a fixed amplitude signal with a Brownian-motion phase process. Classically, a first-order phase-lock loop (PLL) is used; here, the problem is treated via estimation of the quadrature signal components. In this space, the state dynamics are linear with white multiplicative noise. Therefore, linear minimum-variance filters, which have a particularly simple mechanization, are suggested. The resulting error dynamics are linear at any signal/noise ratio, unlike the classical PLL. During synchronization, and above threshold, this filter with constant gains degrades by 3 per cent in output rms phase error with respect to the classical loop. However, up to 80 per cent of the maximum possible noise improvement is obtained below threshold, where the classical loop is nonoptimum, as demonstrated by a Monte Carlo analysis. Filter mechanizations are presented for both carrier and baseband operation.
A new on-line minimum variance filter for the identification of systems with additive and multiplicative noise is described which embodies both accuracy and computational efficiency. The resulting filter is shown to use both the covariance of the parameter vector itself and the covariance of the error in identification. A bias reduction scheme can be used to yield asymptotically unbiased estimates. Experimental results for simulated linearized lateral aircraft motion in a digital closed loop mode are presented, showing the utility of the identification schemes.
An on-line minimum variance parameter identifier is developed which embodies both accuracy and computational efficiency. The formulation results in a linear estimation problem with both additive and multiplicative noise. The resulting filter which utilizes both the covariance of the parameter vector itself and the covariance of the error in identification is proven to be mean square convergent and mean square consistent. The MV parameter identification scheme is then used to construct a stable state and parameter estimation algorithm.
An on-line minimum variance (MV) parameter identifier is developed which embodies both accuracy and computational efficiency. The formulation results in a linear estimation problem with both additive and multiplicative noise (AMN). The resulting filter which utilizes both the covariance of the parameter vector itself and the covariance of the error in identification is proven to be mean-square convergent and mean-square consistent. The MV parameter identification scheme is then used to construct a stable state and parameter estimation algorithm.
Computer program incorporates technique that reduces variance of forward Monte Carlo method for given amount of computer time in determining radiation environment in complex organic and inorganic systems exposed to significant amounts of radiation.
Errors associated with the minimum variance analysis of directional discontinuity normal components were investigated using both computer simulation of discontinuities with controlled properties and the examination of current sheets observed by the Mariner 10 spacecraft. The simulated discontinuities were created by adding fluctuations, represented by isotropic noise, to exactly known but varying (in a plane) magnetic field components. An empirical expression for the magnitude of the error in an estimated discontinuity normal component, relative to the total field across the discontinuity, was derived, as well as other relevant statistical properties. Use of the empirical relation in the analysis of 644 discontinuities observed by Mariner 10 provides a more precise, but probably conservative, estimate of an upper bound on the relative normal component value for tangential discontinuities that can be used to separate rotational from tangential discontinuities in studies using only magnetic field data from a single spacecraft, at least for the interplanetary region of space considered.
The paper deals with a statistical analysis of the errors associated with a minimum variance analysis of directional discontinuities by use of an idealized model of these discontinuities and various simulations, and also by an examination of actual Mariner 10 interplanetary magnetic field data. An empirical expression is derived for the magnitude of the error in an estimated discontinuity normal component, relative to the total field across the directional discontinuity. The analysis was performed primarily to aid in differentiating between interplanetary tangential and rotational discontinuities observed by Mariner 10.
An interactive computer program (in FORTRAN) for the variance covariance analysis of VLBI experiments is presented for use in experiment planning, simulation studies and optimal design problems. The interactive mode is especially suited to these types of analyses providing ease of operation as well as savings in time and cost. The geodetic parameters include baseline vector parameters and variations in polar motion and Earth rotation. A discussion of the theroy on which the program is based provides an overview of the VLBI process emphasizing the areas of interest to geodesy. Special emphasis is placed on the problem of determining correlations between simultaneous observations from a network of stations. A model suitable for covariance analyses is presented. Suggestions towards developing optimal observation schedules are included.
Resource inventories require samples with geographical scatter, sometimes not as widely spaced as would be hoped. A simple model of correlation over distances is used to create a minimum variance unbiased estimate population means. The fitting procedure is illustrated from data used to estimate Missouri corn acreage.
One of the most important parameters of a Differential Absorption Lidar (DIAL) system is the delay time between the on and off resonant pulses. It is important that this delay time is sufficiently small to ensure that the atmosphere is effectively frozen between the pulses. Therefore, most Dial systems were designed with two lasers firing alternately less than 1 msec apart. Despite the importance of this parameter in the design of DIAL systems and its contribution to the overall error of a column measurement, very little is known about the size of the error for the case of a direct-detection system using atmospheric backscatter. The ultraviolet DIAL system uses two independent YAG/dye lasers and is therefore suitable for measuring the effects of different pulse delays on the variance of column measurements for a variety of atmospheric conditions. A set of DIAL returns were acquired with the two lasers tuned to the same wavelength and with a range of pulse delay times between 250 microseconds and several minutes. This data set was recorded in full on a computer and was used both to test different averaging techniques and also to evaluate atmospheric contributions to DIAL columns.
Well-known theoretical predictions suggest that variances of large-scale atmospheric fluctuations, for scales somewhat shorter than those of the forcing mechanisms, should obey a power law, k exp -b, where k is the horizontal wave number. The invariance of the exponent b with season, hemisphere, and latitude is tested using satellite-measured brightness temperatures. Global grids of Tiros-N Microwave Sounding Unit channel 4 measurements, which closely approximate the 30- to 150-mbar layer mean temperature, are constructed for January, March, and August 1979. These grids are zonally Fourier-transformed, and the resulting spectra are averaged over four 18 deg-wide latitude bands. Fits of a power law to these spectra over wave numbers 10-36 and 10-26 give values of b which are independent of season, hemisphere, and latitude band, to within statistical uncertainties. The observed values of b are about 3.6, as compared to theoretical predictions of 3.
Chi and Mendel (1984) analyzed the performance of minimum-variance deconvolution (MVD). In this correspondence, a further analysis of the performance of the MVD filter is presented. It is shown that the MVD filter performs like an inverse filter and a whitening filter as SNR goes to infinity, and like a matched filter as SNR goes to zero. The estimation error of the MVD filter is colored noise, but it becomes white when SNR goes to zero. This analysis also conects the error power-spectral density of the MVD filter with the spectrum of the causal-prediction error filter.
An indirect method of estimating the surface heat flux from observations of vertical velocity variance at the lower mid-levels of the convective atmospheric boundary layer is described. Comparison of surface heat flux estimates with those from boundary-layer heating rates is good, and this method seems to be especially suitable for inhomogeneous terrain for which the surface-layer profile method cannot be used.
A comparison has been made between the predictions of the theory for radial variations of both Alfvenic fluctuations and solar wind proton temperatures proposed by Tu (1987, 1988) and the statistical results of hourly averaged plasma and magnetic field data observed by Helios 1 and 2 from launch through 1980 for different solar wind speed regimes. The comparison shows that for speed ranges between 500-800 km/s, the radial variation of the proton temperature between 0.3 and 1 AU can be explained by heating from the cascade energy which is determined by the radial variation of the total variance of magnetic field vector.
A uniform derivation of minimum-variance reduced-order (MVRO) filter-smoother algorithms from Pontrygin's Minimum Principle is presented. An appropriate performance index for a general class of reduced order estimation problem is formulated herein to yield optimal results over the entire time interval of estimation. These results provide quantitative criteria for measuring the performance of certain classes of heuristically designed, suboptimal reduced-order estimators as well as explicit guidance to the suboptimal filter design process with both continuous and discrete filter-smoother algorithms being considered. By the duality principle, the algorithms of reduced-order estimation can be easily extended to the deterministic problems of optimal control (i.e., the regulator and linear tracking problem).
A study is made of the use of a series of variance and bias confidence criteria recently developed for the eigensystem realization algorithm (ERA) identification technique. The criteria are shown to be very effective, not only for indicating the accuracy of the identification results (especially in terms of confidence intervals), but also for helping the ERA user to obtain better results. They help determine the best sample interval, the true system order, how much data to use and whether to introduce gaps in the data used, what dimension Hankel matrix to use, and how to limit the bias or correct for bias in the estimates.
Increasingly, aircraft system designs require that aerodynamic parameters derived from pneumatic measurements be employed as control-system feedbacks. Such high frequency pressure measurements' accuracy is compromised by pressure distortion due to frictional attenuation and pneumatic resonance within the sensing system. A pneumatic distortion model is here formulated and reduced to a low-order state-variable model which retains most of the full model's dynamic characteristics. This reduced-order model is coupled with standard results from minimum variance estimation theory to develop an algorithm to compensate for pneumatic-distortion effects.