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.
In many pattern recognition problems, data vectors are classified although one or more of the data vector elements are missing. This problem occurs in remote sensing when the ground is obscured by clouds. Optimal linear discrimination procedures for classifying imcomplete data vectors are discussed.
An inconsistency is examined in generic top-down design methods and standards employed in the implementation of reliable software. Many design approaches adopt top-down ordering when defining the structure, interfaces, and processing of a system. However, strict adherence to a top-down sequencing does not permit accurate description of a system's error handling functions. The design of the system response to errors is becoming critical as the reliability requirements of systems increase. How top-down methods such as object oriented design and structured design do not adequately address the issues of error handling is described, and it is suggested using a bottom-up substep within these methods to eliminate the problem.
Machine learning (ML) has become an invaluable tool across a wide array of domains in science as researchers find new ways to leverage its predictive power. This is especially true in chemistry, where ML is used to fit chemical properties or desirable attributes to the local structure of molecules and materials. In the pursuit of greater accuracy, it is relatively simple to increase the size or complexity of such models, although this often requires simultaneously seeking larger datasets in order to both fit and interpret the larger number of parameters. However, it is equally important to assess the quality and relative importance of the data and how these factors impact the training process. We, therefore, investigate the impact of using different loss functions for training neural network potentials (NNPs), as the loss function defines the error and parameter gradients used to train the NNP. In particular, we test the mean-squared error and Huber loss functions and, using insight from these functions, derive a new loss function based on the Asinh function, which yields significant improvement in the accuracy and generality of NNPs. We show that by discounting/minimizing errors and anomalies in the optimization process, both the Huber and Asinh loss functions improve the training of NNPs, leading to a final potential with a greater effective dimensionality.
Error acquisition probability in Mark 1 ranging subsystem as function of signal-to-noise ratio and integration time used in correlation
Double stochastic approximation algorithm for minimizing mean square error in finite expansion of unknown probability distribution functions
A description is presented of high-dispersion spectroscopic observations of eight chemically peculiar stars including Kappa Cnc and Mu Lep, and of one normal comparison star, Nu Cap, in the vicinity of 1362 A. The reported data were obtained with the International Ultraviolet Explorer during the first 15 months of its operation. The initial reduction of the data, which were originally described in a qualitative summary by Leckrone (1980), was affected by a serious error in the calibration of the instrument's intensity transfer function. This error has been corrected in the currently presented data. The results of spectrum syntheses of the 1362 A feature are also described for several of the stars.
Certain physical processes are modeled by partial differential equations which are parabolic over part of the domain and elliptic over the remainder. A family of semi-implicit algorithms which are well suited to initial-boundary value problems of this mixed type is discussed. One important feature of these algorithms is the use of an approximate inverse for the solution of the implicit linear system. A strong error analysis results in an estimate of the total error as a function of approximate inverse error e and time step h.
Given the ubiquity of electronic displays integration in human-based systems, the impending mission critical, space-based applications of electronic displays will necessitate SEE assessment of components unique to electronic displays. A commercially available DDIC designed to drive a small form factor organic light emitting diode (OLED) was visually monitored during heavy ion irradiation to catalogue radiation induced persistent visual error signatures that require manual intervention (i.e., power cycling) to return to nominal function. These error signatures were able to be reproduced via modification of configuration register values utilizing the instruction set intended for interfacing a microcontroller with the DDIC. This approach to emulation of heavy-ion induced errors on a table-top assists with human perception-based criticality analysis as well as development of mitigation techniques.
Analyses of AGC analog data from the Small Astronomy Satellite (SAS) at VHF (136 MHz) have shown amplitude fluctuations in the signals received at several equatorial tracking sites operated by NASA. Errors in data are shown to increase markedly for nighttime satellite passes around the periods of the two equinoxes as compared to the errors observed during other periods. Qualitative and quantitative correlations are established between the ionospheric scintillation phenomenon and the occurrence of errors. Strong time and geographic cause-and-effect relationships are established by determining the statistical correlation between scintillation indices and bit error rates. Detailed analyses of the AGC fading statistics and the fine grain error statistics, such as the block-error distribution and the bit-error autocorrelation function, establish the strong correlation between deep fades and their effects on the burst error structure associated with the transionospheric channel and the communication receiver being employed.
Mariner IV midcourse 50-lb thrust pressure- controlled monopropellant hydrazine propellant system, predicting impulse and velocity error as function of burn time
An error propagation model has been developed for multimodule computing systems in which the main parameters are the distribution functions of error propagation times. A digraph model is used to represent a multimodule computing system, and error propagation in the system is modeled by general distributions of error propagation times between all pairs of modules. Two algorithms are developed to compute systematically and efficiently the distributions of error propagation times. Experiments are also conducted to measure the distributions of error propagation times with the fault-tolerant microprocessor (FTMP). Statistical analysis of experimental data shows that the error propagation times in FTMP do not follow a well-known distribution, thus justifying the use of general distributions in the present model.
This paper develops a method to tune fuzzy controllers using numerical optimization. The main attribute of this approach is that it allows fuzzy logic controllers to be tuned to achieve global performance requirements. Furthermore, this approach allows design constraints to be implemented during the tuning process. The method tunes the controller by parameterizing the membership functions for error, change-in-error and control output. The resulting parameters form a design vector which is iteratively changed to minimize an objective function. The minimal objective function results in an optimal performance of the system. A spacecraft mounted science instrument line-of-sight pointing control is used to demonstrate results.
A quasi-geostrophic (QG) stream function is analyzed by optimal interpolation (OI) over a 59-day period in a 150-km-square domain off northern California. Hydrographic observations acquired over five surveys were assimilated into a QG open boundary ocean model. Assimilation experiments were conducted separately for individual surveys to investigate the sensitivity of the OI analyses to parameters defining the decorrelation scale of an assumed error covariance function. The analyses were intercompared through dynamical hindcasts between surveys. The best hindcast was obtained using the smooth analyses produced with assumed error decorrelation scales identical to those of the observed stream function. The rms difference between the hindcast stream function and the final analysis was only 23 percent of the observation standard deviation. The two sets of OI analyses were temporally smoother than the fields from statistical objective analysis and in good agreement with the only independent data available for comparison.
The determination of maximum endpoint tracking error due to uncompensated inertial payload dynamics is investigated. The problem is formulated on the basis of well-known kinetic and kinematic relationships with constraints on maximum joint positions and velocities as well as the generalized force/speed constraint relationship for each actuator. An endpoint tracking error objective function is formed, and numerical optimization is used to determine the maximum. An example computation is presented to show the tracking error under progressively increasing payload using a PUMA 560 manipulator with its first three joints in motion.
Existence of q-ary uniform convolutional codes where q is any prime-power, noting error- correcting ability comparable to familiar maximal- length block codes
A method is developed for remote sensing of aerosol absorption from satellite images of the earth's surface. The method is based on the measurement of the change in the upward radiances between a clear and a hazy day over a varying surface reflectance. For a zero change balance between brightening due to scattering and darkening due to absorption and scattering is reached. This balance is utilized via a radiative transfer model to derive the aerosol single-scattering albedo. A sensitivity study is performed, and the method is tested against laboratory measurements. It is suggested that for the case of haze introduced on top of an existing background aerosol and with a fair estimate of the scattering phase function, the error in the remotely sensed single-scattering albedo is in the range of 0.03-0.05. The main errors in the method arise from variations in the surface reflectance between the clear and the hazy days, uncertainty in the scattering phase function, and variation of the aerosol and gaseous absorption between these two days. If the satellite calibration varies with time, the measurements of single-scattering albedo can be substantially affected.
In the study of motion effects on tracking performance, a problem of interest is the determination of what sensory inputs a human uses in controlling his tracking task. In the approach presented here a simple canonical model (FID or a proportional, integral, derivative structure) is used to model the human's input-output time series. A study of significant changes in reduction of the output error loss functional is conducted as different permutations of parameters are considered. Since this canonical model includes parameters which are related to inputs to the human (such as the error signal, its derivatives and integration), the study of model order is equivalent to the study of which sensory inputs are being used by the tracker. The parameters are obtained which have the greatest effect on reducing the loss function significantly. In this manner the identification procedure converts the problem of testing for model order into the problem of determining sensory inputs.
The use of a flow-dependent correlation function to improve the accuracy of an optimum interpolation (OI) scheme is examined. The development of the correlation function for the OI analysis scheme used for numerical weather prediction is described. The scheme uses a multivariate surface analysis over the oceans to model the pressure-wind error cross-correlation and it has the ability to use an error correlation function that is flow- and geographically-dependent. A series of four-day data assimilation experiments, conducted from January 5-9, 1979, were used to investigate the effect of the different features of the OI scheme (error correlation) on forecast skill for the barotropic lows and highs. The skill of the OI was compared with that of a successive correlation method (SCM) of analysis. It is observed that the largest difference in the correlation statistics occurred in barotropic and baroclinic lows and highs. The comparison reveals that the OI forecasts were more accurate than the SCM forecasts.