Search NASA⌕ Search

SEARCH · Search NASA

Results for “statistical processing methods”

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 1,207 records · Page 67

Analysis of Time Dependent Data and PRA

In the past decades, several numerical simulation codes have been employed to simulate accident dynamics (e.g., RELAP5-3D, MELCOR, MAAP). In order to evaluate the impact of uncertainties into accident dynamics, several stochastic methodologies have been coupled with these codes. These stochastic methods range from classical Monte-Carlo and Latin Hypercube sampling to stochastic polynomial methods. Similar approaches have been introduced into the risk and safety community where stochastic methods (such as RAVEN, ADAPT, MCDET, ADS) have been coupled with safety analysis codes in order to evaluate the safety impact of timing and sequencing of events. These approaches are usually called Dynamic PRA or simulation-based PRA methods. These uncertainties and safety methods usually generate a large number of simulation runs (database storage may be on the order of gigabytes or higher). The scope of this paper is to present a broad overview of methods and algorithms that can be used to analyze and extract information from large data sets containing time dependent data. In this context, 'extracting information' means constructing input-output correlations, finding commonalities, and identifying outliers. Some of the algorithms presented here have been developed or are under development within the RAVEN statistical framework. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Local orderings in long-range-disordered bismuth-layered intergrowth structure

A series of intergrowth bismuth-layered (Bi{sub 3}TiNbO{sub 9}){sub 2}(Bi{sub 4}Ti{sub 3}O{sub 12}) (2{sub 2}3) ceramics were prepared by conventional solid-state reaction to study the characteristics of the local orderings in long-range-disordered intergrowth structures. High-resolution high-angle annular dark-field (HAADF) imaging reveals the intergrowth structure composed of mixtures of -23-, -223-, -2223- and -22- sequences, while the -223- structure is the thermodynamic stable state of this intergrowth system. It was confirmed by the crystals of recurrent -223- structure prepared by self-flux method and the nature of the local ordering was discussed from their differences in repeating units. The statistics show that when repeating units reach 4 or higher, the independent -223- intergrowth ordering emerges clearly among the competing associated orderings. We infer it is the kinetic factor that induces local compositional variance to result in long-range disordered intergrowth structures. - Graphical abstract: The long-range-disordered intergrowth structure in a (Bi{sub 3}TiNbO{sub 9}){sub 2}(Bi{sub 4}Ti{sub 3}O{sub 12}) (2{sub 2}3) grain, which is composed of various types of local orderings, such as -22-, -23- and -223-. - Highlights: • The characteristic of the long-range-disordered (Bi{sub 3}TiNbO{sub 9}){sub 2}(Bi{sub 4}Ti{sub 3}O{sub 12}) (2{sub 2}3) structure was statistically analyzed, and the ordered -223- structure was speculated to be the thermodynamic stable state of the system. • The crystals of the -223- structure were successfully prepared for the first time by self-melt method. • The lower limit of the repeating units (L) to uniquely determine an independent intergrowth structure was speculated to be L=4. • The analysis inferred that the kinetic process is the controlling factor to limit the structural continuity and induce the long-range-disordered intergrowth structure.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Practical rare event sampling for extreme mesoscale weather

Extreme mesoscale weather, including tropical cyclones, squall lines, and floods, can be enormously damaging and yet challenging to simulate; hence, there is a pressing need for more efficient simulation strategies. Here, we present a new rare event sampling algorithm called quantile diffusion Monte Carlo (quantile DMC). Quantile DMC is a simple-to-use algorithm that can sample extreme tail behavior for a wide class of processes. We demonstrate the advantages of quantile DMC compared to other sampling methods and discuss practical aspects of implementing quantile DMC. To test the feasibility of quantile DMC for extreme mesoscale weather, we sample extremely intense realizations of two historical tropical cyclones, 2010 Hurricane Earl and 2015 Hurricane Joaquin. Our results demonstrate quantile DMC’s potential to provide low-variance extreme weather statistics while highlighting the work that is necessary for quantile DMC to attain greater efficiency in future applications.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

SOURCE DETECTION IN INTERFEROMETRIC VISIBILITY DATA. I. FUNDAMENTAL ESTIMATION LIMITS

Transient radio signals of astrophysical origin present an avenue for studying the dynamic universe. With the next generation of radio interferometers being planned and built, there is great potential for detecting and studying large samples of radio transients. Currently used image-based techniques for detecting radio sources have not been demonstrated to be optimal, and there is a need for development of more sophisticated algorithms and methodology for comparing different detection techniques. A visibility-space detector benefits from our good understanding of visibility-space noise properties and does not suffer from the image artifacts and need for deconvolution in image-space detectors. In this paper, we propose a method for designing optimal source detectors using visibility data, building on statistical decision theory. The approach is substantially different to conventional radio astronomy source detection. Optimal detection requires an accurate model for the data, and we present a realistic model for the likelihood function of radio interferometric data, including the effects of calibration, signal confusion, and atmospheric phase fluctuations. As part of this process, we derive fundamental limits on the calibration of an interferometric array, including the case where many relatively weak 'in-beam' calibrators are used. These limits are then applied, along with a model for atmospheric phase fluctuations, to determine the limits on measuring source position, flux density, and spectral index, in the general case. We then present an optimal visibility-space detector using realistic models for an interferometer.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Stochastic Price Generation for Evaluating Wholesale Electricity Market Bidding Strategies

This work presents a novel method for generating electricity price scenarios from statistical properties of past electricity prices using a hybrid statistical and reduced-form stochastic model. Previous work in applying stochastic differential equations (SDE) to model electricity prices has focused on daily average prices. To extend stochastic price generation methods to hourly or sub-hourly pricing, we address several weaknesses in the state-of-the-art: (1) we replace the mean-reversion component of the SDE with an ARIMA process that is better able to characterize the daily and weekly trends; (2) we extend the price-spike, or jump process to account for conditional probabilities of price spikes occurring in consecutive time steps by replacing the traditional Poisson process for modeling jumps with a generalized point process model inspired by brain neuron models; and (3) we replace the traditional method of estimating spike intensity with empirical variance with a Markov process based on observed price spike intensity transitions. The method is demonstrated with electricity prices from the US ERCOT market and a use-case example is provided for bidding an energy storage unit into the day-ahead and real-time energy markets of ERCOT using stochastic optimization methods. Results show that the the synthetic price model out performs a (naive) persistence forecast model by resulting in 24% to 47% more in profits over 168 simulated days.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

A Ground Flash Fraction Retrieval Algorithm for GLM

A Bayesian inversion method is introduced for retrieving the fraction of ground flashes in a set of N lightning observed by a satellite lightning imager (such as the Geostationary Lightning Mapper, GLM). An exponential model is applied as a physically reasonable constraint to describe the measured lightning optical parameter distributions. Population statistics (i.e., the mean and variance) are invoked to add additional constraints to the retrieval process. The Maximum A Posteriori (MAP) solution is employed. The approach is tested by performing simulated retrievals, and retrieval error statistics are provided. The approach is feasible for N greater than 2000, and retrieval errors decrease as N is increased.

Koshak, William J.↗

Use of Design of Experiments in Determining Neural Network Architectures for Loss of Control Detection

We describe empirical methods for selecting a neural network architecture to implement belief state inference on generic commercial transport aircraft. We highlight a case study on the planning, execution, and analysis of a set of experiments to determine the configurations of a conditional variational autoencoder (CVAE). Our main contribution is the application of a structured method that can be used for machine learning in many aerospace applications. This method optimizes the structure and training parameters of a neural network for belief state inference, using Design of Experiments (DOE) statistical methodologies. The motivation for this specific DOE analysis was to identify the appropriate hyperparameters for measuring the CVAE reconstruction probability and latent space, such that the measurements can be used to infer qualitative state changes for the aircraft. We demonstrate that this process yields information about a trained neural network’s utility for this specific application, along with a quantifiable range of certainty. We execute 84 experiments using loss-of-control flight maneuver data from the NASA T-2 aircraft, demonstrating that this empirical process allows us to construct cheap and simple models with specific attributes amenable to belief state inference in aerospace applications.

Loss of Control↗

Use of Design of Experiments in Determining Neural Network Architectures for Loss of Control Detection

We describe empirical methods for selecting a neural network architecture to implement belief state inference on generic commercial transport aircraft. We highlight a case study on the planning, execution, and analysis of a set of experiments to determine the configurations of a conditional variational autoencoder (CVAE). Our main contribution is the application of a structured method that can be used for machine learning in many aerospace applications. This method optimizes the structure and training parameters of a neural network for belief state inference, using Design of Experiments (DOE) statistical methodologies. The motivation for this specific DOE analysis was to identify the appropriate hyperparameters for measuring the CVAE reconstruction probability and latent space, such that the measurements can be used to infer qualitative state changes for the aircraft. We demonstrate that this process yields information about a trained neural network’s utility for this specific application, along with a quantifiable range of certainty. We execute 84 experiments using loss-of-control flight maneuver data from the NASA T 2 aircraft, demonstrating that this empirical process allows us to construct cheap and simple models with specific attributes amenable to belief state inference in aerospace applications.

Loss of Control↗

Use of Design of Experiments in Determining Neural Network Architectures for Loss of Control Detection

Abstract—We describe empirical methods for selecting a neural network architecture to implement belief state inference on generic commercial transport aircraft. We highlight a case study on the planning, execution, and analysis of a set of experiments to determine the configurations of a conditional variational autoencoder (CVAE). Our main contribution is the application of a structured method that can be used for machine learning in many aerospace applications. This method optimizes the structure and training parameters of a neural network for belief state inference, using Design of Experiments (DOE) statistical methodologies. The motivation for this specific DOE analysis was to identify the appropriate hyperparameters for measuring the CVAE reconstruction probability and latent space, such that the measurements can be used to infer qualitative state changes for the aircraft. We demonstrate that this process yields information about a trained neural network’s utility for this specific application, along with a quantifiable range of certainty. We execute 84 experiments using loss-of-control flight maneuver data from the NASA T-2 aircraft, demonstrating that this empirical process allows us to construct cheap and simple models with specific attributes amenable to belief state inference in aerospace applications.

neural networks↗

Predicting Time Series Outputs and Time-to-Failure for an Aircraft Controller Using Bayesian Modeling

Safety of unmanned aerial systems (UAS) is paramount, but the large number of dynamically changing controller parameters makes it hard to determine if the system is currently stable, and the time before loss of control if not. We propose a hierarchical statistical model using Treed Gaussian Processes to predict (i) whether a flight will be stable (success) or become unstable (failure), (ii) the time-to-failure if unstable, and (iii) time series outputs for flight variables. We first classify the current flight input into success or failure types, and then use separate models for each class to predict the time-to-failure and time series outputs. As different inputs may cause failures at different times, we have to model variable length output curves. We use a basis representation for curves and learn the mappings from input to basis coefficients. We demonstrate the effectiveness of our prediction methods on a NASA neuro-adaptive flight control system.

Statistics↗

Natural Language Processing-Enhanced Nuclear Industry Operating Experience Data Analysis: Aggregation and Interpretation of Multi-Report Analysis Results

Industry-wide operating experience is a critical source of raw data for reliability and risk model parameter estimations for nuclear power plants. A large portion of operating experience data are failure events stored as reports that contain unstructured data, such as narratives. In current practice, a failure report is usually reviewed and manually coded by analysts. The coding is based on extracting several event characteristics such as system name, component type, sub-part type, failure mode, and failure cause. Event narratives are mostly used to help understand events and extract their characteristics. In this line of research, we aim to maximize the usage of event narratives by leveraging natural language processing (NLP) methods to automatically convert an event narrative to a causal graph. This research has promise to improve physical understanding of failure initiation and propagation and to facilitate use of non-failure data (e.g., near-misses and degradations) to complement the limited data pool of failures. In our previous work, we developed an NLP tool and applied it to analyze a number of licensee event reports submitted by U.S. nuclear power plants to the Nuclear Regulatory Commission. In this paper, we will report our recent research progress in aggregating the results of multiple reports, developing network model(s), and drawing statistical insights.

99 GENERAL AND MISCELLANEOUS↗

Examining Anomalous Network Performance with Confidence

Variability in network performance is a major obstacle in effectively analyzing the throughput of modern high performance computer systems. High performance interconnection networks offer excellent best-case network latencies; however, highly parallel applications running on parallel machines typically require consistently high levels of performance to adequately leverage the massive amounts of available computing power. Performance analysts have usually quantified network performance using traditional summary statistics that assume the observational data is sampled from a normal distribution. In our examinations of network performance, we have found this method of analysis often provides too little data to understand anomalous network performance. In particular, we examine a multi-modal performance scenario encountered with an Infiniband interconnection network and we explore the performance repeatability on the custom Cray SeaStar2 interconnection network after a set of software and driver updates.

99 GENERAL AND MISCELLANEOUS↗

Observational evidence against strongly stabilizing tropical cloud feedbacks

In this paper, we present a method to attribute cloud radiative feedbacks to convective processes, using subcloud layer buoyancy as a diagnostic of stable and deep convective regimes. Applying this approach to tropical remote sensing measurements over years 2000–2016 shows that an inferred negative short-term cloud feedback from deep convection was nearly offset by a positive cloud feedback from stable regimes. The net cloud feedback was within statistical uncertainty of the National Center for Atmospheric Research Community Atmosphere Model (CAM5) with historical forcings, with discrepancies in the partitioning of the cloud feedback into convective regimes. Compensation between high-cloud responses to tropics-wide warming in stable and unstable regimes resulted in smaller net changes in high-cloud fraction with warming. In addition, deep convection and associated high clouds set in at warmer temperatures in response to warming, as a consequence of nearly invariant subcloud buoyancy. Lastly, this invariance further constrained the magnitude of cloud radiative feedbacks and is consistent with climate model projections.

54 ENVIRONMENTAL SCIENCES↗

Exploiting Machine Learning in Multiscale Modelling of Materials

Recent developments in efficient machine learning algorithms have spurred significant interest in the materials community. The inherently complex and multiscale problems in Materials Science and Engineering pose a formidable challenge. The present scenario of machine learning research in Materials Science has a clear lacunae, where efficient algorithms are being developed as a separate endeavour, while such methods are being applied as ‘black-box’ models by others. The present article aims to discuss pertinent issues related to the development and application of machine learning algorithms for various aspects of multiscale materials modelling. The authors present an overview of machine learning of equivariant properties, machine learning-aided statistical mechanics, the incorporation of ab initio approaches in multiscale models of materials processing and application of machine learning in uncertainty quantification. In addition to the above, the applicability of Bayesian approach for multiscale modelling will be discussed. Critical issues related to the multiscale materials modelling are also discussed.

42 ENGINEERING↗

Identifying Predictive Factors for Incident Reports in Patients Receiving Radiation Therapy

Purpose: To describe radiation therapy cases during which voluntary incident reporting occurred; and identify patient- or treatment-specific factors that place patients at higher risk for incidents. Methods and Materials: We used our institution's incident learning system to build a database of patients with incident reports filed between January 2011 and December 2013. Patient- and treatment-specific data were reviewed for all patients with reported incidents, which were classified by step in the process and root cause. A control group of patients without events was generated for comparison. Summary statistics, likelihood ratios, and mixed-effect logistic regression models were used for group comparisons. Results: The incident and control groups comprised 794 and 499 patients, respectively. Common root causes included documentation errors (26.5%), communication (22.5%), technical treatment planning (37.5%), and technical treatment delivery (13.5%). Incidents were more frequently reported in minors (age <18 years) than in adult patients (37.7% vs 0.4%, P<.001). Patients with head and neck (16% vs 8%, P<.001) and breast (20% vs 15%, P=.03) primaries more frequently had incidents, whereas brain (18% vs 24%, P=.008) primaries were less frequent. Larger tumors (17% vs 10% had T4 lesions, P=.02), and cases on protocol (9% vs 5%, P=.005) or with intensity modulated radiation therapy/image guided intensity modulated radiation therapy (52% vs 43%, P=.001) were more likely to have incidents. Conclusions: We found several treatment- and patient-specific variables associated with incidents. These factors should be considered by treatment teams at the time of peer review to identify patients at higher risk. Larger datasets are required to recommend changes in care process standards, to minimize safety risks.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Physical model assisted probability of detection in nondestructive evaluation

Nondestructive evaluation is used widely in many engineering and industrial areas to detect defects or flaws such as cracks inside parts or structures during manufacturing or for products in service. The standard statistical model is a simple empirical linear regression between the (possibly transformed) signal response variables and the (possibly transformed) explanatory variables. For some applications, such a simple empirical approach is inadequate. An important alternative approach is to use knowledge of the physics of the inspection process to provide information about the underlying relationship between the response and explanatory variables. Use of such knowledge can greatly increase the power and accuracy of the statistical analysis and enable, when needed, proper extrapolation outside the range of the observed explanatory variables. This paper describes a set of physical model-assisted analyses to study the capability of two different ultrasonic testing inspection methods to detect synthetic hard alpha inclusion and flat-bottom hole defects in a titanium forging disk.

42 ENGINEERING↗

A symmetrical quasi-classical windowing model for the molecular dynamics treatment of non-adiabatic processes involving many electronic states

In the previous work of Cotton and Miller [J. Chem. Phys. 145, 144108 (2016)], an improved symmetrical quasi-classical (SQC) windowing model for the molecular dynamics treatment of electronically non-adiabatic processes was developed in order to extend the original SQC approach to the regime of weak-coupling between the electronic states. Furthermore, the improved SQC model—based on triangular-shaped window functions—handled the weak-coupling limit as intended and, as a bonus, was shown to be universally superior to the original square/histogram SQC windowing model over all coupling regimes, but only for treating systems of two electronic states, as no higher-dimensional generalization was evident. This paper, therefore, provides a generalized version for treating an arbitrary number of electronic states. By construction, the benefits of the two-state triangle model—seamless treatment of weak-coupling and improved accuracy in all coupling regimes—carry over to the generalized version. Far more significant, however, is that the new model provides vastly improved windowing statistics in higher dimensions, enabling the SQC simulation of electronically non-adiabatic processes involving many more relevant electronic states than was previously practical. Capabilities are demonstrated with respect to a 24 pigment trimer model of the Fenna-Matthews-Olson light-harvesting complex, as well as treating similar 48- and 96-electronic state model problems, illustrating the scaling properties of the new method.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

PHYSICS OF NON-GAUSSIAN FIELDS AND THE COSMOLOGICAL GENUS STATISTIC

We report a technique to calculate the impact of distinct physical processes inducing non-Gaussianity on the cosmological density field. A natural decomposition of the cosmic genus statistic into an orthogonal polynomial sequence allows complete expression of the scale-dependent evolution of the topology of large-scale structure, in which effects including galaxy bias, nonlinear gravitational evolution, and primordial non-Gaussianity may be delineated. The relationship of this decomposition to previous methods for analyzing the genus statistic is briefly considered and the following applications are made: (1) the expression of certain systematics affecting topological measurements, (2) the quantification of broad deformations from Gaussianity that appear in the genus statistic as measured in the Horizon Run simulation, and (3) the study of the evolution of the genus curve for simulations with primordial non-Gaussianity. These advances improve the treatment of flux-limited galaxy catalogs for use with this measurement and further the use of the genus statistic as a tool for exploring non-Gaussianity.

79 ASTRONOMY AND ASTROPHYSICS↗