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.
The BPP package contains R code to perform Bayesian projection pursuit regression under various regression settings. Bayesian projection pursuit regression learns the functional relationship between inputs and outputs by considering a sum of functions of one dimensional projections of the inputs. This is useful as a general machine learning software.
Using observed and reanalysis research, the pronounced interdecadal variations of Lake Qinghai (LQH) water levels and associated climate factors were diagnosed. From the 1960s to the early 2000s, the water level of LQH in the Tibetan Plateau has experienced a continual decline of 3 m but has since increased considerably. A water budget assessment of the LQH watershed implied that the water vapor flux divergence ∇•Q is the dominant atmospheric process modulating precipitation and subsequently the lake volume change ΔV. The marked interdecadal variability in ΔV and ∇•Q and was found to be related to the North Pacific (NP) and Pacific decadal oscillation (PDO) modes during the cold season (November–March). Through empirical orthogonal function (EOF) and regression analyses, the water vapor sink over the LQH watershed also responds critically to the summer Eurasian wave train modulated by the low-frequency variability associated with the cold season NP and PDO modes. Removal of these variability modes (NP, PDO, and the Eurasian wave train) led to a residual uptrend in the hydrological variables of ΔV, ∇•Q and precipitation, corresponding to the net water level increase. Attribution analysis using the Coupled Model Intercomparison Project phase 5 (CMIP5) single-forcing experiments shows that the simulations driven by greenhouse gas forcing produced a significant increase in the LQH precipitation, while anthropogenic aerosols generated a minor wetting trend as well.
Accurate prediction of an off-normal event in a nuclear reactor is dependent upon the availability of sensory data, reactor core physical condition, and understanding of the underlying phenomenon. This work presents a method to project the data from some discrete sensory locations to the overall reactor domain during conduction cooldown scenarios similar to High Temperature Gas-cooled Reactors (HTGRs). The existing models for conductive cooldown in a heterogeneous multi-body system, such as an assembly of prismatic blocks or pebble beds relies on knowledge of the thermal contact conductance, requiring significant knowledge of local thermal contacts and heat transport possibilities across those contacts. With a priori knowledge of bulk geometry features and some discrete sensors, a machine learning approach was devised. The presented work uses an experimental facility to mimic conduction cooldown with an assembly of 68 cylindrical rods initially heated to 1200 K. High-fidelity temperature data were collected using an infrared (IR) camera to provide training data to the model and validate the predicted temperature data. The machine learning approach used here first converts the macroscopic bulk geometry information into Graph-Laplacian, and then uses the eigenvectors of the Graph-Laplacian to develop Kernel functions. Support vector regression (SVR) was implemented on the obtained Kernels and used to predict the thermal response in a packed rod assembly during a conduction cooldown experiment. The usage of SVR modeling differs from most models today because of its representation of thermal coupling between rods in the core. When trained with thermographic data, the average normalized error is less than 2% over 400 s, during which temperatures of the assembly have dropped by more than 500 K. The rod temperature prediction performance was significantly better for rods in the interior of the assembly compared to those near the exterior, likely due to the model simplification of the surroundings.
The author has identified the following significant results. The distribution and abundance of white marlin correlated with the chlorophyll, water temperature, and Secchi depth sea truth measurements. Results of correlation analyses for dolphin were inconclusive. Predicition models for white marlin were developed using stepwise multiple regression and discriminant function analysis techniques which demonstrated a potential for increasing the probability of game fishing success. The S190A and B imagery was density sliced/color enhanced with white marlin location superimposed on the image, but no density/white marlin relationship could be established.
High-spectral resolution measurements (in the IR region) from space born infrared sensors can provide sufficient spectral information about cloud properties like cloud top pressure, cloud particle size and cloud optical thickness. Empirical orthogonal function (EOF) based regression scheme was used to retrieve cloud properties along with atmospheric state profiles. A channel based physical inversion scheme was also tried on cloud-contaminated radiances to improve retrieval accuracy achieved by regression scheme. Recently developed principle component (PC) based physical retrieval algorithm utilized the EOF transformation to compress both hyperspectral radiance spectra and atmospherical profiles into PC scores to greatly boost the computational speed while maintaining the information content. The accuracy of the (PC) based radiation transfer model (PCRTM) to simulate top-of the atmosphere(TOA) high-spectral resolution radiance spectra for both clear and cloudy sky conditions has been well tested by comparing with the other line-by-line radiation transfer model and using the observation data. The application of the PCRTM based physical inversion method on simulated IASI radiance under 'cloudy' conditions will be a good integrity test for the new algorithm.
With increasing penetrations of wind power on electric grids, the stability and reliability of interconnected power systems may be impacted. In some countries that have developed renewable energy sources and systems, grid codes have been revised to require wind power plants (WPPs) to provide ancillary services to support the power system frequency in case of severe grid events. To do this, wind turbine generators (WTGs) should be deloaded to reserve a certain amount of active power for primary frequency response; however, deloading curtails annual energy production, and the market for this type of service needs to be further developed. In this report, we focus on the temporary frequency support provided by WTGs through inertial response. WTGs have potential to provide inertial response, but appropriate control methods should be implemented. With the implemented inertial control methods, wind turbines are capable of increasing their active power output by releasing some of their stored kinetic energy when a frequency excursion occurs. Active power can be temporarily boosted above the maximum power points, after which the rotor speed decelerates, and subsequently an active power output reduction restores the kinetic energy. In this report, we develop two types of models for wind power systems: the first is common, based on the wind power aerodynamic equation, and the power coefficient can be regressed using nonlinear functions; the second is much more complicated, wherein the wind turbine system is modeled using the Fatigue, Aerodynamics, Structures, and Turbulence Modeling (FAST) tool with several degrees of freedoms. A nine-bus test power system is built in Simulink and the Real-Time Digital Simulator, respectively, which are used to evaluate the frequency support performance of the WPPs. We implement two distinct types of inertial control methods in the modeled wind turbines: frequency-based inertial control (FBIC) and stepwise inertial control (SIC). We compare the performances of the two methods in terms of their frequency nadirs, rates of change of frequency, and recovery times. We conclude the results under various wind speeds and penetration cases, which provide insight into designing the inertial response of WTGs. Further, we discuss the impact of the parameters on the performance of the inertial control methods. We evaluate both the scaling factors for the FBIC method and the slope values for the TLIC methods. The simulation work shows the characteristics of different inertial responses compared to conventional synchronous generators. Based on the simulation results, we modify, improve, and test the inertial control methods under a more realistic wind turbine model based on FAST. We then validate the inertial responses under highly turbulent wind conditions generated by TurbSim, and we examine their influences on the turbine mechanical components. The extensive simulation proves the effectiveness of the proposed inertial control methods as well as the nine-bus test power system. We then reconsider the parameters. We rebuild the same test power system using Real time Simulator Computer Aided Design (RSCAD), and we implement the inertial control methods in the real Controls Advanced Research Turbine (CART3), which is prepared for the hardware-in-the-loop field-test simulation. After the setups for the hardware and software hybrid simulation platform are complete, the inertial response is further tested on a real wind turbine for the first time, in which CART3 release the controlled inertial response against the emulated frequency excursion, provided by the real-time simulated power system test bed in RTDS.
The integration of Radial Basis Function Networks and Back Propagation Neural Networks with the Multiple Linear Regression has been accomplished to map nonlinear response surfaces over a wide range of independent variables in the process of the Modem Design of Experiments. The integrated method is capable to estimate the precision intervals including confidence and predicted intervals. The power of the innovative method has been demonstrated by applying to a set of wind tunnel test data in construction of response surface and estimation of precision interval.
The study of climate change and its impacts depends on generating projections of future temperature and other climate variables. For detailed studies, these projections usually require some combination of numerical simulation and observations, given that simulations of even the current climate do not perfectly reproduce local conditions. We present a methodology for generating future climate projections that takes advantage of the emergence of climate model ensembles, whose large amounts of data allow for detailed modeling of the probability distribution of temperature or other climate variables. The procedure gives us estimated changes in model distributions that are then applied to observations to yield projections that preserve the spatiotemporal dependence in the observations. We use quantile regression to estimate a discrete set of quantiles of daily temperature as a function of seasonality and long-term change, with smooth spline functions of season, long-term trends, and their interactions used as basis functions for the quantile regression. A particular innovation is that more extreme quantiles are modeled as exceedances above less extreme quantiles in a nested fashion, so that the complexity of the model for exceedances decreases the further out into the tail of the distribution one goes. We apply this method to two large ensembles of model runs using the same forcing scenario, both based on versions of the Community Earth System Model (CESM), run at different resolutions. The approach generates observation-based future simulations with no processing or modeling of the observed climate needed other than a simple linear rescaling. The resulting quantile maps illuminate substantial differences between the climate model ensembles, including differences in warming in the Pacific Northwest that are particularly large in the lower quantiles during winter. We show how the availability of two ensembles allows the efficacy of the method to be tested with a “perfect model” approach, in which we estimate transformations using the lower-resolution ensemble and then apply the estimated transformations to single runs from the high-resolution ensemble. Finally, we describe and implement a simple method for adjusting a transformation estimated from a large ensemble of one climate model using only a single run of a second, but hopefully more realistic, climate model.
Changes in urine chemistry, during and post flight, potentially increases the risk of renal stones in astronauts. Although much is known about the effects of space flight on urine chemistry, no inflight incidence of renal stones in US astronauts exists and the question "How much does this risk change with space flight?" remains difficult to accurately quantify. In this discussion, we tackle this question utilizing a combination of deterministic and probabilistic modeling that implements the physics behind free stone growth and agglomeration, speciation of urine chemistry and published observations of population renal stone incidences to estimate changes in the rate of renal stone presentation. The modeling process utilizes a Population Balance Equation based model developed in the companion IWS abstract by Kassemi et al. (2016) to evaluate the maximum growth and agglomeration potential from a specified set of urine chemistry values. Changes in renal stone occurrence rates are obtained from this model in a probabilistic simulation that interrogates the range of possible urine chemistries using Monte Carlo techniques. Subsequently, each randomly sampled urine chemistry undergoes speciation analysis using the well-established Joint Expert Speciation System (JESS) code to calculate critical values, such as ionic strength and relative supersaturation. The Kassemi model utilizes this information to predict the mean and maximum stone size. We close the assessment loop by using a transfer function that estimates the rate of stone formation from combining the relative supersaturation and both the mean and maximum free stone growth sizes. The transfer function is established by a simulation analysis which combines population stone formation rates and Poisson regression. Training this transfer function requires using the output of the aforementioned assessment steps with inputs from known non-stone-former and known stone-former urine chemistries. Established in a Monte Carlo system, the entire renal stone analysis model produces a probability distribution of the stone formation rate and an expected uncertainty in the estimate. The utility of this analysis will be demonstrated by showing the change in renal stone occurrence predicted by this method using urine chemistry distributions published in Whitson et al. 2009. A comparison to the model predictions to previous assessments of renal stone risk will be used to illustrate initial validation of the model.
This study discusses a rainfall algorithm utilizing six channels of microwave radiance data from the Nimbus-7 Scanning Multifrequency Microwave Radiometer. The algorithm is intended for short-term climate studies over the ocean at low latitudes. To find a set of functional relationships, rain rates are regressed on brightness temperatures for each channel. Next, these functions are integrated over a class of rain-rate distributions to find relations between mean brightness temperatures and mean rain rates. This step accounts for beam filling. Finally, weights are obtained for combining the rain rates from the individual channels. The weights vary with the rain rates, so that the optimum combination of channels is always used. Results are stored in a database grid 1 deg latitude x 1 deg longitude by one month. To test the algorithm, three years (1979-81) of data from the Indian Ocean are processed.
This report summarizes some of the ongoing work related to the development of an expert-elicitation-digital-twin framework for real time damage state prediction in heat exchanger components of a nuclear reactor. The framework is targeted towards predicting damage associated with coupled low cycle fatigue (associated with regular heat-up, cool-down and power operation transients) and high cycle fatigue (associated with flow induced vibration transients). The overall framework will be based on a NoSQL based database, physics-infused-geometry-dependent virtual-sensor data, different AI/ML techniques-based data-driven-predictive-model applications (Apps) and real-time plant sensor measurements available through few existing sensors. Towards this overall goal, this report updates some of the ongoing work, such as on implementation of a NoSQL Database (such as MongoDB), FE based heat transfer analysis of a heat exchanger (e.g. of a PWR steam generator) for generating geometry-dependent virtual sensor data and evaluation of various AI/ML models such as based on multivariate linear regression, ensembled decision-tree based Random-Forest and Gradient-Boosting regression and high-dimensional-kernel-function-transformation based Support-Vector-Machine regression models. The AI/ML models were evaluated for predicting multi-time-series thermal states at thousands of 3D point-clouds
A new diagnostic tool is developed for examining relationships between the synoptic scale circulation and regional temperature distributions in GCMs. The 4 x 5 deg GISS GCM is shown to produce accurate simulations of the variance in the synoptic scale sea level pressure distribution over the U.S. An analysis of the observational data set from the National Meteorological Center (NMC) also shows a strong relationship between the synoptic circulation and grid point temperatures. This relationship is demonstrated by deriving transfer functions between a time-series of circulation parameters and temperatures at individual grid points. The circulation parameters are derived using rotated principal components analysis, and the temperature transfer functions are based on multivariate polynomial regression models. The application of these transfer functions to the GCM circulation indicates that there is considerable spatial bias present in the GCM temperature distributions. The transfer functions are also used to indicate the possible changes in U.S. regional temperatures that could result from differences in synoptic scale circulation between a 1XCO2 and a 2xCO2 climate, using a doubled CO2 version of the same GISS GCM.
The static response of sea level to the forcing of atmospheric pressure, the so-called inverted barometer (IB) effect, is investigated using TOPEX/POSEIDON data. This response, characterized by the rise and fall of sea level to compensate for the change of atmospheric pressure at a rate of -1 cm/mbar, is not associated with any ocean currents and hence is normally treated as an error to be removed from sea level observation. Linear regression and spectral transfer function analyses are applied to sea level and pressure to examine the validity of the IB effect. In regions outside the tropics, the regression coefficient is found to be consistently close to the theoretical value except for the regions of western boundary currents, where the mesoscale variability interferes with the IB effect. The spectral transfer function shows near IB response at periods of 30 degrees is -0.84 +/- 0.29 cm/mbar (1 standard deviation). The deviation from = 1 cm /mbar is shown to be caused primarily by the effect of wind forcing on sea level, based on multivariate linear regression model involving both pressure and wind forcing. The regression coefficient for pressure resulting from the multivariate analysis is -0.96 +/- 0.32 cm/mbar. In the tropics the multivariate analysis fails because sea level in the tropics is primarily responding to remote wind forcing. However, after removing from the data the wind-forced sea level estimated by a dynamic model of the tropical Pacific, the pressure regression coefficient improves from -1.22 +/- 0.69 cm/mbar to -0.99 +/- 0.46 cm/mbar, clearly revealing an IB response. The result of the study suggests that with a proper removal of the effect of wind forcing the IB effect is valid in most of the open ocean at periods longer than 20 days and spatial scales larger than 500 km.
Advanced manufacturing technologies are enabling the next generation of superalloys for extreme temperatures, pressures, and longevity. There is a need for Accelerated Creep Test (ACT) methods to reduce the time needed to implement new creep resistant superalloys. Without the development of ACTs, the qualification of new superalloys using conventional creep tests can take 11+ years corresponding to the 100,000-hour service life of long-lived turbomachinery components. A refined stress relaxation test (SRT) is introduced to succeed real-time conventional creep tests (CCTs). The stress relaxation test generates extensive stress versus creep-strain-rate data within a single specimen. Since the specimen is held below the elastic limit and relatively little creep deformation has occurred, several SRTs at different temperatures can be performed using a single specimen. To obtain data, specimens are rapidly loaded in force-control to just below the elastic limit. The specimen is switched to displacement-control and held at a fixed displacement while stress relaxation occurs. This generates a stress versus time curve that must be converted into stress versus creep-strain-rate. The stress versus creep-strain-rate curves are utilized to calculate creep activation energy and produce a creep-strain-rate master curve. The objective of this study is to determine the best calibration approach for SRT: regression of stress, regression of creep strain, and the finite difference of creep strain. It is recommended that the differentiation be applied; however, finite difference applied to the raw data produces inconsistent calculations of the creep-strain-rate. In this study, a MATLAB algorithm is written to evaluate all three calibration options where, stress and creep deformation are regressed into natural logarithm functions. The resulting creep-strain-rate curves are compared to determine the most reliable and consistent method. Analyzing and evaluating the results the regression of stress and regression or creep strain calibration methods prove to provide accurate results. These methods will help replace CCTs by conducting less expensive ACTs, while increasing the overall efficiency of the development of new materials.
It is shown that the least squares collocation approach to estimating geodetic parameters is identical to conventional minimum variance estimation. Hence, the least squares collocation estimator can be derived either by minimizing the usual least squares quadratic loss function or by computing a conditional expectation by means of the regression equation. When a deterministic functional relationship between the data and the parameters to be estimated is available, one can implement a least squares solution using the functional relation to obtain an equation of condition. It is proved the solution so obtained is identical to what is obtained through least squares collocation. The implications of this equivalance for the estimation of mean gravity anomalies are discussed.
Many biological processes display oscillatory behaviour based on an approximately 24 h internal timing system specific to each individual. One process of particular interest is gene expression, for which several circadian transcriptomic studies have identified associations between gene expression during a 24 h period and an individual's health. A challenge with analysing data from these studies is that each individual's internal timing system is offset relative to the 24 h day-night cycle, where day–night cycle time is recorded for each collected sample. Laboratory procedures can accurately determine each individual's offset and determine the internal time of sample collection. However, these laboratory procedures are labour-intensive and expensive. Here, in this paper, we propose a corrected score function framework to obtain a regression model of gene expression given internal time when the offset of each individual is too burdensome to determine. A feature of this framework is that it does not require the probability distribution generating offsets to be symmetric with a mean of zero. Simulation studies validate the use of this corrected score function framework for cosinor regression, which is prevalent in circadian transcriptomic studies. Illustrations with data from three circadian transcriptomic studies further demonstrate that the proposed framework consistently mitigates bias relative to using a score function that does not account for this offset.
Here, we explore the potential of a data-driven approach to the design of exchange-correlation (xc) functionals. The approach, inspired by convolutional filters in computer vision and surrogate functions from optimization, utilizes convolutions of the electron density to form a feature space to represent local electronic environments and neural networks to map the features to the exchange-correlation energy density. These features are orbital free, and provide a systematic route to including information at various length scales. This work shows that convolutional descriptors are theoretically capable of an exact representation of the electron density, and proposes Maxwell-Cartesian spherical harmonic kernels as a class of rotationally invariant descriptors for the construction of machine learned functionals. The approach is demonstrated using data from the B3LYP functional on a number of small molecules containing C, H, O, and N along with a neural network regression model. The machine learned functionals are compared to standard physical approximations and the accuracy is assessed for the absolute energy of each molecular system as well as formation energies. The results indicate that it is possible to reproduce the exchange-correlation portion of B3LYP formation energies to within chemical accuracy using orbital-free descriptors with a spatial extent of 0.2 Å. The findings provide empirical insight into the spatial range of electron exchange, and suggest that the combination of convolutional descriptors and machine learning regression models is a promising framework for xc functional design, although challenges remain in obtaining training data and generating models consistent with pseudopotentials.
This study extends the accurate and transferable molecular-orbital-based machine learning (MOB-ML) approach to modeling the contribution of electron correlation to dipole moments at the cost of Hartree–Fock computations. A MOB pairwise decomposition of the correlation part of the dipole moment is applied, and these pair dipole moments could be further regressed as a universal function of MOs. The dipole MOB features consist of the energy MOB features and their responses to electric fields. An interpretable and rotationally equivariant derivative kernel for Gaussian process regression (GPR) is introduced to learn the dipole moment more efficiently. The proposed problem setup, feature design, and ML algorithm are shown to provide highly accurate models for both dipole moments and energies on water and 14 small molecules. To demonstrate the ability of MOB-ML to function as generalized density-matrix functionals for molecular dipole moments and energies of organic molecules, we further apply the proposed MOB-ML approach to train and test the molecules from the QM9 dataset. The application of local scalable GPR with Gaussian mixture model unsupervised clustering GPR scales up MOB-ML to a large-data regime while retaining the prediction accuracy. In addition, compared with the literature results, MOB-ML provides the best test mean absolute errors of 4.21 mD and 0.045 kcal/mol for dipole moment and energy models, respectively, when training on 110 000 QM9 molecules. The excellent transferability of the resulting QM9 models is also illustrated by the accurate predictions for four different series of peptides.