Search NASA⌕ Search

SEARCH · Search NASA

Results for “multiple regression”

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 757 records · Page 42

Reinterpreting fundamental plane correlations with machine learning

ABSTRACT This work explores the relationships between galaxy sizes and related observable galaxy properties in a large volume cosmological hydrodynamical simulation. The objectives of this work are to develop a better understanding of both the correlations between galaxy properties and the influence of environment on galaxy physics in order to build an improved model for the galaxy sizes, building off of the fundamental plane. With an accurate intrinsic galaxy size predictor, the residuals in the observed galaxy sizes can potentially be used for multiple cosmological applications, including making measurements of galaxy velocities in spectroscopic samples, estimating the rate of cosmic expansion, and constraining the uncertainties in the photometric redshifts of galaxies. Using projection pursuit regression, the model accurately predicts intrinsic galaxy sizes and have residuals which have limited correlation with galaxy properties. The model decreases the spatial correlation of galaxy size residuals by a factor of ∼5 at small scales compared to the baseline correlation when the mean size is used as a predictor.

Astronomy & Astrophysics↗

Remote sensing of biomass and annual net aerial primary productivity of a salt marsh

Net aerial primary productivity is the rate of storage of organic matter in above-ground plant issues exceeding the respiratory use by the plants during the period of measurement. It is pointed out that this plant tissue represents the fixed carbon available for transfer to and consumption by the heterotrophic organisms in a salt marsh or the estuary. One method of estimating annual net aerial primary productivity (NAPP) required multiple harvesting of the marsh vegetation. A rapid nondestructive remote sensing technique for estimating biomass and NAPP would, therefore, be a significant asset. The present investigation was designed to employ simple regression models, equating spectral radiance indices with Spartina alterniflora biomass to nondestructively estimate salt marsh biomass. The results of the study showed that the considered approach can be successfully used to estimate salt marsh biomass.

Hardisky, M. A.↗

Development and testing of 11- and 24-inch hybrid motors under the joint government/industry IR&D program

Establishment of a test facility and associated 11-in.-diameter motor for hybrid propulsion technology development at NASA's George C. Marshall Space Flight Center is discussed in this paper. Results of twenty 11-in.-diameter motor tests with a UTF-29901 (60 percent polycyclopentadiene, 40 percent hydroxyl-terminated polybutadiene)/gaseous oxygen propellant system are presented. Tests at this scale have developed fuel regression correlations for comparison with results of yet-to-be-completed, 24-in.-diameter motor tests; demonstrated combustion efficiency levels in the 95 percent range for both single- and multiple-port grain configurations; have shown smooth and stable throttling characteristics over flight-type throttle ranges; and have begun to establish criteria for stable combustion in hybrid motors. The testing of 24-in. motors has not as yet been initiated and is not addressed.

Boardman, T. A.↗

Estimating Uncertainty in Long Term Total Ozone Records from Multiple Sources

Total ozone measurements derived from the TOMS and SBUV backscattered solar UV instrument series cover the period from late 1978 to the present. As the SBUV series of instruments comes to an end, we look to the 10 years of data from the AURA Ozone Monitoring Instrument (OMI) and two years of data from the Ozone Mapping Profiler Suite (OMPS) on board the Suomi National Polar-orbiting Partnership satellite to continue the record. When combining these records to construct a single long-term data set for analysis we must estimate the uncertainty in the record resulting from potential biases and drifts in the individual measurement records. In this study we present a Monte Carlo analysis used to estimate uncertainties in the Merged Ozone Dataset (MOD), constructed from the Version 8.6 SBUV2 series of instruments. We extend this analysis to incorporate OMI and OMPS total ozone data into the record and investigate the impact of multiple overlapping measurements on the estimated error. We also present an updated column ozone trend analysis and compare the size of statistical error (error from variability not explained by our linear regression model) to that from instrument uncertainty.

TOMS and SBUV↗

Online evolutionary neural architecture search for multivariate non-stationary time series forecasting

Time series forecasting (TSF) is one of the most important tasks in data science. TSF models are usually pre-trained with historical data and then applied on future unseen datapoints. However, real-world time series data is usually non-stationary and models trained offline usually face problems from data drift. Models trained and designed in an offline fashion can not quickly adapt to changes quickly or be deployed in real-time. To address these issues, this work presents the Online NeuroEvolution-based Neural Architecture Search (ONE-NAS) algorithm, which is a novel neural architecture search method capable of automatically designing and dynamically training recurrent neural networks (RNNs) for online forecasting tasks. Without any pre-training, ONE-NAS utilizes populations of RNNs that are continuously updated with new network structures and weights in response to new multivariate input data. ONE-NAS is tested on real-world, large-scale multivariate wind turbine data as well as the univariate Dow Jones Industrial Average (DJIA) dataset. These results demonstrate that ONE-NAS outperforms traditional statistical time series forecasting methods, including online linear regression, fixed long short-term memory (LSTM) and gated recurrent unit (GRU) models trained online, as well as state-of-the-art, online ARIMA strategies. Additionally, results show that utilizing multiple populations of RNNs which are periodically repopulated provide significant performance improvements, allowing this online neural network architecture design and training to be successful.

97 MATHEMATICS AND COMPUTING↗

Modeling Heavy/Medium-Duty Fuel Consumption Based on Drive Cycle Properties

This paper presents multiple methods for predicting heavy/medium-duty vehicle fuel consumption based on driving cycle information. A polynomial model, a black box artificial neural net model, a polynomial neural network model, and a multivariate adaptive regression splines (MARS) model were developed and verified using data collected from chassis testing performed on a parcel delivery diesel truck operating over the Heavy Heavy-Duty Diesel Truck (HHDDT), City Suburban Heavy Vehicle Cycle (CSHVC), New York Composite Cycle (NYCC), and hydraulic hybrid vehicle (HHV) drive cycles. Each model was trained using one of four drive cycles as a training cycle and the other three as testing cycles. By comparing the training and testing results, a representative training cycle was chosen and used to further tune each method. HHDDT as the training cycle gave the best predictive results, because HHDDT contains a variety of drive characteristics, such as high speed, acceleration, idling, and deceleration. Among the four model approaches, MARS gave the best predictive performance, with an average absolute percent error of -1.84% over the four chassis dynamometer drive cycles. To further evaluate the accuracy of the predictive models, the approaches were first applied to real-world data. MARS outperformed the other three approaches, providing an average absolute percent error of -2.2% of four real-world road segments. The MARS model performance was then compared to HHDDT, CSHVC, NYCC, and HHV drive cycles with the performance from Future Automotive System Technology Simulator (FASTSim). The results indicated that the MARS method achieved a comparative predictive performance with FASTSim.

33 ADVANCED PROPULSION SYSTEMS↗

On the method of Ermakov and Zolotukhin for multiple integration

The method of Ermakov and Zolotukhin is discussed along with its later developments. By introducing the idea of pseudo-implementation a practical assessment of the method is made. The performance of the method is found to be unimpressive in comparison with a recent regression method.

Cranley, R.↗

SRBench++: Principled Benchmarking of Symbolic Regression With Domain-Expert Interpretation

Symbolic regression searches for analytic expressions that accurately describe studied phenomena. The main promise of this approach is that it may return an interpretable model that can be insightful to users, while maintaining high accuracy. The current standard for benchmarking these algorithms is SRBench, which evaluates methods on hundreds of datasets that are a mix of real-world and simulated processes spanning multiple domains. At present, the ability of SRBench to evaluate interpretability is limited to measuring the size of expressions on real-world data, and the exactness of model forms on synthetic data. In practice, model size is only one of many factors used by subject experts to determine how interpretable a model truly is. Furthermore, SRBench does not characterize algorithm performance on specific, challenging sub-tasks of regression such as feature selection and evasion of local minima. In this work, we propose and evaluate an approach to benchmarking SR algorithms that addresses these limitations of SRBench by 1) incorporating expert evaluations of interpretability on a domain-specific task, and 2) evaluating algorithms over distinct properties of data science tasks. We evaluate 12 modern symbolic regression algorithms on these benchmarks and present an in-depth analysis of the results, discuss current challenges of symbolic regression algorithms and highlight possible improvements for the benchmark itself.

97 MATHEMATICS AND COMPUTING↗

Modern sensitivity analysis - 15651

Sensitivity analysis has been applied to complex models, such as performance assessment models, since at least the advent of computational tools to support such an approach to model evaluation. Initially, sensitivity analysis methods were applied to deterministic models and focused on changing one input variable at a time. Probabilistic modeling opened the door to evaluating changes in all input variables simultaneously. Standardized rank regression methods were primarily used in the last century to evaluate probabilistic models. These methods assume linearity in the rank response of the model to the inputs. At the very least this assumes monotonicity in the response, but for the most part the response needs to be near-linear. This is highly unlikely for performance assessment models that are based on differential equations, multiplication and division of variables that are usually not correlated, and conservatism that is not well understood. There are better statistical methods for evaluating highly non-linear and dynamic probabilistic models. Some of these methods were founded in the data mining industry. This century has seen a rapid development of the internet that led directly to the development of data mining techniques for massive databases. Although the goals were different, these techniques can be adapted to sensitivity analysis for complex probabilistic models, which, also, form large, or even massive, databases. These modern sensitivity analysis procedures assess the importance of every explanatory variable (input parameters) used in a probabilistic model. In practice, this means that for each endpoint of interest, every explanatory variable in the model is associated with a quantitative measure of importance. For a given explanatory variable, the importance value depends on the endpoint of interest. This global approach to sensitivity analysis allows all levels of interactions to be evaluated, so that the variation in the output can be measured at every level of possible interaction. The effects are collected together during sensitivity analysis processing, and then separated out to attribute the overall contribution of each input parameter to the model output. These global sensitivity analysis procedures have the ability to characterize non-linear and non-monotonic relationships between explanatory variables and endpoints of interest. This is critically important because of the need to characterize complex interactions among multiple explanatory variables in PA models that often have non-linear and non-monotonic effects. The options for sensitivity analysis are fully explored, including an evaluation of several methods when applying them to a well-known complex non-linear non-monotonic model. The results clearly demonstrate the relative benefit of using data mining methods over simple statistical approaches such as standardized rank regression. (authors)

97 MATHEMATICS AND COMPUTING↗

Comparison of CT methods for determining graft steatosis in living donor liver transplantation

Purpose: To evaluate and compare the diagnostic performance of non-enhanced computed tomography (NECT) and contrast-enhanced CT (CECT) attenuation indices in the assessment of hepatic steatosis by using biopsy as the reference standard. Materials and methods: This retrospective study was approved by our Institutional Review Board. 55 Potential donors who underwent both NECT and triphasic CECT and core liver biopsy, were included the study. Average attenuation measurements that were obtained from multiple regions in liver, spleen, and psoas muscle on both unenhanced and CECT were used for analysis. Hepatic attenuation measurements were analyzed with and without normalization with the spleen and psoas muscle. Linear regression and receiver operating characteristic (ROC) curve analysis were used to evaluate the statistical association between CT indices and steatosis at histology. Results: Linear regression analysis confirmed the strongest correlation between steatosis and normalized measurements of hepatic attenuation with splenic attenuations on hepatic venous phase of CECT scan (R 0.821; R{sup 2} 0.674 and R 0.816; R{sup 2} 0.665, respectively). The use of ROC curve analysis also demonstrated that normalized measurements of hepatic attenuation with splenic attenuations on hepatic venous phase of CECT showed high diagnostic performance regarding the qualitative distinction of steatosis (AUC values greater than 0.9). Conclusion: Attenuation measurements of liver normalized with spleen on hepatic venous phase may be useful in evaluating steatosis in donor candidates with moderate to severe steatosis who are unacceptable for liver donation. In this manner unnecessary liver biopsy may be avoided in those donor candidates.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Mapping Multivariate Influence of Alloying Elements on Creep Behavior for Design of New Martensitic Steels

Here, heritage data for the class of 9 to 12 wt pct Cr steels are studied using data science to quantify the statistically significant relationships among multiple processing/microstructure and performance variables. The effort is undertaken to find new martensitic steels for creep life of 10 5 hours or greater at 650 °C and 100 MPa using machine learning. Linear regression and lasso regression were utilized to identify alloying elements that contribute towards better creep strength. Visualization techniques such as t-distributed stochastic neighbor embedding and pair-wire element specific comparisons were utilized to explore information gaps that exist within the data and are in conflict with existing domain knowledge. Lastly, cfombining all results suggest that the next alloy design to be explored should be 9 wt pct Cr with high W (2 to 3 wt pct) and high Co (2 to 3 wt pct) for creep life of 10 5 hours or greater at 650 °C, 100 MPa.

36 MATERIALS SCIENCE↗

Lethality is Local, but Survival is Systemic: Temporal and Multi-Organ Responses to Chlorine Gas Exposure in a Murine Model

Chlorine gas (Cl2) is a highly toxic chemical associated with both localized lung injury and systemic health effects. While pulmonary damage has been well characterized, the systemic inflammatory and metabolic responses remain poorly understood. We aimed to define the temporal and multi-organ responses to Cl2 exposure in a murine model, with a focus on identifying spatiotemporal inflammation and its impact on survival and lethality. SKH1 mice were exposed for 10 min to varying concentrations of Cl2 (94.4–810 ppm, representative of non-lethal, LD10, and LD50 doses) and monitored for respiratory function, perfusion, and acidosis using organ-specific imaging. At multiple time points (40 min, 6 h, 24 h, and 7 d), we measured phosphoproteins, cytokines, chemokines, growth factors, and metabolic hormones in the lungs, heart, cortex, and plasma. Statistical modeling and logistic regression were used to identify biomarkers associated with lethality and survival. We found that lung injury was the primary cause of potential lethality, particularly via early phosphoprotein signaling disruptions. However, survival correlated with early systemic coordination of inflammatory and metabolic signals across organs. Perfusion and acidosis imaging were strongly associated with chemokine and hormone responses. Key survival-associated plasma biomarkers included decreased insulin, increased ghrelin, and decreased eotaxin. While potential lethality from Cl2 exposure is locally driven by pulmonary injury, survival depends on systemic, multi-organ responses that occur rapidly post-exposure. Within this model, our findings identify a potential therapeutic window to enhance survival and suggest candidate biomarkers that may be explored translationally for both triage and treatment of chlorine-related incidents.

chlorine gas↗

Operator learning for predicting multiscale bubble growth dynamics

We report simulating and predicting multiscale problems that couple multiple physics and dynamics across many orders of spatiotemporal scales is a great challenge that has not been investigated systematically by deep neural networks (DNNs). Herein, we develop a framework based on operator regression, the so-called deep operator network (DeepONet), with the long-term objective to simplify multiscale modeling by avoiding the fragile and time-consuming “hand-shaking” interface algorithms for stitching together heterogeneous descriptions of multiscale phenomena. To this end, as a first step, we investigate if a DeepONet can learn the dynamics of different scale regimes, one at the deterministic macroscale and the other at the stochastic microscale regime with inherent thermal fluctuations. Specifically, we test the effectiveness and accuracy of the DeepONet in predicting multirate bubble growth dynamics, which is described by a Rayleigh–Plesset (R–P) equation at the macroscale and modeled as a stochastic nucleation and cavitation process at the microscale by dissipative particle dynamics (DPD). First, we generate data using the R–P equation for multirate bubble growth dynamics caused by randomly time-varying liquid pressures drawn from Gaussian random fields (GRFs). Our results show that properly trained DeepONets can accurately predict the macroscale bubble growth dynamics and can outperform long short-term memory networks. We also demonstrate that the DeepONet can extrapolate accurately outside the input distribution using only very few new measurements. Subsequently, we train the DeepONet with DPD data corresponding to stochastic bubble growth dynamics. Although the DPD data are noisy and we only collect sparse data points on the trajectories, the trained DeepONet model is able to predict accurately the mean bubble dynamics for time-varying GRF pressures. Taken together, our findings demonstrate that DeepONets can be employed to unify the macroscale and microscale models of the multirate bubble growth problem, hence providing new insight into the role of operator regression via DNNs in tackling realistic multiscale problems and in simplifying modeling with heterogeneous descriptions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Decayheatml

This code is designed to predict and analyze the decay heat generated in molten salt reactors (MSRs) using a hybrid approach that combines machine learning and segmented polynomial fitting. The accurate prediction of decay heat is essential for reactor safety and the optimization of spent fuel storage. The code operates through several key components: 1) Data Architecture: It incorporates a modular data architecture that handles various MSR-specific operational parameters such as power density, humidity content, and air ingress. These parameters are sampled using Sobol sequences to ensure comprehensive coverage of operational uncertainties. 2) Machine Learning Framework: The code employs a diverse set of machine learning models, including polynomial regression, decision trees, random forests, gradient boosting, support vector regression, k-nearest neighbors, multi-layer perceptrons, and symbolic regression. These models are trained to predict decay heat over a wide temporal range, from immediate shutdown up to 10,000 years. 3) Region-Optimized Training: The temporal domain is divided into multiple regions, each modeled separately to capture distinct decay heat characteristics across different time scales. This approach significantly improves the accuracy and interpretability of predictions. 4) Segmented Polynomial Interpretation (SPI): The SPI method translates machine learning predictions into piecewise polynomial equations. These equations are physically interpretable and can be directly integrated into existing engineering workflows and safety analyses. 5) Front-End Interfaces: The code includes both a Jupyter notebook interface for research development and a Streamlit web application for operational deployment. These interfaces allow users to interactively explore decay heat predictions, adjust operational parameters, and visualize results in real-time. 6) Applications: The framework supports various applications, including safety system validation and spent fuel container optimization. It enables real-time evaluation of worst-case decay heat scenarios, informing the design of passive safety systems and optimizing container designs for long-term storage. Overall, this code provides a robust, accurate, and user-friendly tool for predicting decay heat in MSRs, enhancing reactor safety, and optimizing spent fuel management.

Retamales, Mauricio Eduardo Tano [Idaho National L↗

Deep machine learning-assisted multiphoton microscopy to reduce light exposure and expedite imaging

Two-photon excitation fluorescence (2PEF) allows imaging of tissue up to about one millimeter in thickness. Typically, reducing fluorescence excitation exposure reduces the quality of the image. However, using deep learning super resolution techniques, these low-resolution images can be converted to high-resolution images. This work explores improving human tissue imaging by applying deep learning to maximize image quality while reducing fluorescence excitation exposure. We analyze two methods: a method based on U-Net, and a patch-based regression method. Both methods are evaluated on a skin dataset and an eye dataset. The eye dataset includes 1200 paired high power and low power images of retinal organoids. The skin dataset contains multiple frames of each sample of human skin. High-resolution images were formed by averaging 70 frames for each sample and low-resolution images were formed by averaging the first 7 and 15 frames for each sample. The skin dataset includes 550 images for each of the resolution levels. We track two measures of performance for the two methods: mean squared error (MSE) and structural similarity index measure (SSIM). For the eye dataset, the patches method achieves an average MSE of 27,611 compared to 146,855 for the U-Net method, and an average SSIM of 0.636 compared to 0.607 for the U-Net method. For the skin dataset, the patches method achieves an average MSE of 3.768 compared to 4.032 for the U-Net method, and an average SSIM of 0.824 compared to 0.783 for the U-Net method. Despite better performance on image quality, the patches method is worse than the U-Net method when comparing the speed of prediction, taking 303 seconds to predict one image compared to less than one second for the U-Net method.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Diagnosis of Corrosion Process in Nuclear Power Plant Secondary Piping Structures

Corrosion inside secondary pipe structures in Nuclear Power Plants is difficult to detect without intrusion. Statistical and machine learning techniques are used to predict corrosion using pipe vibration data. 3D accelerometers were used to record pipe vibration data in three directions using six sensors. Vibration data were processed using the Hilbert-Huang Transformation, which is known for its adaptable basis function to non-linear and non-stationary signals. Hilbert-Huang Transformation using empirical mode functions generates multiple intrinsic mode functions of different amplitude and frequency levels from raw vibration data. From intrinsic mode functions features are extracted which include phase, energy, entropy, mean, standard deviation, skewness, and kurtosis. Logistic regression using only phase information correctly distinguished mass additions from baseline data in roughly 83% and 85% of the cases. Phase information varied substantially among different baseline runs and within trials of the same run. Random Forest resulted in >95% testing accuracy for both mass addition and mass removal data. Support Vector Machine had ~98% testing accuracy for mass removal and ~92% testing accuracy for mass additions when using Permutation Importance feature selection. When using Principle Component Analysis with Support Vector Machine, testing accuracy was ~68% testing accuracy for mass additions and ~85% testing accuracy for mass.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Remote sensing of changes in morphology and physiology of trees under stress

Results of continuing studies of forest trees subjected to varying types of stress are reported. Both greenhouse and field studies are included. Greenhouse work with tree seedlings exposed to varying levels of NaCl and CaCl2 in the soil indicated that, in the initial stages, palisade cells shrink and the amount of air space in the leaf increases. As the severity of damage increases, the cells of the spongy mesophyll shrink and flatten, and the amount of air space in the leaf decreases. Statistical analysis of foliar reflectance and associated moisture content data led to a series of regression equations for predicting foliar moisture content from reflectance data. Equations were calculated for three species, yellow birch (Betula alleghaniensis Britton), sugar maple (Acer saccharum Marsh.) and white ash (Fraxinus americana L.) having multiple correlation coefficients of 0.98, 0.94 and 0.93 respectively. Interpretation of multispectral imagery of the Ann Arbor Forestry Test Site (NASA Site 190) provided evidence that infections of Fomes annosus can be detected in the early stages. Infections of two needle cast diseases were also detected in conifer plantations in the test site. A study of automatic interpretation of multispectral scanner imagery for tree species recognition provided encouraging results.

Olson, C. E., Jr.↗

Attribution of Recovery in Lower-Stratospheric Ozone

Multiple satellite and ground-based observations provide consistent evidence that the thickness of Earth's protective ozone layer has stopped declining since 1997, close to the time of peak stratospheric halogen loading. Regression analyses with Effective Equivalent Stratospheric Chlorine (EESC) in conjunction with further analyses using more sophisticated photochemical model calculations constrained by satellite data demonstrate that the cessation of ozone depletion between 18-25 km altitude is consistent with a leveling off of stratospheric abundances of chlorine and bromine, due to the Montreal Protocol and its amendments. However, ozone increases in the lowest part of the stratosphere, from the tropopause to 18 km, account for about half of the improvement in total column ozone during the past 9 years at northern hemisphere mid-latitudes. The increase in ozone for altitudes below 18 km is most likely driven by changes in transport, rather than driven by declining chlorine and bromine. Even with this evidence that the Montreal Protocol and its amendments are having the desired, positive effect on ozone above 18 km, total column ozone is recovering faster than expected due to the apparent transport driven changes at lower altitudes. Accurate prediction of future levels of stratospheric ozone will require comprehensive understanding of the factors that drive temporal changes at various altitudes, and partitioning of the recent transport-driven increases between natural variability and changes in atmospheric structure perhaps related to anthropogenic climate change.

Yang, Eun-Su↗