Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “machine”

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 1,621 records · Page 90

Prediction of cardiac death after adenosine myocardial perfusion SPECT based on machine learning

Background: We developed machine-learning (ML) models to estimate a patient’s risk of cardiac death based on adenosine myocardial perfusion SPECT (MPS) and associated clinical data, and compared their performance to baseline logistic regression (LR). We demonstrated an approach to visually convey the reasoning behind a patient’s risk to provide insight to clinicians beyond that of a “black box.” Methods: We trained multiple models using 122 potential clinical predictors (features) for 8321 patients, including 551 cases of subsequent cardiac death. Accuracy was measured by area under the ROC curve (AUC), computed within a cross-validation framework. We developed a method to display the model’s rationale to facilitate clinical interpretation. Results: The baseline LR (AUC = 0.76; 14 features) was outperformed by all other methods. A least absolute shrinkage and selection operator (LASSO) model (AUC = 0.77; p = .045; 6 features) required the fewest features. A support vector machine (SVM) model (AUC = 0.83; p < .0001; 49 features) provided the highest accuracy. Conclusions: LASSO outperformed LR in both accuracy and simplicity (number of features), with SVM yielding best AUC for prediction of cardiac death in patients undergoing MPS. Combined with presenting the reasoning behind the risk scores, our results suggest that ML can be more effective than LR for this application.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

A Comparative Study of the Efficacy of Local/Global and Parametric/Nonparametric Machine Learning Methods for Establishing Structure–Property Linkages in High-Contrast 3D Elastic Composites

Reduced-order structure–property (S-P) linkages play a pivotal role in the tailored design of materials for advanced engineering components. There is a critical need to distill these from the simulation datasets aggregated using sophisticated, computationally expensive, physics-based numerical tools (e.g., finite element methods). The recent emergence of materials data science approaches has opened new avenues for addressing this challenge. In this paper, we critically compare the relative merits of the application of four distinct machine learning approaches for their efficacy in extracting microstructure-property linkages from the finite element simulation data aggregated on high-contrast elastic composites with different microstructures. The machine learning approaches selected for the study have included different combinations of local/global and parametric/nonparametric approaches. Furthermore, the nonparametric approaches selected for this study are based on Gaussian Process (GP) models that allow for a formal treatment of uncertainty quantification in the predicted values. The predictive performances of these different approaches have been compared against each other using rigorous cross-validation error metrics. Furthermore, their sensitivity to both the dataset size and dimensionality has been investigated.

36 MATERIALS SCIENCE↗

Compressing unstructured mesh data from simulations using machine learning

The amount of data output from a computer simulation has grown to terabytes and petabytes as increasingly complex simulations are being run on massively parallel systems. As we approach exaflop computing in the next decade, it is expected that the I/O subsystem will not be able to write out these large volumes of data. In this paper, we explore the use of machine learning to compress the data before it is written out. Despite the computational constraints that limit us to using very simple learning algorithms, our results show that machine learning is a viable option for compressing unstructured data. Furthermore, we demonstrate that by simply using a better sampling algorithm to generate the training set, we can obtain more accurate results compared to random sampling, but at no extra cost. Further, by carefully selecting and incorporating points with high prediction error, we can improve reconstruction accuracy without sacrificing the compression rate.

97 MATHEMATICS AND COMPUTING↗

Reducing systematic bias in machine learning applications to J/ψ signal extraction in high-energy nuclear physics

Machine learning techniques are increasingly used in high-energy nuclear physics because they can exploit multivariate correlations more efficiently than conventional cut-based analyses. A central challenge is the construction of training samples that faithfully reproduce the detector response observed in data. Signal samples are usually derived from detector simulations; therefore, mismatches between simulation and data can degrade classifier performance and introduce systematic biases. This work presents two practical correction procedures, namely cumulative distribution function (CDF) mapping and a shift-and-scale transformation, to align simulated signal features with those measured in data. Their performance is demonstrated with $J$/$\psi$ yield measurements in $\sqrt{s_{nn}}$ = 200 GeV Ru+Ru and Zr+Zr collisions recorded by STAR. A set of self-consistency tests shows that these procedures substantially suppress the systematic bias associated with data-simulation discrepancies in machine-learning-based signal extraction.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

FPGA-Accelerated Machine Learning Inference as a Service for Particle Physics Computing

Large-scale particle physics experiments face challenging demands for high-throughput computing resources both now and in the future. New heterogeneous computing paradigms on dedicated hardware with increased parallelization, such as Field Programmable Gate Arrays (FPGAs), offer exciting solutions with large potential gains. The growing applications of machine learning algorithms in particle physics for simulation, reconstruction, and analysis are naturally deployed on such platforms. We demonstrate that the acceleration of machine learning inference as a web service represents a heterogeneous computing solution for particle physics experiments that potentially requires minimal modification to the current computing model. As examples, we retrain the ResNet-50 convolutional neural network to demonstrate state-of-the-art performance for top quark jet tagging at the LHC and apply a ResNet-50 model with transfer learning for neutrino event classification. Using Project Brainwave by Microsoft to accelerate the ResNet-50 image classification model, we achieve average inference times of 60 (10) ms with our experimental physics software framework using Brainwave as a cloud (edge or on-premises) service, representing an improvement by a factor of approximately 30 (175) in model inference latency over traditional CPU inference in current experimental hardware. A single FPGA service accessed by many CPUs achieves a throughput of 600–700 inferences per second using an image batch of one, comparable to large batch-size GPU throughput and significantly better than small batch-size GPU throughput. Deployed as an edge or cloud service for the particle physics computing model, coprocessor accelerators can have a higher duty cycle and are potentially much more cost-effective.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Modification of a Superficial X-Ray Therapy Machine for Rectal Contact Therapy

X-ray therapy of superficial rectal cancers using a hand-held 50 kV contact unit (Philips RT-50) in a technique first described by Papillon had reached a point of widening clinical acceptability when the manufacturer of this equipment discontinued its production. To pursue this endocavitary approach to rectal therapy, technical modifications have to be made to conventional superficial x-ray therapy machines. Advantages over the original Papillon method include remote viewing of the lesion through the proctoscopic cone and a lower radiation exposure for the operator. We have evaluated a Bucky Combination Therapy Unit under conditions in which the operating voltage (65 kV), target skin distance (23.6 cm), and added filtration (0.39 mm Al) were selected in order to match as closely as possible the beam penetration characteristics of the “standard” (Papillon) technique. With this equipment, the thermal characteristics of the tube anode and housing limit the amount of radiation that can be delivered before a “rest period” for the machine is needed. In practice, 3 minutes of irradiation at an exposure rate of 500 R/min can be performed followed by an interval of 3 minutes before irradiation can be resumed.

61 RADIATION PROTECTION AND DOSIMETRY↗

Part-scale microstructure prediction for laser powder bed fusion Ti-6Al-4V using a hybrid mechanistic and machine learning model

Laser powder bed fusion (LPBF) Ti-6Al-4V is widely studied for use in structural applications in aerospace and medical industries, but mechanical anisotropy and microstructural inhomogeneity prohibits its wider adoption. Although successful microstructure prediction models have been developed, a remaining challenge is their limited integration across length/time scales and validation by experimental studies. Here, this work proposes a physics-augmented machine learning surrogate model to unite predictions of LPBF temperature, β phase morphology and texture, and α/α’ formation into a single framework that is calibrated and validated with experiments. First, a phase field (PF) model of the martensitic β→α’ transformation is developed and calibrated using data from in-situ synchrotron cyclic heating/cooling studies quantifying the variation of α phase fraction with time. In parallel, an established finite difference-Monte Carlo (FDMC) model predicts the part-scale temperature profile and β grain formation during solidification. A dataset is developed using LPBF cyclic temperature descriptors from the FDMC model as inputs and corresponding α/α’ phase fraction and width from the PF model as outputs. Five machine learning (ML) regression models are tested and optimized, having mean absolute error in testing ≤ 4 %, and the k-nearest neighbors (KNN) model is selected as the best performing. The KNN model is called at the nodal level during post-processing of the FDMC model to replace and downscale the response of the PF model. The combined agility and accuracy of the hybrid FDMC-ML model enables part-scale microstructure predictions that can be further used for property predictions to accelerate AM process optimization.

36 MATERIALS SCIENCE↗

Pre-trained network-based transfer learning: A small-sample machine learning approach to nuclear power plant classification problem

Some research topics belonging to classification problems in the nuclear industry, such as fault diagnosis and accident identification, can be solved by feature extraction and subsequent application of statistical machine learning classifiers. Recently, deep neural network-based methods with automatic feature extraction and high accuracy have gained wide attention. They usually require large-scale training data, however, plant fault or accident data are scarce or difficult to obtain. Here this paper proposes a convolutional network (CNN)-based transfer learning method to solve this problem. The network's shallow layer is derived from a pre-trained CNN based on the ImageNet database to automatically extract features, and the deep layer is customized to match the classification problem. Data in non-image formats are converted to image formats and subsequently used to train the network. Case studies of rotating machines fault diagnosis show that the proposed method requires only limited training data to achieve high accuracy.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Photometric redshifts for the S-PLUS Survey: Is machine learning up to the task?

The Southern Photometric Local Universe Survey (S-PLUS) is a novel project that aims to map the Southern Hemisphere using a twelve filter system, comprising five broad-band SDSS-like filters and seven narrow-band filters optimized for important stellar features in the local universe. In this paper we use the photometry and morphological information from the first S-PLUS data release (S-PLUS DR1) cross-matched to unWISE data and spectroscopic redshifts from Sloan Digital Sky Survey DR15. We explore three different machine learning methods (Gaussian Processes with GPz and two Deep Learning models made with TensorFlow) and compare them with the currently used template-fitting method in the S-PLUS DR1 to address whether machine learning methods can take advantage of the twelve filter system for photometric redshift prediction. Using tests for accuracy for both single-point estimates such as the calculation of the scatter, bias, and outlier fraction, and probability distribution functions (PDFs) such as the Probability Integral Transform (PIT), the Continuous Ranked Probability Score (CRPS) and the Odds distribution, we conclude that a deep-learning method using a combination of a Bayesian Neural Network and a Mixture Density Network offers the most accurate photometric redshifts for the current test sample. In conclusion, it achieves single-point photometric redshifts with scatter (σ NMAD ) of 0.023, normalized bias of -0.001, and outlier fraction of 0.64% for galaxies with r_auto magnitudes between 16 and 21.

79 ASTRONOMY AND ASTROPHYSICS↗

Occupancy prediction through machine learning and data fusion of environmental sensing and Wi-Fi sensing in buildings

Occupancy information is crucial to building facility design, operation, and energy efficiency. Many studies propose the use of environmental sensors (such as carbon dioxide, air temperature, and relative humidity sensors) and radio-frequency sensors (Wi-Fi networks) to monitor, assess, and predict occupancy information for buildings. As many methods have been developed and a variety of sensory data sources are available, establishing a proper selection of model and data source is critical to the successful implementation of occupancy prediction systems. This study compared three popular machine learning algorithms, including k-nearest neighbors (kNN), support vector machine (SVM), and artificial neural network (ANN), combined with three data sources, including environmental data, Wi-Fi data, and fused data, to optimize the occupancy models’ performance in various scenarios. Three error measurement metrics, the mean average error (MAE), mean average percentage error (MAPE), and root mean squared error (RMSE), have been employed to compare the models’ accuracies. Examined with an on-site experiment, the results suggest that the ANN-based model with fused data has the best performance, while the SVM model is more suitable with Wi-Fi data. The results also indicate that, comparing with independent data sources, the fused data set does not necessarily improve model accuracy but shows a better robustness for occupancy prediction.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Leveraging knowledge engineering and machine learning for microbial bio-manufacturing

Genome scale modeling (GSM) predicts the performance of microbial workhorses and helps identify beneficial gene targets. GSM integrated with intracellular flux dynamics, omics, and thermodynamics have shown remarkable progress in both elucidating complex cellular phenomena and computational strain design (CSD). Nonetheless, these models still show high uncertainty due to a poor understanding of innate pathway regulations, metabolic burdens, and other factors (such as stress tolerance and metabolite channeling). Besides, the engineered hosts may have genetic mutations or non-genetic variations in bioreactor conditions and thus CSD rarely foresees fermentation rate and titer. Metabolic models play important role in design-build-test-learn cycles for strain improvement, and machine learning (ML) may provide a viable complementary approach for driving strain design and deciphering cellular processes. In order to develop quality ML models, knowledge engineering leverages and standardizes the wealth of information in literature (e.g., genomic/phenomic data, synthetic biology strategies, and bioprocess variables). Data driven frameworks can offer new constraints for mechanistic models to describe cellular regulations, to design pathways, to search gene targets, and to estimate fermentation titer/rate/yield under specified growth conditions (e.g., mixing, nutrients, and O 2 ). Furthermore this review highlights the scope of information collections, database constructions, and machine learning techniques (such as deep learning and transfer learning), which may facilitate "Learn and Design" for strain development.

59 BASIC BIOLOGICAL SCIENCES↗

Accelerating geostatistical modeling using geostatistics-informed machine Learning

Ordinary Kriging (OK) is a popular geostatistical algorithm for spatial interpolation and estimation. The computational complexity of OK changes quadratically and cubically for memory and speed, respectively, given the number of data. Therefore, it is computationally intensive and also challenging to process a large set of data, especially in three-dimensional (3D) cases. This paper develops a geostatistics-informed machine learning (GIML) model to improve the efficiency of OK by reducing the number of points required to be estimated using OK. Specifically, only a very few of the unknown points are estimated by OK to get the weights and estimations, which are used as the training dataset. Moreover, the governing equations of OK are used to guide our proposed machine learning to better reproduce the spatial distributions. Our results show that the proposed GIML can reduce the computational time of OK by at least one order of magnitude. The effectiveness of the GIML is evaluated and compared using a 2D case. Furthermore, we demonstrate its efficiency and robustness by considering a different number of training samples on various 3D simulation grids.

58 GEOSCIENCES↗

Insights from machine learning of carbon electrodes for electric double layer capacitors

Recent years have witnessed the broad use of carbon electrodes for electric double layer capacitors (EDLCs) because of large surface area, high porosity and low cost. Whereas experimental investigations are mostly focused on the device performance, computational studies have been rarely concerned with electrochemical properties at conditions remote from equilibrium, limiting their direct applications to materials design. Through a comprehensive analysis of extensive experimental data with various machine-learning methods, we report herein quantitative correlations between the structural features of carbon electrodes and the in-operando behavior of EDLCs including energy and power density. Machine learning allows us to identify important characteristics of activated carbons useful to optimize their efficiency in energy storage.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Structures of Asymmetric ClpX Hexamers Reveal Nucleotide-Dependent Motions in a AAA+ Protein-Unfolding Machine

ClpX is a AAA+ machine that uses the energy of ATP binding and hydrolysis to unfold native proteins and translocate unfolded polypeptides into the ClpP peptidase. The crystal structures presented here reveal striking asymmetry in ring hexamers of nucleotide-free and nucleotide-bound ClpX. Asymmetry arises from large changes in rotation between the large and small AAA+ domains of individual subunits. These differences prevent nucleotide binding to two subunits, generate a staggered arrangement of ClpX subunits and pore loops around the hexameric ring, and provide a mechanism for coupling conformational changes caused by ATP binding or hydrolysis in one subunit to flexing motions of the entire ring. Our structures explain numerous solution studies of ClpX function, predict mechanisms for pore elasticity during translocation of irregular polypeptides, and suggest how repetitive conformational changes might be coupled to mechanical work during the ATPase cycle of ClpX and related molecular machines.

36 MATERIALS SCIENCE↗

Machine learning-based ethylene concentration estimation, real-time optimization and feedback control of an experimental electrochemical reactor

With the increase in electricity supply from clean energy sources, electrochemical reduction of carbon dioxide (CO 2 ) has received increasing attention as an alternative source of carbon-based fuels. As CO 2 reduction is becoming a stronger alternative for the clean production of chemicals, the need to model, optimize and control the electrochemical reduction of the CO 2 process becomes inevitable. However, on one hand, a first-principles model to represent the electrochemical CO 2 reduction has not been fully developed yet because of the complexity of its reaction mechanism, which makes it challenging to define a precise state-space model for the control system. On the other hand, the unavailability of efficient concentration measurement sensors continues to challenge our ability to develop feedback control systems. Gas chromatography (GC) is the most common equipment for monitoring the gas product composition, but it requires a period of time to analyze the sample, which means that GC can provide only delayed measurements during the operation. Moreover, the electrochemical CO 2 reduction process is catalyzed by a fast-deactivating copper catalyst and undergoes a selectivity shift from the product-of-interest at the later stages of experiments, which can pose a challenge for conventional control methods. To this end, machine learning (ML) techniques provide a potential approach to overcome those difficulties due to their demonstrated ability to capture the dynamic behavior of a chemical process from data. Motivated by the above considerations, we propose a machine learning-based modeling methodology that integrates support vector regression and first-principles modeling to capture the dynamic behavior of an experimental electrochemical reactor; this model, together with limited gas chromatography measurements, is employed to predict the evolution of gas-phase ethylene concentration. The model prediction is directly used in a proportional-integral (PI) controller that manipulates the applied potential to regulate the gas-phase ethylene concentration at energy-optimal set-point values computed by a real-time process optimizer (RTO). Specifically, the RTO calculates the operation set-point by solving an optimization problem to maximize the economic benefit of the reactor. Finally, suitable compensation methods are introduced to further account for the experimental uncertainties and handle catalyst deactivation. The proposed modeling, optimization, and control approaches are the first demonstration of active control for a CO 2 electrolyzer and contribute to the automation and scale-up efforts for electrified manufacturing of fuels and chemicals starting from CO 2 .

42 ENGINEERING↗

Machining-inducted residual strain and ductile-brittle transition in sapphire revealed by X-ray nanoprobe diffraction microscopy

Hard X-ray nanoprobe diffraction microscopy is employed to investigate machining-induced residual lattice strain in single-crystal sapphire following ultra-precision orthogonal plunge cutting. In this study, tilt-series and single-angle acquisition strategies are systematically compared to evaluate their capability for resolving lattice variations associated with the machining process. An analytical framework is developed to enable the efficient and reliable extraction of strains, which serve as strain-based indicators of changes in the dominant deformation mechanism. The measured high-dimensional strain exhibits spatial heterogeneity and systematic evolution along the cutting path, consistent with deformation-regime changes from ductile response to crack-prone behavior.

Liang, Rui↗

Microstructure optimization with constrained design objectives using machine learning-based feedback-aware data-generation

Microstructure sensitive design has a critical impact on the performance of engineering materials. The safety and performance requirements of critical components, as well as the cost of material and machining of Titanium components, make dovetailing of the microstructure imperative. This paper addresses the optimization of several microstructure design problems for Titanium components under specific design constraints using a feedback-aware data-driven solution methodology. In this study, the microstructure is modeled with an orientation distribution function (ODF) that measures the volumes of different crystallographic orientations. Two algorithms are used to sample the entire microstructure space followed by machine learning-aided identification of a minimal subset of ODF dimensions which is subsequently explored by targeted sampling. Conventional optimization methods lead to a unique microstructure rather than yielding a comprehensive space of optimal or near-optimal microstructures. Multiple solutions are crucial for the deployment of materials design for manufacturing as traditional manufacturing processes can only generate a limited set of microstructures. Our data sampling-based methodology not only outperforms or is on par with other optimization techniques in terms of the optimal property value, but also provides numerous near-optimal solutions, 3–4 orders of magnitude more than previous methods. Consequently, the proposed framework delivers a spectrum of optimal solutions in the microstructure space which can accelerate materials development and reduce manufacturing costs.

36 MATERIALS SCIENCE↗

A combined machine learning and density functional theory study of binary Ti-Nb and Ti-Zr alloys: Stability and Young’s modulus

The multicomponent Ti alloys, specifically the -phase, have experienced a strong growth over the last decades, due to their outstanding properties of ultra-high strength and low Young’s modulus. These properties play a significant role in many aerospace and biomedical applications. Selection and optimization of multicomponent alloys is challenging due to the vast chemical and compositional space. Here we investigate the use of machine learning techniques informed by density functional calculations to guide the selection of Nb- and Zr-based Ti binary alloys. From the cubic structures obtained from high throughput calculations and literature, we identify several structures with Young’s moduli below 40 GPa. The multivariant decision tree methods provide efficient surrogate models to identify structure variables have high influences on the energetic stability and Young’s modulus. Furthermore, we implement a workflow of incorporating DFT provided results and machine learning method to explore the chemical and composition space of other binary and multicomponent alloys, to eventually accelerate the material design via taking advantages of identified key variables.

36 MATERIALS SCIENCE↗