Search NASA⌕ Search

Engineering topics

Kim, Jungkwun

Publications and source records attributed to Kim, Jungkwun.

A General Spatiotemporal Imputation Framework for Missing Sensor Data

Many applications from precision agriculture, environmental monitoring and transportation networks rely on data collected across space and time over a large geographic area. Missing data poses a significant challenge for any data-driven inference and control tasks. Data imputation or the estimation of missing data can help fill these gaps by utilizing inherent spatial relationships and temporal patterns. A variety of spatiotemporal imputation models have been developed to address missing data in spatiotemporal datasets. However, these classical methods rely on the assumption that the underlying data follows a smooth trend and fail to provide accurate estimates when there is a large number of missing points in the data. Even though there are machine learning driven tensor completion approaches such as convolutional neural network based tensor completion (CoSTCo) that capture the non-linear relationships in the dataset, the transductive nature makes the algorithm less scalable. Thus, existing approaches for estimating the missing information do not effectively capture all dimensions of the spatiotemporal data structure, resulting in erroneous predictions and poor performance. The main contributions of this paper are: (1) We propose a novel inductive framework (G-LSTM) for missing data imputation that integrates a graph neural network with LSTMs to effectively capture both spatial and temporal dependencies. (2) Experimental results on a traffic dataset demonstrate that the proposed GNN integrated with an LSTM framework achieves improved imputation and maintains steady performance even when there are extreme missing conditions in comparison with the state-of-the-art imputation framework (i.e, CoSTCo). (3) The simulation results on a traffic network show up to 69% reduction in mean absolute error and 61% reduction in root mean square error when compared to CoSTCo.

Tharzeen, Aabila↗

Alternate current magnetic property characterization of nonstoichiometric zinc ferrite nanocrystals for inductor fabrication via a solution based process

We investigate the ac magnetic behavior of solution processable, non-stoichiometric zinc ferrite nanocrystals with a series of sizes and zinc concentrations. Nearly monodisperse Zn{sub x}Fe{sub 3−x}O{sub 4} nanocrystals (x = 0–0.25) with an average size ranging from 7.4 nm to 13.8 nm are synthesized by using a solvothermal method. All the nanocrystals are in a superparamagnetic state at 300 K, which is confirmed by Superconductive Quantum Interference Device magnetometry. Due to the doping of non-magnetic Zn{sup 2+} into A site of ferrite, the saturation magnetization of nanocrystals increases as the size and Zn concentration increases. The ac magnetic permeability measurements at radio frequencies reveal that the real part of the magnetic permeability of similarly sized ferrite nanocrystals can be enhanced by almost twofold as the Zn{sup 2+} doping level increases from 0 to 0.25. The integration of 12.3 nm Zn{sub 0.25}Fe{sub 2.75}O{sub 4} nanocrystals into a toroidal inductor and a solenoid inductor prepared via a simple solution cast process yields a higher quality factors than air core inductors with the same geometries up to 5 MHz and 9 MHz, respectively, which is in the regime of the switching frequencies for the advanced integrated power converters.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Highly-­laminated, high-­saturation-flux-­density, magnetic cores for on-­chip inductors in power converter applications

The objective of this project is development of a manufacturing process to realize thick highly-laminated metallic magnetic cores with suppressed eddy current losses for power applications and demonstration of integrated magnetic components using the process. The continual reduction in size and increase in functionality of portable electronic devices drives the development of ultracompact electrical power converters and therefore their associated inductors, which typically occupy the largest volume within the converter. This inductor size reduction can be achieved by increasing both the converter switching frequency as well as inductor operating flux density by utilizing appropriate magnetic core materials. Magnetically soft metallic alloys (e.g., NiFe and CoNiFe) are good core material candidates due to their superior magnetic properties (e.g., high saturation flux density and low coercivity) compared to conventional ferrite. However, to minimize eddy current losses, the core thickness is limited to the skin depth (typically less than 3 µm for 10 MHz operation); cores of this thickness are likely insufficient for handling high power (> 10W). Therefore, there is great interest in the development of highly-laminated metallic alloy cores with single lamination thickness in the sub-micron range to achieve suppressed eddy-current losses and high power handling capacity simultaneously. During the project, highly-laminated metallic alloy cores comprising up to 2000 layers of permalloy (Ni 80 Fe 20 ) or CoNiFe materials with single lamination thickness less than 1 µm have been fabricated based on automated multilayer electrodeposition technology compatible with CMOS manufacturing. The developed highly-laminated metallic cores preserve the high saturation flux density and low coercivity characteristics of metallic alloys, while suppressing eddy-current losses less than 1 W/cm3 at operation flux density of 0.5 T and frequency over 1 MHz. The developed highly-laminated metallic alloy cores have been further integrated into microfabricated windings to realize compact, on-chip inductors. Utilizing CMOS-compatible integration techniques, circular (toroid) and bar-shape (solenoid) microinductors have been fabricated exhibiting a constant inductance higher than 1 µH up to 50 MHz with peak quality factor as high as 35 at 5 MHz. The microfabricated inductor with highly-laminated core was tested in a power converter and demonstrated a converter efficiency of 93% at 5.5 MHz and 42 W under 100V input and 35V output operation.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗