Title:
A representation method for large and complex engineering design datasets with sequential outputs

dc.contributor.advisor Mavris, Dimitri N.
dc.contributor.author Iwata, Curtis
dc.contributor.committeeMember Schrage, Daniel
dc.contributor.committeeMember Volovoi, Vitali
dc.contributor.committeeMember German, Brian
dc.contributor.committeeMember Balestrini-Robinson, Santiago
dc.contributor.department Aerospace Engineering
dc.date.accessioned 2014-01-13T16:21:28Z
dc.date.available 2014-01-13T16:21:28Z
dc.date.created 2013-12
dc.date.issued 2013-08-22
dc.date.submitted December 2013
dc.date.updated 2014-01-13T16:21:28Z
dc.description.abstract This research addresses the problem of creating surrogate models of high-level operations and sustainment (O&S) simulations with time sequential (TS) outputs. O&S is a continuous process of using and maintaining assets such as a fleet of aircraft, and the infrastructure to support this process is the O&S system. To track the performance of the O&S system, metrics such as operational availability are recorded and reported as a time history. Modeling and simulation (M&S) is often used as a preliminary tool to study the impact of implementing changes to O&S systems such as investing in new technologies and changing the inventory policies. A visual analytics (VA) interface is useful to navigate the data from the M&S process so that these options can be compared, and surrogate modeling enables some key features of the VA interface such as interpolation and interactivity. Fitting a surrogate model is difficult to TS data because of its size and nonlinear behavior. The Surrogate Modeling and Regression of Time Sequences (SMARTS) methodology was proposed to address this problem. An intermediate domain Z was calculated from the simulation output data in a way that a point in Z corresponds to a unique TS shape or pattern. A regression was then fit to capture the entire range of possible TS shapes using Z as the inputs, and a separate regression was fit to transform the inputs into the Z. The method was tested on output data from an O&S simulation model and compared against other regression methods for statistical accuracy and visual consistency. The proposed methodology was shown to be conditionally better than the other methodologies.
dc.description.degree Ph.D.
dc.format.mimetype application/pdf
dc.identifier.uri http://hdl.handle.net/1853/50266
dc.language.iso en_US
dc.publisher Georgia Institute of Technology
dc.subject Regression
dc.subject Surrogate modeling
dc.subject Time series
dc.subject Operations and sustainment
dc.subject Design
dc.subject Visualization
dc.subject Simulation
dc.subject.lcsh Computer simulation
dc.subject.lcsh Mathematical models
dc.subject.lcsh Engineering design
dc.title A representation method for large and complex engineering design datasets with sequential outputs
dc.type Text
dc.type.genre Dissertation
dspace.entity.type Publication
local.contributor.advisor Mavris, Dimitri N.
local.contributor.corporatename Daniel Guggenheim School of Aerospace Engineering
local.contributor.corporatename Aerospace Systems Design Laboratory (ASDL)
local.contributor.corporatename College of Engineering
local.relation.ispartofseries Doctor of Philosophy with a Major in Aerospace Engineering
relation.isAdvisorOfPublication d355c865-c3df-4bfe-8328-24541ea04f62
relation.isOrgUnitOfPublication a348b767-ea7e-4789-af1f-1f1d5925fb65
relation.isOrgUnitOfPublication a8736075-ffb0-4c28-aa40-2160181ead8c
relation.isOrgUnitOfPublication 7c022d60-21d5-497c-b552-95e489a06569
relation.isSeriesOfPublication f6a932db-1cde-43b5-bcab-bf573da55ed6
thesis.degree.level Doctoral
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
IWATA-DISSERTATION-2013.pdf
Size:
16.8 MB
Format:
Adobe Portable Document Format
Description:
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
LICENSE.txt
Size:
3.86 KB
Format:
Plain Text
Description: