Show simple item record

dc.contributor.advisorSchrage, Daniel
dc.contributor.authorPatterson, Frank H.
dc.date.accessioned2016-01-07T17:36:13Z
dc.date.available2016-01-07T17:36:13Z
dc.date.created2015-12
dc.date.issued2015-11-10
dc.date.submittedDecember 2015
dc.identifier.urihttp://hdl.handle.net/1853/54413
dc.description.abstractAn evolving set of modern physics-based, multi-disciplinary conceptual design methods seek to explore the feasibility of a new generation of systems, with new capabilities, capable of missions that conventional vehicles cannot be empirically redesigned to perform. These methods provide a more complete understanding of a concept's design space, forecasting the feasibility of uncertain systems, but are often computationally expensive and time consuming to prepare. This trend creates a unique and critical need to identify a manageable number of capable concept alternatives early in the design process. Ongoing efforts attempting to stretch capability through new architectures, like the U.S. Army's Future Vertical Lift effort and DARPA's Vertical Takeoff and Landing (VTOL) X-plane program highlight this need. The process of identifying and selecting a concept configuration is often given insufficient attention, especially when a small subset of favorable concept families is not immediately apparent. Commonly utilized methods for concept generation, like filtered morphological analysis, often identify an exponential number of alternatives. Simple approaches to concept selection then rely on designers to identify a relatively small subset of alternatives for comparison through simple methods regularly related to decision matrices (Pugh, TOPSIS, AHP, etc.). More in-depth approaches utilize modeling and simulation to compare concepts with techniques such as stochastic optimization or probabilistic decision making, but a complicated setup limits these approaches to just a discrete few alternatives. A new framework to identify and select promising, robust concept configurations utilizing fuzzy methods is proposed in this research and applied to the example problem of concept selection for DARPA's VTOL Xplane program. The framework leverages fuzzy systems in conjunction with morphological analysis to assess large design spaces of potential architecture alternatives while capturing the inherent uncertainty and ambiguity in the evaluation of these early concepts. Experiments show how various fuzzy systems can be utilized for evaluating criteria of interest across disparate architectures by modeling expert knowledge as well as simple physics-based data. The models are integrated into a single environment and variations on multi-criteria optimization are tested to demonstrate an ability to identify a non-dominated set of architectural families in a large combinatorial design space. The resulting framework is shown to provide an approach to quickly identify promising concepts in the face of uncertainty early in the design process.An evolving set of modern physics-based, multi-disciplinary conceptual design methods seek to explore the feasibility of a new generation of systems, with new capabilities, capable of missions that conventional vehicles cannot be empirically redesigned to perform. These methods provide a more complete understanding of a concept's design space, forecasting the feasibility of uncertain systems, but are often computationally expensive and time consuming to prepare. This trend creates a unique and critical need to identify a manageable number of capable concept alternatives early in the design process. Ongoing efforts attempting to stretch capability through new architectures, like the U.S. Army's Future Vertical Lift effort and DARPA's Vertical Takeoff and Landing (VTOL) X-plane program highlight this need. The process of identifying and selecting a concept configuration is often given insufficient attention, especially when a small subset of favorable concept families is not immediately apparent. Commonly utilized methods for concept generation, like filtered morphological analysis, often identify an exponential number of alternatives. Simple approaches to concept selection then rely on designers to identify a relatively small subset of alternatives for comparison through simple methods regularly related to decision matrices (Pugh, TOPSIS, AHP, etc.). More in-depth approaches utilize modeling and simulation to compare concepts with techniques such as stochastic optimization or probabilistic decision making, but a complicated setup limits these approaches to just a discrete few alternatives. A new framework to identify and select promising, robust concept configurations utilizing fuzzy methods is proposed in this research and applied to the example problem of concept selection for DARPA's VTOL Xplane program. The framework leverages fuzzy systems in conjunction with morphological analysis to assess large design spaces of potential architecture alternatives while capturing the inherent uncertainty and ambiguity in the evaluation of these early concepts. Experiments show how various fuzzy systems can be utilized for evaluating criteria of interest across disparate architectures by modeling expert knowledge as well as simple physics-based data. The models are integrated into a single environment and variations on multi-criteria optimization are tested to demonstrate an ability to identify a non-dominated set of architectural families in a large combinatorial design space. The resulting framework is shown to provide an approach to quickly identify promising concepts in the face of uncertainty early in the design process.
dc.format.mimetypeapplication/pdf
dc.language.isoen_US
dc.publisherGeorgia Institute of Technology
dc.subjectSystems engineering
dc.subjectConcept selection
dc.subjectEngineering design
dc.subjectFuzzy
dc.subjectDecision making
dc.subjectSystems architecture
dc.subjectKnowledge based engineering
dc.titleFuzzy framework for robust architecture identification in concept selection
dc.typeDissertation
dc.description.degreePh.D.
dc.contributor.departmentAerospace Engineering
thesis.degree.levelDoctoral
dc.contributor.committeeMemberMavris, Dimitri
dc.contributor.committeeMemberVachtsevanos, George
dc.contributor.committeeMemberAshok, Sylvester
dc.contributor.committeeMemberGerman, Brian
dc.date.updated2016-01-07T17:36:13Z


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record