A unified approach to multiple objective design of engineering systems
Abstract
The design and performance of a complex engineering system depends on a number of different, often conflicting, criteria which cannot be combined into a single measure of performance. This research presents solution methodologies for computer-aided design of engineering systems in the presence of multiple objective functions. The concept of a Pareto-optimal solution in the context of multiobjective design problems is introduced, and several techniques for generating Pareto-optimal solutions are discussed. A comparison is made between the relative efficiency of various techniques with regards to the required preference information, as well as the quality of solutions generated by each technique. Design optimization problems where the objective functions, constraints, and design data are not known precisely are also addressed under the purview of this research. The tools of fuzzy set theory are employed to solve these ill-structured optimization problems. The fundamental assumption in fuzzy mathematical programming applications involving the use of linear membership functions is critically examined. Several nonlinear shapes for the membership functions are proposed consistent with varying perceptions of the designer, and are analyzed to determine their impact on the overall design process. The concepts in cooperative game theory and fuzzy set theory have been combined to yield robust and computationally efficient design optimization methodologies, referred to herein as, cooperative fuzzy games. The concept of cooperative fuzzy games is illustrated via an application to three design problems including an integrated structure and control design of flexible space structures. Two search procedures based on the mechanics of natural selection and neural networks are also examined with regards to their potential in solving design optimization problems. The optimum results obtained using these biological information processing schemes are frequently seen to outperform the optimum solutions using gradient based search procedures. The multiple objective optimization techniques presented throughout this thesis offer promising avenues for an effective consideration of all the mutually conflicting requirements inherent in the design problem; and are expected to provide a systematic methodology to formulate and solve multiobjective optimization problems in a form directly applicable to engineering design.
Degree
Ph.D.
Advisors
Rao, Purdue University.
Subject Area
Mechanical engineering|Systems design|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.