Evolutionary Multi-Objective Decision Support Systems for Conceptual Design

In this thesis the problem of conceptual engineering design and the possible use of adaptive search techniques and other machine based methods therein are expl or d. For the multi–objective optimisation (MOO) within conceptual design problem, genetic a lgorithms (GA) adapted to MOO are used and various techniques explored: weighted sums, lexic ographic order, Pareto method with and without ranking, VEGA–like approaches etc. Large number of uns are performed for finding the optimal configuration and setting of the GA parameters. A nov el method,weighted Pareto method is introduced and applied to a real–world optimisation proble m. Decision support methods within conceptual engineering de sign framework are discussed and a new preference method developed. The preference method for tra nslating vague qualitative categories (such as “more important”, “much less important” etc.) into quantitative values (numbers) is based on fuzzy preferences and graph theory methods. Several appl ic tions of preferences are presented and discussed: in weighted sum based optimisation methods; in weighted Pareto method; for ordering and manipulating constraints and scenarios; for a co-evolutionary, distributive GA–based MOO method; The issue of complexity and sensitivity is addressed as well as potential generalisations of presented preference methods. Interactive dynamical constraints in the form of design scenarios are introduced. These are based on a propositional logic and a fairly rich mat hematical language. They can be added, deleted and modified on–line during the design session witho ut need for recompiling the code. The use of machine–based agents in conceptual design proces s is investigated. They are classified into several different categories (e.g. interface agents, search agents, information agents). Several different categories of agents performing various special ised task are developed (mostly dealing with preferences, but also some filtering ones). They are integra t d with the conceptual engineering design system to form a closed loop system that includes both comput er and designer. All these different aspects of conceptual engineering desi gn are applied within Plymouth Engineering Design Centre / British Aerospace conceptual airframe desi gn project.