Resumen de Towards an Optimal Engineering Design Process Using Appropriate Adaptive Search Strategies
The paper introduces current research relating to the integration of adaptive search ( AS) techniques such as the genetic algorithm ( GA) with engineering design practice. Relevant AS strategies and complementary techniques that contribute at each stage of the design process are identified and results from their initial implementation are presented. The paper attempts to raise the perception of these techniques as global search algorithms that can assist at every level of design in addition to their specific abilities relating to optimization. The paper therefore illustrates how AS can satisfactorily: manipulate high-level conceptual design grammars; provide concurrent multi-level processing of whole-system design hierarchies; identify high-performance regions of a conceptual/preliminary design space; provide relevant design information during the search process; improve the calibration of preliminary design software to reduce associated risk; access remote feasible regions of complex, non-linear design spaces that are discontinuous, multi-modal and characterized by heavy, non-linear constraint; and manipulate computationally expensive analysis software to provide optimal detailed design solutions within a realistic period of time. The overall objective of the complementary projects described in the paper is the development of highly interactive designer/AS strategies that will contribute significantly to the engineering design process. The resulting tools must be considered as powerful extensions of the design team stimulating innovative reasoning at the higher conceptual levels of the design process, providing diverse, high-performance solutions to support decision making during preliminary design and acting as powerful global optimizers that can operate successfully within highly complex domains during detailed design. The paper discusses successful implementation at each level and identifies commonalities both in the design characteristics of each design stage and in the various search strategies. Cooperative frameworks involving a number of search strategies/optimization techniques that operate concurrently within a single search environment are introduced.
