Competitive control in systems engineering.

Abstract

The three main difficulties of the product development process are non-convexity, uncertainty, and multiobjectivity. Because of these problems, the product development process is characterized using two metaphors. The first metaphor is navigating in uncharted waters. The second is the development process as a competition among the various objectives that must be satisfied, including the need to minimize costs and maximize performance. Both of these metaphors lead to the concept of 'good directions' for guiding the development process. Methods of obtaining 'good directions' are described, and fuzziness is recruited to deal with the uncertainties in the direction-finding process. The two main approaches to product development are the optimizing approach and the combinative approach. Experimental designs and genetic algorithms are suggested for efficiently investigating the design space through the correct use of models and prototypes. Experimental designs can be used as part of either the optimizing or the combinative approaches, while the genetic algorithm is more effective for the combinative approach. An innovative approach to robust design is introduced which emphasizes three-level factorial experimental designs. In addition, models of competition are examined as a means of understanding the concepts of competition which are applicable to system theory and systems engineering. Case studies describing the Manhattan Project and the Polaris Program help to illustrate the role of competition in product development.