Strategies for Modeling, Approximation, and Decomposition in Genetic Algorithms Based Multidisciplinary Design

This chapter discusses the applicability of new computational paradigms motivated by biological processes, in the realm of multidisciplinary engineering design, and particularly, in the context of using formal methods of design optimization. The computational models considered in this discussion include genetic algorithms, neural networks, and a modeling of the biological immune system. The focus of the chapter is two-fold. First, it introduces the reader to the implementation of these newly emergent methods. Second, it describes how the use of these methods alleviates some of the difficulties associated with the application of formal optimization methods in practical design problems. Such problems are typically characterized by the presence of a large number of design variables and constraints, the need to consider multiple objective criterion, and, in some cases, a fuzzy description of design specifications. The analysis associated with the multidisciplinary design problem is both complex and computationally expensive. The discussion focuses on methods to reduce the computational effort through development of efficient optimal search algorithms, and in the efficient management of couplings in the analysis problem.