Coevolutionary Genetic Adaptation: A New Paradigm For Distributed Multidisciplinary Design Optimization

This paper introduces a new paradigm for distributed multidisciplinary design optimization (MDO) of complex coupled systems. It is inspired by the phenomena of coevolutionary adaptation occurring in ecological systems. The focus of the paper is to develop flexible design architectures for addressing the organizational and computational challenges involved in application of formal optimization techniques to large-scale multidisciplinary systems. The distinguishing advantages offered by the proposed MDO architectures include retainment of disciplinary autonomy and decentralized decision making, massive parallelism, reduced software integration and interdisciplinary communication overheads, accommodation of variable-complexity design parameterization with a mix of discrete and continuous variables, and robustness in nonconvex design spaces. The niche for coevolutionary adaptation based search strategies in the spectrum of MDO methodologies is discussed. Problems from the domain of structu..