A Distributed Coevolutionary Multidisciplinary Design Optimization Algorithm

• Published in: 2010 Third International Joint Conference on Computational Sciences and Optimization Vol. 2; pp. 77 - 80
• Main Authors: Yonggang Xing, Shuo Tang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2010
• Subjects: Biological system modeling; competetive-cooperative; complex coupled systems; Computational efficiency; Computer architecture; Computer network management; Design methodology; Design optimization; distributed coevolutionary algorithm; Distributed computing; Educational institutions; multidisciplinary design optimization; Optimization methods; Robustness
• ISBN: 1424468124; 9781424468126
• DOI: 10.1109/CSO.2010.16

In order to provide efficient algorithm for multi-disciplinary design optimization of complex coupled systems, a distributed coevolutionary multidisciplinary design optimization algorithm is proposed. The algorithm imitates the biological competitive and cooperative coevolutionary process in ecological systems. The ideas of decomposition and cooperation in coevolutionary algorithm combine with the ideas of decomposition and synergism in MDO. Based on the method of domain decomposition and the implicit iteration strategy, the complex coupled system is decomposed into relatively independent and autonomic multidisciplinary systems. Each discipline is modeled as a species. Thus the competitive-cooperative adaptive coevolutionary multidisciplinary optimization process has been modeled amongst the populations of multidisciplinary species. For illustrating the proposed algorithm, a multidisciplinary design optimization test problem generated by a robust simulator called CASCADE is utilized to simulate. Experimental results reveal the proposed algorithm has good search capability and convergence performance. Hence, the presented algorithm is efficient and robust in solving multidisciplinary design optimization problem of complex coupled systems.