Authors: Hong Yoon Kim; Nickolas Vlahopoulos
Addresses: Department of Naval Architecture and Marine Engineering, The University of Michigan, Ann Arbor, MI, USA ' Department of Naval Architecture and Marine Engineering, The University of Michigan, Ann Arbor, MI, USA
Abstract: An optimisation algorithm that uses a bi-level structure in order to consolidate the solution among multiple discipline optimisations with both local and global design variables is presented. This algorithm is following the bi-level integrated system synthesis decomposition scheme but each discipline level optimisation has its own distinct objective function and constraints. The constraints from all the disciplines are accumulated at the system level optimisation in order to ensure that the final solution is feasible for all disciplines. The structure and the computational process of the algorithm are presented. Two analytical examples are solved for demonstrating the utility of the new algorithm. In all analyses, the optimal results converge on the Pareto front. It is further demonstrated that utilising different weights in the system level objective function allows constructing points on the Pareto front.
Keywords: multi-objective optimisation; bi-level optimisation; global design variables; local design variables; design variable decomposition; Pareto front.
International Journal of Mathematical Modelling and Numerical Optimisation, 2015 Vol.6 No.4, pp.265 - 281
Received: 02 Jan 2015
Accepted: 10 Jul 2015
Published online: 31 Dec 2015 *